diff --git a/bower.json b/bower.json index e008038a0..f003c7b69 100644 --- a/bower.json +++ b/bower.json @@ -1,6 +1,6 @@ { "name": "pdfjs-dist", - "version": "1.4.76", + "version": "1.4.79", "main": [ "build/pdf.js", "build/pdf.worker.js" diff --git a/build/pdf.combined.js b/build/pdf.combined.js index 6e3f774ec..b040b741b 100644 --- a/build/pdf.combined.js +++ b/build/pdf.combined.js @@ -28,8 +28,8 @@ factory((root.pdfjsDistBuildPdfCombined = {})); // Use strict in our context only - users might not want it 'use strict'; -var pdfjsVersion = '1.4.76'; -var pdfjsBuild = '45fea88'; +var pdfjsVersion = '1.4.79'; +var pdfjsBuild = '9eedfc1'; var pdfjsFilePath = typeof document !== 'undefined' && document.currentScript ? @@ -31559,12474 +31559,12484 @@ exports.IdentityCMap = IdentityCMap; (function (root, factory) { { - factory((root.pdfjsCoreObj = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreCrypto, root.pdfjsCoreParser, - root.pdfjsCoreChunkedStream); + factory((root.pdfjsCorePsParser = {}), root.pdfjsSharedUtil, + root.pdfjsCoreParser); } -}(this, function (exports, sharedUtil, corePrimitives, coreCrypto, coreParser, - coreChunkedStream) { +}(this, function (exports, sharedUtil, coreParser) { -var InvalidPDFException = sharedUtil.InvalidPDFException; -var MissingDataException = sharedUtil.MissingDataException; -var XRefParseException = sharedUtil.XRefParseException; -var assert = sharedUtil.assert; -var bytesToString = sharedUtil.bytesToString; -var createPromiseCapability = sharedUtil.createPromiseCapability; var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isInt = sharedUtil.isInt; -var isString = sharedUtil.isString; -var shadow = sharedUtil.shadow; -var stringToPDFString = sharedUtil.stringToPDFString; -var stringToUTF8String = sharedUtil.stringToUTF8String; -var warn = sharedUtil.warn; -var isValidUrl = sharedUtil.isValidUrl; -var Util = sharedUtil.Util; -var Ref = corePrimitives.Ref; -var RefSet = corePrimitives.RefSet; -var RefSetCache = corePrimitives.RefSetCache; -var isName = corePrimitives.isName; -var isCmd = corePrimitives.isCmd; -var isDict = corePrimitives.isDict; -var isRef = corePrimitives.isRef; -var isStream = corePrimitives.isStream; -var CipherTransformFactory = coreCrypto.CipherTransformFactory; +var EOF = coreParser.EOF; var Lexer = coreParser.Lexer; -var Parser = coreParser.Parser; -var ChunkedStream = coreChunkedStream.ChunkedStream; - -var Catalog = (function CatalogClosure() { - function Catalog(pdfManager, xref, pageFactory) { - this.pdfManager = pdfManager; - this.xref = xref; - this.catDict = xref.getCatalogObj(); - this.fontCache = new RefSetCache(); - assert(isDict(this.catDict), - 'catalog object is not a dictionary'); - // TODO refactor to move getPage() to the PDFDocument. - this.pageFactory = pageFactory; - this.pagePromises = []; +var PostScriptParser = (function PostScriptParserClosure() { + function PostScriptParser(lexer) { + this.lexer = lexer; + this.operators = []; + this.token = null; + this.prev = null; } - - Catalog.prototype = { - get metadata() { - var streamRef = this.catDict.getRaw('Metadata'); - if (!isRef(streamRef)) { - return shadow(this, 'metadata', null); - } - - var encryptMetadata = (!this.xref.encrypt ? false : - this.xref.encrypt.encryptMetadata); - - var stream = this.xref.fetch(streamRef, !encryptMetadata); - var metadata; - if (stream && isDict(stream.dict)) { - var type = stream.dict.get('Type'); - var subtype = stream.dict.get('Subtype'); - - if (isName(type) && isName(subtype) && - type.name === 'Metadata' && subtype.name === 'XML') { - // XXX: This should examine the charset the XML document defines, - // however since there are currently no real means to decode - // arbitrary charsets, let's just hope that the author of the PDF - // was reasonable enough to stick with the XML default charset, - // which is UTF-8. - try { - metadata = stringToUTF8String(bytesToString(stream.getBytes())); - } catch (e) { - info('Skipping invalid metadata.'); - } - } - } - - return shadow(this, 'metadata', metadata); - }, - get toplevelPagesDict() { - var pagesObj = this.catDict.get('Pages'); - assert(isDict(pagesObj), 'invalid top-level pages dictionary'); - // shadow the prototype getter - return shadow(this, 'toplevelPagesDict', pagesObj); + PostScriptParser.prototype = { + nextToken: function PostScriptParser_nextToken() { + this.prev = this.token; + this.token = this.lexer.getToken(); }, - get documentOutline() { - var obj = null; - try { - obj = this.readDocumentOutline(); - } catch (ex) { - if (ex instanceof MissingDataException) { - throw ex; - } - warn('Unable to read document outline'); + accept: function PostScriptParser_accept(type) { + if (this.token.type === type) { + this.nextToken(); + return true; } - return shadow(this, 'documentOutline', obj); + return false; }, - readDocumentOutline: function Catalog_readDocumentOutline() { - var xref = this.xref; - var obj = this.catDict.get('Outlines'); - var root = { items: [] }; - if (isDict(obj)) { - obj = obj.getRaw('First'); - var processed = new RefSet(); - if (isRef(obj)) { - var queue = [{obj: obj, parent: root}]; - // to avoid recursion keeping track of the items - // in the processed dictionary - processed.put(obj); - while (queue.length > 0) { - var i = queue.shift(); - var outlineDict = xref.fetchIfRef(i.obj); - if (outlineDict === null) { - continue; - } - if (!outlineDict.has('Title')) { - error('Invalid outline item'); - } - var actionDict = outlineDict.get('A'), dest = null, url = null; - if (actionDict) { - var destEntry = actionDict.get('D'); - if (destEntry) { - dest = destEntry; - } else { - var uriEntry = actionDict.get('URI'); - if (isString(uriEntry) && isValidUrl(uriEntry, false)) { - url = uriEntry; - } - } - } else if (outlineDict.has('Dest')) { - dest = outlineDict.getRaw('Dest'); - if (isName(dest)) { - dest = dest.name; - } - } - var title = outlineDict.get('Title'); - var outlineItem = { - dest: dest, - url: url, - title: stringToPDFString(title), - color: outlineDict.get('C') || [0, 0, 0], - count: outlineDict.get('Count'), - bold: !!(outlineDict.get('F') & 2), - italic: !!(outlineDict.get('F') & 1), - items: [] - }; - i.parent.items.push(outlineItem); - obj = outlineDict.getRaw('First'); - if (isRef(obj) && !processed.has(obj)) { - queue.push({obj: obj, parent: outlineItem}); - processed.put(obj); - } - obj = outlineDict.getRaw('Next'); - if (isRef(obj) && !processed.has(obj)) { - queue.push({obj: obj, parent: i.parent}); - processed.put(obj); - } - } - } + expect: function PostScriptParser_expect(type) { + if (this.accept(type)) { + return true; } - return (root.items.length > 0 ? root.items : null); + error('Unexpected symbol: found ' + this.token.type + ' expected ' + + type + '.'); }, - get numPages() { - var obj = this.toplevelPagesDict.get('Count'); - assert( - isInt(obj), - 'page count in top level pages object is not an integer' - ); - // shadow the prototype getter - return shadow(this, 'num', obj); + parse: function PostScriptParser_parse() { + this.nextToken(); + this.expect(PostScriptTokenTypes.LBRACE); + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + return this.operators; }, - get destinations() { - function fetchDestination(dest) { - return isDict(dest) ? dest.get('D') : dest; - } - - var xref = this.xref; - var dests = {}, nameTreeRef, nameDictionaryRef; - var obj = this.catDict.get('Names'); - if (obj && obj.has('Dests')) { - nameTreeRef = obj.getRaw('Dests'); - } else if (this.catDict.has('Dests')) { - nameDictionaryRef = this.catDict.get('Dests'); - } - - if (nameDictionaryRef) { - // reading simple destination dictionary - obj = nameDictionaryRef; - obj.forEach(function catalogForEach(key, value) { - if (!value) { - return; - } - dests[key] = fetchDestination(value); - }); - } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - var names = nameTree.getAll(); - for (var name in names) { - dests[name] = fetchDestination(names[name]); + parseBlock: function PostScriptParser_parseBlock() { + while (true) { + if (this.accept(PostScriptTokenTypes.NUMBER)) { + this.operators.push(this.prev.value); + } else if (this.accept(PostScriptTokenTypes.OPERATOR)) { + this.operators.push(this.prev.value); + } else if (this.accept(PostScriptTokenTypes.LBRACE)) { + this.parseCondition(); + } else { + return; } } - return shadow(this, 'destinations', dests); }, - getDestination: function Catalog_getDestination(destinationId) { - function fetchDestination(dest) { - return isDict(dest) ? dest.get('D') : dest; - } - - var xref = this.xref; - var dest = null, nameTreeRef, nameDictionaryRef; - var obj = this.catDict.get('Names'); - if (obj && obj.has('Dests')) { - nameTreeRef = obj.getRaw('Dests'); - } else if (this.catDict.has('Dests')) { - nameDictionaryRef = this.catDict.get('Dests'); - } + parseCondition: function PostScriptParser_parseCondition() { + // Add two place holders that will be updated later + var conditionLocation = this.operators.length; + this.operators.push(null, null); - if (nameDictionaryRef) { // Simple destination dictionary. - var value = nameDictionaryRef.get(destinationId); - if (value) { - dest = fetchDestination(value); - } - } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - dest = fetchDestination(nameTree.get(destinationId)); - } - return dest; - }, + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + if (this.accept(PostScriptTokenTypes.IF)) { + // The true block is right after the 'if' so it just falls through on + // true else it jumps and skips the true block. + this.operators[conditionLocation] = this.operators.length; + this.operators[conditionLocation + 1] = 'jz'; + } else if (this.accept(PostScriptTokenTypes.LBRACE)) { + var jumpLocation = this.operators.length; + this.operators.push(null, null); + var endOfTrue = this.operators.length; + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + this.expect(PostScriptTokenTypes.IFELSE); + // The jump is added at the end of the true block to skip the false + // block. + this.operators[jumpLocation] = this.operators.length; + this.operators[jumpLocation + 1] = 'j'; - get pageLabels() { - var obj = null; - try { - obj = this.readPageLabels(); - } catch (ex) { - if (ex instanceof MissingDataException) { - throw ex; - } - warn('Unable to read page labels.'); - } - return shadow(this, 'pageLabels', obj); - }, - readPageLabels: function Catalog_readPageLabels() { - var obj = this.catDict.getRaw('PageLabels'); - if (!obj) { - return null; + this.operators[conditionLocation] = endOfTrue; + this.operators[conditionLocation + 1] = 'jz'; + } else { + error('PS Function: error parsing conditional.'); } - var pageLabels = new Array(this.numPages); - var style = null; - var prefix = ''; - var start = 1; - - var numberTree = new NumberTree(obj, this.xref); - var nums = numberTree.getAll(); - var currentLabel = '', currentIndex = 1; - - for (var i = 0, ii = this.numPages; i < ii; i++) { - if (i in nums) { - var labelDict = nums[i]; - assert(isDict(labelDict), 'The PageLabel is not a dictionary.'); + } + }; + return PostScriptParser; +})(); - var type = labelDict.get('Type'); - assert(!type || (isName(type) && type.name === 'PageLabel'), - 'Invalid type in PageLabel dictionary.'); +var PostScriptTokenTypes = { + LBRACE: 0, + RBRACE: 1, + NUMBER: 2, + OPERATOR: 3, + IF: 4, + IFELSE: 5 +}; - var s = labelDict.get('S'); - assert(!s || isName(s), 'Invalid style in PageLabel dictionary.'); - style = (s ? s.name : null); +var PostScriptToken = (function PostScriptTokenClosure() { + function PostScriptToken(type, value) { + this.type = type; + this.value = value; + } - prefix = labelDict.get('P') || ''; - assert(isString(prefix), 'Invalid prefix in PageLabel dictionary.'); + var opCache = Object.create(null); - start = labelDict.get('St') || 1; - assert(isInt(start), 'Invalid start in PageLabel dictionary.'); - currentIndex = start; - } + PostScriptToken.getOperator = function PostScriptToken_getOperator(op) { + var opValue = opCache[op]; + if (opValue) { + return opValue; + } + return opCache[op] = new PostScriptToken(PostScriptTokenTypes.OPERATOR, op); + }; - switch (style) { - case 'D': - currentLabel = currentIndex; - break; - case 'R': - case 'r': - currentLabel = Util.toRoman(currentIndex, style === 'r'); - break; - case 'A': - case 'a': - var LIMIT = 26; // Use only the characters A--Z, or a--z. - var A_UPPER_CASE = 0x41, A_LOWER_CASE = 0x61; + PostScriptToken.LBRACE = new PostScriptToken(PostScriptTokenTypes.LBRACE, + '{'); + PostScriptToken.RBRACE = new PostScriptToken(PostScriptTokenTypes.RBRACE, + '}'); + PostScriptToken.IF = new PostScriptToken(PostScriptTokenTypes.IF, 'IF'); + PostScriptToken.IFELSE = new PostScriptToken(PostScriptTokenTypes.IFELSE, + 'IFELSE'); + return PostScriptToken; +})(); - var baseCharCode = (style === 'a' ? A_LOWER_CASE : A_UPPER_CASE); - var letterIndex = currentIndex - 1; - var character = String.fromCharCode(baseCharCode + - (letterIndex % LIMIT)); - var charBuf = []; - for (var j = 0, jj = (letterIndex / LIMIT) | 0; j <= jj; j++) { - charBuf.push(character); - } - currentLabel = charBuf.join(''); - break; - default: - assert(!style, - 'Invalid style "' + style + '" in PageLabel dictionary.'); - } - pageLabels[i] = prefix + currentLabel; +var PostScriptLexer = (function PostScriptLexerClosure() { + function PostScriptLexer(stream) { + this.stream = stream; + this.nextChar(); - currentLabel = ''; - currentIndex++; - } - return pageLabels; + this.strBuf = []; + } + PostScriptLexer.prototype = { + nextChar: function PostScriptLexer_nextChar() { + return (this.currentChar = this.stream.getByte()); }, + getToken: function PostScriptLexer_getToken() { + var comment = false; + var ch = this.currentChar; - get attachments() { - var xref = this.xref; - var attachments = null, nameTreeRef; - var obj = this.catDict.get('Names'); - if (obj) { - nameTreeRef = obj.getRaw('EmbeddedFiles'); - } + // skip comments + while (true) { + if (ch < 0) { + return EOF; + } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - var names = nameTree.getAll(); - for (var name in names) { - var fs = new FileSpec(names[name], xref); - if (!attachments) { - attachments = Object.create(null); + if (comment) { + if (ch === 0x0A || ch === 0x0D) { + comment = false; } - attachments[stringToPDFString(name)] = fs.serializable; + } else if (ch === 0x25) { // '%' + comment = true; + } else if (!Lexer.isSpace(ch)) { + break; } + ch = this.nextChar(); } - return shadow(this, 'attachments', attachments); - }, - get javaScript() { - var xref = this.xref; - var obj = this.catDict.get('Names'); + switch (ch | 0) { + case 0x30: case 0x31: case 0x32: case 0x33: case 0x34: // '0'-'4' + case 0x35: case 0x36: case 0x37: case 0x38: case 0x39: // '5'-'9' + case 0x2B: case 0x2D: case 0x2E: // '+', '-', '.' + return new PostScriptToken(PostScriptTokenTypes.NUMBER, + this.getNumber()); + case 0x7B: // '{' + this.nextChar(); + return PostScriptToken.LBRACE; + case 0x7D: // '}' + this.nextChar(); + return PostScriptToken.RBRACE; + } + // operator + var strBuf = this.strBuf; + strBuf.length = 0; + strBuf[0] = String.fromCharCode(ch); - var javaScript = []; - function appendIfJavaScriptDict(jsDict) { - var type = jsDict.get('S'); - if (!isName(type) || type.name !== 'JavaScript') { - return; - } - var js = jsDict.get('JS'); - if (isStream(js)) { - js = bytesToString(js.getBytes()); - } else if (!isString(js)) { - return; - } - javaScript.push(stringToPDFString(js)); + while ((ch = this.nextChar()) >= 0 && // and 'A'-'Z', 'a'-'z' + ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { + strBuf.push(String.fromCharCode(ch)); } - if (obj && obj.has('JavaScript')) { - var nameTree = new NameTree(obj.getRaw('JavaScript'), xref); - var names = nameTree.getAll(); - for (var name in names) { - // We don't really use the JavaScript right now. This code is - // defensive so we don't cause errors on document load. - var jsDict = names[name]; - if (isDict(jsDict)) { - appendIfJavaScriptDict(jsDict); - } - } + var str = strBuf.join(''); + switch (str.toLowerCase()) { + case 'if': + return PostScriptToken.IF; + case 'ifelse': + return PostScriptToken.IFELSE; + default: + return PostScriptToken.getOperator(str); } + }, + getNumber: function PostScriptLexer_getNumber() { + var ch = this.currentChar; + var strBuf = this.strBuf; + strBuf.length = 0; + strBuf[0] = String.fromCharCode(ch); - // Append OpenAction actions to javaScript array - var openactionDict = this.catDict.get('OpenAction'); - if (isDict(openactionDict, 'Action')) { - var actionType = openactionDict.get('S'); - if (isName(actionType) && actionType.name === 'Named') { - // The named Print action is not a part of the PDF 1.7 specification, - // but is supported by many PDF readers/writers (including Adobe's). - var action = openactionDict.get('N'); - if (isName(action) && action.name === 'Print') { - javaScript.push('print({});'); - } + while ((ch = this.nextChar()) >= 0) { + if ((ch >= 0x30 && ch <= 0x39) || // '0'-'9' + ch === 0x2D || ch === 0x2E) { // '-', '.' + strBuf.push(String.fromCharCode(ch)); } else { - appendIfJavaScriptDict(openactionDict); + break; } } - - return shadow(this, 'javaScript', javaScript); - }, - - cleanup: function Catalog_cleanup() { - var promises = []; - this.fontCache.forEach(function (promise) { - promises.push(promise); - }); - return Promise.all(promises).then(function (translatedFonts) { - for (var i = 0, ii = translatedFonts.length; i < ii; i++) { - var font = translatedFonts[i].dict; - delete font.translated; - } - this.fontCache.clear(); - }.bind(this)); - }, - - getPage: function Catalog_getPage(pageIndex) { - if (!(pageIndex in this.pagePromises)) { - this.pagePromises[pageIndex] = this.getPageDict(pageIndex).then( - function (a) { - var dict = a[0]; - var ref = a[1]; - return this.pageFactory.createPage(pageIndex, dict, ref, - this.fontCache); - }.bind(this) - ); + var value = parseFloat(strBuf.join('')); + if (isNaN(value)) { + error('Invalid floating point number: ' + value); } - return this.pagePromises[pageIndex]; - }, + return value; + } + }; + return PostScriptLexer; +})(); - getPageDict: function Catalog_getPageDict(pageIndex) { - var capability = createPromiseCapability(); - var nodesToVisit = [this.catDict.getRaw('Pages')]; - var currentPageIndex = 0; - var xref = this.xref; - var checkAllKids = false; +exports.PostScriptLexer = PostScriptLexer; +exports.PostScriptParser = PostScriptParser; +})); - function next() { - while (nodesToVisit.length) { - var currentNode = nodesToVisit.pop(); - if (isRef(currentNode)) { - xref.fetchAsync(currentNode).then(function (obj) { - if (isDict(obj, 'Page') || (isDict(obj) && !obj.has('Kids'))) { - if (pageIndex === currentPageIndex) { - capability.resolve([obj, currentNode]); - } else { - currentPageIndex++; - next(); - } - return; - } - nodesToVisit.push(obj); - next(); - }, capability.reject); - return; - } +(function (root, factory) { + { + factory((root.pdfjsDisplayAPI = {}), root.pdfjsSharedUtil, + root.pdfjsDisplayFontLoader, root.pdfjsDisplayCanvas, + root.pdfjsDisplayMetadata, root.pdfjsSharedGlobal); + } +}(this, function (exports, sharedUtil, displayFontLoader, displayCanvas, + displayMetadata, sharedGlobal, amdRequire) { - // Must be a child page dictionary. - assert( - isDict(currentNode), - 'page dictionary kid reference points to wrong type of object' - ); - var count = currentNode.get('Count'); - // If the current node doesn't have any children, avoid getting stuck - // in an empty node further down in the tree (see issue5644.pdf). - if (count === 0) { - checkAllKids = true; - } - // Skip nodes where the page can't be. - if (currentPageIndex + count <= pageIndex) { - currentPageIndex += count; - continue; - } +var InvalidPDFException = sharedUtil.InvalidPDFException; +var MessageHandler = sharedUtil.MessageHandler; +var MissingPDFException = sharedUtil.MissingPDFException; +var PasswordResponses = sharedUtil.PasswordResponses; +var PasswordException = sharedUtil.PasswordException; +var StatTimer = sharedUtil.StatTimer; +var UnexpectedResponseException = sharedUtil.UnexpectedResponseException; +var UnknownErrorException = sharedUtil.UnknownErrorException; +var Util = sharedUtil.Util; +var createPromiseCapability = sharedUtil.createPromiseCapability; +var combineUrl = sharedUtil.combineUrl; +var error = sharedUtil.error; +var deprecated = sharedUtil.deprecated; +var info = sharedUtil.info; +var isArrayBuffer = sharedUtil.isArrayBuffer; +var isSameOrigin = sharedUtil.isSameOrigin; +var loadJpegStream = sharedUtil.loadJpegStream; +var stringToBytes = sharedUtil.stringToBytes; +var warn = sharedUtil.warn; +var FontFaceObject = displayFontLoader.FontFaceObject; +var FontLoader = displayFontLoader.FontLoader; +var CanvasGraphics = displayCanvas.CanvasGraphics; +var createScratchCanvas = displayCanvas.createScratchCanvas; +var Metadata = displayMetadata.Metadata; +var PDFJS = sharedGlobal.PDFJS; +var globalScope = sharedGlobal.globalScope; - var kids = currentNode.get('Kids'); - assert(isArray(kids), 'page dictionary kids object is not an array'); - if (!checkAllKids && count === kids.length) { - // Nodes that don't have the page have been skipped and this is the - // bottom of the tree which means the page requested must be a - // descendant of this pages node. Ideally we would just resolve the - // promise with the page ref here, but there is the case where more - // pages nodes could link to single a page (see issue 3666 pdf). To - // handle this push it back on the queue so if it is a pages node it - // will be descended into. - nodesToVisit = [kids[pageIndex - currentPageIndex]]; - currentPageIndex = pageIndex; - continue; - } else { - for (var last = kids.length - 1; last >= 0; last--) { - nodesToVisit.push(kids[last]); - } - } - } - capability.reject('Page index ' + pageIndex + ' not found.'); - } - next(); - return capability.promise; - }, +var DEFAULT_RANGE_CHUNK_SIZE = 65536; // 2^16 = 65536 - getPageIndex: function Catalog_getPageIndex(ref) { - // The page tree nodes have the count of all the leaves below them. To get - // how many pages are before we just have to walk up the tree and keep - // adding the count of siblings to the left of the node. - var xref = this.xref; - function pagesBeforeRef(kidRef) { - var total = 0; - var parentRef; - return xref.fetchAsync(kidRef).then(function (node) { - if (!node) { - return null; - } - parentRef = node.getRaw('Parent'); - return node.getAsync('Parent'); - }).then(function (parent) { - if (!parent) { - return null; - } - return parent.getAsync('Kids'); - }).then(function (kids) { - if (!kids) { - return null; - } - var kidPromises = []; - var found = false; - for (var i = 0; i < kids.length; i++) { - var kid = kids[i]; - assert(isRef(kid), 'kids must be a ref'); - if (kid.num === kidRef.num) { - found = true; - break; - } - kidPromises.push(xref.fetchAsync(kid).then(function (kid) { - if (kid.has('Count')) { - var count = kid.get('Count'); - total += count; - } else { // page leaf node - total++; - } - })); - } - if (!found) { - error('kid ref not found in parents kids'); - } - return Promise.all(kidPromises).then(function () { - return [total, parentRef]; - }); - }); - } - var total = 0; - function next(ref) { - return pagesBeforeRef(ref).then(function (args) { - if (!args) { - return total; - } - var count = args[0]; - var parentRef = args[1]; - total += count; - return next(parentRef); - }); - } +/** + * The maximum allowed image size in total pixels e.g. width * height. Images + * above this value will not be drawn. Use -1 for no limit. + * @var {number} + */ +PDFJS.maxImageSize = (PDFJS.maxImageSize === undefined ? + -1 : PDFJS.maxImageSize); - return next(ref); - } - }; +/** + * The url of where the predefined Adobe CMaps are located. Include trailing + * slash. + * @var {string} + */ +PDFJS.cMapUrl = (PDFJS.cMapUrl === undefined ? null : PDFJS.cMapUrl); - return Catalog; -})(); +/** + * Specifies if CMaps are binary packed. + * @var {boolean} + */ +PDFJS.cMapPacked = PDFJS.cMapPacked === undefined ? false : PDFJS.cMapPacked; -var XRef = (function XRefClosure() { - function XRef(stream, password) { - this.stream = stream; - this.entries = []; - this.xrefstms = Object.create(null); - // prepare the XRef cache - this.cache = []; - this.password = password; - this.stats = { - streamTypes: [], - fontTypes: [] - }; - } +/** + * By default fonts are converted to OpenType fonts and loaded via font face + * rules. If disabled, the font will be rendered using a built in font renderer + * that constructs the glyphs with primitive path commands. + * @var {boolean} + */ +PDFJS.disableFontFace = (PDFJS.disableFontFace === undefined ? + false : PDFJS.disableFontFace); - XRef.prototype = { - setStartXRef: function XRef_setStartXRef(startXRef) { - // Store the starting positions of xref tables as we process them - // so we can recover from missing data errors - this.startXRefQueue = [startXRef]; - }, +/** + * Path for image resources, mainly for annotation icons. Include trailing + * slash. + * @var {string} + */ +PDFJS.imageResourcesPath = (PDFJS.imageResourcesPath === undefined ? + '' : PDFJS.imageResourcesPath); - parse: function XRef_parse(recoveryMode) { - var trailerDict; - if (!recoveryMode) { - trailerDict = this.readXRef(); - } else { - warn('Indexing all PDF objects'); - trailerDict = this.indexObjects(); - } - trailerDict.assignXref(this); - this.trailer = trailerDict; - var encrypt = trailerDict.get('Encrypt'); - if (encrypt) { - var ids = trailerDict.get('ID'); - var fileId = (ids && ids.length) ? ids[0] : ''; - this.encrypt = new CipherTransformFactory(encrypt, fileId, - this.password); - } +/** + * Disable the web worker and run all code on the main thread. This will happen + * automatically if the browser doesn't support workers or sending typed arrays + * to workers. + * @var {boolean} + */ +PDFJS.disableWorker = (PDFJS.disableWorker === undefined ? + false : PDFJS.disableWorker); - // get the root dictionary (catalog) object - if (!(this.root = trailerDict.get('Root'))) { - error('Invalid root reference'); - } - }, +/** + * Path and filename of the worker file. Required when the worker is enabled in + * development mode. If unspecified in the production build, the worker will be + * loaded based on the location of the pdf.js file. It is recommended that + * the workerSrc is set in a custom application to prevent issues caused by + * third-party frameworks and libraries. + * @var {string} + */ +PDFJS.workerSrc = (PDFJS.workerSrc === undefined ? null : PDFJS.workerSrc); - processXRefTable: function XRef_processXRefTable(parser) { - if (!('tableState' in this)) { - // Stores state of the table as we process it so we can resume - // from middle of table in case of missing data error - this.tableState = { - entryNum: 0, - streamPos: parser.lexer.stream.pos, - parserBuf1: parser.buf1, - parserBuf2: parser.buf2 - }; - } +/** + * Disable range request loading of PDF files. When enabled and if the server + * supports partial content requests then the PDF will be fetched in chunks. + * Enabled (false) by default. + * @var {boolean} + */ +PDFJS.disableRange = (PDFJS.disableRange === undefined ? + false : PDFJS.disableRange); - var obj = this.readXRefTable(parser); +/** + * Disable streaming of PDF file data. By default PDF.js attempts to load PDF + * in chunks. This default behavior can be disabled. + * @var {boolean} + */ +PDFJS.disableStream = (PDFJS.disableStream === undefined ? + false : PDFJS.disableStream); - // Sanity check - if (!isCmd(obj, 'trailer')) { - error('Invalid XRef table: could not find trailer dictionary'); - } - // Read trailer dictionary, e.g. - // trailer - // << /Size 22 - // /Root 20R - // /Info 10R - // /ID [ <81b14aafa313db63dbd6f981e49f94f4> ] - // >> - // The parser goes through the entire stream << ... >> and provides - // a getter interface for the key-value table - var dict = parser.getObj(); +/** + * Disable pre-fetching of PDF file data. When range requests are enabled PDF.js + * will automatically keep fetching more data even if it isn't needed to display + * the current page. This default behavior can be disabled. + * + * NOTE: It is also necessary to disable streaming, see above, + * in order for disabling of pre-fetching to work correctly. + * @var {boolean} + */ +PDFJS.disableAutoFetch = (PDFJS.disableAutoFetch === undefined ? + false : PDFJS.disableAutoFetch); - // The pdflib PDF generator can generate a nested trailer dictionary - if (!isDict(dict) && dict.dict) { - dict = dict.dict; - } - if (!isDict(dict)) { - error('Invalid XRef table: could not parse trailer dictionary'); - } - delete this.tableState; +/** + * Enables special hooks for debugging PDF.js. + * @var {boolean} + */ +PDFJS.pdfBug = (PDFJS.pdfBug === undefined ? false : PDFJS.pdfBug); - return dict; - }, +/** + * Enables transfer usage in postMessage for ArrayBuffers. + * @var {boolean} + */ +PDFJS.postMessageTransfers = (PDFJS.postMessageTransfers === undefined ? + true : PDFJS.postMessageTransfers); - readXRefTable: function XRef_readXRefTable(parser) { - // Example of cross-reference table: - // xref - // 0 1 <-- subsection header (first obj #, obj count) - // 0000000000 65535 f <-- actual object (offset, generation #, f/n) - // 23 2 <-- subsection header ... and so on ... - // 0000025518 00002 n - // 0000025635 00000 n - // trailer - // ... +/** + * Disables URL.createObjectURL usage. + * @var {boolean} + */ +PDFJS.disableCreateObjectURL = (PDFJS.disableCreateObjectURL === undefined ? + false : PDFJS.disableCreateObjectURL); - var stream = parser.lexer.stream; - var tableState = this.tableState; - stream.pos = tableState.streamPos; - parser.buf1 = tableState.parserBuf1; - parser.buf2 = tableState.parserBuf2; +/** + * Disables WebGL usage. + * @var {boolean} + */ +PDFJS.disableWebGL = (PDFJS.disableWebGL === undefined ? + true : PDFJS.disableWebGL); - // Outer loop is over subsection headers - var obj; +/** + * Disables fullscreen support, and by extension Presentation Mode, + * in browsers which support the fullscreen API. + * @var {boolean} + */ +PDFJS.disableFullscreen = (PDFJS.disableFullscreen === undefined ? + false : PDFJS.disableFullscreen); - while (true) { - if (!('firstEntryNum' in tableState) || !('entryCount' in tableState)) { - if (isCmd(obj = parser.getObj(), 'trailer')) { - break; - } - tableState.firstEntryNum = obj; - tableState.entryCount = parser.getObj(); - } +/** + * Enables CSS only zooming. + * @var {boolean} + */ +PDFJS.useOnlyCssZoom = (PDFJS.useOnlyCssZoom === undefined ? + false : PDFJS.useOnlyCssZoom); - var first = tableState.firstEntryNum; - var count = tableState.entryCount; - if (!isInt(first) || !isInt(count)) { - error('Invalid XRef table: wrong types in subsection header'); - } - // Inner loop is over objects themselves - for (var i = tableState.entryNum; i < count; i++) { - tableState.streamPos = stream.pos; - tableState.entryNum = i; - tableState.parserBuf1 = parser.buf1; - tableState.parserBuf2 = parser.buf2; +/** + * Controls the logging level. + * The constants from PDFJS.VERBOSITY_LEVELS should be used: + * - errors + * - warnings [default] + * - infos + * @var {number} + */ +PDFJS.verbosity = (PDFJS.verbosity === undefined ? + PDFJS.VERBOSITY_LEVELS.warnings : PDFJS.verbosity); - var entry = {}; - entry.offset = parser.getObj(); - entry.gen = parser.getObj(); - var type = parser.getObj(); +/** + * The maximum supported canvas size in total pixels e.g. width * height. + * The default value is 4096 * 4096. Use -1 for no limit. + * @var {number} + */ +PDFJS.maxCanvasPixels = (PDFJS.maxCanvasPixels === undefined ? + 16777216 : PDFJS.maxCanvasPixels); - if (isCmd(type, 'f')) { - entry.free = true; - } else if (isCmd(type, 'n')) { - entry.uncompressed = true; - } +/** + * (Deprecated) Opens external links in a new window if enabled. + * The default behavior opens external links in the PDF.js window. + * + * NOTE: This property has been deprecated, please use + * `PDFJS.externalLinkTarget = PDFJS.LinkTarget.BLANK` instead. + * @var {boolean} + */ +PDFJS.openExternalLinksInNewWindow = ( + PDFJS.openExternalLinksInNewWindow === undefined ? + false : PDFJS.openExternalLinksInNewWindow); - // Validate entry obj - if (!isInt(entry.offset) || !isInt(entry.gen) || - !(entry.free || entry.uncompressed)) { - error('Invalid entry in XRef subsection: ' + first + ', ' + count); - } +/** + * Specifies the |target| attribute for external links. + * The constants from PDFJS.LinkTarget should be used: + * - NONE [default] + * - SELF + * - BLANK + * - PARENT + * - TOP + * @var {number} + */ +PDFJS.externalLinkTarget = (PDFJS.externalLinkTarget === undefined ? + PDFJS.LinkTarget.NONE : PDFJS.externalLinkTarget); - if (!this.entries[i + first]) { - this.entries[i + first] = entry; - } - } +/** + * Specifies the |rel| attribute for external links. Defaults to stripping + * the referrer. + * @var {string} + */ +PDFJS.externalLinkRel = (PDFJS.externalLinkRel === undefined ? + 'noreferrer' : PDFJS.externalLinkRel); - tableState.entryNum = 0; - tableState.streamPos = stream.pos; - tableState.parserBuf1 = parser.buf1; - tableState.parserBuf2 = parser.buf2; - delete tableState.firstEntryNum; - delete tableState.entryCount; - } +/** + * Determines if we can eval strings as JS. Primarily used to improve + * performance for font rendering. + * @var {boolean} + */ +PDFJS.isEvalSupported = (PDFJS.isEvalSupported === undefined ? + true : PDFJS.isEvalSupported); - // Per issue 3248: hp scanners generate bad XRef - if (first === 1 && this.entries[1] && this.entries[1].free) { - // shifting the entries - this.entries.shift(); - } +/** + * Document initialization / loading parameters object. + * + * @typedef {Object} DocumentInitParameters + * @property {string} url - The URL of the PDF. + * @property {TypedArray|Array|string} data - Binary PDF data. Use typed arrays + * (Uint8Array) to improve the memory usage. If PDF data is BASE64-encoded, + * use atob() to convert it to a binary string first. + * @property {Object} httpHeaders - Basic authentication headers. + * @property {boolean} withCredentials - Indicates whether or not cross-site + * Access-Control requests should be made using credentials such as cookies + * or authorization headers. The default is false. + * @property {string} password - For decrypting password-protected PDFs. + * @property {TypedArray} initialData - A typed array with the first portion or + * all of the pdf data. Used by the extension since some data is already + * loaded before the switch to range requests. + * @property {number} length - The PDF file length. It's used for progress + * reports and range requests operations. + * @property {PDFDataRangeTransport} range + * @property {number} rangeChunkSize - Optional parameter to specify + * maximum number of bytes fetched per range request. The default value is + * 2^16 = 65536. + * @property {PDFWorker} worker - The worker that will be used for the loading + * and parsing of the PDF data. + */ - // Sanity check: as per spec, first object must be free - if (this.entries[0] && !this.entries[0].free) { - error('Invalid XRef table: unexpected first object'); - } - return obj; - }, +/** + * @typedef {Object} PDFDocumentStats + * @property {Array} streamTypes - Used stream types in the document (an item + * is set to true if specific stream ID was used in the document). + * @property {Array} fontTypes - Used font type in the document (an item is set + * to true if specific font ID was used in the document). + */ - processXRefStream: function XRef_processXRefStream(stream) { - if (!('streamState' in this)) { - // Stores state of the stream as we process it so we can resume - // from middle of stream in case of missing data error - var streamParameters = stream.dict; - var byteWidths = streamParameters.get('W'); - var range = streamParameters.get('Index'); - if (!range) { - range = [0, streamParameters.get('Size')]; - } +/** + * This is the main entry point for loading a PDF and interacting with it. + * NOTE: If a URL is used to fetch the PDF data a standard XMLHttpRequest(XHR) + * is used, which means it must follow the same origin rules that any XHR does + * e.g. No cross domain requests without CORS. + * + * @param {string|TypedArray|DocumentInitParameters|PDFDataRangeTransport} src + * Can be a url to where a PDF is located, a typed array (Uint8Array) + * already populated with data or parameter object. + * + * @param {PDFDataRangeTransport} pdfDataRangeTransport (deprecated) It is used + * if you want to manually serve range requests for data in the PDF. + * + * @param {function} passwordCallback (deprecated) It is used to request a + * password if wrong or no password was provided. The callback receives two + * parameters: function that needs to be called with new password and reason + * (see {PasswordResponses}). + * + * @param {function} progressCallback (deprecated) It is used to be able to + * monitor the loading progress of the PDF file (necessary to implement e.g. + * a loading bar). The callback receives an {Object} with the properties: + * {number} loaded and {number} total. + * + * @return {PDFDocumentLoadingTask} + */ +PDFJS.getDocument = function getDocument(src, + pdfDataRangeTransport, + passwordCallback, + progressCallback) { + var task = new PDFDocumentLoadingTask(); - this.streamState = { - entryRanges: range, - byteWidths: byteWidths, - entryNum: 0, - streamPos: stream.pos - }; + // Support of the obsolete arguments (for compatibility with API v1.0) + if (arguments.length > 1) { + deprecated('getDocument is called with pdfDataRangeTransport, ' + + 'passwordCallback or progressCallback argument'); + } + if (pdfDataRangeTransport) { + if (!(pdfDataRangeTransport instanceof PDFDataRangeTransport)) { + // Not a PDFDataRangeTransport instance, trying to add missing properties. + pdfDataRangeTransport = Object.create(pdfDataRangeTransport); + pdfDataRangeTransport.length = src.length; + pdfDataRangeTransport.initialData = src.initialData; + if (!pdfDataRangeTransport.abort) { + pdfDataRangeTransport.abort = function () {}; } - this.readXRefStream(stream); - delete this.streamState; - - return stream.dict; - }, - - readXRefStream: function XRef_readXRefStream(stream) { - var i, j; - var streamState = this.streamState; - stream.pos = streamState.streamPos; - - var byteWidths = streamState.byteWidths; - var typeFieldWidth = byteWidths[0]; - var offsetFieldWidth = byteWidths[1]; - var generationFieldWidth = byteWidths[2]; - - var entryRanges = streamState.entryRanges; - while (entryRanges.length > 0) { - var first = entryRanges[0]; - var n = entryRanges[1]; + } + src = Object.create(src); + src.range = pdfDataRangeTransport; + } + task.onPassword = passwordCallback || null; + task.onProgress = progressCallback || null; - if (!isInt(first) || !isInt(n)) { - error('Invalid XRef range fields: ' + first + ', ' + n); - } - if (!isInt(typeFieldWidth) || !isInt(offsetFieldWidth) || - !isInt(generationFieldWidth)) { - error('Invalid XRef entry fields length: ' + first + ', ' + n); - } - for (i = streamState.entryNum; i < n; ++i) { - streamState.entryNum = i; - streamState.streamPos = stream.pos; + var source; + if (typeof src === 'string') { + source = { url: src }; + } else if (isArrayBuffer(src)) { + source = { data: src }; + } else if (src instanceof PDFDataRangeTransport) { + source = { range: src }; + } else { + if (typeof src !== 'object') { + error('Invalid parameter in getDocument, need either Uint8Array, ' + + 'string or a parameter object'); + } + if (!src.url && !src.data && !src.range) { + error('Invalid parameter object: need either .data, .range or .url'); + } - var type = 0, offset = 0, generation = 0; - for (j = 0; j < typeFieldWidth; ++j) { - type = (type << 8) | stream.getByte(); - } - // if type field is absent, its default value is 1 - if (typeFieldWidth === 0) { - type = 1; - } - for (j = 0; j < offsetFieldWidth; ++j) { - offset = (offset << 8) | stream.getByte(); - } - for (j = 0; j < generationFieldWidth; ++j) { - generation = (generation << 8) | stream.getByte(); - } - var entry = {}; - entry.offset = offset; - entry.gen = generation; - switch (type) { - case 0: - entry.free = true; - break; - case 1: - entry.uncompressed = true; - break; - case 2: - break; - default: - error('Invalid XRef entry type: ' + type); - } - if (!this.entries[first + i]) { - this.entries[first + i] = entry; - } - } + source = src; + } - streamState.entryNum = 0; - streamState.streamPos = stream.pos; - entryRanges.splice(0, 2); + var params = {}; + var rangeTransport = null; + var worker = null; + for (var key in source) { + if (key === 'url' && typeof window !== 'undefined') { + // The full path is required in the 'url' field. + params[key] = combineUrl(window.location.href, source[key]); + continue; + } else if (key === 'range') { + rangeTransport = source[key]; + continue; + } else if (key === 'worker') { + worker = source[key]; + continue; + } else if (key === 'data' && !(source[key] instanceof Uint8Array)) { + // Converting string or array-like data to Uint8Array. + var pdfBytes = source[key]; + if (typeof pdfBytes === 'string') { + params[key] = stringToBytes(pdfBytes); + } else if (typeof pdfBytes === 'object' && pdfBytes !== null && + !isNaN(pdfBytes.length)) { + params[key] = new Uint8Array(pdfBytes); + } else if (isArrayBuffer(pdfBytes)) { + params[key] = new Uint8Array(pdfBytes); + } else { + error('Invalid PDF binary data: either typed array, string or ' + + 'array-like object is expected in the data property.'); } - }, + continue; + } + params[key] = source[key]; + } - indexObjects: function XRef_indexObjects() { - // Simple scan through the PDF content to find objects, - // trailers and XRef streams. - var TAB = 0x9, LF = 0xA, CR = 0xD, SPACE = 0x20; - var PERCENT = 0x25, LT = 0x3C; + params.rangeChunkSize = params.rangeChunkSize || DEFAULT_RANGE_CHUNK_SIZE; - function readToken(data, offset) { - var token = '', ch = data[offset]; - while (ch !== LF && ch !== CR && ch !== LT) { - if (++offset >= data.length) { - break; - } - token += String.fromCharCode(ch); - ch = data[offset]; - } - return token; - } - function skipUntil(data, offset, what) { - var length = what.length, dataLength = data.length; - var skipped = 0; - // finding byte sequence - while (offset < dataLength) { - var i = 0; - while (i < length && data[offset + i] === what[i]) { - ++i; - } - if (i >= length) { - break; // sequence found - } - offset++; - skipped++; - } - return skipped; + if (!worker) { + // Worker was not provided -- creating and owning our own. + worker = new PDFWorker(); + task._worker = worker; + } + var docId = task.docId; + worker.promise.then(function () { + if (task.destroyed) { + throw new Error('Loading aborted'); + } + return _fetchDocument(worker, params, rangeTransport, docId).then( + function (workerId) { + if (task.destroyed) { + throw new Error('Loading aborted'); } - var objRegExp = /^(\d+)\s+(\d+)\s+obj\b/; - var trailerBytes = new Uint8Array([116, 114, 97, 105, 108, 101, 114]); - var startxrefBytes = new Uint8Array([115, 116, 97, 114, 116, 120, 114, - 101, 102]); - var endobjBytes = new Uint8Array([101, 110, 100, 111, 98, 106]); - var xrefBytes = new Uint8Array([47, 88, 82, 101, 102]); + var messageHandler = new MessageHandler(docId, workerId, worker.port); + var transport = new WorkerTransport(messageHandler, task, rangeTransport); + task._transport = transport; + messageHandler.send('Ready', null); + }); + }).catch(task._capability.reject); - // Clear out any existing entries, since they may be bogus. - this.entries.length = 0; + return task; +}; - var stream = this.stream; - stream.pos = 0; - var buffer = stream.getBytes(); - var position = stream.start, length = buffer.length; - var trailers = [], xrefStms = []; - while (position < length) { - var ch = buffer[position]; - if (ch === TAB || ch === LF || ch === CR || ch === SPACE) { - ++position; - continue; - } - if (ch === PERCENT) { // %-comment - do { - ++position; - if (position >= length) { - break; - } - ch = buffer[position]; - } while (ch !== LF && ch !== CR); - continue; - } - var token = readToken(buffer, position); - var m; - if (token.indexOf('xref') === 0 && - (token.length === 4 || /\s/.test(token[4]))) { - position += skipUntil(buffer, position, trailerBytes); - trailers.push(position); - position += skipUntil(buffer, position, startxrefBytes); - } else if ((m = objRegExp.exec(token))) { - if (typeof this.entries[m[1]] === 'undefined') { - this.entries[m[1]] = { - offset: position - stream.start, - gen: m[2] | 0, - uncompressed: true - }; - } - var contentLength = skipUntil(buffer, position, endobjBytes) + 7; - var content = buffer.subarray(position, position + contentLength); +/** + * Starts fetching of specified PDF document/data. + * @param {PDFWorker} worker + * @param {Object} source + * @param {PDFDataRangeTransport} pdfDataRangeTransport + * @param {string} docId Unique document id, used as MessageHandler id. + * @returns {Promise} The promise, which is resolved when worker id of + * MessageHandler is known. + * @private + */ +function _fetchDocument(worker, source, pdfDataRangeTransport, docId) { + if (worker.destroyed) { + return Promise.reject(new Error('Worker was destroyed')); + } - // checking XRef stream suspect - // (it shall have '/XRef' and next char is not a letter) - var xrefTagOffset = skipUntil(content, 0, xrefBytes); - if (xrefTagOffset < contentLength && - content[xrefTagOffset + 5] < 64) { - xrefStms.push(position - stream.start); - this.xrefstms[position - stream.start] = 1; // Avoid recursion - } + source.disableAutoFetch = PDFJS.disableAutoFetch; + source.disableStream = PDFJS.disableStream; + source.chunkedViewerLoading = !!pdfDataRangeTransport; + if (pdfDataRangeTransport) { + source.length = pdfDataRangeTransport.length; + source.initialData = pdfDataRangeTransport.initialData; + } + return worker.messageHandler.sendWithPromise('GetDocRequest', { + docId: docId, + source: source, + disableRange: PDFJS.disableRange, + maxImageSize: PDFJS.maxImageSize, + cMapUrl: PDFJS.cMapUrl, + cMapPacked: PDFJS.cMapPacked, + disableFontFace: PDFJS.disableFontFace, + disableCreateObjectURL: PDFJS.disableCreateObjectURL, + verbosity: PDFJS.verbosity + }).then(function (workerId) { + if (worker.destroyed) { + throw new Error('Worker was destroyed'); + } + return workerId; + }); +} - position += contentLength; - } else if (token.indexOf('trailer') === 0 && - (token.length === 7 || /\s/.test(token[7]))) { - trailers.push(position); - position += skipUntil(buffer, position, startxrefBytes); - } else { - position += token.length + 1; - } - } - // reading XRef streams - var i, ii; - for (i = 0, ii = xrefStms.length; i < ii; ++i) { - this.startXRefQueue.push(xrefStms[i]); - this.readXRef(/* recoveryMode */ true); - } - // finding main trailer - var dict; - for (i = 0, ii = trailers.length; i < ii; ++i) { - stream.pos = trailers[i]; - var parser = new Parser(new Lexer(stream), true, this); - var obj = parser.getObj(); - if (!isCmd(obj, 'trailer')) { - continue; - } - // read the trailer dictionary - if (!isDict(dict = parser.getObj())) { - continue; - } - // taking the first one with 'ID' - if (dict.has('ID')) { - return dict; - } - } - // no tailer with 'ID', taking last one (if exists) - if (dict) { - return dict; - } - // nothing helps - // calling error() would reject worker with an UnknownErrorException. - throw new InvalidPDFException('Invalid PDF structure'); - }, +/** + * PDF document loading operation. + * @class + * @alias PDFDocumentLoadingTask + */ +var PDFDocumentLoadingTask = (function PDFDocumentLoadingTaskClosure() { + var nextDocumentId = 0; - readXRef: function XRef_readXRef(recoveryMode) { - var stream = this.stream; + /** @constructs PDFDocumentLoadingTask */ + function PDFDocumentLoadingTask() { + this._capability = createPromiseCapability(); + this._transport = null; + this._worker = null; - try { - while (this.startXRefQueue.length) { - var startXRef = this.startXRefQueue[0]; + /** + * Unique document loading task id -- used in MessageHandlers. + * @type {string} + */ + this.docId = 'd' + (nextDocumentId++); - stream.pos = startXRef + stream.start; + /** + * Shows if loading task is destroyed. + * @type {boolean} + */ + this.destroyed = false; - var parser = new Parser(new Lexer(stream), true, this); - var obj = parser.getObj(); - var dict; + /** + * Callback to request a password if wrong or no password was provided. + * The callback receives two parameters: function that needs to be called + * with new password and reason (see {PasswordResponses}). + */ + this.onPassword = null; - // Get dictionary - if (isCmd(obj, 'xref')) { - // Parse end-of-file XRef - dict = this.processXRefTable(parser); - if (!this.topDict) { - this.topDict = dict; - } + /** + * Callback to be able to monitor the loading progress of the PDF file + * (necessary to implement e.g. a loading bar). The callback receives + * an {Object} with the properties: {number} loaded and {number} total. + */ + this.onProgress = null; - // Recursively get other XRefs 'XRefStm', if any - obj = dict.get('XRefStm'); - if (isInt(obj)) { - var pos = obj; - // ignore previously loaded xref streams - // (possible infinite recursion) - if (!(pos in this.xrefstms)) { - this.xrefstms[pos] = 1; - this.startXRefQueue.push(pos); - } - } - } else if (isInt(obj)) { - // Parse in-stream XRef - if (!isInt(parser.getObj()) || - !isCmd(parser.getObj(), 'obj') || - !isStream(obj = parser.getObj())) { - error('Invalid XRef stream'); - } - dict = this.processXRefStream(obj); - if (!this.topDict) { - this.topDict = dict; - } - if (!dict) { - error('Failed to read XRef stream'); - } - } else { - error('Invalid XRef stream header'); - } + /** + * Callback to when unsupported feature is used. The callback receives + * an {PDFJS.UNSUPPORTED_FEATURES} argument. + */ + this.onUnsupportedFeature = null; + } - // Recursively get previous dictionary, if any - obj = dict.get('Prev'); - if (isInt(obj)) { - this.startXRefQueue.push(obj); - } else if (isRef(obj)) { - // The spec says Prev must not be a reference, i.e. "/Prev NNN" - // This is a fallback for non-compliant PDFs, i.e. "/Prev NNN 0 R" - this.startXRefQueue.push(obj.num); - } + PDFDocumentLoadingTask.prototype = + /** @lends PDFDocumentLoadingTask.prototype */ { + /** + * @return {Promise} + */ + get promise() { + return this._capability.promise; + }, - this.startXRefQueue.shift(); - } + /** + * Aborts all network requests and destroys worker. + * @return {Promise} A promise that is resolved after destruction activity + * is completed. + */ + destroy: function () { + this.destroyed = true; - return this.topDict; - } catch (e) { - if (e instanceof MissingDataException) { - throw e; + var transportDestroyed = !this._transport ? Promise.resolve() : + this._transport.destroy(); + return transportDestroyed.then(function () { + this._transport = null; + if (this._worker) { + this._worker.destroy(); + this._worker = null; } - info('(while reading XRef): ' + e); - } - - if (recoveryMode) { - return; - } - throw new XRefParseException(); + }.bind(this)); }, - getEntry: function XRef_getEntry(i) { - var xrefEntry = this.entries[i]; - if (xrefEntry && !xrefEntry.free && xrefEntry.offset) { - return xrefEntry; - } - return null; - }, + /** + * Registers callbacks to indicate the document loading completion. + * + * @param {function} onFulfilled The callback for the loading completion. + * @param {function} onRejected The callback for the loading failure. + * @return {Promise} A promise that is resolved after the onFulfilled or + * onRejected callback. + */ + then: function PDFDocumentLoadingTask_then(onFulfilled, onRejected) { + return this.promise.then.apply(this.promise, arguments); + } + }; - fetchIfRef: function XRef_fetchIfRef(obj) { - if (!isRef(obj)) { - return obj; - } - return this.fetch(obj); - }, + return PDFDocumentLoadingTask; +})(); - fetch: function XRef_fetch(ref, suppressEncryption) { - assert(isRef(ref), 'ref object is not a reference'); - var num = ref.num; - if (num in this.cache) { - var cacheEntry = this.cache[num]; - return cacheEntry; - } +/** + * Abstract class to support range requests file loading. + * @class + * @alias PDFJS.PDFDataRangeTransport + * @param {number} length + * @param {Uint8Array} initialData + */ +var PDFDataRangeTransport = (function pdfDataRangeTransportClosure() { + function PDFDataRangeTransport(length, initialData) { + this.length = length; + this.initialData = initialData; - var xrefEntry = this.getEntry(num); + this._rangeListeners = []; + this._progressListeners = []; + this._progressiveReadListeners = []; + this._readyCapability = createPromiseCapability(); + } + PDFDataRangeTransport.prototype = + /** @lends PDFDataRangeTransport.prototype */ { + addRangeListener: + function PDFDataRangeTransport_addRangeListener(listener) { + this._rangeListeners.push(listener); + }, - // the referenced entry can be free - if (xrefEntry === null) { - return (this.cache[num] = null); - } + addProgressListener: + function PDFDataRangeTransport_addProgressListener(listener) { + this._progressListeners.push(listener); + }, - if (xrefEntry.uncompressed) { - xrefEntry = this.fetchUncompressed(ref, xrefEntry, suppressEncryption); - } else { - xrefEntry = this.fetchCompressed(xrefEntry, suppressEncryption); - } - if (isDict(xrefEntry)){ - xrefEntry.objId = ref.toString(); - } else if (isStream(xrefEntry)) { - xrefEntry.dict.objId = ref.toString(); - } - return xrefEntry; + addProgressiveReadListener: + function PDFDataRangeTransport_addProgressiveReadListener(listener) { + this._progressiveReadListeners.push(listener); }, - fetchUncompressed: function XRef_fetchUncompressed(ref, xrefEntry, - suppressEncryption) { - var gen = ref.gen; - var num = ref.num; - if (xrefEntry.gen !== gen) { - error('inconsistent generation in XRef'); - } - var stream = this.stream.makeSubStream(xrefEntry.offset + - this.stream.start); - var parser = new Parser(new Lexer(stream), true, this); - var obj1 = parser.getObj(); - var obj2 = parser.getObj(); - var obj3 = parser.getObj(); - if (!isInt(obj1) || parseInt(obj1, 10) !== num || - !isInt(obj2) || parseInt(obj2, 10) !== gen || - !isCmd(obj3)) { - error('bad XRef entry'); - } - if (!isCmd(obj3, 'obj')) { - // some bad PDFs use "obj1234" and really mean 1234 - if (obj3.cmd.indexOf('obj') === 0) { - num = parseInt(obj3.cmd.substring(3), 10); - if (!isNaN(num)) { - return num; - } - } - error('bad XRef entry'); - } - if (this.encrypt && !suppressEncryption) { - xrefEntry = parser.getObj(this.encrypt.createCipherTransform(num, gen)); - } else { - xrefEntry = parser.getObj(); - } - if (!isStream(xrefEntry)) { - this.cache[num] = xrefEntry; + onDataRange: function PDFDataRangeTransport_onDataRange(begin, chunk) { + var listeners = this._rangeListeners; + for (var i = 0, n = listeners.length; i < n; ++i) { + listeners[i](begin, chunk); } - return xrefEntry; }, - fetchCompressed: function XRef_fetchCompressed(xrefEntry, - suppressEncryption) { - var tableOffset = xrefEntry.offset; - var stream = this.fetch(new Ref(tableOffset, 0)); - if (!isStream(stream)) { - error('bad ObjStm stream'); - } - var first = stream.dict.get('First'); - var n = stream.dict.get('N'); - if (!isInt(first) || !isInt(n)) { - error('invalid first and n parameters for ObjStm stream'); - } - var parser = new Parser(new Lexer(stream), false, this); - parser.allowStreams = true; - var i, entries = [], num, nums = []; - // read the object numbers to populate cache - for (i = 0; i < n; ++i) { - num = parser.getObj(); - if (!isInt(num)) { - error('invalid object number in the ObjStm stream: ' + num); - } - nums.push(num); - var offset = parser.getObj(); - if (!isInt(offset)) { - error('invalid object offset in the ObjStm stream: ' + offset); + onDataProgress: function PDFDataRangeTransport_onDataProgress(loaded) { + this._readyCapability.promise.then(function () { + var listeners = this._progressListeners; + for (var i = 0, n = listeners.length; i < n; ++i) { + listeners[i](loaded); } - } - // read stream objects for cache - for (i = 0; i < n; ++i) { - entries.push(parser.getObj()); - num = nums[i]; - var entry = this.entries[num]; - if (entry && entry.offset === tableOffset && entry.gen === i) { - this.cache[num] = entries[i]; + }.bind(this)); + }, + + onDataProgressiveRead: + function PDFDataRangeTransport_onDataProgress(chunk) { + this._readyCapability.promise.then(function () { + var listeners = this._progressiveReadListeners; + for (var i = 0, n = listeners.length; i < n; ++i) { + listeners[i](chunk); } - } - xrefEntry = entries[xrefEntry.gen]; - if (xrefEntry === undefined) { - error('bad XRef entry for compressed object'); - } - return xrefEntry; + }.bind(this)); }, - fetchIfRefAsync: function XRef_fetchIfRefAsync(obj) { - if (!isRef(obj)) { - return Promise.resolve(obj); - } - return this.fetchAsync(obj); + transportReady: function PDFDataRangeTransport_transportReady() { + this._readyCapability.resolve(); }, - fetchAsync: function XRef_fetchAsync(ref, suppressEncryption) { - var streamManager = this.stream.manager; - var xref = this; - return new Promise(function tryFetch(resolve, reject) { - try { - resolve(xref.fetch(ref, suppressEncryption)); - } catch (e) { - if (e instanceof MissingDataException) { - streamManager.requestRange(e.begin, e.end).then(function () { - tryFetch(resolve, reject); - }, reject); - return; - } - reject(e); - } - }); + requestDataRange: + function PDFDataRangeTransport_requestDataRange(begin, end) { + throw new Error('Abstract method PDFDataRangeTransport.requestDataRange'); }, - getCatalogObj: function XRef_getCatalogObj() { - return this.root; + abort: function PDFDataRangeTransport_abort() { } }; - - return XRef; + return PDFDataRangeTransport; })(); +PDFJS.PDFDataRangeTransport = PDFDataRangeTransport; + /** - * A NameTree/NumberTree is like a Dict but has some advantageous properties, - * see the specification (7.9.6 and 7.9.7) for additional details. - * TODO: implement all the Dict functions and make this more efficient. + * Proxy to a PDFDocument in the worker thread. Also, contains commonly used + * properties that can be read synchronously. + * @class + * @alias PDFDocumentProxy */ -var NameOrNumberTree = (function NameOrNumberTreeClosure() { - function NameOrNumberTree(root, xref) { - throw new Error('Cannot initialize NameOrNumberTree.'); +var PDFDocumentProxy = (function PDFDocumentProxyClosure() { + function PDFDocumentProxy(pdfInfo, transport, loadingTask) { + this.pdfInfo = pdfInfo; + this.transport = transport; + this.loadingTask = loadingTask; } - - NameOrNumberTree.prototype = { - getAll: function NameOrNumberTree_getAll() { - var dict = Object.create(null); - if (!this.root) { - return dict; - } - var xref = this.xref; - // Reading Name/Number tree. - var processed = new RefSet(); - processed.put(this.root); - var queue = [this.root]; - while (queue.length > 0) { - var i, n; - var obj = xref.fetchIfRef(queue.shift()); - if (!isDict(obj)) { - continue; - } - if (obj.has('Kids')) { - var kids = obj.get('Kids'); - for (i = 0, n = kids.length; i < n; i++) { - var kid = kids[i]; - assert(!processed.has(kid), - 'Duplicate entry in "' + this._type + '" tree.'); - queue.push(kid); - processed.put(kid); - } - continue; - } - var entries = obj.get(this._type); - if (isArray(entries)) { - for (i = 0, n = entries.length; i < n; i += 2) { - dict[xref.fetchIfRef(entries[i])] = xref.fetchIfRef(entries[i + 1]); - } - } - } - return dict; + PDFDocumentProxy.prototype = /** @lends PDFDocumentProxy.prototype */ { + /** + * @return {number} Total number of pages the PDF contains. + */ + get numPages() { + return this.pdfInfo.numPages; }, - - get: function NameOrNumberTree_get(key) { - if (!this.root) { - return null; - } - - var xref = this.xref; - var kidsOrEntries = xref.fetchIfRef(this.root); - var loopCount = 0; - var MAX_LEVELS = 10; - var l, r, m; - - // Perform a binary search to quickly find the entry that - // contains the key we are looking for. - while (kidsOrEntries.has('Kids')) { - if (++loopCount > MAX_LEVELS) { - warn('Search depth limit reached for "' + this._type + '" tree.'); - return null; - } - - var kids = kidsOrEntries.get('Kids'); - if (!isArray(kids)) { - return null; - } - - l = 0; - r = kids.length - 1; - while (l <= r) { - m = (l + r) >> 1; - var kid = xref.fetchIfRef(kids[m]); - var limits = kid.get('Limits'); - - if (key < xref.fetchIfRef(limits[0])) { - r = m - 1; - } else if (key > xref.fetchIfRef(limits[1])) { - l = m + 1; - } else { - kidsOrEntries = xref.fetchIfRef(kids[m]); - break; - } - } - if (l > r) { - return null; - } - } - - // If we get here, then we have found the right entry. Now go through the - // entries in the dictionary until we find the key we're looking for. - var entries = kidsOrEntries.get(this._type); - if (isArray(entries)) { - // Perform a binary search to reduce the lookup time. - l = 0; - r = entries.length - 2; - while (l <= r) { - // Check only even indices (0, 2, 4, ...) because the - // odd indices contain the actual data. - m = (l + r) & ~1; - var currentKey = xref.fetchIfRef(entries[m]); - if (key < currentKey) { - r = m - 2; - } else if (key > currentKey) { - l = m + 2; - } else { - return xref.fetchIfRef(entries[m + 1]); - } - } - } - return null; + /** + * @return {string} A unique ID to identify a PDF. Not guaranteed to be + * unique. + */ + get fingerprint() { + return this.pdfInfo.fingerprint; + }, + /** + * @param {number} pageNumber The page number to get. The first page is 1. + * @return {Promise} A promise that is resolved with a {@link PDFPageProxy} + * object. + */ + getPage: function PDFDocumentProxy_getPage(pageNumber) { + return this.transport.getPage(pageNumber); + }, + /** + * @param {{num: number, gen: number}} ref The page reference. Must have + * the 'num' and 'gen' properties. + * @return {Promise} A promise that is resolved with the page index that is + * associated with the reference. + */ + getPageIndex: function PDFDocumentProxy_getPageIndex(ref) { + return this.transport.getPageIndex(ref); + }, + /** + * @return {Promise} A promise that is resolved with a lookup table for + * mapping named destinations to reference numbers. + * + * This can be slow for large documents: use getDestination instead + */ + getDestinations: function PDFDocumentProxy_getDestinations() { + return this.transport.getDestinations(); + }, + /** + * @param {string} id The named destination to get. + * @return {Promise} A promise that is resolved with all information + * of the given named destination. + */ + getDestination: function PDFDocumentProxy_getDestination(id) { + return this.transport.getDestination(id); + }, + /** + * @return {Promise} A promise that is resolved with: + * an Array containing the pageLabels that correspond to the pageIndexes, + * or `null` when no pageLabels are present in the PDF file. + */ + getPageLabels: function PDFDocumentProxy_getPageLabels() { + return this.transport.getPageLabels(); + }, + /** + * @return {Promise} A promise that is resolved with a lookup table for + * mapping named attachments to their content. + */ + getAttachments: function PDFDocumentProxy_getAttachments() { + return this.transport.getAttachments(); + }, + /** + * @return {Promise} A promise that is resolved with an array of all the + * JavaScript strings in the name tree. + */ + getJavaScript: function PDFDocumentProxy_getJavaScript() { + return this.transport.getJavaScript(); + }, + /** + * @return {Promise} A promise that is resolved with an {Array} that is a + * tree outline (if it has one) of the PDF. The tree is in the format of: + * [ + * { + * title: string, + * bold: boolean, + * italic: boolean, + * color: rgb Uint8Array, + * dest: dest obj, + * url: string, + * items: array of more items like this + * }, + * ... + * ]. + */ + getOutline: function PDFDocumentProxy_getOutline() { + return this.transport.getOutline(); + }, + /** + * @return {Promise} A promise that is resolved with an {Object} that has + * info and metadata properties. Info is an {Object} filled with anything + * available in the information dictionary and similarly metadata is a + * {Metadata} object with information from the metadata section of the PDF. + */ + getMetadata: function PDFDocumentProxy_getMetadata() { + return this.transport.getMetadata(); + }, + /** + * @return {Promise} A promise that is resolved with a TypedArray that has + * the raw data from the PDF. + */ + getData: function PDFDocumentProxy_getData() { + return this.transport.getData(); + }, + /** + * @return {Promise} A promise that is resolved when the document's data + * is loaded. It is resolved with an {Object} that contains the length + * property that indicates size of the PDF data in bytes. + */ + getDownloadInfo: function PDFDocumentProxy_getDownloadInfo() { + return this.transport.downloadInfoCapability.promise; + }, + /** + * @return {Promise} A promise this is resolved with current stats about + * document structures (see {@link PDFDocumentStats}). + */ + getStats: function PDFDocumentProxy_getStats() { + return this.transport.getStats(); + }, + /** + * Cleans up resources allocated by the document, e.g. created @font-face. + */ + cleanup: function PDFDocumentProxy_cleanup() { + this.transport.startCleanup(); + }, + /** + * Destroys current document instance and terminates worker. + */ + destroy: function PDFDocumentProxy_destroy() { + return this.loadingTask.destroy(); } }; - return NameOrNumberTree; + return PDFDocumentProxy; })(); -var NameTree = (function NameTreeClosure() { - function NameTree(root, xref) { - this.root = root; - this.xref = xref; - this._type = 'Names'; - } - - Util.inherit(NameTree, NameOrNumberTree, {}); - - return NameTree; -})(); +/** + * Page getTextContent parameters. + * + * @typedef {Object} getTextContentParameters + * @param {boolean} normalizeWhitespace - replaces all occurrences of + * whitespace with standard spaces (0x20). The default value is `false`. + */ -var NumberTree = (function NumberTreeClosure() { - function NumberTree(root, xref) { - this.root = root; - this.xref = xref; - this._type = 'Nums'; - } +/** + * Page text content. + * + * @typedef {Object} TextContent + * @property {array} items - array of {@link TextItem} + * @property {Object} styles - {@link TextStyles} objects, indexed by font + * name. + */ - Util.inherit(NumberTree, NameOrNumberTree, {}); +/** + * Page text content part. + * + * @typedef {Object} TextItem + * @property {string} str - text content. + * @property {string} dir - text direction: 'ttb', 'ltr' or 'rtl'. + * @property {array} transform - transformation matrix. + * @property {number} width - width in device space. + * @property {number} height - height in device space. + * @property {string} fontName - font name used by pdf.js for converted font. + */ - return NumberTree; -})(); +/** + * Text style. + * + * @typedef {Object} TextStyle + * @property {number} ascent - font ascent. + * @property {number} descent - font descent. + * @property {boolean} vertical - text is in vertical mode. + * @property {string} fontFamily - possible font family + */ /** - * "A PDF file can refer to the contents of another file by using a File - * Specification (PDF 1.1)", see the spec (7.11) for more details. - * NOTE: Only embedded files are supported (as part of the attachments support) - * TODO: support the 'URL' file system (with caching if !/V), portable - * collections attributes and related files (/RF) + * Page annotation parameters. + * + * @typedef {Object} GetAnnotationsParameters + * @param {string} intent - Determines the annotations that will be fetched, + * can be either 'display' (viewable annotations) or 'print' + * (printable annotations). + * If the parameter is omitted, all annotations are fetched. */ -var FileSpec = (function FileSpecClosure() { - function FileSpec(root, xref) { - if (!root || !isDict(root)) { - return; - } - this.xref = xref; - this.root = root; - if (root.has('FS')) { - this.fs = root.get('FS'); - } - this.description = root.has('Desc') ? - stringToPDFString(root.get('Desc')) : - ''; - if (root.has('RF')) { - warn('Related file specifications are not supported'); - } - this.contentAvailable = true; - if (!root.has('EF')) { - this.contentAvailable = false; - warn('Non-embedded file specifications are not supported'); - } - } - - function pickPlatformItem(dict) { - // Look for the filename in this order: - // UF, F, Unix, Mac, DOS - if (dict.has('UF')) { - return dict.get('UF'); - } else if (dict.has('F')) { - return dict.get('F'); - } else if (dict.has('Unix')) { - return dict.get('Unix'); - } else if (dict.has('Mac')) { - return dict.get('Mac'); - } else if (dict.has('DOS')) { - return dict.get('DOS'); - } else { - return null; - } - } - FileSpec.prototype = { - get filename() { - if (!this._filename && this.root) { - var filename = pickPlatformItem(this.root) || 'unnamed'; - this._filename = stringToPDFString(filename). - replace(/\\\\/g, '\\'). - replace(/\\\//g, '/'). - replace(/\\/g, '/'); - } - return this._filename; - }, - get content() { - if (!this.contentAvailable) { - return null; - } - if (!this.contentRef && this.root) { - this.contentRef = pickPlatformItem(this.root.get('EF')); - } - var content = null; - if (this.contentRef) { - var xref = this.xref; - var fileObj = xref.fetchIfRef(this.contentRef); - if (fileObj && isStream(fileObj)) { - content = fileObj.getBytes(); - } else { - warn('Embedded file specification points to non-existing/invalid ' + - 'content'); - } - } else { - warn('Embedded file specification does not have a content'); - } - return content; - }, - get serializable() { - return { - filename: this.filename, - content: this.content - }; - } - }; - return FileSpec; -})(); +/** + * Page render parameters. + * + * @typedef {Object} RenderParameters + * @property {Object} canvasContext - A 2D context of a DOM Canvas object. + * @property {PDFJS.PageViewport} viewport - Rendering viewport obtained by + * calling of PDFPage.getViewport method. + * @property {string} intent - Rendering intent, can be 'display' or 'print' + * (default value is 'display'). + * @property {Array} transform - (optional) Additional transform, applied + * just before viewport transform. + * @property {Object} imageLayer - (optional) An object that has beginLayout, + * endLayout and appendImage functions. + * @property {function} continueCallback - (deprecated) A function that will be + * called each time the rendering is paused. To continue + * rendering call the function that is the first argument + * to the callback. + */ /** - * A helper for loading missing data in object graphs. It traverses the graph - * depth first and queues up any objects that have missing data. Once it has - * has traversed as many objects that are available it attempts to bundle the - * missing data requests and then resume from the nodes that weren't ready. + * PDF page operator list. * - * NOTE: It provides protection from circular references by keeping track of - * of loaded references. However, you must be careful not to load any graphs - * that have references to the catalog or other pages since that will cause the - * entire PDF document object graph to be traversed. + * @typedef {Object} PDFOperatorList + * @property {Array} fnArray - Array containing the operator functions. + * @property {Array} argsArray - Array containing the arguments of the + * functions. */ -var ObjectLoader = (function() { - function mayHaveChildren(value) { - return isRef(value) || isDict(value) || isArray(value) || isStream(value); - } - function addChildren(node, nodesToVisit) { - var value; - if (isDict(node) || isStream(node)) { - var map; - if (isDict(node)) { - map = node.map; - } else { - map = node.dict.map; - } - for (var key in map) { - value = map[key]; - if (mayHaveChildren(value)) { - nodesToVisit.push(value); - } +/** + * Proxy to a PDFPage in the worker thread. + * @class + * @alias PDFPageProxy + */ +var PDFPageProxy = (function PDFPageProxyClosure() { + function PDFPageProxy(pageIndex, pageInfo, transport) { + this.pageIndex = pageIndex; + this.pageInfo = pageInfo; + this.transport = transport; + this.stats = new StatTimer(); + this.stats.enabled = !!globalScope.PDFJS.enableStats; + this.commonObjs = transport.commonObjs; + this.objs = new PDFObjects(); + this.cleanupAfterRender = false; + this.pendingCleanup = false; + this.intentStates = Object.create(null); + this.destroyed = false; + } + PDFPageProxy.prototype = /** @lends PDFPageProxy.prototype */ { + /** + * @return {number} Page number of the page. First page is 1. + */ + get pageNumber() { + return this.pageIndex + 1; + }, + /** + * @return {number} The number of degrees the page is rotated clockwise. + */ + get rotate() { + return this.pageInfo.rotate; + }, + /** + * @return {Object} The reference that points to this page. It has 'num' and + * 'gen' properties. + */ + get ref() { + return this.pageInfo.ref; + }, + /** + * @return {Array} An array of the visible portion of the PDF page in the + * user space units - [x1, y1, x2, y2]. + */ + get view() { + return this.pageInfo.view; + }, + /** + * @param {number} scale The desired scale of the viewport. + * @param {number} rotate Degrees to rotate the viewport. If omitted this + * defaults to the page rotation. + * @return {PDFJS.PageViewport} Contains 'width' and 'height' properties + * along with transforms required for rendering. + */ + getViewport: function PDFPageProxy_getViewport(scale, rotate) { + if (arguments.length < 2) { + rotate = this.rotate; } - } else if (isArray(node)) { - for (var i = 0, ii = node.length; i < ii; i++) { - value = node[i]; - if (mayHaveChildren(value)) { - nodesToVisit.push(value); - } + return new PDFJS.PageViewport(this.view, scale, rotate, 0, 0); + }, + /** + * @param {GetAnnotationsParameters} params - Annotation parameters. + * @return {Promise} A promise that is resolved with an {Array} of the + * annotation objects. + */ + getAnnotations: function PDFPageProxy_getAnnotations(params) { + var intent = (params && params.intent) || null; + + if (!this.annotationsPromise || this.annotationsIntent !== intent) { + this.annotationsPromise = this.transport.getAnnotations(this.pageIndex, + intent); + this.annotationsIntent = intent; } - } - } + return this.annotationsPromise; + }, + /** + * Begins the process of rendering a page to the desired context. + * @param {RenderParameters} params Page render parameters. + * @return {RenderTask} An object that contains the promise, which + * is resolved when the page finishes rendering. + */ + render: function PDFPageProxy_render(params) { + var stats = this.stats; + stats.time('Overall'); - function ObjectLoader(obj, keys, xref) { - this.obj = obj; - this.keys = keys; - this.xref = xref; - this.refSet = null; - this.capability = null; - } + // If there was a pending destroy cancel it so no cleanup happens during + // this call to render. + this.pendingCleanup = false; - ObjectLoader.prototype = { - load: function ObjectLoader_load() { - var keys = this.keys; - this.capability = createPromiseCapability(); - // Don't walk the graph if all the data is already loaded. - if (!(this.xref.stream instanceof ChunkedStream) || - this.xref.stream.getMissingChunks().length === 0) { - this.capability.resolve(); - return this.capability.promise; + var renderingIntent = (params.intent === 'print' ? 'print' : 'display'); + + if (!this.intentStates[renderingIntent]) { + this.intentStates[renderingIntent] = Object.create(null); } + var intentState = this.intentStates[renderingIntent]; - this.refSet = new RefSet(); - // Setup the initial nodes to visit. - var nodesToVisit = []; - for (var i = 0; i < keys.length; i++) { - nodesToVisit.push(this.obj[keys[i]]); + // If there's no displayReadyCapability yet, then the operatorList + // was never requested before. Make the request and create the promise. + if (!intentState.displayReadyCapability) { + intentState.receivingOperatorList = true; + intentState.displayReadyCapability = createPromiseCapability(); + intentState.operatorList = { + fnArray: [], + argsArray: [], + lastChunk: false + }; + + this.stats.time('Page Request'); + this.transport.messageHandler.send('RenderPageRequest', { + pageIndex: this.pageNumber - 1, + intent: renderingIntent + }); } - this._walk(nodesToVisit); - return this.capability.promise; - }, + var internalRenderTask = new InternalRenderTask(complete, params, + this.objs, + this.commonObjs, + intentState.operatorList, + this.pageNumber); + internalRenderTask.useRequestAnimationFrame = renderingIntent !== 'print'; + if (!intentState.renderTasks) { + intentState.renderTasks = []; + } + intentState.renderTasks.push(internalRenderTask); + var renderTask = internalRenderTask.task; - _walk: function ObjectLoader_walk(nodesToVisit) { - var nodesToRevisit = []; - var pendingRequests = []; - // DFS walk of the object graph. - while (nodesToVisit.length) { - var currentNode = nodesToVisit.pop(); + // Obsolete parameter support + if (params.continueCallback) { + deprecated('render is used with continueCallback parameter'); + renderTask.onContinue = params.continueCallback; + } - // Only references or chunked streams can cause missing data exceptions. - if (isRef(currentNode)) { - // Skip nodes that have already been visited. - if (this.refSet.has(currentNode)) { - continue; - } - try { - var ref = currentNode; - this.refSet.put(ref); - currentNode = this.xref.fetch(currentNode); - } catch (e) { - if (!(e instanceof MissingDataException)) { - throw e; - } - nodesToRevisit.push(currentNode); - pendingRequests.push({ begin: e.begin, end: e.end }); + var self = this; + intentState.displayReadyCapability.promise.then( + function pageDisplayReadyPromise(transparency) { + if (self.pendingCleanup) { + complete(); + return; } + stats.time('Rendering'); + internalRenderTask.initalizeGraphics(transparency); + internalRenderTask.operatorListChanged(); + }, + function pageDisplayReadPromiseError(reason) { + complete(reason); } - if (currentNode && currentNode.getBaseStreams) { - var baseStreams = currentNode.getBaseStreams(); - var foundMissingData = false; - for (var i = 0; i < baseStreams.length; i++) { - var stream = baseStreams[i]; - if (stream.getMissingChunks && stream.getMissingChunks().length) { - foundMissingData = true; - pendingRequests.push({ - begin: stream.start, - end: stream.end - }); - } - } - if (foundMissingData) { - nodesToRevisit.push(currentNode); - } + ); + + function complete(error) { + var i = intentState.renderTasks.indexOf(internalRenderTask); + if (i >= 0) { + intentState.renderTasks.splice(i, 1); } - addChildren(currentNode, nodesToVisit); - } + if (self.cleanupAfterRender) { + self.pendingCleanup = true; + } + self._tryCleanup(); - if (pendingRequests.length) { - this.xref.stream.manager.requestRanges(pendingRequests).then( - function pendingRequestCallback() { - nodesToVisit = nodesToRevisit; - for (var i = 0; i < nodesToRevisit.length; i++) { - var node = nodesToRevisit[i]; - // Remove any reference nodes from the currrent refset so they - // aren't skipped when we revist them. - if (isRef(node)) { - this.refSet.remove(node); - } - } - this._walk(nodesToVisit); - }.bind(this), this.capability.reject); - return; + if (error) { + internalRenderTask.capability.reject(error); + } else { + internalRenderTask.capability.resolve(); + } + stats.timeEnd('Rendering'); + stats.timeEnd('Overall'); } - // Everything is loaded. - this.refSet = null; - this.capability.resolve(); - } - }; - return ObjectLoader; -})(); + return renderTask; + }, -exports.Catalog = Catalog; -exports.ObjectLoader = ObjectLoader; -exports.XRef = XRef; -})); + /** + * @return {Promise} A promise resolved with an {@link PDFOperatorList} + * object that represents page's operator list. + */ + getOperatorList: function PDFPageProxy_getOperatorList() { + function operatorListChanged() { + if (intentState.operatorList.lastChunk) { + intentState.opListReadCapability.resolve(intentState.operatorList); + var i = intentState.renderTasks.indexOf(opListTask); + if (i >= 0) { + intentState.renderTasks.splice(i, 1); + } + } + } -(function (root, factory) { - { - factory((root.pdfjsCorePsParser = {}), root.pdfjsSharedUtil, - root.pdfjsCoreParser); - } -}(this, function (exports, sharedUtil, coreParser) { + var renderingIntent = 'oplist'; + if (!this.intentStates[renderingIntent]) { + this.intentStates[renderingIntent] = Object.create(null); + } + var intentState = this.intentStates[renderingIntent]; + var opListTask; -var error = sharedUtil.error; -var EOF = coreParser.EOF; -var Lexer = coreParser.Lexer; + if (!intentState.opListReadCapability) { + opListTask = {}; + opListTask.operatorListChanged = operatorListChanged; + intentState.receivingOperatorList = true; + intentState.opListReadCapability = createPromiseCapability(); + intentState.renderTasks = []; + intentState.renderTasks.push(opListTask); + intentState.operatorList = { + fnArray: [], + argsArray: [], + lastChunk: false + }; -var PostScriptParser = (function PostScriptParserClosure() { - function PostScriptParser(lexer) { - this.lexer = lexer; - this.operators = []; - this.token = null; - this.prev = null; - } - PostScriptParser.prototype = { - nextToken: function PostScriptParser_nextToken() { - this.prev = this.token; - this.token = this.lexer.getToken(); - }, - accept: function PostScriptParser_accept(type) { - if (this.token.type === type) { - this.nextToken(); - return true; - } - return false; - }, - expect: function PostScriptParser_expect(type) { - if (this.accept(type)) { - return true; + this.transport.messageHandler.send('RenderPageRequest', { + pageIndex: this.pageIndex, + intent: renderingIntent + }); } - error('Unexpected symbol: found ' + this.token.type + ' expected ' + - type + '.'); + return intentState.opListReadCapability.promise; }, - parse: function PostScriptParser_parse() { - this.nextToken(); - this.expect(PostScriptTokenTypes.LBRACE); - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - return this.operators; + + /** + * @param {getTextContentParameters} params - getTextContent parameters. + * @return {Promise} That is resolved a {@link TextContent} + * object that represent the page text content. + */ + getTextContent: function PDFPageProxy_getTextContent(params) { + var normalizeWhitespace = (params && params.normalizeWhitespace) || false; + + return this.transport.messageHandler.sendWithPromise('GetTextContent', { + pageIndex: this.pageNumber - 1, + normalizeWhitespace: normalizeWhitespace, + }); }, - parseBlock: function PostScriptParser_parseBlock() { - while (true) { - if (this.accept(PostScriptTokenTypes.NUMBER)) { - this.operators.push(this.prev.value); - } else if (this.accept(PostScriptTokenTypes.OPERATOR)) { - this.operators.push(this.prev.value); - } else if (this.accept(PostScriptTokenTypes.LBRACE)) { - this.parseCondition(); - } else { + + /** + * Destroys page object. + */ + _destroy: function PDFPageProxy_destroy() { + this.destroyed = true; + this.transport.pageCache[this.pageIndex] = null; + + var waitOn = []; + Object.keys(this.intentStates).forEach(function(intent) { + if (intent === 'oplist') { + // Avoid errors below, since the renderTasks are just stubs. return; } - } + var intentState = this.intentStates[intent]; + intentState.renderTasks.forEach(function(renderTask) { + var renderCompleted = renderTask.capability.promise. + catch(function () {}); // ignoring failures + waitOn.push(renderCompleted); + renderTask.cancel(); + }); + }, this); + this.objs.clear(); + this.annotationsPromise = null; + this.pendingCleanup = false; + return Promise.all(waitOn); }, - parseCondition: function PostScriptParser_parseCondition() { - // Add two place holders that will be updated later - var conditionLocation = this.operators.length; - this.operators.push(null, null); - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - if (this.accept(PostScriptTokenTypes.IF)) { - // The true block is right after the 'if' so it just falls through on - // true else it jumps and skips the true block. - this.operators[conditionLocation] = this.operators.length; - this.operators[conditionLocation + 1] = 'jz'; - } else if (this.accept(PostScriptTokenTypes.LBRACE)) { - var jumpLocation = this.operators.length; - this.operators.push(null, null); - var endOfTrue = this.operators.length; - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - this.expect(PostScriptTokenTypes.IFELSE); - // The jump is added at the end of the true block to skip the false - // block. - this.operators[jumpLocation] = this.operators.length; - this.operators[jumpLocation + 1] = 'j'; + /** + * Cleans up resources allocated by the page. (deprecated) + */ + destroy: function() { + deprecated('page destroy method, use cleanup() instead'); + this.cleanup(); + }, - this.operators[conditionLocation] = endOfTrue; - this.operators[conditionLocation + 1] = 'jz'; - } else { - error('PS Function: error parsing conditional.'); + /** + * Cleans up resources allocated by the page. + */ + cleanup: function PDFPageProxy_cleanup() { + this.pendingCleanup = true; + this._tryCleanup(); + }, + /** + * For internal use only. Attempts to clean up if rendering is in a state + * where that's possible. + * @ignore + */ + _tryCleanup: function PDFPageProxy_tryCleanup() { + if (!this.pendingCleanup || + Object.keys(this.intentStates).some(function(intent) { + var intentState = this.intentStates[intent]; + return (intentState.renderTasks.length !== 0 || + intentState.receivingOperatorList); + }, this)) { + return; + } + + Object.keys(this.intentStates).forEach(function(intent) { + delete this.intentStates[intent]; + }, this); + this.objs.clear(); + this.annotationsPromise = null; + this.pendingCleanup = false; + }, + /** + * For internal use only. + * @ignore + */ + _startRenderPage: function PDFPageProxy_startRenderPage(transparency, + intent) { + var intentState = this.intentStates[intent]; + // TODO Refactor RenderPageRequest to separate rendering + // and operator list logic + if (intentState.displayReadyCapability) { + intentState.displayReadyCapability.resolve(transparency); + } + }, + /** + * For internal use only. + * @ignore + */ + _renderPageChunk: function PDFPageProxy_renderPageChunk(operatorListChunk, + intent) { + var intentState = this.intentStates[intent]; + var i, ii; + // Add the new chunk to the current operator list. + for (i = 0, ii = operatorListChunk.length; i < ii; i++) { + intentState.operatorList.fnArray.push(operatorListChunk.fnArray[i]); + intentState.operatorList.argsArray.push( + operatorListChunk.argsArray[i]); + } + intentState.operatorList.lastChunk = operatorListChunk.lastChunk; + + // Notify all the rendering tasks there are more operators to be consumed. + for (i = 0; i < intentState.renderTasks.length; i++) { + intentState.renderTasks[i].operatorListChanged(); + } + + if (operatorListChunk.lastChunk) { + intentState.receivingOperatorList = false; + this._tryCleanup(); } } }; - return PostScriptParser; + return PDFPageProxy; })(); -var PostScriptTokenTypes = { - LBRACE: 0, - RBRACE: 1, - NUMBER: 2, - OPERATOR: 3, - IF: 4, - IFELSE: 5 -}; +/** + * PDF.js web worker abstraction, it controls instantiation of PDF documents and + * WorkerTransport for them. If creation of a web worker is not possible, + * a "fake" worker will be used instead. + * @class + */ +var PDFWorker = (function PDFWorkerClosure() { + var nextFakeWorkerId = 0; -var PostScriptToken = (function PostScriptTokenClosure() { - function PostScriptToken(type, value) { - this.type = type; - this.value = value; + function getWorkerSrc() { + if (PDFJS.workerSrc) { + return PDFJS.workerSrc; + } + if (pdfjsFilePath) { + return pdfjsFilePath.replace(/\.js$/i, '.worker.js'); + } + error('No PDFJS.workerSrc specified'); } - var opCache = Object.create(null); - - PostScriptToken.getOperator = function PostScriptToken_getOperator(op) { - var opValue = opCache[op]; - if (opValue) { - return opValue; + // Loads worker code into main thread. + function setupFakeWorkerGlobal() { + if (!PDFJS.fakeWorkerFilesLoadedCapability) { + PDFJS.fakeWorkerFilesLoadedCapability = createPromiseCapability(); + // In the developer build load worker_loader which in turn loads all the + // other files and resolves the promise. In production only the + // pdf.worker.js file is needed. + PDFJS.fakeWorkerFilesLoadedCapability.resolve(); } - return opCache[op] = new PostScriptToken(PostScriptTokenTypes.OPERATOR, op); - }; + return PDFJS.fakeWorkerFilesLoadedCapability.promise; + } - PostScriptToken.LBRACE = new PostScriptToken(PostScriptTokenTypes.LBRACE, - '{'); - PostScriptToken.RBRACE = new PostScriptToken(PostScriptTokenTypes.RBRACE, - '}'); - PostScriptToken.IF = new PostScriptToken(PostScriptTokenTypes.IF, 'IF'); - PostScriptToken.IFELSE = new PostScriptToken(PostScriptTokenTypes.IFELSE, - 'IFELSE'); - return PostScriptToken; -})(); + function createCDNWrapper(url) { + // We will rely on blob URL's property to specify origin. + // We want this function to fail in case if createObjectURL or Blob do not + // exist or fail for some reason -- our Worker creation will fail anyway. + var wrapper = 'importScripts(\'' + url + '\');'; + return URL.createObjectURL(new Blob([wrapper])); + } -var PostScriptLexer = (function PostScriptLexerClosure() { - function PostScriptLexer(stream) { - this.stream = stream; - this.nextChar(); + function PDFWorker(name) { + this.name = name; + this.destroyed = false; - this.strBuf = []; + this._readyCapability = createPromiseCapability(); + this._port = null; + this._webWorker = null; + this._messageHandler = null; + this._initialize(); } - PostScriptLexer.prototype = { - nextChar: function PostScriptLexer_nextChar() { - return (this.currentChar = this.stream.getByte()); + + PDFWorker.prototype = /** @lends PDFWorker.prototype */ { + get promise() { + return this._readyCapability.promise; }, - getToken: function PostScriptLexer_getToken() { - var comment = false; - var ch = this.currentChar; - // skip comments - while (true) { - if (ch < 0) { - return EOF; - } + get port() { + return this._port; + }, - if (comment) { - if (ch === 0x0A || ch === 0x0D) { - comment = false; - } - } else if (ch === 0x25) { // '%' - comment = true; - } else if (!Lexer.isSpace(ch)) { - break; - } - ch = this.nextChar(); - } - switch (ch | 0) { - case 0x30: case 0x31: case 0x32: case 0x33: case 0x34: // '0'-'4' - case 0x35: case 0x36: case 0x37: case 0x38: case 0x39: // '5'-'9' - case 0x2B: case 0x2D: case 0x2E: // '+', '-', '.' - return new PostScriptToken(PostScriptTokenTypes.NUMBER, - this.getNumber()); - case 0x7B: // '{' - this.nextChar(); - return PostScriptToken.LBRACE; - case 0x7D: // '}' - this.nextChar(); - return PostScriptToken.RBRACE; - } - // operator - var strBuf = this.strBuf; - strBuf.length = 0; - strBuf[0] = String.fromCharCode(ch); + get messageHandler() { + return this._messageHandler; + }, - while ((ch = this.nextChar()) >= 0 && // and 'A'-'Z', 'a'-'z' - ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { - strBuf.push(String.fromCharCode(ch)); - } - var str = strBuf.join(''); - switch (str.toLowerCase()) { - case 'if': - return PostScriptToken.IF; - case 'ifelse': - return PostScriptToken.IFELSE; - default: - return PostScriptToken.getOperator(str); - } + _initialize: function PDFWorker_initialize() { + // If worker support isn't disabled explicit and the browser has worker + // support, create a new web worker and test if it/the browser fulfills + // all requirements to run parts of pdf.js in a web worker. + // Right now, the requirement is, that an Uint8Array is still an + // Uint8Array as it arrives on the worker. (Chrome added this with v.15.) + // Either workers are disabled, not supported or have thrown an exception. + // Thus, we fallback to a faked worker. + this._setupFakeWorker(); }, - getNumber: function PostScriptLexer_getNumber() { - var ch = this.currentChar; - var strBuf = this.strBuf; - strBuf.length = 0; - strBuf[0] = String.fromCharCode(ch); - while ((ch = this.nextChar()) >= 0) { - if ((ch >= 0x30 && ch <= 0x39) || // '0'-'9' - ch === 0x2D || ch === 0x2E) { // '-', '.' - strBuf.push(String.fromCharCode(ch)); - } else { - break; - } - } - var value = parseFloat(strBuf.join('')); - if (isNaN(value)) { - error('Invalid floating point number: ' + value); + _setupFakeWorker: function PDFWorker_setupFakeWorker() { + if (!globalScope.PDFJS.disableWorker) { + warn('Setting up fake worker.'); + globalScope.PDFJS.disableWorker = true; } - return value; - } - }; - return PostScriptLexer; -})(); -exports.PostScriptLexer = PostScriptLexer; -exports.PostScriptParser = PostScriptParser; -})); + setupFakeWorkerGlobal().then(function () { + if (this.destroyed) { + this._readyCapability.reject(new Error('Worker was destroyed')); + return; + } + // If we don't use a worker, just post/sendMessage to the main thread. + var port = { + _listeners: [], + postMessage: function (obj) { + var e = {data: obj}; + this._listeners.forEach(function (listener) { + listener.call(this, e); + }, this); + }, + addEventListener: function (name, listener) { + this._listeners.push(listener); + }, + removeEventListener: function (name, listener) { + var i = this._listeners.indexOf(listener); + this._listeners.splice(i, 1); + }, + terminate: function () {} + }; + this._port = port; -(function (root, factory) { - { - factory((root.pdfjsDisplayAPI = {}), root.pdfjsSharedUtil, - root.pdfjsDisplayFontLoader, root.pdfjsDisplayCanvas, - root.pdfjsDisplayMetadata, root.pdfjsSharedGlobal); - } -}(this, function (exports, sharedUtil, displayFontLoader, displayCanvas, - displayMetadata, sharedGlobal, amdRequire) { + // All fake workers use the same port, making id unique. + var id = 'fake' + (nextFakeWorkerId++); -var InvalidPDFException = sharedUtil.InvalidPDFException; -var MessageHandler = sharedUtil.MessageHandler; -var MissingPDFException = sharedUtil.MissingPDFException; -var PasswordResponses = sharedUtil.PasswordResponses; -var PasswordException = sharedUtil.PasswordException; -var StatTimer = sharedUtil.StatTimer; -var UnexpectedResponseException = sharedUtil.UnexpectedResponseException; -var UnknownErrorException = sharedUtil.UnknownErrorException; -var Util = sharedUtil.Util; -var createPromiseCapability = sharedUtil.createPromiseCapability; -var combineUrl = sharedUtil.combineUrl; -var error = sharedUtil.error; -var deprecated = sharedUtil.deprecated; -var info = sharedUtil.info; -var isArrayBuffer = sharedUtil.isArrayBuffer; -var isSameOrigin = sharedUtil.isSameOrigin; -var loadJpegStream = sharedUtil.loadJpegStream; -var stringToBytes = sharedUtil.stringToBytes; -var warn = sharedUtil.warn; -var FontFaceObject = displayFontLoader.FontFaceObject; -var FontLoader = displayFontLoader.FontLoader; -var CanvasGraphics = displayCanvas.CanvasGraphics; -var createScratchCanvas = displayCanvas.createScratchCanvas; -var Metadata = displayMetadata.Metadata; -var PDFJS = sharedGlobal.PDFJS; -var globalScope = sharedGlobal.globalScope; + // If the main thread is our worker, setup the handling for the + // messages -- the main thread sends to it self. + var workerHandler = new MessageHandler(id + '_worker', id, port); + PDFJS.WorkerMessageHandler.setup(workerHandler, port); -var DEFAULT_RANGE_CHUNK_SIZE = 65536; // 2^16 = 65536 + var messageHandler = new MessageHandler(id, id + '_worker', port); + this._messageHandler = messageHandler; + this._readyCapability.resolve(); + }.bind(this)); + }, + /** + * Destroys the worker instance. + */ + destroy: function PDFWorker_destroy() { + this.destroyed = true; + if (this._webWorker) { + // We need to terminate only web worker created resource. + this._webWorker.terminate(); + this._webWorker = null; + } + this._port = null; + if (this._messageHandler) { + this._messageHandler.destroy(); + this._messageHandler = null; + } + } + }; -/** - * The maximum allowed image size in total pixels e.g. width * height. Images - * above this value will not be drawn. Use -1 for no limit. - * @var {number} - */ -PDFJS.maxImageSize = (PDFJS.maxImageSize === undefined ? - -1 : PDFJS.maxImageSize); + return PDFWorker; +})(); +PDFJS.PDFWorker = PDFWorker; /** - * The url of where the predefined Adobe CMaps are located. Include trailing - * slash. - * @var {string} + * For internal use only. + * @ignore */ -PDFJS.cMapUrl = (PDFJS.cMapUrl === undefined ? null : PDFJS.cMapUrl); +var WorkerTransport = (function WorkerTransportClosure() { + function WorkerTransport(messageHandler, loadingTask, pdfDataRangeTransport) { + this.messageHandler = messageHandler; + this.loadingTask = loadingTask; + this.pdfDataRangeTransport = pdfDataRangeTransport; + this.commonObjs = new PDFObjects(); + this.fontLoader = new FontLoader(loadingTask.docId); -/** - * Specifies if CMaps are binary packed. - * @var {boolean} - */ -PDFJS.cMapPacked = PDFJS.cMapPacked === undefined ? false : PDFJS.cMapPacked; + this.destroyed = false; + this.destroyCapability = null; -/** - * By default fonts are converted to OpenType fonts and loaded via font face - * rules. If disabled, the font will be rendered using a built in font renderer - * that constructs the glyphs with primitive path commands. - * @var {boolean} - */ -PDFJS.disableFontFace = (PDFJS.disableFontFace === undefined ? - false : PDFJS.disableFontFace); + this.pageCache = []; + this.pagePromises = []; + this.downloadInfoCapability = createPromiseCapability(); -/** - * Path for image resources, mainly for annotation icons. Include trailing - * slash. - * @var {string} - */ -PDFJS.imageResourcesPath = (PDFJS.imageResourcesPath === undefined ? - '' : PDFJS.imageResourcesPath); + this.setupMessageHandler(); + } + WorkerTransport.prototype = { + destroy: function WorkerTransport_destroy() { + if (this.destroyCapability) { + return this.destroyCapability.promise; + } -/** - * Disable the web worker and run all code on the main thread. This will happen - * automatically if the browser doesn't support workers or sending typed arrays - * to workers. - * @var {boolean} - */ -PDFJS.disableWorker = (PDFJS.disableWorker === undefined ? - false : PDFJS.disableWorker); + this.destroyed = true; + this.destroyCapability = createPromiseCapability(); -/** - * Path and filename of the worker file. Required when the worker is enabled in - * development mode. If unspecified in the production build, the worker will be - * loaded based on the location of the pdf.js file. It is recommended that - * the workerSrc is set in a custom application to prevent issues caused by - * third-party frameworks and libraries. - * @var {string} - */ -PDFJS.workerSrc = (PDFJS.workerSrc === undefined ? null : PDFJS.workerSrc); + var waitOn = []; + // We need to wait for all renderings to be completed, e.g. + // timeout/rAF can take a long time. + this.pageCache.forEach(function (page) { + if (page) { + waitOn.push(page._destroy()); + } + }); + this.pageCache = []; + this.pagePromises = []; + var self = this; + // We also need to wait for the worker to finish its long running tasks. + var terminated = this.messageHandler.sendWithPromise('Terminate', null); + waitOn.push(terminated); + Promise.all(waitOn).then(function () { + self.fontLoader.clear(); + if (self.pdfDataRangeTransport) { + self.pdfDataRangeTransport.abort(); + self.pdfDataRangeTransport = null; + } + if (self.messageHandler) { + self.messageHandler.destroy(); + self.messageHandler = null; + } + self.destroyCapability.resolve(); + }, this.destroyCapability.reject); + return this.destroyCapability.promise; + }, -/** - * Disable range request loading of PDF files. When enabled and if the server - * supports partial content requests then the PDF will be fetched in chunks. - * Enabled (false) by default. - * @var {boolean} - */ -PDFJS.disableRange = (PDFJS.disableRange === undefined ? - false : PDFJS.disableRange); + setupMessageHandler: + function WorkerTransport_setupMessageHandler() { + var messageHandler = this.messageHandler; -/** - * Disable streaming of PDF file data. By default PDF.js attempts to load PDF - * in chunks. This default behavior can be disabled. - * @var {boolean} - */ -PDFJS.disableStream = (PDFJS.disableStream === undefined ? - false : PDFJS.disableStream); + function updatePassword(password) { + messageHandler.send('UpdatePassword', password); + } -/** - * Disable pre-fetching of PDF file data. When range requests are enabled PDF.js - * will automatically keep fetching more data even if it isn't needed to display - * the current page. This default behavior can be disabled. - * - * NOTE: It is also necessary to disable streaming, see above, - * in order for disabling of pre-fetching to work correctly. - * @var {boolean} - */ -PDFJS.disableAutoFetch = (PDFJS.disableAutoFetch === undefined ? - false : PDFJS.disableAutoFetch); + var pdfDataRangeTransport = this.pdfDataRangeTransport; + if (pdfDataRangeTransport) { + pdfDataRangeTransport.addRangeListener(function(begin, chunk) { + messageHandler.send('OnDataRange', { + begin: begin, + chunk: chunk + }); + }); -/** - * Enables special hooks for debugging PDF.js. - * @var {boolean} - */ -PDFJS.pdfBug = (PDFJS.pdfBug === undefined ? false : PDFJS.pdfBug); + pdfDataRangeTransport.addProgressListener(function(loaded) { + messageHandler.send('OnDataProgress', { + loaded: loaded + }); + }); -/** - * Enables transfer usage in postMessage for ArrayBuffers. - * @var {boolean} - */ -PDFJS.postMessageTransfers = (PDFJS.postMessageTransfers === undefined ? - true : PDFJS.postMessageTransfers); + pdfDataRangeTransport.addProgressiveReadListener(function(chunk) { + messageHandler.send('OnDataRange', { + chunk: chunk + }); + }); -/** - * Disables URL.createObjectURL usage. - * @var {boolean} - */ -PDFJS.disableCreateObjectURL = (PDFJS.disableCreateObjectURL === undefined ? - false : PDFJS.disableCreateObjectURL); - -/** - * Disables WebGL usage. - * @var {boolean} - */ -PDFJS.disableWebGL = (PDFJS.disableWebGL === undefined ? - true : PDFJS.disableWebGL); + messageHandler.on('RequestDataRange', + function transportDataRange(data) { + pdfDataRangeTransport.requestDataRange(data.begin, data.end); + }, this); + } -/** - * Disables fullscreen support, and by extension Presentation Mode, - * in browsers which support the fullscreen API. - * @var {boolean} - */ -PDFJS.disableFullscreen = (PDFJS.disableFullscreen === undefined ? - false : PDFJS.disableFullscreen); + messageHandler.on('GetDoc', function transportDoc(data) { + var pdfInfo = data.pdfInfo; + this.numPages = data.pdfInfo.numPages; + var loadingTask = this.loadingTask; + var pdfDocument = new PDFDocumentProxy(pdfInfo, this, loadingTask); + this.pdfDocument = pdfDocument; + loadingTask._capability.resolve(pdfDocument); + }, this); -/** - * Enables CSS only zooming. - * @var {boolean} - */ -PDFJS.useOnlyCssZoom = (PDFJS.useOnlyCssZoom === undefined ? - false : PDFJS.useOnlyCssZoom); + messageHandler.on('NeedPassword', + function transportNeedPassword(exception) { + var loadingTask = this.loadingTask; + if (loadingTask.onPassword) { + return loadingTask.onPassword(updatePassword, + PasswordResponses.NEED_PASSWORD); + } + loadingTask._capability.reject( + new PasswordException(exception.message, exception.code)); + }, this); -/** - * Controls the logging level. - * The constants from PDFJS.VERBOSITY_LEVELS should be used: - * - errors - * - warnings [default] - * - infos - * @var {number} - */ -PDFJS.verbosity = (PDFJS.verbosity === undefined ? - PDFJS.VERBOSITY_LEVELS.warnings : PDFJS.verbosity); + messageHandler.on('IncorrectPassword', + function transportIncorrectPassword(exception) { + var loadingTask = this.loadingTask; + if (loadingTask.onPassword) { + return loadingTask.onPassword(updatePassword, + PasswordResponses.INCORRECT_PASSWORD); + } + loadingTask._capability.reject( + new PasswordException(exception.message, exception.code)); + }, this); -/** - * The maximum supported canvas size in total pixels e.g. width * height. - * The default value is 4096 * 4096. Use -1 for no limit. - * @var {number} - */ -PDFJS.maxCanvasPixels = (PDFJS.maxCanvasPixels === undefined ? - 16777216 : PDFJS.maxCanvasPixels); + messageHandler.on('InvalidPDF', function transportInvalidPDF(exception) { + this.loadingTask._capability.reject( + new InvalidPDFException(exception.message)); + }, this); -/** - * (Deprecated) Opens external links in a new window if enabled. - * The default behavior opens external links in the PDF.js window. - * - * NOTE: This property has been deprecated, please use - * `PDFJS.externalLinkTarget = PDFJS.LinkTarget.BLANK` instead. - * @var {boolean} - */ -PDFJS.openExternalLinksInNewWindow = ( - PDFJS.openExternalLinksInNewWindow === undefined ? - false : PDFJS.openExternalLinksInNewWindow); + messageHandler.on('MissingPDF', function transportMissingPDF(exception) { + this.loadingTask._capability.reject( + new MissingPDFException(exception.message)); + }, this); -/** - * Specifies the |target| attribute for external links. - * The constants from PDFJS.LinkTarget should be used: - * - NONE [default] - * - SELF - * - BLANK - * - PARENT - * - TOP - * @var {number} - */ -PDFJS.externalLinkTarget = (PDFJS.externalLinkTarget === undefined ? - PDFJS.LinkTarget.NONE : PDFJS.externalLinkTarget); + messageHandler.on('UnexpectedResponse', + function transportUnexpectedResponse(exception) { + this.loadingTask._capability.reject( + new UnexpectedResponseException(exception.message, exception.status)); + }, this); -/** - * Specifies the |rel| attribute for external links. Defaults to stripping - * the referrer. - * @var {string} - */ -PDFJS.externalLinkRel = (PDFJS.externalLinkRel === undefined ? - 'noreferrer' : PDFJS.externalLinkRel); + messageHandler.on('UnknownError', + function transportUnknownError(exception) { + this.loadingTask._capability.reject( + new UnknownErrorException(exception.message, exception.details)); + }, this); -/** - * Determines if we can eval strings as JS. Primarily used to improve - * performance for font rendering. - * @var {boolean} - */ -PDFJS.isEvalSupported = (PDFJS.isEvalSupported === undefined ? - true : PDFJS.isEvalSupported); + messageHandler.on('DataLoaded', function transportPage(data) { + this.downloadInfoCapability.resolve(data); + }, this); -/** - * Document initialization / loading parameters object. - * - * @typedef {Object} DocumentInitParameters - * @property {string} url - The URL of the PDF. - * @property {TypedArray|Array|string} data - Binary PDF data. Use typed arrays - * (Uint8Array) to improve the memory usage. If PDF data is BASE64-encoded, - * use atob() to convert it to a binary string first. - * @property {Object} httpHeaders - Basic authentication headers. - * @property {boolean} withCredentials - Indicates whether or not cross-site - * Access-Control requests should be made using credentials such as cookies - * or authorization headers. The default is false. - * @property {string} password - For decrypting password-protected PDFs. - * @property {TypedArray} initialData - A typed array with the first portion or - * all of the pdf data. Used by the extension since some data is already - * loaded before the switch to range requests. - * @property {number} length - The PDF file length. It's used for progress - * reports and range requests operations. - * @property {PDFDataRangeTransport} range - * @property {number} rangeChunkSize - Optional parameter to specify - * maximum number of bytes fetched per range request. The default value is - * 2^16 = 65536. - * @property {PDFWorker} worker - The worker that will be used for the loading - * and parsing of the PDF data. - */ + messageHandler.on('PDFManagerReady', function transportPage(data) { + if (this.pdfDataRangeTransport) { + this.pdfDataRangeTransport.transportReady(); + } + }, this); -/** - * @typedef {Object} PDFDocumentStats - * @property {Array} streamTypes - Used stream types in the document (an item - * is set to true if specific stream ID was used in the document). - * @property {Array} fontTypes - Used font type in the document (an item is set - * to true if specific font ID was used in the document). - */ + messageHandler.on('StartRenderPage', function transportRender(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } + var page = this.pageCache[data.pageIndex]; -/** - * This is the main entry point for loading a PDF and interacting with it. - * NOTE: If a URL is used to fetch the PDF data a standard XMLHttpRequest(XHR) - * is used, which means it must follow the same origin rules that any XHR does - * e.g. No cross domain requests without CORS. - * - * @param {string|TypedArray|DocumentInitParameters|PDFDataRangeTransport} src - * Can be a url to where a PDF is located, a typed array (Uint8Array) - * already populated with data or parameter object. - * - * @param {PDFDataRangeTransport} pdfDataRangeTransport (deprecated) It is used - * if you want to manually serve range requests for data in the PDF. - * - * @param {function} passwordCallback (deprecated) It is used to request a - * password if wrong or no password was provided. The callback receives two - * parameters: function that needs to be called with new password and reason - * (see {PasswordResponses}). - * - * @param {function} progressCallback (deprecated) It is used to be able to - * monitor the loading progress of the PDF file (necessary to implement e.g. - * a loading bar). The callback receives an {Object} with the properties: - * {number} loaded and {number} total. - * - * @return {PDFDocumentLoadingTask} - */ -PDFJS.getDocument = function getDocument(src, - pdfDataRangeTransport, - passwordCallback, - progressCallback) { - var task = new PDFDocumentLoadingTask(); + page.stats.timeEnd('Page Request'); + page._startRenderPage(data.transparency, data.intent); + }, this); - // Support of the obsolete arguments (for compatibility with API v1.0) - if (arguments.length > 1) { - deprecated('getDocument is called with pdfDataRangeTransport, ' + - 'passwordCallback or progressCallback argument'); - } - if (pdfDataRangeTransport) { - if (!(pdfDataRangeTransport instanceof PDFDataRangeTransport)) { - // Not a PDFDataRangeTransport instance, trying to add missing properties. - pdfDataRangeTransport = Object.create(pdfDataRangeTransport); - pdfDataRangeTransport.length = src.length; - pdfDataRangeTransport.initialData = src.initialData; - if (!pdfDataRangeTransport.abort) { - pdfDataRangeTransport.abort = function () {}; - } - } - src = Object.create(src); - src.range = pdfDataRangeTransport; - } - task.onPassword = passwordCallback || null; - task.onProgress = progressCallback || null; + messageHandler.on('RenderPageChunk', function transportRender(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } + var page = this.pageCache[data.pageIndex]; - var source; - if (typeof src === 'string') { - source = { url: src }; - } else if (isArrayBuffer(src)) { - source = { data: src }; - } else if (src instanceof PDFDataRangeTransport) { - source = { range: src }; - } else { - if (typeof src !== 'object') { - error('Invalid parameter in getDocument, need either Uint8Array, ' + - 'string or a parameter object'); - } - if (!src.url && !src.data && !src.range) { - error('Invalid parameter object: need either .data, .range or .url'); - } + page._renderPageChunk(data.operatorList, data.intent); + }, this); - source = src; - } + messageHandler.on('commonobj', function transportObj(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } - var params = {}; - var rangeTransport = null; - var worker = null; - for (var key in source) { - if (key === 'url' && typeof window !== 'undefined') { - // The full path is required in the 'url' field. - params[key] = combineUrl(window.location.href, source[key]); - continue; - } else if (key === 'range') { - rangeTransport = source[key]; - continue; - } else if (key === 'worker') { - worker = source[key]; - continue; - } else if (key === 'data' && !(source[key] instanceof Uint8Array)) { - // Converting string or array-like data to Uint8Array. - var pdfBytes = source[key]; - if (typeof pdfBytes === 'string') { - params[key] = stringToBytes(pdfBytes); - } else if (typeof pdfBytes === 'object' && pdfBytes !== null && - !isNaN(pdfBytes.length)) { - params[key] = new Uint8Array(pdfBytes); - } else if (isArrayBuffer(pdfBytes)) { - params[key] = new Uint8Array(pdfBytes); - } else { - error('Invalid PDF binary data: either typed array, string or ' + - 'array-like object is expected in the data property.'); - } - continue; - } - params[key] = source[key]; - } + var id = data[0]; + var type = data[1]; + if (this.commonObjs.hasData(id)) { + return; + } - params.rangeChunkSize = params.rangeChunkSize || DEFAULT_RANGE_CHUNK_SIZE; + switch (type) { + case 'Font': + var exportedData = data[2]; - if (!worker) { - // Worker was not provided -- creating and owning our own. - worker = new PDFWorker(); - task._worker = worker; - } - var docId = task.docId; - worker.promise.then(function () { - if (task.destroyed) { - throw new Error('Loading aborted'); - } - return _fetchDocument(worker, params, rangeTransport, docId).then( - function (workerId) { - if (task.destroyed) { - throw new Error('Loading aborted'); - } - var messageHandler = new MessageHandler(docId, workerId, worker.port); - var transport = new WorkerTransport(messageHandler, task, rangeTransport); - task._transport = transport; - messageHandler.send('Ready', null); - }); - }).catch(task._capability.reject); + var font; + if ('error' in exportedData) { + var error = exportedData.error; + warn('Error during font loading: ' + error); + this.commonObjs.resolve(id, error); + break; + } else { + font = new FontFaceObject(exportedData); + } - return task; -}; + this.fontLoader.bind( + [font], + function fontReady(fontObjs) { + this.commonObjs.resolve(id, font); + }.bind(this) + ); + break; + case 'FontPath': + this.commonObjs.resolve(id, data[2]); + break; + default: + error('Got unknown common object type ' + type); + } + }, this); -/** - * Starts fetching of specified PDF document/data. - * @param {PDFWorker} worker - * @param {Object} source - * @param {PDFDataRangeTransport} pdfDataRangeTransport - * @param {string} docId Unique document id, used as MessageHandler id. - * @returns {Promise} The promise, which is resolved when worker id of - * MessageHandler is known. - * @private - */ -function _fetchDocument(worker, source, pdfDataRangeTransport, docId) { - if (worker.destroyed) { - return Promise.reject(new Error('Worker was destroyed')); - } + messageHandler.on('obj', function transportObj(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } - source.disableAutoFetch = PDFJS.disableAutoFetch; - source.disableStream = PDFJS.disableStream; - source.chunkedViewerLoading = !!pdfDataRangeTransport; - if (pdfDataRangeTransport) { - source.length = pdfDataRangeTransport.length; - source.initialData = pdfDataRangeTransport.initialData; - } - return worker.messageHandler.sendWithPromise('GetDocRequest', { - docId: docId, - source: source, - disableRange: PDFJS.disableRange, - maxImageSize: PDFJS.maxImageSize, - cMapUrl: PDFJS.cMapUrl, - cMapPacked: PDFJS.cMapPacked, - disableFontFace: PDFJS.disableFontFace, - disableCreateObjectURL: PDFJS.disableCreateObjectURL, - verbosity: PDFJS.verbosity - }).then(function (workerId) { - if (worker.destroyed) { - throw new Error('Worker was destroyed'); - } - return workerId; - }); -} + var id = data[0]; + var pageIndex = data[1]; + var type = data[2]; + var pageProxy = this.pageCache[pageIndex]; + var imageData; + if (pageProxy.objs.hasData(id)) { + return; + } -/** - * PDF document loading operation. - * @class - * @alias PDFDocumentLoadingTask - */ -var PDFDocumentLoadingTask = (function PDFDocumentLoadingTaskClosure() { - var nextDocumentId = 0; + switch (type) { + case 'JpegStream': + imageData = data[3]; + loadJpegStream(id, imageData, pageProxy.objs); + break; + case 'Image': + imageData = data[3]; + pageProxy.objs.resolve(id, imageData); - /** @constructs PDFDocumentLoadingTask */ - function PDFDocumentLoadingTask() { - this._capability = createPromiseCapability(); - this._transport = null; - this._worker = null; + // heuristics that will allow not to store large data + var MAX_IMAGE_SIZE_TO_STORE = 8000000; + if (imageData && 'data' in imageData && + imageData.data.length > MAX_IMAGE_SIZE_TO_STORE) { + pageProxy.cleanupAfterRender = true; + } + break; + default: + error('Got unknown object type ' + type); + } + }, this); - /** - * Unique document loading task id -- used in MessageHandlers. - * @type {string} - */ - this.docId = 'd' + (nextDocumentId++); + messageHandler.on('DocProgress', function transportDocProgress(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } - /** - * Shows if loading task is destroyed. - * @type {boolean} - */ - this.destroyed = false; + var loadingTask = this.loadingTask; + if (loadingTask.onProgress) { + loadingTask.onProgress({ + loaded: data.loaded, + total: data.total + }); + } + }, this); - /** - * Callback to request a password if wrong or no password was provided. - * The callback receives two parameters: function that needs to be called - * with new password and reason (see {PasswordResponses}). - */ - this.onPassword = null; + messageHandler.on('PageError', function transportError(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } - /** - * Callback to be able to monitor the loading progress of the PDF file - * (necessary to implement e.g. a loading bar). The callback receives - * an {Object} with the properties: {number} loaded and {number} total. - */ - this.onProgress = null; + var page = this.pageCache[data.pageNum - 1]; + var intentState = page.intentStates[data.intent]; + if (intentState.displayReadyCapability) { + intentState.displayReadyCapability.reject(data.error); + } else { + error(data.error); + } + }, this); - /** - * Callback to when unsupported feature is used. The callback receives - * an {PDFJS.UNSUPPORTED_FEATURES} argument. - */ - this.onUnsupportedFeature = null; - } + messageHandler.on('UnsupportedFeature', + function transportUnsupportedFeature(data) { + if (this.destroyed) { + return; // Ignore any pending requests if the worker was terminated. + } + var featureId = data.featureId; + var loadingTask = this.loadingTask; + if (loadingTask.onUnsupportedFeature) { + loadingTask.onUnsupportedFeature(featureId); + } + PDFJS.UnsupportedManager.notify(featureId); + }, this); - PDFDocumentLoadingTask.prototype = - /** @lends PDFDocumentLoadingTask.prototype */ { - /** - * @return {Promise} - */ - get promise() { - return this._capability.promise; - }, + messageHandler.on('JpegDecode', function(data) { + if (this.destroyed) { + return Promise.reject('Worker was terminated'); + } - /** - * Aborts all network requests and destroys worker. - * @return {Promise} A promise that is resolved after destruction activity - * is completed. - */ - destroy: function () { - this.destroyed = true; + var imageUrl = data[0]; + var components = data[1]; + if (components !== 3 && components !== 1) { + return Promise.reject( + new Error('Only 3 components or 1 component can be returned')); + } - var transportDestroyed = !this._transport ? Promise.resolve() : - this._transport.destroy(); - return transportDestroyed.then(function () { - this._transport = null; - if (this._worker) { - this._worker.destroy(); - this._worker = null; + return new Promise(function (resolve, reject) { + var img = new Image(); + img.onload = function () { + var width = img.width; + var height = img.height; + var size = width * height; + var rgbaLength = size * 4; + var buf = new Uint8Array(size * components); + var tmpCanvas = createScratchCanvas(width, height); + var tmpCtx = tmpCanvas.getContext('2d'); + tmpCtx.drawImage(img, 0, 0); + var data = tmpCtx.getImageData(0, 0, width, height).data; + var i, j; + + if (components === 3) { + for (i = 0, j = 0; i < rgbaLength; i += 4, j += 3) { + buf[j] = data[i]; + buf[j + 1] = data[i + 1]; + buf[j + 2] = data[i + 2]; + } + } else if (components === 1) { + for (i = 0, j = 0; i < rgbaLength; i += 4, j++) { + buf[j] = data[i]; + } + } + resolve({ data: buf, width: width, height: height}); + }; + img.onerror = function () { + reject(new Error('JpegDecode failed to load image')); + }; + img.src = imageUrl; + }); + }, this); + }, + + getData: function WorkerTransport_getData() { + return this.messageHandler.sendWithPromise('GetData', null); + }, + + getPage: function WorkerTransport_getPage(pageNumber, capability) { + if (pageNumber <= 0 || pageNumber > this.numPages || + (pageNumber|0) !== pageNumber) { + return Promise.reject(new Error('Invalid page request')); + } + + var pageIndex = pageNumber - 1; + if (pageIndex in this.pagePromises) { + return this.pagePromises[pageIndex]; + } + var promise = this.messageHandler.sendWithPromise('GetPage', { + pageIndex: pageIndex + }).then(function (pageInfo) { + if (this.destroyed) { + throw new Error('Transport destroyed'); } + var page = new PDFPageProxy(pageIndex, pageInfo, this); + this.pageCache[pageIndex] = page; + return page; }.bind(this)); + this.pagePromises[pageIndex] = promise; + return promise; }, - /** - * Registers callbacks to indicate the document loading completion. - * - * @param {function} onFulfilled The callback for the loading completion. - * @param {function} onRejected The callback for the loading failure. - * @return {Promise} A promise that is resolved after the onFulfilled or - * onRejected callback. - */ - then: function PDFDocumentLoadingTask_then(onFulfilled, onRejected) { - return this.promise.then.apply(this.promise, arguments); - } - }; - - return PDFDocumentLoadingTask; -})(); + getPageIndex: function WorkerTransport_getPageIndexByRef(ref) { + return this.messageHandler.sendWithPromise('GetPageIndex', { ref: ref }); + }, -/** - * Abstract class to support range requests file loading. - * @class - * @alias PDFJS.PDFDataRangeTransport - * @param {number} length - * @param {Uint8Array} initialData - */ -var PDFDataRangeTransport = (function pdfDataRangeTransportClosure() { - function PDFDataRangeTransport(length, initialData) { - this.length = length; - this.initialData = initialData; + getAnnotations: function WorkerTransport_getAnnotations(pageIndex, intent) { + return this.messageHandler.sendWithPromise('GetAnnotations', { + pageIndex: pageIndex, + intent: intent, + }); + }, - this._rangeListeners = []; - this._progressListeners = []; - this._progressiveReadListeners = []; - this._readyCapability = createPromiseCapability(); - } - PDFDataRangeTransport.prototype = - /** @lends PDFDataRangeTransport.prototype */ { - addRangeListener: - function PDFDataRangeTransport_addRangeListener(listener) { - this._rangeListeners.push(listener); + getDestinations: function WorkerTransport_getDestinations() { + return this.messageHandler.sendWithPromise('GetDestinations', null); }, - addProgressListener: - function PDFDataRangeTransport_addProgressListener(listener) { - this._progressListeners.push(listener); + getDestination: function WorkerTransport_getDestination(id) { + return this.messageHandler.sendWithPromise('GetDestination', { id: id }); }, - addProgressiveReadListener: - function PDFDataRangeTransport_addProgressiveReadListener(listener) { - this._progressiveReadListeners.push(listener); + getPageLabels: function WorkerTransport_getPageLabels() { + return this.messageHandler.sendWithPromise('GetPageLabels', null); }, - onDataRange: function PDFDataRangeTransport_onDataRange(begin, chunk) { - var listeners = this._rangeListeners; - for (var i = 0, n = listeners.length; i < n; ++i) { - listeners[i](begin, chunk); - } + getAttachments: function WorkerTransport_getAttachments() { + return this.messageHandler.sendWithPromise('GetAttachments', null); }, - onDataProgress: function PDFDataRangeTransport_onDataProgress(loaded) { - this._readyCapability.promise.then(function () { - var listeners = this._progressListeners; - for (var i = 0, n = listeners.length; i < n; ++i) { - listeners[i](loaded); - } - }.bind(this)); + getJavaScript: function WorkerTransport_getJavaScript() { + return this.messageHandler.sendWithPromise('GetJavaScript', null); }, - onDataProgressiveRead: - function PDFDataRangeTransport_onDataProgress(chunk) { - this._readyCapability.promise.then(function () { - var listeners = this._progressiveReadListeners; - for (var i = 0, n = listeners.length; i < n; ++i) { - listeners[i](chunk); - } - }.bind(this)); + getOutline: function WorkerTransport_getOutline() { + return this.messageHandler.sendWithPromise('GetOutline', null); }, - transportReady: function PDFDataRangeTransport_transportReady() { - this._readyCapability.resolve(); + getMetadata: function WorkerTransport_getMetadata() { + return this.messageHandler.sendWithPromise('GetMetadata', null). + then(function transportMetadata(results) { + return { + info: results[0], + metadata: (results[1] ? new Metadata(results[1]) : null) + }; + }); }, - requestDataRange: - function PDFDataRangeTransport_requestDataRange(begin, end) { - throw new Error('Abstract method PDFDataRangeTransport.requestDataRange'); + getStats: function WorkerTransport_getStats() { + return this.messageHandler.sendWithPromise('GetStats', null); }, - abort: function PDFDataRangeTransport_abort() { + startCleanup: function WorkerTransport_startCleanup() { + this.messageHandler.sendWithPromise('Cleanup', null). + then(function endCleanup() { + for (var i = 0, ii = this.pageCache.length; i < ii; i++) { + var page = this.pageCache[i]; + if (page) { + page.cleanup(); + } + } + this.commonObjs.clear(); + this.fontLoader.clear(); + }.bind(this)); } }; - return PDFDataRangeTransport; -})(); + return WorkerTransport; -PDFJS.PDFDataRangeTransport = PDFDataRangeTransport; +})(); /** - * Proxy to a PDFDocument in the worker thread. Also, contains commonly used - * properties that can be read synchronously. - * @class - * @alias PDFDocumentProxy + * A PDF document and page is built of many objects. E.g. there are objects + * for fonts, images, rendering code and such. These objects might get processed + * inside of a worker. The `PDFObjects` implements some basic functions to + * manage these objects. + * @ignore */ -var PDFDocumentProxy = (function PDFDocumentProxyClosure() { - function PDFDocumentProxy(pdfInfo, transport, loadingTask) { - this.pdfInfo = pdfInfo; - this.transport = transport; - this.loadingTask = loadingTask; +var PDFObjects = (function PDFObjectsClosure() { + function PDFObjects() { + this.objs = Object.create(null); } - PDFDocumentProxy.prototype = /** @lends PDFDocumentProxy.prototype */ { - /** - * @return {number} Total number of pages the PDF contains. - */ - get numPages() { - return this.pdfInfo.numPages; - }, - /** - * @return {string} A unique ID to identify a PDF. Not guaranteed to be - * unique. - */ - get fingerprint() { - return this.pdfInfo.fingerprint; - }, - /** - * @param {number} pageNumber The page number to get. The first page is 1. - * @return {Promise} A promise that is resolved with a {@link PDFPageProxy} - * object. - */ - getPage: function PDFDocumentProxy_getPage(pageNumber) { - return this.transport.getPage(pageNumber); - }, + + PDFObjects.prototype = { /** - * @param {{num: number, gen: number}} ref The page reference. Must have - * the 'num' and 'gen' properties. - * @return {Promise} A promise that is resolved with the page index that is - * associated with the reference. + * Internal function. + * Ensures there is an object defined for `objId`. */ - getPageIndex: function PDFDocumentProxy_getPageIndex(ref) { - return this.transport.getPageIndex(ref); + ensureObj: function PDFObjects_ensureObj(objId) { + if (this.objs[objId]) { + return this.objs[objId]; + } + + var obj = { + capability: createPromiseCapability(), + data: null, + resolved: false + }; + this.objs[objId] = obj; + + return obj; }, + /** - * @return {Promise} A promise that is resolved with a lookup table for - * mapping named destinations to reference numbers. + * If called *without* callback, this returns the data of `objId` but the + * object needs to be resolved. If it isn't, this function throws. * - * This can be slow for large documents: use getDestination instead - */ - getDestinations: function PDFDocumentProxy_getDestinations() { - return this.transport.getDestinations(); - }, - /** - * @param {string} id The named destination to get. - * @return {Promise} A promise that is resolved with all information - * of the given named destination. - */ - getDestination: function PDFDocumentProxy_getDestination(id) { - return this.transport.getDestination(id); - }, - /** - * @return {Promise} A promise that is resolved with: - * an Array containing the pageLabels that correspond to the pageIndexes, - * or `null` when no pageLabels are present in the PDF file. - */ - getPageLabels: function PDFDocumentProxy_getPageLabels() { - return this.transport.getPageLabels(); - }, - /** - * @return {Promise} A promise that is resolved with a lookup table for - * mapping named attachments to their content. + * If called *with* a callback, the callback is called with the data of the + * object once the object is resolved. That means, if you call this + * function and the object is already resolved, the callback gets called + * right away. */ - getAttachments: function PDFDocumentProxy_getAttachments() { - return this.transport.getAttachments(); + get: function PDFObjects_get(objId, callback) { + // If there is a callback, then the get can be async and the object is + // not required to be resolved right now + if (callback) { + this.ensureObj(objId).capability.promise.then(callback); + return null; + } + + // If there isn't a callback, the user expects to get the resolved data + // directly. + var obj = this.objs[objId]; + + // If there isn't an object yet or the object isn't resolved, then the + // data isn't ready yet! + if (!obj || !obj.resolved) { + error('Requesting object that isn\'t resolved yet ' + objId); + } + + return obj.data; }, + /** - * @return {Promise} A promise that is resolved with an array of all the - * JavaScript strings in the name tree. + * Resolves the object `objId` with optional `data`. */ - getJavaScript: function PDFDocumentProxy_getJavaScript() { - return this.transport.getJavaScript(); + resolve: function PDFObjects_resolve(objId, data) { + var obj = this.ensureObj(objId); + + obj.resolved = true; + obj.data = data; + obj.capability.resolve(data); }, - /** - * @return {Promise} A promise that is resolved with an {Array} that is a - * tree outline (if it has one) of the PDF. The tree is in the format of: - * [ - * { - * title: string, - * bold: boolean, - * italic: boolean, - * color: rgb array, - * dest: dest obj, - * url: string, - * items: array of more items like this - * }, - * ... - * ]. - */ - getOutline: function PDFDocumentProxy_getOutline() { - return this.transport.getOutline(); + + isResolved: function PDFObjects_isResolved(objId) { + var objs = this.objs; + + if (!objs[objId]) { + return false; + } else { + return objs[objId].resolved; + } }, - /** - * @return {Promise} A promise that is resolved with an {Object} that has - * info and metadata properties. Info is an {Object} filled with anything - * available in the information dictionary and similarly metadata is a - * {Metadata} object with information from the metadata section of the PDF. - */ - getMetadata: function PDFDocumentProxy_getMetadata() { - return this.transport.getMetadata(); + + hasData: function PDFObjects_hasData(objId) { + return this.isResolved(objId); }, + /** - * @return {Promise} A promise that is resolved with a TypedArray that has - * the raw data from the PDF. + * Returns the data of `objId` if object exists, null otherwise. */ - getData: function PDFDocumentProxy_getData() { - return this.transport.getData(); + getData: function PDFObjects_getData(objId) { + var objs = this.objs; + if (!objs[objId] || !objs[objId].resolved) { + return null; + } else { + return objs[objId].data; + } }, + + clear: function PDFObjects_clear() { + this.objs = Object.create(null); + } + }; + return PDFObjects; +})(); + +/** + * Allows controlling of the rendering tasks. + * @class + * @alias RenderTask + */ +var RenderTask = (function RenderTaskClosure() { + function RenderTask(internalRenderTask) { + this._internalRenderTask = internalRenderTask; + /** - * @return {Promise} A promise that is resolved when the document's data - * is loaded. It is resolved with an {Object} that contains the length - * property that indicates size of the PDF data in bytes. + * Callback for incremental rendering -- a function that will be called + * each time the rendering is paused. To continue rendering call the + * function that is the first argument to the callback. + * @type {function} */ - getDownloadInfo: function PDFDocumentProxy_getDownloadInfo() { - return this.transport.downloadInfoCapability.promise; - }, + this.onContinue = null; + } + + RenderTask.prototype = /** @lends RenderTask.prototype */ { /** - * @return {Promise} A promise this is resolved with current stats about - * document structures (see {@link PDFDocumentStats}). + * Promise for rendering task completion. + * @return {Promise} */ - getStats: function PDFDocumentProxy_getStats() { - return this.transport.getStats(); + get promise() { + return this._internalRenderTask.capability.promise; }, + /** - * Cleans up resources allocated by the document, e.g. created @font-face. + * Cancels the rendering task. If the task is currently rendering it will + * not be cancelled until graphics pauses with a timeout. The promise that + * this object extends will resolved when cancelled. */ - cleanup: function PDFDocumentProxy_cleanup() { - this.transport.startCleanup(); + cancel: function RenderTask_cancel() { + this._internalRenderTask.cancel(); }, + /** - * Destroys current document instance and terminates worker. + * Registers callbacks to indicate the rendering task completion. + * + * @param {function} onFulfilled The callback for the rendering completion. + * @param {function} onRejected The callback for the rendering failure. + * @return {Promise} A promise that is resolved after the onFulfilled or + * onRejected callback. */ - destroy: function PDFDocumentProxy_destroy() { - return this.loadingTask.destroy(); + then: function RenderTask_then(onFulfilled, onRejected) { + return this.promise.then.apply(this.promise, arguments); } }; - return PDFDocumentProxy; + + return RenderTask; })(); /** - * Page getTextContent parameters. - * - * @typedef {Object} getTextContentParameters - * @param {boolean} normalizeWhitespace - replaces all occurrences of - * whitespace with standard spaces (0x20). The default value is `false`. + * For internal use only. + * @ignore */ +var InternalRenderTask = (function InternalRenderTaskClosure() { -/** - * Page text content. - * - * @typedef {Object} TextContent - * @property {array} items - array of {@link TextItem} - * @property {Object} styles - {@link TextStyles} objects, indexed by font - * name. - */ + function InternalRenderTask(callback, params, objs, commonObjs, operatorList, + pageNumber) { + this.callback = callback; + this.params = params; + this.objs = objs; + this.commonObjs = commonObjs; + this.operatorListIdx = null; + this.operatorList = operatorList; + this.pageNumber = pageNumber; + this.running = false; + this.graphicsReadyCallback = null; + this.graphicsReady = false; + this.useRequestAnimationFrame = false; + this.cancelled = false; + this.capability = createPromiseCapability(); + this.task = new RenderTask(this); + // caching this-bound methods + this._continueBound = this._continue.bind(this); + this._scheduleNextBound = this._scheduleNext.bind(this); + this._nextBound = this._next.bind(this); + } -/** - * Page text content part. - * - * @typedef {Object} TextItem - * @property {string} str - text content. - * @property {string} dir - text direction: 'ttb', 'ltr' or 'rtl'. - * @property {array} transform - transformation matrix. - * @property {number} width - width in device space. - * @property {number} height - height in device space. - * @property {string} fontName - font name used by pdf.js for converted font. - */ + InternalRenderTask.prototype = { -/** - * Text style. - * - * @typedef {Object} TextStyle - * @property {number} ascent - font ascent. - * @property {number} descent - font descent. - * @property {boolean} vertical - text is in vertical mode. - * @property {string} fontFamily - possible font family - */ + initalizeGraphics: + function InternalRenderTask_initalizeGraphics(transparency) { -/** - * Page annotation parameters. - * - * @typedef {Object} GetAnnotationsParameters - * @param {string} intent - Determines the annotations that will be fetched, - * can be either 'display' (viewable annotations) or 'print' - * (printable annotations). - * If the parameter is omitted, all annotations are fetched. - */ + if (this.cancelled) { + return; + } + if (PDFJS.pdfBug && 'StepperManager' in globalScope && + globalScope.StepperManager.enabled) { + this.stepper = globalScope.StepperManager.create(this.pageNumber - 1); + this.stepper.init(this.operatorList); + this.stepper.nextBreakPoint = this.stepper.getNextBreakPoint(); + } -/** - * Page render parameters. - * - * @typedef {Object} RenderParameters - * @property {Object} canvasContext - A 2D context of a DOM Canvas object. - * @property {PDFJS.PageViewport} viewport - Rendering viewport obtained by - * calling of PDFPage.getViewport method. - * @property {string} intent - Rendering intent, can be 'display' or 'print' - * (default value is 'display'). - * @property {Array} transform - (optional) Additional transform, applied - * just before viewport transform. - * @property {Object} imageLayer - (optional) An object that has beginLayout, - * endLayout and appendImage functions. - * @property {function} continueCallback - (deprecated) A function that will be - * called each time the rendering is paused. To continue - * rendering call the function that is the first argument - * to the callback. - */ - -/** - * PDF page operator list. - * - * @typedef {Object} PDFOperatorList - * @property {Array} fnArray - Array containing the operator functions. - * @property {Array} argsArray - Array containing the arguments of the - * functions. - */ + var params = this.params; + this.gfx = new CanvasGraphics(params.canvasContext, this.commonObjs, + this.objs, params.imageLayer); -/** - * Proxy to a PDFPage in the worker thread. - * @class - * @alias PDFPageProxy - */ -var PDFPageProxy = (function PDFPageProxyClosure() { - function PDFPageProxy(pageIndex, pageInfo, transport) { - this.pageIndex = pageIndex; - this.pageInfo = pageInfo; - this.transport = transport; - this.stats = new StatTimer(); - this.stats.enabled = !!globalScope.PDFJS.enableStats; - this.commonObjs = transport.commonObjs; - this.objs = new PDFObjects(); - this.cleanupAfterRender = false; - this.pendingCleanup = false; - this.intentStates = Object.create(null); - this.destroyed = false; - } - PDFPageProxy.prototype = /** @lends PDFPageProxy.prototype */ { - /** - * @return {number} Page number of the page. First page is 1. - */ - get pageNumber() { - return this.pageIndex + 1; - }, - /** - * @return {number} The number of degrees the page is rotated clockwise. - */ - get rotate() { - return this.pageInfo.rotate; - }, - /** - * @return {Object} The reference that points to this page. It has 'num' and - * 'gen' properties. - */ - get ref() { - return this.pageInfo.ref; - }, - /** - * @return {Array} An array of the visible portion of the PDF page in the - * user space units - [x1, y1, x2, y2]. - */ - get view() { - return this.pageInfo.view; - }, - /** - * @param {number} scale The desired scale of the viewport. - * @param {number} rotate Degrees to rotate the viewport. If omitted this - * defaults to the page rotation. - * @return {PDFJS.PageViewport} Contains 'width' and 'height' properties - * along with transforms required for rendering. - */ - getViewport: function PDFPageProxy_getViewport(scale, rotate) { - if (arguments.length < 2) { - rotate = this.rotate; + this.gfx.beginDrawing(params.transform, params.viewport, transparency); + this.operatorListIdx = 0; + this.graphicsReady = true; + if (this.graphicsReadyCallback) { + this.graphicsReadyCallback(); } - return new PDFJS.PageViewport(this.view, scale, rotate, 0, 0); }, - /** - * @param {GetAnnotationsParameters} params - Annotation parameters. - * @return {Promise} A promise that is resolved with an {Array} of the - * annotation objects. - */ - getAnnotations: function PDFPageProxy_getAnnotations(params) { - var intent = (params && params.intent) || null; - if (!this.annotationsPromise || this.annotationsIntent !== intent) { - this.annotationsPromise = this.transport.getAnnotations(this.pageIndex, - intent); - this.annotationsIntent = intent; - } - return this.annotationsPromise; + cancel: function InternalRenderTask_cancel() { + this.running = false; + this.cancelled = true; + this.callback('cancelled'); }, - /** - * Begins the process of rendering a page to the desired context. - * @param {RenderParameters} params Page render parameters. - * @return {RenderTask} An object that contains the promise, which - * is resolved when the page finishes rendering. - */ - render: function PDFPageProxy_render(params) { - var stats = this.stats; - stats.time('Overall'); - - // If there was a pending destroy cancel it so no cleanup happens during - // this call to render. - this.pendingCleanup = false; - - var renderingIntent = (params.intent === 'print' ? 'print' : 'display'); - - if (!this.intentStates[renderingIntent]) { - this.intentStates[renderingIntent] = Object.create(null); - } - var intentState = this.intentStates[renderingIntent]; - - // If there's no displayReadyCapability yet, then the operatorList - // was never requested before. Make the request and create the promise. - if (!intentState.displayReadyCapability) { - intentState.receivingOperatorList = true; - intentState.displayReadyCapability = createPromiseCapability(); - intentState.operatorList = { - fnArray: [], - argsArray: [], - lastChunk: false - }; - - this.stats.time('Page Request'); - this.transport.messageHandler.send('RenderPageRequest', { - pageIndex: this.pageNumber - 1, - intent: renderingIntent - }); - } - var internalRenderTask = new InternalRenderTask(complete, params, - this.objs, - this.commonObjs, - intentState.operatorList, - this.pageNumber); - internalRenderTask.useRequestAnimationFrame = renderingIntent !== 'print'; - if (!intentState.renderTasks) { - intentState.renderTasks = []; + operatorListChanged: function InternalRenderTask_operatorListChanged() { + if (!this.graphicsReady) { + if (!this.graphicsReadyCallback) { + this.graphicsReadyCallback = this._continueBound; + } + return; } - intentState.renderTasks.push(internalRenderTask); - var renderTask = internalRenderTask.task; - // Obsolete parameter support - if (params.continueCallback) { - deprecated('render is used with continueCallback parameter'); - renderTask.onContinue = params.continueCallback; + if (this.stepper) { + this.stepper.updateOperatorList(this.operatorList); } - var self = this; - intentState.displayReadyCapability.promise.then( - function pageDisplayReadyPromise(transparency) { - if (self.pendingCleanup) { - complete(); - return; - } - stats.time('Rendering'); - internalRenderTask.initalizeGraphics(transparency); - internalRenderTask.operatorListChanged(); - }, - function pageDisplayReadPromiseError(reason) { - complete(reason); - } - ); - - function complete(error) { - var i = intentState.renderTasks.indexOf(internalRenderTask); - if (i >= 0) { - intentState.renderTasks.splice(i, 1); - } - - if (self.cleanupAfterRender) { - self.pendingCleanup = true; - } - self._tryCleanup(); - - if (error) { - internalRenderTask.capability.reject(error); - } else { - internalRenderTask.capability.resolve(); - } - stats.timeEnd('Rendering'); - stats.timeEnd('Overall'); + if (this.running) { + return; } - - return renderTask; + this._continue(); }, - /** - * @return {Promise} A promise resolved with an {@link PDFOperatorList} - * object that represents page's operator list. - */ - getOperatorList: function PDFPageProxy_getOperatorList() { - function operatorListChanged() { - if (intentState.operatorList.lastChunk) { - intentState.opListReadCapability.resolve(intentState.operatorList); - - var i = intentState.renderTasks.indexOf(opListTask); - if (i >= 0) { - intentState.renderTasks.splice(i, 1); - } - } - } - - var renderingIntent = 'oplist'; - if (!this.intentStates[renderingIntent]) { - this.intentStates[renderingIntent] = Object.create(null); + _continue: function InternalRenderTask__continue() { + this.running = true; + if (this.cancelled) { + return; } - var intentState = this.intentStates[renderingIntent]; - var opListTask; - - if (!intentState.opListReadCapability) { - opListTask = {}; - opListTask.operatorListChanged = operatorListChanged; - intentState.receivingOperatorList = true; - intentState.opListReadCapability = createPromiseCapability(); - intentState.renderTasks = []; - intentState.renderTasks.push(opListTask); - intentState.operatorList = { - fnArray: [], - argsArray: [], - lastChunk: false - }; - - this.transport.messageHandler.send('RenderPageRequest', { - pageIndex: this.pageIndex, - intent: renderingIntent - }); + if (this.task.onContinue) { + this.task.onContinue.call(this.task, this._scheduleNextBound); + } else { + this._scheduleNext(); } - return intentState.opListReadCapability.promise; - }, - - /** - * @param {getTextContentParameters} params - getTextContent parameters. - * @return {Promise} That is resolved a {@link TextContent} - * object that represent the page text content. - */ - getTextContent: function PDFPageProxy_getTextContent(params) { - var normalizeWhitespace = (params && params.normalizeWhitespace) || false; - - return this.transport.messageHandler.sendWithPromise('GetTextContent', { - pageIndex: this.pageNumber - 1, - normalizeWhitespace: normalizeWhitespace, - }); - }, - - /** - * Destroys page object. - */ - _destroy: function PDFPageProxy_destroy() { - this.destroyed = true; - this.transport.pageCache[this.pageIndex] = null; - - var waitOn = []; - Object.keys(this.intentStates).forEach(function(intent) { - if (intent === 'oplist') { - // Avoid errors below, since the renderTasks are just stubs. - return; - } - var intentState = this.intentStates[intent]; - intentState.renderTasks.forEach(function(renderTask) { - var renderCompleted = renderTask.capability.promise. - catch(function () {}); // ignoring failures - waitOn.push(renderCompleted); - renderTask.cancel(); - }); - }, this); - this.objs.clear(); - this.annotationsPromise = null; - this.pendingCleanup = false; - return Promise.all(waitOn); }, - /** - * Cleans up resources allocated by the page. (deprecated) - */ - destroy: function() { - deprecated('page destroy method, use cleanup() instead'); - this.cleanup(); + _scheduleNext: function InternalRenderTask__scheduleNext() { + if (this.useRequestAnimationFrame) { + window.requestAnimationFrame(this._nextBound); + } else { + Promise.resolve(undefined).then(this._nextBound); + } }, - /** - * Cleans up resources allocated by the page. - */ - cleanup: function PDFPageProxy_cleanup() { - this.pendingCleanup = true; - this._tryCleanup(); - }, - /** - * For internal use only. Attempts to clean up if rendering is in a state - * where that's possible. - * @ignore - */ - _tryCleanup: function PDFPageProxy_tryCleanup() { - if (!this.pendingCleanup || - Object.keys(this.intentStates).some(function(intent) { - var intentState = this.intentStates[intent]; - return (intentState.renderTasks.length !== 0 || - intentState.receivingOperatorList); - }, this)) { + _next: function InternalRenderTask__next() { + if (this.cancelled) { return; } - - Object.keys(this.intentStates).forEach(function(intent) { - delete this.intentStates[intent]; - }, this); - this.objs.clear(); - this.annotationsPromise = null; - this.pendingCleanup = false; - }, - /** - * For internal use only. - * @ignore - */ - _startRenderPage: function PDFPageProxy_startRenderPage(transparency, - intent) { - var intentState = this.intentStates[intent]; - // TODO Refactor RenderPageRequest to separate rendering - // and operator list logic - if (intentState.displayReadyCapability) { - intentState.displayReadyCapability.resolve(transparency); - } - }, - /** - * For internal use only. - * @ignore - */ - _renderPageChunk: function PDFPageProxy_renderPageChunk(operatorListChunk, - intent) { - var intentState = this.intentStates[intent]; - var i, ii; - // Add the new chunk to the current operator list. - for (i = 0, ii = operatorListChunk.length; i < ii; i++) { - intentState.operatorList.fnArray.push(operatorListChunk.fnArray[i]); - intentState.operatorList.argsArray.push( - operatorListChunk.argsArray[i]); + this.operatorListIdx = this.gfx.executeOperatorList(this.operatorList, + this.operatorListIdx, + this._continueBound, + this.stepper); + if (this.operatorListIdx === this.operatorList.argsArray.length) { + this.running = false; + if (this.operatorList.lastChunk) { + this.gfx.endDrawing(); + this.callback(); + } } - intentState.operatorList.lastChunk = operatorListChunk.lastChunk; + } - // Notify all the rendering tasks there are more operators to be consumed. - for (i = 0; i < intentState.renderTasks.length; i++) { - intentState.renderTasks[i].operatorListChanged(); - } + }; - if (operatorListChunk.lastChunk) { - intentState.receivingOperatorList = false; - this._tryCleanup(); + return InternalRenderTask; +})(); + +/** + * (Deprecated) Global observer of unsupported feature usages. Use + * onUnsupportedFeature callback of the {PDFDocumentLoadingTask} instance. + */ +PDFJS.UnsupportedManager = (function UnsupportedManagerClosure() { + var listeners = []; + return { + listen: function (cb) { + deprecated('Global UnsupportedManager.listen is used: ' + + ' use PDFDocumentLoadingTask.onUnsupportedFeature instead'); + listeners.push(cb); + }, + notify: function (featureId) { + for (var i = 0, ii = listeners.length; i < ii; i++) { + listeners[i](featureId); } } }; - return PDFPageProxy; })(); -/** - * PDF.js web worker abstraction, it controls instantiation of PDF documents and - * WorkerTransport for them. If creation of a web worker is not possible, - * a "fake" worker will be used instead. - * @class - */ -var PDFWorker = (function PDFWorkerClosure() { - var nextFakeWorkerId = 0; +exports.getDocument = PDFJS.getDocument; +exports.PDFDataRangeTransport = PDFDataRangeTransport; +exports.PDFDocumentProxy = PDFDocumentProxy; +exports.PDFPageProxy = PDFPageProxy; +})); - function getWorkerSrc() { - if (PDFJS.workerSrc) { - return PDFJS.workerSrc; - } - if (pdfjsFilePath) { - return pdfjsFilePath.replace(/\.js$/i, '.worker.js'); - } - error('No PDFJS.workerSrc specified'); - } - // Loads worker code into main thread. - function setupFakeWorkerGlobal() { - if (!PDFJS.fakeWorkerFilesLoadedCapability) { - PDFJS.fakeWorkerFilesLoadedCapability = createPromiseCapability(); - // In the developer build load worker_loader which in turn loads all the - // other files and resolves the promise. In production only the - // pdf.worker.js file is needed. - PDFJS.fakeWorkerFilesLoadedCapability.resolve(); - } - return PDFJS.fakeWorkerFilesLoadedCapability.promise; +(function (root, factory) { + { + factory((root.pdfjsCoreFonts = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreStream, root.pdfjsCoreParser, + root.pdfjsCoreCMap, root.pdfjsCoreGlyphList, root.pdfjsCoreCharsets, + root.pdfjsCoreFontRenderer, root.pdfjsCoreEncodings, + root.pdfjsCoreStandardFonts, root.pdfjsCoreUnicode); } +}(this, function (exports, sharedUtil, corePrimitives, coreStream, coreParser, + coreCMap, coreGlyphList, coreCharsets, coreFontRenderer, + coreEncodings, coreStandardFonts, coreUnicode) { - function createCDNWrapper(url) { - // We will rely on blob URL's property to specify origin. - // We want this function to fail in case if createObjectURL or Blob do not - // exist or fail for some reason -- our Worker creation will fail anyway. - var wrapper = 'importScripts(\'' + url + '\');'; - return URL.createObjectURL(new Blob([wrapper])); - } +var FONT_IDENTITY_MATRIX = sharedUtil.FONT_IDENTITY_MATRIX; +var FontType = sharedUtil.FontType; +var Util = sharedUtil.Util; +var assert = sharedUtil.assert; +var bytesToString = sharedUtil.bytesToString; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isInt = sharedUtil.isInt; +var isNum = sharedUtil.isNum; +var readUint32 = sharedUtil.readUint32; +var shadow = sharedUtil.shadow; +var stringToBytes = sharedUtil.stringToBytes; +var string32 = sharedUtil.string32; +var warn = sharedUtil.warn; +var Name = corePrimitives.Name; +var Stream = coreStream.Stream; +var Lexer = coreParser.Lexer; +var CMapFactory = coreCMap.CMapFactory; +var IdentityCMap = coreCMap.IdentityCMap; +var getGlyphsUnicode = coreGlyphList.getGlyphsUnicode; +var getDingbatsGlyphsUnicode = coreGlyphList.getDingbatsGlyphsUnicode; +var ISOAdobeCharset = coreCharsets.ISOAdobeCharset; +var ExpertCharset = coreCharsets.ExpertCharset; +var ExpertSubsetCharset = coreCharsets.ExpertSubsetCharset; +var FontRendererFactory = coreFontRenderer.FontRendererFactory; +var WinAnsiEncoding = coreEncodings.WinAnsiEncoding; +var StandardEncoding = coreEncodings.StandardEncoding; +var MacRomanEncoding = coreEncodings.MacRomanEncoding; +var SymbolSetEncoding = coreEncodings.SymbolSetEncoding; +var ZapfDingbatsEncoding = coreEncodings.ZapfDingbatsEncoding; +var ExpertEncoding = coreEncodings.ExpertEncoding; +var getEncoding = coreEncodings.getEncoding; +var getStdFontMap = coreStandardFonts.getStdFontMap; +var getNonStdFontMap = coreStandardFonts.getNonStdFontMap; +var getGlyphMapForStandardFonts = coreStandardFonts.getGlyphMapForStandardFonts; +var getSupplementalGlyphMapForArialBlack = + coreStandardFonts.getSupplementalGlyphMapForArialBlack; +var getUnicodeRangeFor = coreUnicode.getUnicodeRangeFor; +var mapSpecialUnicodeValues = coreUnicode.mapSpecialUnicodeValues; - function PDFWorker(name) { - this.name = name; - this.destroyed = false; +// Unicode Private Use Area +var PRIVATE_USE_OFFSET_START = 0xE000; +var PRIVATE_USE_OFFSET_END = 0xF8FF; +var SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = false; - this._readyCapability = createPromiseCapability(); - this._port = null; - this._webWorker = null; - this._messageHandler = null; - this._initialize(); - } +// PDF Glyph Space Units are one Thousandth of a TextSpace Unit +// except for Type 3 fonts +var PDF_GLYPH_SPACE_UNITS = 1000; - PDFWorker.prototype = /** @lends PDFWorker.prototype */ { - get promise() { - return this._readyCapability.promise; - }, +// Hinting is currently disabled due to unknown problems on windows +// in tracemonkey and various other pdfs with type1 fonts. +var HINTING_ENABLED = false; - get port() { - return this._port; - }, +// Accented charactars are not displayed properly on windows, using this flag +// to control analysis of seac charstrings. +var SEAC_ANALYSIS_ENABLED = false; - get messageHandler() { - return this._messageHandler; - }, +// Maximum subroutine call depth of type 2 chartrings. Matches OTS. +var MAX_SUBR_NESTING = 10; - _initialize: function PDFWorker_initialize() { - // If worker support isn't disabled explicit and the browser has worker - // support, create a new web worker and test if it/the browser fulfills - // all requirements to run parts of pdf.js in a web worker. - // Right now, the requirement is, that an Uint8Array is still an - // Uint8Array as it arrives on the worker. (Chrome added this with v.15.) - // Either workers are disabled, not supported or have thrown an exception. - // Thus, we fallback to a faked worker. - this._setupFakeWorker(); - }, +var FontFlags = { + FixedPitch: 1, + Serif: 2, + Symbolic: 4, + Script: 8, + Nonsymbolic: 32, + Italic: 64, + AllCap: 65536, + SmallCap: 131072, + ForceBold: 262144 +}; - _setupFakeWorker: function PDFWorker_setupFakeWorker() { - if (!globalScope.PDFJS.disableWorker) { - warn('Setting up fake worker.'); - globalScope.PDFJS.disableWorker = true; - } +var MacStandardGlyphOrdering = [ + '.notdef', '.null', 'nonmarkingreturn', 'space', 'exclam', 'quotedbl', + 'numbersign', 'dollar', 'percent', 'ampersand', 'quotesingle', 'parenleft', + 'parenright', 'asterisk', 'plus', 'comma', 'hyphen', 'period', 'slash', + 'zero', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', + 'nine', 'colon', 'semicolon', 'less', 'equal', 'greater', 'question', 'at', + 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', + 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'bracketleft', + 'backslash', 'bracketright', 'asciicircum', 'underscore', 'grave', 'a', 'b', + 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', + 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 'braceleft', 'bar', 'braceright', + 'asciitilde', 'Adieresis', 'Aring', 'Ccedilla', 'Eacute', 'Ntilde', + 'Odieresis', 'Udieresis', 'aacute', 'agrave', 'acircumflex', 'adieresis', + 'atilde', 'aring', 'ccedilla', 'eacute', 'egrave', 'ecircumflex', 'edieresis', + 'iacute', 'igrave', 'icircumflex', 'idieresis', 'ntilde', 'oacute', 'ograve', + 'ocircumflex', 'odieresis', 'otilde', 'uacute', 'ugrave', 'ucircumflex', + 'udieresis', 'dagger', 'degree', 'cent', 'sterling', 'section', 'bullet', + 'paragraph', 'germandbls', 'registered', 'copyright', 'trademark', 'acute', + 'dieresis', 'notequal', 'AE', 'Oslash', 'infinity', 'plusminus', 'lessequal', + 'greaterequal', 'yen', 'mu', 'partialdiff', 'summation', 'product', 'pi', + 'integral', 'ordfeminine', 'ordmasculine', 'Omega', 'ae', 'oslash', + 'questiondown', 'exclamdown', 'logicalnot', 'radical', 'florin', + 'approxequal', 'Delta', 'guillemotleft', 'guillemotright', 'ellipsis', + 'nonbreakingspace', 'Agrave', 'Atilde', 'Otilde', 'OE', 'oe', 'endash', + 'emdash', 'quotedblleft', 'quotedblright', 'quoteleft', 'quoteright', + 'divide', 'lozenge', 'ydieresis', 'Ydieresis', 'fraction', 'currency', + 'guilsinglleft', 'guilsinglright', 'fi', 'fl', 'daggerdbl', 'periodcentered', + 'quotesinglbase', 'quotedblbase', 'perthousand', 'Acircumflex', + 'Ecircumflex', 'Aacute', 'Edieresis', 'Egrave', 'Iacute', 'Icircumflex', + 'Idieresis', 'Igrave', 'Oacute', 'Ocircumflex', 'apple', 'Ograve', 'Uacute', + 'Ucircumflex', 'Ugrave', 'dotlessi', 'circumflex', 'tilde', 'macron', + 'breve', 'dotaccent', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', + 'Lslash', 'lslash', 'Scaron', 'scaron', 'Zcaron', 'zcaron', 'brokenbar', + 'Eth', 'eth', 'Yacute', 'yacute', 'Thorn', 'thorn', 'minus', 'multiply', + 'onesuperior', 'twosuperior', 'threesuperior', 'onehalf', 'onequarter', + 'threequarters', 'franc', 'Gbreve', 'gbreve', 'Idotaccent', 'Scedilla', + 'scedilla', 'Cacute', 'cacute', 'Ccaron', 'ccaron', 'dcroat']; - setupFakeWorkerGlobal().then(function () { - if (this.destroyed) { - this._readyCapability.reject(new Error('Worker was destroyed')); - return; - } +function adjustWidths(properties) { + if (!properties.fontMatrix) { + return; + } + if (properties.fontMatrix[0] === FONT_IDENTITY_MATRIX[0]) { + return; + } + // adjusting width to fontMatrix scale + var scale = 0.001 / properties.fontMatrix[0]; + var glyphsWidths = properties.widths; + for (var glyph in glyphsWidths) { + glyphsWidths[glyph] *= scale; + } + properties.defaultWidth *= scale; +} - // If we don't use a worker, just post/sendMessage to the main thread. - var port = { - _listeners: [], - postMessage: function (obj) { - var e = {data: obj}; - this._listeners.forEach(function (listener) { - listener.call(this, e); - }, this); - }, - addEventListener: function (name, listener) { - this._listeners.push(listener); - }, - removeEventListener: function (name, listener) { - var i = this._listeners.indexOf(listener); - this._listeners.splice(i, 1); - }, - terminate: function () {} - }; - this._port = port; +function getFontType(type, subtype) { + switch (type) { + case 'Type1': + return subtype === 'Type1C' ? FontType.TYPE1C : FontType.TYPE1; + case 'CIDFontType0': + return subtype === 'CIDFontType0C' ? FontType.CIDFONTTYPE0C : + FontType.CIDFONTTYPE0; + case 'OpenType': + return FontType.OPENTYPE; + case 'TrueType': + return FontType.TRUETYPE; + case 'CIDFontType2': + return FontType.CIDFONTTYPE2; + case 'MMType1': + return FontType.MMTYPE1; + case 'Type0': + return FontType.TYPE0; + default: + return FontType.UNKNOWN; + } +} - // All fake workers use the same port, making id unique. - var id = 'fake' + (nextFakeWorkerId++); +var Glyph = (function GlyphClosure() { + function Glyph(fontChar, unicode, accent, width, vmetric, operatorListId, + isSpace) { + this.fontChar = fontChar; + this.unicode = unicode; + this.accent = accent; + this.width = width; + this.vmetric = vmetric; + this.operatorListId = operatorListId; + this.isSpace = isSpace; + } - // If the main thread is our worker, setup the handling for the - // messages -- the main thread sends to it self. - var workerHandler = new MessageHandler(id + '_worker', id, port); - PDFJS.WorkerMessageHandler.setup(workerHandler, port); + Glyph.prototype.matchesForCache = function(fontChar, unicode, accent, width, + vmetric, operatorListId, isSpace) { + return this.fontChar === fontChar && + this.unicode === unicode && + this.accent === accent && + this.width === width && + this.vmetric === vmetric && + this.operatorListId === operatorListId && + this.isSpace === isSpace; + }; - var messageHandler = new MessageHandler(id, id + '_worker', port); - this._messageHandler = messageHandler; - this._readyCapability.resolve(); - }.bind(this)); + return Glyph; +})(); + +var ToUnicodeMap = (function ToUnicodeMapClosure() { + function ToUnicodeMap(cmap) { + // The elements of this._map can be integers or strings, depending on how + // |cmap| was created. + this._map = cmap; + } + + ToUnicodeMap.prototype = { + get length() { + return this._map.length; }, - /** - * Destroys the worker instance. - */ - destroy: function PDFWorker_destroy() { - this.destroyed = true; - if (this._webWorker) { - // We need to terminate only web worker created resource. - this._webWorker.terminate(); - this._webWorker = null; - } - this._port = null; - if (this._messageHandler) { - this._messageHandler.destroy(); - this._messageHandler = null; + forEach: function(callback) { + for (var charCode in this._map) { + callback(charCode, this._map[charCode].charCodeAt(0)); } - } - }; + }, - return PDFWorker; -})(); -PDFJS.PDFWorker = PDFWorker; + has: function(i) { + return this._map[i] !== undefined; + }, -/** - * For internal use only. - * @ignore - */ -var WorkerTransport = (function WorkerTransportClosure() { - function WorkerTransport(messageHandler, loadingTask, pdfDataRangeTransport) { - this.messageHandler = messageHandler; - this.loadingTask = loadingTask; - this.pdfDataRangeTransport = pdfDataRangeTransport; - this.commonObjs = new PDFObjects(); - this.fontLoader = new FontLoader(loadingTask.docId); + get: function(i) { + return this._map[i]; + }, - this.destroyed = false; - this.destroyCapability = null; + charCodeOf: function(v) { + return this._map.indexOf(v); + } + }; - this.pageCache = []; - this.pagePromises = []; - this.downloadInfoCapability = createPromiseCapability(); + return ToUnicodeMap; +})(); - this.setupMessageHandler(); +var IdentityToUnicodeMap = (function IdentityToUnicodeMapClosure() { + function IdentityToUnicodeMap(firstChar, lastChar) { + this.firstChar = firstChar; + this.lastChar = lastChar; } - WorkerTransport.prototype = { - destroy: function WorkerTransport_destroy() { - if (this.destroyCapability) { - return this.destroyCapability.promise; - } - this.destroyed = true; - this.destroyCapability = createPromiseCapability(); + IdentityToUnicodeMap.prototype = { + get length() { + return (this.lastChar + 1) - this.firstChar; + }, - var waitOn = []; - // We need to wait for all renderings to be completed, e.g. - // timeout/rAF can take a long time. - this.pageCache.forEach(function (page) { - if (page) { - waitOn.push(page._destroy()); - } - }); - this.pageCache = []; - this.pagePromises = []; - var self = this; - // We also need to wait for the worker to finish its long running tasks. - var terminated = this.messageHandler.sendWithPromise('Terminate', null); - waitOn.push(terminated); - Promise.all(waitOn).then(function () { - self.fontLoader.clear(); - if (self.pdfDataRangeTransport) { - self.pdfDataRangeTransport.abort(); - self.pdfDataRangeTransport = null; - } - if (self.messageHandler) { - self.messageHandler.destroy(); - self.messageHandler = null; - } - self.destroyCapability.resolve(); - }, this.destroyCapability.reject); - return this.destroyCapability.promise; + forEach: function (callback) { + for (var i = this.firstChar, ii = this.lastChar; i <= ii; i++) { + callback(i, i); + } }, - setupMessageHandler: - function WorkerTransport_setupMessageHandler() { - var messageHandler = this.messageHandler; + has: function (i) { + return this.firstChar <= i && i <= this.lastChar; + }, - function updatePassword(password) { - messageHandler.send('UpdatePassword', password); + get: function (i) { + if (this.firstChar <= i && i <= this.lastChar) { + return String.fromCharCode(i); } + return undefined; + }, - var pdfDataRangeTransport = this.pdfDataRangeTransport; - if (pdfDataRangeTransport) { - pdfDataRangeTransport.addRangeListener(function(begin, chunk) { - messageHandler.send('OnDataRange', { - begin: begin, - chunk: chunk - }); - }); + charCodeOf: function (v) { + return (isInt(v) && v >= this.firstChar && v <= this.lastChar) ? v : -1; + } + }; - pdfDataRangeTransport.addProgressListener(function(loaded) { - messageHandler.send('OnDataProgress', { - loaded: loaded - }); - }); + return IdentityToUnicodeMap; +})(); - pdfDataRangeTransport.addProgressiveReadListener(function(chunk) { - messageHandler.send('OnDataRange', { - chunk: chunk - }); - }); +var OpenTypeFileBuilder = (function OpenTypeFileBuilderClosure() { + function writeInt16(dest, offset, num) { + dest[offset] = (num >> 8) & 0xFF; + dest[offset + 1] = num & 0xFF; + } - messageHandler.on('RequestDataRange', - function transportDataRange(data) { - pdfDataRangeTransport.requestDataRange(data.begin, data.end); - }, this); + function writeInt32(dest, offset, num) { + dest[offset] = (num >> 24) & 0xFF; + dest[offset + 1] = (num >> 16) & 0xFF; + dest[offset + 2] = (num >> 8) & 0xFF; + dest[offset + 3] = num & 0xFF; + } + + function writeData(dest, offset, data) { + var i, ii; + if (data instanceof Uint8Array) { + dest.set(data, offset); + } else if (typeof data === 'string') { + for (i = 0, ii = data.length; i < ii; i++) { + dest[offset++] = data.charCodeAt(i) & 0xFF; } + } else { + // treating everything else as array + for (i = 0, ii = data.length; i < ii; i++) { + dest[offset++] = data[i] & 0xFF; + } + } + } - messageHandler.on('GetDoc', function transportDoc(data) { - var pdfInfo = data.pdfInfo; - this.numPages = data.pdfInfo.numPages; - var loadingTask = this.loadingTask; - var pdfDocument = new PDFDocumentProxy(pdfInfo, this, loadingTask); - this.pdfDocument = pdfDocument; - loadingTask._capability.resolve(pdfDocument); - }, this); + function OpenTypeFileBuilder(sfnt) { + this.sfnt = sfnt; + this.tables = Object.create(null); + } - messageHandler.on('NeedPassword', - function transportNeedPassword(exception) { - var loadingTask = this.loadingTask; - if (loadingTask.onPassword) { - return loadingTask.onPassword(updatePassword, - PasswordResponses.NEED_PASSWORD); - } - loadingTask._capability.reject( - new PasswordException(exception.message, exception.code)); - }, this); + OpenTypeFileBuilder.getSearchParams = + function OpenTypeFileBuilder_getSearchParams(entriesCount, entrySize) { + var maxPower2 = 1, log2 = 0; + while ((maxPower2 ^ entriesCount) > maxPower2) { + maxPower2 <<= 1; + log2++; + } + var searchRange = maxPower2 * entrySize; + return { + range: searchRange, + entry: log2, + rangeShift: entrySize * entriesCount - searchRange + }; + }; - messageHandler.on('IncorrectPassword', - function transportIncorrectPassword(exception) { - var loadingTask = this.loadingTask; - if (loadingTask.onPassword) { - return loadingTask.onPassword(updatePassword, - PasswordResponses.INCORRECT_PASSWORD); - } - loadingTask._capability.reject( - new PasswordException(exception.message, exception.code)); - }, this); + var OTF_HEADER_SIZE = 12; + var OTF_TABLE_ENTRY_SIZE = 16; - messageHandler.on('InvalidPDF', function transportInvalidPDF(exception) { - this.loadingTask._capability.reject( - new InvalidPDFException(exception.message)); - }, this); + OpenTypeFileBuilder.prototype = { + toArray: function OpenTypeFileBuilder_toArray() { + var sfnt = this.sfnt; - messageHandler.on('MissingPDF', function transportMissingPDF(exception) { - this.loadingTask._capability.reject( - new MissingPDFException(exception.message)); - }, this); + // Tables needs to be written by ascendant alphabetic order + var tables = this.tables; + var tablesNames = Object.keys(tables); + tablesNames.sort(); + var numTables = tablesNames.length; - messageHandler.on('UnexpectedResponse', - function transportUnexpectedResponse(exception) { - this.loadingTask._capability.reject( - new UnexpectedResponseException(exception.message, exception.status)); - }, this); + var i, j, jj, table, tableName; + // layout the tables data + var offset = OTF_HEADER_SIZE + numTables * OTF_TABLE_ENTRY_SIZE; + var tableOffsets = [offset]; + for (i = 0; i < numTables; i++) { + table = tables[tablesNames[i]]; + var paddedLength = ((table.length + 3) & ~3) >>> 0; + offset += paddedLength; + tableOffsets.push(offset); + } - messageHandler.on('UnknownError', - function transportUnknownError(exception) { - this.loadingTask._capability.reject( - new UnknownErrorException(exception.message, exception.details)); - }, this); + var file = new Uint8Array(offset); + // write the table data first (mostly for checksum) + for (i = 0; i < numTables; i++) { + table = tables[tablesNames[i]]; + writeData(file, tableOffsets[i], table); + } - messageHandler.on('DataLoaded', function transportPage(data) { - this.downloadInfoCapability.resolve(data); - }, this); + // sfnt version (4 bytes) + if (sfnt === 'true') { + // Windows hates the Mac TrueType sfnt version number + sfnt = string32(0x00010000); + } + file[0] = sfnt.charCodeAt(0) & 0xFF; + file[1] = sfnt.charCodeAt(1) & 0xFF; + file[2] = sfnt.charCodeAt(2) & 0xFF; + file[3] = sfnt.charCodeAt(3) & 0xFF; - messageHandler.on('PDFManagerReady', function transportPage(data) { - if (this.pdfDataRangeTransport) { - this.pdfDataRangeTransport.transportReady(); - } - }, this); + // numTables (2 bytes) + writeInt16(file, 4, numTables); - messageHandler.on('StartRenderPage', function transportRender(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } - var page = this.pageCache[data.pageIndex]; + var searchParams = OpenTypeFileBuilder.getSearchParams(numTables, 16); - page.stats.timeEnd('Page Request'); - page._startRenderPage(data.transparency, data.intent); - }, this); + // searchRange (2 bytes) + writeInt16(file, 6, searchParams.range); + // entrySelector (2 bytes) + writeInt16(file, 8, searchParams.entry); + // rangeShift (2 bytes) + writeInt16(file, 10, searchParams.rangeShift); - messageHandler.on('RenderPageChunk', function transportRender(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. + offset = OTF_HEADER_SIZE; + // writing table entries + for (i = 0; i < numTables; i++) { + tableName = tablesNames[i]; + file[offset] = tableName.charCodeAt(0) & 0xFF; + file[offset + 1] = tableName.charCodeAt(1) & 0xFF; + file[offset + 2] = tableName.charCodeAt(2) & 0xFF; + file[offset + 3] = tableName.charCodeAt(3) & 0xFF; + + // checksum + var checksum = 0; + for (j = tableOffsets[i], jj = tableOffsets[i + 1]; j < jj; j += 4) { + var quad = (file[j] << 24) + (file[j + 1] << 16) + + (file[j + 2] << 8) + file[j + 3]; + checksum = (checksum + quad) | 0; } - var page = this.pageCache[data.pageIndex]; + writeInt32(file, offset + 4, checksum); - page._renderPageChunk(data.operatorList, data.intent); - }, this); + // offset + writeInt32(file, offset + 8, tableOffsets[i]); + // length + writeInt32(file, offset + 12, tables[tableName].length); - messageHandler.on('commonobj', function transportObj(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } + offset += OTF_TABLE_ENTRY_SIZE; + } + return file; + }, - var id = data[0]; - var type = data[1]; - if (this.commonObjs.hasData(id)) { - return; - } + addTable: function OpenTypeFileBuilder_addTable(tag, data) { + if (tag in this.tables) { + throw new Error('Table ' + tag + ' already exists'); + } + this.tables[tag] = data; + } + }; - switch (type) { - case 'Font': - var exportedData = data[2]; + return OpenTypeFileBuilder; +})(); - var font; - if ('error' in exportedData) { - var error = exportedData.error; - warn('Error during font loading: ' + error); - this.commonObjs.resolve(id, error); - break; - } else { - font = new FontFaceObject(exportedData); - } - - this.fontLoader.bind( - [font], - function fontReady(fontObjs) { - this.commonObjs.resolve(id, font); - }.bind(this) - ); - break; - case 'FontPath': - this.commonObjs.resolve(id, data[2]); - break; - default: - error('Got unknown common object type ' + type); - } - }, this); +// Problematic Unicode characters in the fonts that needs to be moved to avoid +// issues when they are painted on the canvas, e.g. complex-script shaping or +// control/whitespace characters. The ranges are listed in pairs: the first item +// is a code of the first problematic code, the second one is the next +// non-problematic code. The ranges must be in sorted order. +var ProblematicCharRanges = new Int32Array([ + // Control characters. + 0x0000, 0x0020, + 0x007F, 0x00A1, + 0x00AD, 0x00AE, + // Chars that is used in complex-script shaping. + 0x0600, 0x0780, + 0x08A0, 0x10A0, + 0x1780, 0x1800, + // General punctuation chars. + 0x2000, 0x2010, + 0x2011, 0x2012, + 0x2028, 0x2030, + 0x205F, 0x2070, + 0x25CC, 0x25CD, + // Chars that is used in complex-script shaping. + 0xAA60, 0xAA80, + // Specials Unicode block. + 0xFFF0, 0x10000 +]); - messageHandler.on('obj', function transportObj(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } +/** + * 'Font' is the class the outside world should use, it encapsulate all the font + * decoding logics whatever type it is (assuming the font type is supported). + * + * For example to read a Type1 font and to attach it to the document: + * var type1Font = new Font("MyFontName", binaryFile, propertiesObject); + * type1Font.bind(); + */ +var Font = (function FontClosure() { + function Font(name, file, properties) { + var charCode, glyphName, fontChar; - var id = data[0]; - var pageIndex = data[1]; - var type = data[2]; - var pageProxy = this.pageCache[pageIndex]; - var imageData; - if (pageProxy.objs.hasData(id)) { - return; - } + this.name = name; + this.loadedName = properties.loadedName; + this.isType3Font = properties.isType3Font; + this.sizes = []; - switch (type) { - case 'JpegStream': - imageData = data[3]; - loadJpegStream(id, imageData, pageProxy.objs); - break; - case 'Image': - imageData = data[3]; - pageProxy.objs.resolve(id, imageData); + this.glyphCache = Object.create(null); - // heuristics that will allow not to store large data - var MAX_IMAGE_SIZE_TO_STORE = 8000000; - if (imageData && 'data' in imageData && - imageData.data.length > MAX_IMAGE_SIZE_TO_STORE) { - pageProxy.cleanupAfterRender = true; - } - break; - default: - error('Got unknown object type ' + type); - } - }, this); + var names = name.split('+'); + names = names.length > 1 ? names[1] : names[0]; + names = names.split(/[-,_]/g)[0]; + this.isSerifFont = !!(properties.flags & FontFlags.Serif); + this.isSymbolicFont = !!(properties.flags & FontFlags.Symbolic); + this.isMonospace = !!(properties.flags & FontFlags.FixedPitch); - messageHandler.on('DocProgress', function transportDocProgress(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } + var type = properties.type; + var subtype = properties.subtype; + this.type = type; - var loadingTask = this.loadingTask; - if (loadingTask.onProgress) { - loadingTask.onProgress({ - loaded: data.loaded, - total: data.total - }); - } - }, this); + this.fallbackName = (this.isMonospace ? 'monospace' : + (this.isSerifFont ? 'serif' : 'sans-serif')); - messageHandler.on('PageError', function transportError(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } + this.differences = properties.differences; + this.widths = properties.widths; + this.defaultWidth = properties.defaultWidth; + this.composite = properties.composite; + this.wideChars = properties.wideChars; + this.cMap = properties.cMap; + this.ascent = properties.ascent / PDF_GLYPH_SPACE_UNITS; + this.descent = properties.descent / PDF_GLYPH_SPACE_UNITS; + this.fontMatrix = properties.fontMatrix; + this.bbox = properties.bbox; - var page = this.pageCache[data.pageNum - 1]; - var intentState = page.intentStates[data.intent]; - if (intentState.displayReadyCapability) { - intentState.displayReadyCapability.reject(data.error); - } else { - error(data.error); - } - }, this); + this.toUnicode = properties.toUnicode = this.buildToUnicode(properties); - messageHandler.on('UnsupportedFeature', - function transportUnsupportedFeature(data) { - if (this.destroyed) { - return; // Ignore any pending requests if the worker was terminated. - } - var featureId = data.featureId; - var loadingTask = this.loadingTask; - if (loadingTask.onUnsupportedFeature) { - loadingTask.onUnsupportedFeature(featureId); - } - PDFJS.UnsupportedManager.notify(featureId); - }, this); + this.toFontChar = []; - messageHandler.on('JpegDecode', function(data) { - if (this.destroyed) { - return Promise.reject('Worker was terminated'); - } + if (properties.type === 'Type3') { + for (charCode = 0; charCode < 256; charCode++) { + this.toFontChar[charCode] = (this.differences[charCode] || + properties.defaultEncoding[charCode]); + } + this.fontType = FontType.TYPE3; + return; + } - var imageUrl = data[0]; - var components = data[1]; - if (components !== 3 && components !== 1) { - return Promise.reject( - new Error('Only 3 components or 1 component can be returned')); - } + this.cidEncoding = properties.cidEncoding; + this.vertical = properties.vertical; + if (this.vertical) { + this.vmetrics = properties.vmetrics; + this.defaultVMetrics = properties.defaultVMetrics; + } + var glyphsUnicodeMap; + if (!file || file.isEmpty) { + if (file) { + // Some bad PDF generators will include empty font files, + // attempting to recover by assuming that no file exists. + warn('Font file is empty in "' + name + '" (' + this.loadedName + ')'); + } - return new Promise(function (resolve, reject) { - var img = new Image(); - img.onload = function () { - var width = img.width; - var height = img.height; - var size = width * height; - var rgbaLength = size * 4; - var buf = new Uint8Array(size * components); - var tmpCanvas = createScratchCanvas(width, height); - var tmpCtx = tmpCanvas.getContext('2d'); - tmpCtx.drawImage(img, 0, 0); - var data = tmpCtx.getImageData(0, 0, width, height).data; - var i, j; + this.missingFile = true; + // The file data is not specified. Trying to fix the font name + // to be used with the canvas.font. + var fontName = name.replace(/[,_]/g, '-'); + var stdFontMap = getStdFontMap(), nonStdFontMap = getNonStdFontMap(); + var isStandardFont = !!stdFontMap[fontName] || + !!(nonStdFontMap[fontName] && stdFontMap[nonStdFontMap[fontName]]); + fontName = stdFontMap[fontName] || nonStdFontMap[fontName] || fontName; - if (components === 3) { - for (i = 0, j = 0; i < rgbaLength; i += 4, j += 3) { - buf[j] = data[i]; - buf[j + 1] = data[i + 1]; - buf[j + 2] = data[i + 2]; - } - } else if (components === 1) { - for (i = 0, j = 0; i < rgbaLength; i += 4, j++) { - buf[j] = data[i]; - } - } - resolve({ data: buf, width: width, height: height}); - }; - img.onerror = function () { - reject(new Error('JpegDecode failed to load image')); - }; - img.src = imageUrl; - }); - }, this); - }, + this.bold = (fontName.search(/bold/gi) !== -1); + this.italic = ((fontName.search(/oblique/gi) !== -1) || + (fontName.search(/italic/gi) !== -1)); - getData: function WorkerTransport_getData() { - return this.messageHandler.sendWithPromise('GetData', null); - }, + // Use 'name' instead of 'fontName' here because the original + // name ArialBlack for example will be replaced by Helvetica. + this.black = (name.search(/Black/g) !== -1); - getPage: function WorkerTransport_getPage(pageNumber, capability) { - if (pageNumber <= 0 || pageNumber > this.numPages || - (pageNumber|0) !== pageNumber) { - return Promise.reject(new Error('Invalid page request')); + // if at least one width is present, remeasure all chars when exists + this.remeasure = Object.keys(this.widths).length > 0; + if (isStandardFont && type === 'CIDFontType2' && + properties.cidEncoding.indexOf('Identity-') === 0) { + var GlyphMapForStandardFonts = getGlyphMapForStandardFonts(); + // Standard fonts might be embedded as CID font without glyph mapping. + // Building one based on GlyphMapForStandardFonts. + var map = []; + for (charCode in GlyphMapForStandardFonts) { + map[+charCode] = GlyphMapForStandardFonts[charCode]; + } + if (/ArialBlack/i.test(name)) { + var SupplementalGlyphMapForArialBlack = + getSupplementalGlyphMapForArialBlack(); + for (charCode in SupplementalGlyphMapForArialBlack) { + map[+charCode] = SupplementalGlyphMapForArialBlack[charCode]; + } + } + var isIdentityUnicode = this.toUnicode instanceof IdentityToUnicodeMap; + if (!isIdentityUnicode) { + this.toUnicode.forEach(function(charCode, unicodeCharCode) { + map[+charCode] = unicodeCharCode; + }); + } + this.toFontChar = map; + this.toUnicode = new ToUnicodeMap(map); + } else if (/Symbol/i.test(fontName)) { + var symbols = SymbolSetEncoding; + glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode in symbols) { + fontChar = glyphsUnicodeMap[symbols[charCode]]; + if (!fontChar) { + continue; + } + this.toFontChar[charCode] = fontChar; + } + for (charCode in properties.differences) { + fontChar = glyphsUnicodeMap[properties.differences[charCode]]; + if (!fontChar) { + continue; + } + this.toFontChar[charCode] = fontChar; + } + } else if (/Dingbats/i.test(fontName)) { + glyphsUnicodeMap = getDingbatsGlyphsUnicode(); + if (/Wingdings/i.test(name)) { + warn('Wingdings font without embedded font file, ' + + 'falling back to the ZapfDingbats encoding.'); + } + var dingbats = ZapfDingbatsEncoding; + for (charCode in dingbats) { + fontChar = glyphsUnicodeMap[dingbats[charCode]]; + if (!fontChar) { + continue; + } + this.toFontChar[charCode] = fontChar; + } + for (charCode in properties.differences) { + fontChar = glyphsUnicodeMap[properties.differences[charCode]]; + if (!fontChar) { + continue; + } + this.toFontChar[charCode] = fontChar; + } + } else if (isStandardFont) { + this.toFontChar = []; + glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode in properties.defaultEncoding) { + glyphName = (properties.differences[charCode] || + properties.defaultEncoding[charCode]); + this.toFontChar[charCode] = glyphsUnicodeMap[glyphName]; + } + } else { + var unicodeCharCode, notCidFont = (type.indexOf('CIDFontType') === -1); + glyphsUnicodeMap = getGlyphsUnicode(); + this.toUnicode.forEach(function(charCode, unicodeCharCode) { + if (notCidFont) { + glyphName = (properties.differences[charCode] || + properties.defaultEncoding[charCode]); + unicodeCharCode = (glyphsUnicodeMap[glyphName] || unicodeCharCode); + } + this.toFontChar[charCode] = unicodeCharCode; + }.bind(this)); } + this.loadedName = fontName.split('-')[0]; + this.loading = false; + this.fontType = getFontType(type, subtype); + return; + } - var pageIndex = pageNumber - 1; - if (pageIndex in this.pagePromises) { - return this.pagePromises[pageIndex]; + // Some fonts might use wrong font types for Type1C or CIDFontType0C + if (subtype === 'Type1C' && (type !== 'Type1' && type !== 'MMType1')) { + // Some TrueType fonts by mistake claim Type1C + if (isTrueTypeFile(file)) { + subtype = 'TrueType'; + } else { + type = 'Type1'; } - var promise = this.messageHandler.sendWithPromise('GetPage', { - pageIndex: pageIndex - }).then(function (pageInfo) { - if (this.destroyed) { - throw new Error('Transport destroyed'); - } - var page = new PDFPageProxy(pageIndex, pageInfo, this); - this.pageCache[pageIndex] = page; - return page; - }.bind(this)); - this.pagePromises[pageIndex] = promise; - return promise; - }, + } + if (subtype === 'CIDFontType0C' && type !== 'CIDFontType0') { + type = 'CIDFontType0'; + } + if (subtype === 'OpenType') { + type = 'OpenType'; + } + // Some CIDFontType0C fonts by mistake claim CIDFontType0. + if (type === 'CIDFontType0') { + if (isType1File(file)) { + subtype = 'CIDFontType0'; + } else if (isOpenTypeFile(file)) { + // Sometimes the type/subtype can be a complete lie (see issue6782.pdf). + type = subtype = 'OpenType'; + } else { + subtype = 'CIDFontType0C'; + } + } - getPageIndex: function WorkerTransport_getPageIndexByRef(ref) { - return this.messageHandler.sendWithPromise('GetPageIndex', { ref: ref }); - }, + var data; + switch (type) { + case 'MMType1': + info('MMType1 font (' + name + '), falling back to Type1.'); + /* falls through */ + case 'Type1': + case 'CIDFontType0': + this.mimetype = 'font/opentype'; - getAnnotations: function WorkerTransport_getAnnotations(pageIndex, intent) { - return this.messageHandler.sendWithPromise('GetAnnotations', { - pageIndex: pageIndex, - intent: intent, - }); - }, + var cff = (subtype === 'Type1C' || subtype === 'CIDFontType0C') ? + new CFFFont(file, properties) : new Type1Font(name, file, properties); - getDestinations: function WorkerTransport_getDestinations() { - return this.messageHandler.sendWithPromise('GetDestinations', null); - }, + adjustWidths(properties); - getDestination: function WorkerTransport_getDestination(id) { - return this.messageHandler.sendWithPromise('GetDestination', { id: id }); - }, + // Wrap the CFF data inside an OTF font file + data = this.convert(name, cff, properties); + break; - getPageLabels: function WorkerTransport_getPageLabels() { - return this.messageHandler.sendWithPromise('GetPageLabels', null); - }, + case 'OpenType': + case 'TrueType': + case 'CIDFontType2': + this.mimetype = 'font/opentype'; - getAttachments: function WorkerTransport_getAttachments() { - return this.messageHandler.sendWithPromise('GetAttachments', null); - }, + // Repair the TrueType file. It is can be damaged in the point of + // view of the sanitizer + data = this.checkAndRepair(name, file, properties); + if (this.isOpenType) { + adjustWidths(properties); - getJavaScript: function WorkerTransport_getJavaScript() { - return this.messageHandler.sendWithPromise('GetJavaScript', null); - }, + type = 'OpenType'; + } + break; - getOutline: function WorkerTransport_getOutline() { - return this.messageHandler.sendWithPromise('GetOutline', null); - }, + default: + error('Font ' + type + ' is not supported'); + break; + } - getMetadata: function WorkerTransport_getMetadata() { - return this.messageHandler.sendWithPromise('GetMetadata', null). - then(function transportMetadata(results) { - return { - info: results[0], - metadata: (results[1] ? new Metadata(results[1]) : null) - }; - }); - }, + this.data = data; + this.fontType = getFontType(type, subtype); - getStats: function WorkerTransport_getStats() { - return this.messageHandler.sendWithPromise('GetStats', null); - }, + // Transfer some properties again that could change during font conversion + this.fontMatrix = properties.fontMatrix; + this.widths = properties.widths; + this.defaultWidth = properties.defaultWidth; + this.encoding = properties.baseEncoding; + this.seacMap = properties.seacMap; - startCleanup: function WorkerTransport_startCleanup() { - this.messageHandler.sendWithPromise('Cleanup', null). - then(function endCleanup() { - for (var i = 0, ii = this.pageCache.length; i < ii; i++) { - var page = this.pageCache[i]; - if (page) { - page.cleanup(); - } - } - this.commonObjs.clear(); - this.fontLoader.clear(); - }.bind(this)); - } - }; - return WorkerTransport; + this.loading = true; + } -})(); + Font.getFontID = (function () { + var ID = 1; + return function Font_getFontID() { + return String(ID++); + }; + })(); -/** - * A PDF document and page is built of many objects. E.g. there are objects - * for fonts, images, rendering code and such. These objects might get processed - * inside of a worker. The `PDFObjects` implements some basic functions to - * manage these objects. - * @ignore - */ -var PDFObjects = (function PDFObjectsClosure() { - function PDFObjects() { - this.objs = Object.create(null); + function int16(b0, b1) { + return (b0 << 8) + b1; } - PDFObjects.prototype = { - /** - * Internal function. - * Ensures there is an object defined for `objId`. - */ - ensureObj: function PDFObjects_ensureObj(objId) { - if (this.objs[objId]) { - return this.objs[objId]; - } + function int32(b0, b1, b2, b3) { + return (b0 << 24) + (b1 << 16) + (b2 << 8) + b3; + } - var obj = { - capability: createPromiseCapability(), - data: null, - resolved: false - }; - this.objs[objId] = obj; + function string16(value) { + return String.fromCharCode((value >> 8) & 0xff, value & 0xff); + } - return obj; - }, + function safeString16(value) { + // clamp value to the 16-bit int range + value = (value > 0x7FFF ? 0x7FFF : (value < -0x8000 ? -0x8000 : value)); + return String.fromCharCode((value >> 8) & 0xff, value & 0xff); + } - /** - * If called *without* callback, this returns the data of `objId` but the - * object needs to be resolved. If it isn't, this function throws. - * - * If called *with* a callback, the callback is called with the data of the - * object once the object is resolved. That means, if you call this - * function and the object is already resolved, the callback gets called - * right away. - */ - get: function PDFObjects_get(objId, callback) { - // If there is a callback, then the get can be async and the object is - // not required to be resolved right now - if (callback) { - this.ensureObj(objId).capability.promise.then(callback); - return null; - } - - // If there isn't a callback, the user expects to get the resolved data - // directly. - var obj = this.objs[objId]; - - // If there isn't an object yet or the object isn't resolved, then the - // data isn't ready yet! - if (!obj || !obj.resolved) { - error('Requesting object that isn\'t resolved yet ' + objId); - } - - return obj.data; - }, - - /** - * Resolves the object `objId` with optional `data`. - */ - resolve: function PDFObjects_resolve(objId, data) { - var obj = this.ensureObj(objId); + function isTrueTypeFile(file) { + var header = file.peekBytes(4); + return readUint32(header, 0) === 0x00010000; + } - obj.resolved = true; - obj.data = data; - obj.capability.resolve(data); - }, + function isOpenTypeFile(file) { + var header = file.peekBytes(4); + return bytesToString(header) === 'OTTO'; + } - isResolved: function PDFObjects_isResolved(objId) { - var objs = this.objs; + function isType1File(file) { + var header = file.peekBytes(2); + // All Type1 font programs must begin with the comment '%!' (0x25 + 0x21). + if (header[0] === 0x25 && header[1] === 0x21) { + return true; + } + // ... obviously some fonts violate that part of the specification, + // please refer to the comment in |Type1Font| below. + if (header[0] === 0x80 && header[1] === 0x01) { // pfb file header. + return true; + } + return false; + } - if (!objs[objId]) { - return false; + /** + * Helper function for |adjustMapping|. + * @return {boolean} + */ + function isProblematicUnicodeLocation(code) { + // Using binary search to find a range start. + var i = 0, j = ProblematicCharRanges.length - 1; + while (i < j) { + var c = (i + j + 1) >> 1; + if (code < ProblematicCharRanges[c]) { + j = c - 1; } else { - return objs[objId].resolved; + i = c; } - }, - - hasData: function PDFObjects_hasData(objId) { - return this.isResolved(objId); - }, + } + // Even index means code in problematic range. + return !(i & 1); + } - /** - * Returns the data of `objId` if object exists, null otherwise. - */ - getData: function PDFObjects_getData(objId) { - var objs = this.objs; - if (!objs[objId] || !objs[objId].resolved) { - return null; - } else { - return objs[objId].data; + /** + * Rebuilds the char code to glyph ID map by trying to replace the char codes + * with their unicode value. It also moves char codes that are in known + * problematic locations. + * @return {Object} Two properties: + * 'toFontChar' - maps original char codes(the value that will be read + * from commands such as show text) to the char codes that will be used in the + * font that we build + * 'charCodeToGlyphId' - maps the new font char codes to glyph ids + */ + function adjustMapping(charCodeToGlyphId, properties) { + var toUnicode = properties.toUnicode; + var isSymbolic = !!(properties.flags & FontFlags.Symbolic); + var isIdentityUnicode = + properties.toUnicode instanceof IdentityToUnicodeMap; + var newMap = Object.create(null); + var toFontChar = []; + var usedFontCharCodes = []; + var nextAvailableFontCharCode = PRIVATE_USE_OFFSET_START; + for (var originalCharCode in charCodeToGlyphId) { + originalCharCode |= 0; + var glyphId = charCodeToGlyphId[originalCharCode]; + var fontCharCode = originalCharCode; + // First try to map the value to a unicode position if a non identity map + // was created. + if (!isIdentityUnicode && toUnicode.has(originalCharCode)) { + var unicode = toUnicode.get(fontCharCode); + // TODO: Try to map ligatures to the correct spot. + if (unicode.length === 1) { + fontCharCode = unicode.charCodeAt(0); + } } - }, + // Try to move control characters, special characters and already mapped + // characters to the private use area since they will not be drawn by + // canvas if left in their current position. Also, move characters if the + // font was symbolic and there is only an identity unicode map since the + // characters probably aren't in the correct position (fixes an issue + // with firefox and thuluthfont). + if ((usedFontCharCodes[fontCharCode] !== undefined || + isProblematicUnicodeLocation(fontCharCode) || + (isSymbolic && isIdentityUnicode)) && + nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END) { // Room left. + // Loop to try and find a free spot in the private use area. + do { + fontCharCode = nextAvailableFontCharCode++; - clear: function PDFObjects_clear() { - this.objs = Object.create(null); - } - }; - return PDFObjects; -})(); + if (SKIP_PRIVATE_USE_RANGE_F000_TO_F01F && fontCharCode === 0xF000) { + fontCharCode = 0xF020; + nextAvailableFontCharCode = fontCharCode + 1; + } -/** - * Allows controlling of the rendering tasks. - * @class - * @alias RenderTask - */ -var RenderTask = (function RenderTaskClosure() { - function RenderTask(internalRenderTask) { - this._internalRenderTask = internalRenderTask; + } while (usedFontCharCodes[fontCharCode] !== undefined && + nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END); + } - /** - * Callback for incremental rendering -- a function that will be called - * each time the rendering is paused. To continue rendering call the - * function that is the first argument to the callback. - * @type {function} - */ - this.onContinue = null; + newMap[fontCharCode] = glyphId; + toFontChar[originalCharCode] = fontCharCode; + usedFontCharCodes[fontCharCode] = true; + } + return { + toFontChar: toFontChar, + charCodeToGlyphId: newMap, + nextAvailableFontCharCode: nextAvailableFontCharCode + }; } - RenderTask.prototype = /** @lends RenderTask.prototype */ { - /** - * Promise for rendering task completion. - * @return {Promise} - */ - get promise() { - return this._internalRenderTask.capability.promise; - }, - - /** - * Cancels the rendering task. If the task is currently rendering it will - * not be cancelled until graphics pauses with a timeout. The promise that - * this object extends will resolved when cancelled. - */ - cancel: function RenderTask_cancel() { - this._internalRenderTask.cancel(); - }, - - /** - * Registers callbacks to indicate the rendering task completion. - * - * @param {function} onFulfilled The callback for the rendering completion. - * @param {function} onRejected The callback for the rendering failure. - * @return {Promise} A promise that is resolved after the onFulfilled or - * onRejected callback. - */ - then: function RenderTask_then(onFulfilled, onRejected) { - return this.promise.then.apply(this.promise, arguments); + function getRanges(glyphs, numGlyphs) { + // Array.sort() sorts by characters, not numerically, so convert to an + // array of characters. + var codes = []; + for (var charCode in glyphs) { + // Remove an invalid glyph ID mappings to make OTS happy. + if (glyphs[charCode] >= numGlyphs) { + continue; + } + codes.push({ fontCharCode: charCode | 0, glyphId: glyphs[charCode] }); } - }; - - return RenderTask; -})(); + codes.sort(function fontGetRangesSort(a, b) { + return a.fontCharCode - b.fontCharCode; + }); -/** - * For internal use only. - * @ignore - */ -var InternalRenderTask = (function InternalRenderTaskClosure() { + // Split the sorted codes into ranges. + var ranges = []; + var length = codes.length; + for (var n = 0; n < length; ) { + var start = codes[n].fontCharCode; + var codeIndices = [codes[n].glyphId]; + ++n; + var end = start; + while (n < length && end + 1 === codes[n].fontCharCode) { + codeIndices.push(codes[n].glyphId); + ++end; + ++n; + if (end === 0xFFFF) { + break; + } + } + ranges.push([start, end, codeIndices]); + } - function InternalRenderTask(callback, params, objs, commonObjs, operatorList, - pageNumber) { - this.callback = callback; - this.params = params; - this.objs = objs; - this.commonObjs = commonObjs; - this.operatorListIdx = null; - this.operatorList = operatorList; - this.pageNumber = pageNumber; - this.running = false; - this.graphicsReadyCallback = null; - this.graphicsReady = false; - this.useRequestAnimationFrame = false; - this.cancelled = false; - this.capability = createPromiseCapability(); - this.task = new RenderTask(this); - // caching this-bound methods - this._continueBound = this._continue.bind(this); - this._scheduleNextBound = this._scheduleNext.bind(this); - this._nextBound = this._next.bind(this); + return ranges; } - InternalRenderTask.prototype = { + function createCmapTable(glyphs, numGlyphs) { + var ranges = getRanges(glyphs, numGlyphs); + var numTables = ranges[ranges.length - 1][1] > 0xFFFF ? 2 : 1; + var cmap = '\x00\x00' + // version + string16(numTables) + // numTables + '\x00\x03' + // platformID + '\x00\x01' + // encodingID + string32(4 + numTables * 8); // start of the table record - initalizeGraphics: - function InternalRenderTask_initalizeGraphics(transparency) { + var i, ii, j, jj; + for (i = ranges.length - 1; i >= 0; --i) { + if (ranges[i][0] <= 0xFFFF) { break; } + } + var bmpLength = i + 1; - if (this.cancelled) { - return; - } - if (PDFJS.pdfBug && 'StepperManager' in globalScope && - globalScope.StepperManager.enabled) { - this.stepper = globalScope.StepperManager.create(this.pageNumber - 1); - this.stepper.init(this.operatorList); - this.stepper.nextBreakPoint = this.stepper.getNextBreakPoint(); - } + if (ranges[i][0] < 0xFFFF && ranges[i][1] === 0xFFFF) { + ranges[i][1] = 0xFFFE; + } + var trailingRangesCount = ranges[i][1] < 0xFFFF ? 1 : 0; + var segCount = bmpLength + trailingRangesCount; + var searchParams = OpenTypeFileBuilder.getSearchParams(segCount, 2); - var params = this.params; - this.gfx = new CanvasGraphics(params.canvasContext, this.commonObjs, - this.objs, params.imageLayer); + // Fill up the 4 parallel arrays describing the segments. + var startCount = ''; + var endCount = ''; + var idDeltas = ''; + var idRangeOffsets = ''; + var glyphsIds = ''; + var bias = 0; - this.gfx.beginDrawing(params.transform, params.viewport, transparency); - this.operatorListIdx = 0; - this.graphicsReady = true; - if (this.graphicsReadyCallback) { - this.graphicsReadyCallback(); + var range, start, end, codes; + for (i = 0, ii = bmpLength; i < ii; i++) { + range = ranges[i]; + start = range[0]; + end = range[1]; + startCount += string16(start); + endCount += string16(end); + codes = range[2]; + var contiguous = true; + for (j = 1, jj = codes.length; j < jj; ++j) { + if (codes[j] !== codes[j - 1] + 1) { + contiguous = false; + break; + } } - }, + if (!contiguous) { + var offset = (segCount - i) * 2 + bias * 2; + bias += (end - start + 1); - cancel: function InternalRenderTask_cancel() { - this.running = false; - this.cancelled = true; - this.callback('cancelled'); - }, + idDeltas += string16(0); + idRangeOffsets += string16(offset); - operatorListChanged: function InternalRenderTask_operatorListChanged() { - if (!this.graphicsReady) { - if (!this.graphicsReadyCallback) { - this.graphicsReadyCallback = this._continueBound; + for (j = 0, jj = codes.length; j < jj; ++j) { + glyphsIds += string16(codes[j]); } - return; - } - - if (this.stepper) { - this.stepper.updateOperatorList(this.operatorList); - } + } else { + var startCode = codes[0]; - if (this.running) { - return; + idDeltas += string16((startCode - start) & 0xFFFF); + idRangeOffsets += string16(0); } - this._continue(); - }, + } - _continue: function InternalRenderTask__continue() { - this.running = true; - if (this.cancelled) { - return; - } - if (this.task.onContinue) { - this.task.onContinue.call(this.task, this._scheduleNextBound); - } else { - this._scheduleNext(); - } - }, + if (trailingRangesCount > 0) { + endCount += '\xFF\xFF'; + startCount += '\xFF\xFF'; + idDeltas += '\x00\x01'; + idRangeOffsets += '\x00\x00'; + } - _scheduleNext: function InternalRenderTask__scheduleNext() { - if (this.useRequestAnimationFrame) { - window.requestAnimationFrame(this._nextBound); - } else { - Promise.resolve(undefined).then(this._nextBound); - } - }, + var format314 = '\x00\x00' + // language + string16(2 * segCount) + + string16(searchParams.range) + + string16(searchParams.entry) + + string16(searchParams.rangeShift) + + endCount + '\x00\x00' + startCount + + idDeltas + idRangeOffsets + glyphsIds; - _next: function InternalRenderTask__next() { - if (this.cancelled) { - return; - } - this.operatorListIdx = this.gfx.executeOperatorList(this.operatorList, - this.operatorListIdx, - this._continueBound, - this.stepper); - if (this.operatorListIdx === this.operatorList.argsArray.length) { - this.running = false; - if (this.operatorList.lastChunk) { - this.gfx.endDrawing(); - this.callback(); + var format31012 = ''; + var header31012 = ''; + if (numTables > 1) { + cmap += '\x00\x03' + // platformID + '\x00\x0A' + // encodingID + string32(4 + numTables * 8 + + 4 + format314.length); // start of the table record + format31012 = ''; + for (i = 0, ii = ranges.length; i < ii; i++) { + range = ranges[i]; + start = range[0]; + codes = range[2]; + var code = codes[0]; + for (j = 1, jj = codes.length; j < jj; ++j) { + if (codes[j] !== codes[j - 1] + 1) { + end = range[0] + j - 1; + format31012 += string32(start) + // startCharCode + string32(end) + // endCharCode + string32(code); // startGlyphID + start = end + 1; + code = codes[j]; + } } + format31012 += string32(start) + // startCharCode + string32(range[1]) + // endCharCode + string32(code); // startGlyphID } + header31012 = '\x00\x0C' + // format + '\x00\x00' + // reserved + string32(format31012.length + 16) + // length + '\x00\x00\x00\x00' + // language + string32(format31012.length / 12); // nGroups } - }; - - return InternalRenderTask; -})(); + return cmap + '\x00\x04' + // format + string16(format314.length + 4) + // length + format314 + header31012 + format31012; + } -/** - * (Deprecated) Global observer of unsupported feature usages. Use - * onUnsupportedFeature callback of the {PDFDocumentLoadingTask} instance. - */ -PDFJS.UnsupportedManager = (function UnsupportedManagerClosure() { - var listeners = []; - return { - listen: function (cb) { - deprecated('Global UnsupportedManager.listen is used: ' + - ' use PDFDocumentLoadingTask.onUnsupportedFeature instead'); - listeners.push(cb); - }, - notify: function (featureId) { - for (var i = 0, ii = listeners.length; i < ii; i++) { - listeners[i](featureId); - } + function validateOS2Table(os2) { + var stream = new Stream(os2.data); + var version = stream.getUint16(); + // TODO verify all OS/2 tables fields, but currently we validate only those + // that give us issues + stream.getBytes(60); // skipping type, misc sizes, panose, unicode ranges + var selection = stream.getUint16(); + if (version < 4 && (selection & 0x0300)) { + return false; + } + var firstChar = stream.getUint16(); + var lastChar = stream.getUint16(); + if (firstChar > lastChar) { + return false; + } + stream.getBytes(6); // skipping sTypoAscender/Descender/LineGap + var usWinAscent = stream.getUint16(); + if (usWinAscent === 0) { // makes font unreadable by windows + return false; } - }; -})(); -exports.getDocument = PDFJS.getDocument; -exports.PDFDataRangeTransport = PDFDataRangeTransport; -exports.PDFDocumentProxy = PDFDocumentProxy; -exports.PDFPageProxy = PDFPageProxy; -})); + // OS/2 appears to be valid, resetting some fields + os2.data[8] = os2.data[9] = 0; // IE rejects fonts if fsType != 0 + return true; + } + function createOS2Table(properties, charstrings, override) { + override = override || { + unitsPerEm: 0, + yMax: 0, + yMin: 0, + ascent: 0, + descent: 0 + }; -(function (root, factory) { - { - factory((root.pdfjsCoreFonts = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreStream, root.pdfjsCoreParser, - root.pdfjsCoreCMap, root.pdfjsCoreGlyphList, root.pdfjsCoreCharsets, - root.pdfjsCoreFontRenderer, root.pdfjsCoreEncodings, - root.pdfjsCoreStandardFonts, root.pdfjsCoreUnicode); - } -}(this, function (exports, sharedUtil, corePrimitives, coreStream, coreParser, - coreCMap, coreGlyphList, coreCharsets, coreFontRenderer, - coreEncodings, coreStandardFonts, coreUnicode) { + var ulUnicodeRange1 = 0; + var ulUnicodeRange2 = 0; + var ulUnicodeRange3 = 0; + var ulUnicodeRange4 = 0; -var FONT_IDENTITY_MATRIX = sharedUtil.FONT_IDENTITY_MATRIX; -var FontType = sharedUtil.FontType; -var Util = sharedUtil.Util; -var assert = sharedUtil.assert; -var bytesToString = sharedUtil.bytesToString; -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isInt = sharedUtil.isInt; -var isNum = sharedUtil.isNum; -var readUint32 = sharedUtil.readUint32; -var shadow = sharedUtil.shadow; -var stringToBytes = sharedUtil.stringToBytes; -var string32 = sharedUtil.string32; -var warn = sharedUtil.warn; -var Name = corePrimitives.Name; -var Stream = coreStream.Stream; -var Lexer = coreParser.Lexer; -var CMapFactory = coreCMap.CMapFactory; -var IdentityCMap = coreCMap.IdentityCMap; -var getGlyphsUnicode = coreGlyphList.getGlyphsUnicode; -var getDingbatsGlyphsUnicode = coreGlyphList.getDingbatsGlyphsUnicode; -var ISOAdobeCharset = coreCharsets.ISOAdobeCharset; -var ExpertCharset = coreCharsets.ExpertCharset; -var ExpertSubsetCharset = coreCharsets.ExpertSubsetCharset; -var FontRendererFactory = coreFontRenderer.FontRendererFactory; -var WinAnsiEncoding = coreEncodings.WinAnsiEncoding; -var StandardEncoding = coreEncodings.StandardEncoding; -var MacRomanEncoding = coreEncodings.MacRomanEncoding; -var SymbolSetEncoding = coreEncodings.SymbolSetEncoding; -var ZapfDingbatsEncoding = coreEncodings.ZapfDingbatsEncoding; -var ExpertEncoding = coreEncodings.ExpertEncoding; -var getEncoding = coreEncodings.getEncoding; -var getStdFontMap = coreStandardFonts.getStdFontMap; -var getNonStdFontMap = coreStandardFonts.getNonStdFontMap; -var getGlyphMapForStandardFonts = coreStandardFonts.getGlyphMapForStandardFonts; -var getSupplementalGlyphMapForArialBlack = - coreStandardFonts.getSupplementalGlyphMapForArialBlack; -var getUnicodeRangeFor = coreUnicode.getUnicodeRangeFor; -var mapSpecialUnicodeValues = coreUnicode.mapSpecialUnicodeValues; - -// Unicode Private Use Area -var PRIVATE_USE_OFFSET_START = 0xE000; -var PRIVATE_USE_OFFSET_END = 0xF8FF; -var SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = false; - -// PDF Glyph Space Units are one Thousandth of a TextSpace Unit -// except for Type 3 fonts -var PDF_GLYPH_SPACE_UNITS = 1000; - -// Hinting is currently disabled due to unknown problems on windows -// in tracemonkey and various other pdfs with type1 fonts. -var HINTING_ENABLED = false; + var firstCharIndex = null; + var lastCharIndex = 0; -// Accented charactars are not displayed properly on windows, using this flag -// to control analysis of seac charstrings. -var SEAC_ANALYSIS_ENABLED = false; + if (charstrings) { + for (var code in charstrings) { + code |= 0; + if (firstCharIndex > code || !firstCharIndex) { + firstCharIndex = code; + } + if (lastCharIndex < code) { + lastCharIndex = code; + } -// Maximum subroutine call depth of type 2 chartrings. Matches OTS. -var MAX_SUBR_NESTING = 10; + var position = getUnicodeRangeFor(code); + if (position < 32) { + ulUnicodeRange1 |= 1 << position; + } else if (position < 64) { + ulUnicodeRange2 |= 1 << position - 32; + } else if (position < 96) { + ulUnicodeRange3 |= 1 << position - 64; + } else if (position < 123) { + ulUnicodeRange4 |= 1 << position - 96; + } else { + error('Unicode ranges Bits > 123 are reserved for internal usage'); + } + } + } else { + // TODO + firstCharIndex = 0; + lastCharIndex = 255; + } -var FontFlags = { - FixedPitch: 1, - Serif: 2, - Symbolic: 4, - Script: 8, - Nonsymbolic: 32, - Italic: 64, - AllCap: 65536, - SmallCap: 131072, - ForceBold: 262144 -}; + var bbox = properties.bbox || [0, 0, 0, 0]; + var unitsPerEm = (override.unitsPerEm || + 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]); -var MacStandardGlyphOrdering = [ - '.notdef', '.null', 'nonmarkingreturn', 'space', 'exclam', 'quotedbl', - 'numbersign', 'dollar', 'percent', 'ampersand', 'quotesingle', 'parenleft', - 'parenright', 'asterisk', 'plus', 'comma', 'hyphen', 'period', 'slash', - 'zero', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', - 'nine', 'colon', 'semicolon', 'less', 'equal', 'greater', 'question', 'at', - 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', - 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'bracketleft', - 'backslash', 'bracketright', 'asciicircum', 'underscore', 'grave', 'a', 'b', - 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', - 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 'braceleft', 'bar', 'braceright', - 'asciitilde', 'Adieresis', 'Aring', 'Ccedilla', 'Eacute', 'Ntilde', - 'Odieresis', 'Udieresis', 'aacute', 'agrave', 'acircumflex', 'adieresis', - 'atilde', 'aring', 'ccedilla', 'eacute', 'egrave', 'ecircumflex', 'edieresis', - 'iacute', 'igrave', 'icircumflex', 'idieresis', 'ntilde', 'oacute', 'ograve', - 'ocircumflex', 'odieresis', 'otilde', 'uacute', 'ugrave', 'ucircumflex', - 'udieresis', 'dagger', 'degree', 'cent', 'sterling', 'section', 'bullet', - 'paragraph', 'germandbls', 'registered', 'copyright', 'trademark', 'acute', - 'dieresis', 'notequal', 'AE', 'Oslash', 'infinity', 'plusminus', 'lessequal', - 'greaterequal', 'yen', 'mu', 'partialdiff', 'summation', 'product', 'pi', - 'integral', 'ordfeminine', 'ordmasculine', 'Omega', 'ae', 'oslash', - 'questiondown', 'exclamdown', 'logicalnot', 'radical', 'florin', - 'approxequal', 'Delta', 'guillemotleft', 'guillemotright', 'ellipsis', - 'nonbreakingspace', 'Agrave', 'Atilde', 'Otilde', 'OE', 'oe', 'endash', - 'emdash', 'quotedblleft', 'quotedblright', 'quoteleft', 'quoteright', - 'divide', 'lozenge', 'ydieresis', 'Ydieresis', 'fraction', 'currency', - 'guilsinglleft', 'guilsinglright', 'fi', 'fl', 'daggerdbl', 'periodcentered', - 'quotesinglbase', 'quotedblbase', 'perthousand', 'Acircumflex', - 'Ecircumflex', 'Aacute', 'Edieresis', 'Egrave', 'Iacute', 'Icircumflex', - 'Idieresis', 'Igrave', 'Oacute', 'Ocircumflex', 'apple', 'Ograve', 'Uacute', - 'Ucircumflex', 'Ugrave', 'dotlessi', 'circumflex', 'tilde', 'macron', - 'breve', 'dotaccent', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', - 'Lslash', 'lslash', 'Scaron', 'scaron', 'Zcaron', 'zcaron', 'brokenbar', - 'Eth', 'eth', 'Yacute', 'yacute', 'Thorn', 'thorn', 'minus', 'multiply', - 'onesuperior', 'twosuperior', 'threesuperior', 'onehalf', 'onequarter', - 'threequarters', 'franc', 'Gbreve', 'gbreve', 'Idotaccent', 'Scedilla', - 'scedilla', 'Cacute', 'cacute', 'Ccaron', 'ccaron', 'dcroat']; + // if the font units differ to the PDF glyph space units + // then scale up the values + var scale = (properties.ascentScaled ? 1.0 : + unitsPerEm / PDF_GLYPH_SPACE_UNITS); -function adjustWidths(properties) { - if (!properties.fontMatrix) { - return; - } - if (properties.fontMatrix[0] === FONT_IDENTITY_MATRIX[0]) { - return; - } - // adjusting width to fontMatrix scale - var scale = 0.001 / properties.fontMatrix[0]; - var glyphsWidths = properties.widths; - for (var glyph in glyphsWidths) { - glyphsWidths[glyph] *= scale; - } - properties.defaultWidth *= scale; -} + var typoAscent = (override.ascent || + Math.round(scale * (properties.ascent || bbox[3]))); + var typoDescent = (override.descent || + Math.round(scale * (properties.descent || bbox[1]))); + if (typoDescent > 0 && properties.descent > 0 && bbox[1] < 0) { + typoDescent = -typoDescent; // fixing incorrect descent + } + var winAscent = override.yMax || typoAscent; + var winDescent = -override.yMin || -typoDescent; -function getFontType(type, subtype) { - switch (type) { - case 'Type1': - return subtype === 'Type1C' ? FontType.TYPE1C : FontType.TYPE1; - case 'CIDFontType0': - return subtype === 'CIDFontType0C' ? FontType.CIDFONTTYPE0C : - FontType.CIDFONTTYPE0; - case 'OpenType': - return FontType.OPENTYPE; - case 'TrueType': - return FontType.TRUETYPE; - case 'CIDFontType2': - return FontType.CIDFONTTYPE2; - case 'MMType1': - return FontType.MMTYPE1; - case 'Type0': - return FontType.TYPE0; - default: - return FontType.UNKNOWN; + return '\x00\x03' + // version + '\x02\x24' + // xAvgCharWidth + '\x01\xF4' + // usWeightClass + '\x00\x05' + // usWidthClass + '\x00\x00' + // fstype (0 to let the font loads via font-face on IE) + '\x02\x8A' + // ySubscriptXSize + '\x02\xBB' + // ySubscriptYSize + '\x00\x00' + // ySubscriptXOffset + '\x00\x8C' + // ySubscriptYOffset + '\x02\x8A' + // ySuperScriptXSize + '\x02\xBB' + // ySuperScriptYSize + '\x00\x00' + // ySuperScriptXOffset + '\x01\xDF' + // ySuperScriptYOffset + '\x00\x31' + // yStrikeOutSize + '\x01\x02' + // yStrikeOutPosition + '\x00\x00' + // sFamilyClass + '\x00\x00\x06' + + String.fromCharCode(properties.fixedPitch ? 0x09 : 0x00) + + '\x00\x00\x00\x00\x00\x00' + // Panose + string32(ulUnicodeRange1) + // ulUnicodeRange1 (Bits 0-31) + string32(ulUnicodeRange2) + // ulUnicodeRange2 (Bits 32-63) + string32(ulUnicodeRange3) + // ulUnicodeRange3 (Bits 64-95) + string32(ulUnicodeRange4) + // ulUnicodeRange4 (Bits 96-127) + '\x2A\x32\x31\x2A' + // achVendID + string16(properties.italicAngle ? 1 : 0) + // fsSelection + string16(firstCharIndex || + properties.firstChar) + // usFirstCharIndex + string16(lastCharIndex || properties.lastChar) + // usLastCharIndex + string16(typoAscent) + // sTypoAscender + string16(typoDescent) + // sTypoDescender + '\x00\x64' + // sTypoLineGap (7%-10% of the unitsPerEM value) + string16(winAscent) + // usWinAscent + string16(winDescent) + // usWinDescent + '\x00\x00\x00\x00' + // ulCodePageRange1 (Bits 0-31) + '\x00\x00\x00\x00' + // ulCodePageRange2 (Bits 32-63) + string16(properties.xHeight) + // sxHeight + string16(properties.capHeight) + // sCapHeight + string16(0) + // usDefaultChar + string16(firstCharIndex || properties.firstChar) + // usBreakChar + '\x00\x03'; // usMaxContext } -} -var Glyph = (function GlyphClosure() { - function Glyph(fontChar, unicode, accent, width, vmetric, operatorListId, - isSpace) { - this.fontChar = fontChar; - this.unicode = unicode; - this.accent = accent; - this.width = width; - this.vmetric = vmetric; - this.operatorListId = operatorListId; - this.isSpace = isSpace; + function createPostTable(properties) { + var angle = Math.floor(properties.italicAngle * (Math.pow(2, 16))); + return ('\x00\x03\x00\x00' + // Version number + string32(angle) + // italicAngle + '\x00\x00' + // underlinePosition + '\x00\x00' + // underlineThickness + string32(properties.fixedPitch) + // isFixedPitch + '\x00\x00\x00\x00' + // minMemType42 + '\x00\x00\x00\x00' + // maxMemType42 + '\x00\x00\x00\x00' + // minMemType1 + '\x00\x00\x00\x00'); // maxMemType1 } - Glyph.prototype.matchesForCache = function(fontChar, unicode, accent, width, - vmetric, operatorListId, isSpace) { - return this.fontChar === fontChar && - this.unicode === unicode && - this.accent === accent && - this.width === width && - this.vmetric === vmetric && - this.operatorListId === operatorListId && - this.isSpace === isSpace; - }; - - return Glyph; -})(); + function createNameTable(name, proto) { + if (!proto) { + proto = [[], []]; // no strings and unicode strings + } -var ToUnicodeMap = (function ToUnicodeMapClosure() { - function ToUnicodeMap(cmap) { - // The elements of this._map can be integers or strings, depending on how - // |cmap| was created. - this._map = cmap; - } + var strings = [ + proto[0][0] || 'Original licence', // 0.Copyright + proto[0][1] || name, // 1.Font family + proto[0][2] || 'Unknown', // 2.Font subfamily (font weight) + proto[0][3] || 'uniqueID', // 3.Unique ID + proto[0][4] || name, // 4.Full font name + proto[0][5] || 'Version 0.11', // 5.Version + proto[0][6] || '', // 6.Postscript name + proto[0][7] || 'Unknown', // 7.Trademark + proto[0][8] || 'Unknown', // 8.Manufacturer + proto[0][9] || 'Unknown' // 9.Designer + ]; - ToUnicodeMap.prototype = { - get length() { - return this._map.length; - }, + // Mac want 1-byte per character strings while Windows want + // 2-bytes per character, so duplicate the names table + var stringsUnicode = []; + var i, ii, j, jj, str; + for (i = 0, ii = strings.length; i < ii; i++) { + str = proto[1][i] || strings[i]; - forEach: function(callback) { - for (var charCode in this._map) { - callback(charCode, this._map[charCode].charCodeAt(0)); + var strBufUnicode = []; + for (j = 0, jj = str.length; j < jj; j++) { + strBufUnicode.push(string16(str.charCodeAt(j))); } - }, + stringsUnicode.push(strBufUnicode.join('')); + } - has: function(i) { - return this._map[i] !== undefined; - }, + var names = [strings, stringsUnicode]; + var platforms = ['\x00\x01', '\x00\x03']; + var encodings = ['\x00\x00', '\x00\x01']; + var languages = ['\x00\x00', '\x04\x09']; - get: function(i) { - return this._map[i]; - }, + var namesRecordCount = strings.length * platforms.length; + var nameTable = + '\x00\x00' + // format + string16(namesRecordCount) + // Number of names Record + string16(namesRecordCount * 12 + 6); // Storage - charCodeOf: function(v) { - return this._map.indexOf(v); + // Build the name records field + var strOffset = 0; + for (i = 0, ii = platforms.length; i < ii; i++) { + var strs = names[i]; + for (j = 0, jj = strs.length; j < jj; j++) { + str = strs[j]; + var nameRecord = + platforms[i] + // platform ID + encodings[i] + // encoding ID + languages[i] + // language ID + string16(j) + // name ID + string16(str.length) + + string16(strOffset); + nameTable += nameRecord; + strOffset += str.length; + } } - }; - return ToUnicodeMap; -})(); - -var IdentityToUnicodeMap = (function IdentityToUnicodeMapClosure() { - function IdentityToUnicodeMap(firstChar, lastChar) { - this.firstChar = firstChar; - this.lastChar = lastChar; + nameTable += strings.join('') + stringsUnicode.join(''); + return nameTable; } - IdentityToUnicodeMap.prototype = { - get length() { - return (this.lastChar + 1) - this.firstChar; - }, - - forEach: function (callback) { - for (var i = this.firstChar, ii = this.lastChar; i <= ii; i++) { - callback(i, i); - } - }, - - has: function (i) { - return this.firstChar <= i && i <= this.lastChar; + Font.prototype = { + name: null, + font: null, + mimetype: null, + encoding: null, + get renderer() { + var renderer = FontRendererFactory.create(this); + return shadow(this, 'renderer', renderer); }, - get: function (i) { - if (this.firstChar <= i && i <= this.lastChar) { - return String.fromCharCode(i); + exportData: function Font_exportData() { + // TODO remove enumerating of the properties, e.g. hardcode exact names. + var data = {}; + for (var i in this) { + if (this.hasOwnProperty(i)) { + data[i] = this[i]; + } } - return undefined; + return data; }, - charCodeOf: function (v) { - return (isInt(v) && v >= this.firstChar && v <= this.lastChar) ? v : -1; - } - }; + checkAndRepair: function Font_checkAndRepair(name, font, properties) { + function readTableEntry(file) { + var tag = bytesToString(file.getBytes(4)); - return IdentityToUnicodeMap; -})(); + var checksum = file.getInt32(); + var offset = file.getInt32() >>> 0; + var length = file.getInt32() >>> 0; -var OpenTypeFileBuilder = (function OpenTypeFileBuilderClosure() { - function writeInt16(dest, offset, num) { - dest[offset] = (num >> 8) & 0xFF; - dest[offset + 1] = num & 0xFF; - } + // Read the table associated data + var previousPosition = file.pos; + file.pos = file.start ? file.start : 0; + file.skip(offset); + var data = file.getBytes(length); + file.pos = previousPosition; - function writeInt32(dest, offset, num) { - dest[offset] = (num >> 24) & 0xFF; - dest[offset + 1] = (num >> 16) & 0xFF; - dest[offset + 2] = (num >> 8) & 0xFF; - dest[offset + 3] = num & 0xFF; - } + if (tag === 'head') { + // clearing checksum adjustment + data[8] = data[9] = data[10] = data[11] = 0; + data[17] |= 0x20; //Set font optimized for cleartype flag + } - function writeData(dest, offset, data) { - var i, ii; - if (data instanceof Uint8Array) { - dest.set(data, offset); - } else if (typeof data === 'string') { - for (i = 0, ii = data.length; i < ii; i++) { - dest[offset++] = data.charCodeAt(i) & 0xFF; - } - } else { - // treating everything else as array - for (i = 0, ii = data.length; i < ii; i++) { - dest[offset++] = data[i] & 0xFF; + return { + tag: tag, + checksum: checksum, + length: length, + offset: offset, + data: data + }; } - } - } - function OpenTypeFileBuilder(sfnt) { - this.sfnt = sfnt; - this.tables = Object.create(null); - } + function readOpenTypeHeader(ttf) { + return { + version: bytesToString(ttf.getBytes(4)), + numTables: ttf.getUint16(), + searchRange: ttf.getUint16(), + entrySelector: ttf.getUint16(), + rangeShift: ttf.getUint16() + }; + } - OpenTypeFileBuilder.getSearchParams = - function OpenTypeFileBuilder_getSearchParams(entriesCount, entrySize) { - var maxPower2 = 1, log2 = 0; - while ((maxPower2 ^ entriesCount) > maxPower2) { - maxPower2 <<= 1; - log2++; - } - var searchRange = maxPower2 * entrySize; - return { - range: searchRange, - entry: log2, - rangeShift: entrySize * entriesCount - searchRange - }; - }; + /** + * Read the appropriate subtable from the cmap according to 9.6.6.4 from + * PDF spec + */ + function readCmapTable(cmap, font, isSymbolicFont, hasEncoding) { + if (!cmap) { + warn('No cmap table available.'); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } + var segment; + var start = (font.start ? font.start : 0) + cmap.offset; + font.pos = start; - var OTF_HEADER_SIZE = 12; - var OTF_TABLE_ENTRY_SIZE = 16; + var version = font.getUint16(); + var numTables = font.getUint16(); - OpenTypeFileBuilder.prototype = { - toArray: function OpenTypeFileBuilder_toArray() { - var sfnt = this.sfnt; + var potentialTable; + var canBreak = false; + // There's an order of preference in terms of which cmap subtable to + // use: + // - non-symbolic fonts the preference is a 3,1 table then a 1,0 table + // - symbolic fonts the preference is a 3,0 table then a 1,0 table + // The following takes advantage of the fact that the tables are sorted + // to work. + for (var i = 0; i < numTables; i++) { + var platformId = font.getUint16(); + var encodingId = font.getUint16(); + var offset = font.getInt32() >>> 0; + var useTable = false; - // Tables needs to be written by ascendant alphabetic order - var tables = this.tables; - var tablesNames = Object.keys(tables); - tablesNames.sort(); - var numTables = tablesNames.length; + if (platformId === 0 && encodingId === 0) { + useTable = true; + // Continue the loop since there still may be a higher priority + // table. + } else if (platformId === 1 && encodingId === 0) { + useTable = true; + // Continue the loop since there still may be a higher priority + // table. + } else if (platformId === 3 && encodingId === 1 && + ((!isSymbolicFont && hasEncoding) || !potentialTable)) { + useTable = true; + if (!isSymbolicFont) { + canBreak = true; + } + } else if (isSymbolicFont && platformId === 3 && encodingId === 0) { + useTable = true; + canBreak = true; + } - var i, j, jj, table, tableName; - // layout the tables data - var offset = OTF_HEADER_SIZE + numTables * OTF_TABLE_ENTRY_SIZE; - var tableOffsets = [offset]; - for (i = 0; i < numTables; i++) { - table = tables[tablesNames[i]]; - var paddedLength = ((table.length + 3) & ~3) >>> 0; - offset += paddedLength; - tableOffsets.push(offset); - } - - var file = new Uint8Array(offset); - // write the table data first (mostly for checksum) - for (i = 0; i < numTables; i++) { - table = tables[tablesNames[i]]; - writeData(file, tableOffsets[i], table); - } - - // sfnt version (4 bytes) - if (sfnt === 'true') { - // Windows hates the Mac TrueType sfnt version number - sfnt = string32(0x00010000); - } - file[0] = sfnt.charCodeAt(0) & 0xFF; - file[1] = sfnt.charCodeAt(1) & 0xFF; - file[2] = sfnt.charCodeAt(2) & 0xFF; - file[3] = sfnt.charCodeAt(3) & 0xFF; - - // numTables (2 bytes) - writeInt16(file, 4, numTables); - - var searchParams = OpenTypeFileBuilder.getSearchParams(numTables, 16); - - // searchRange (2 bytes) - writeInt16(file, 6, searchParams.range); - // entrySelector (2 bytes) - writeInt16(file, 8, searchParams.entry); - // rangeShift (2 bytes) - writeInt16(file, 10, searchParams.rangeShift); - - offset = OTF_HEADER_SIZE; - // writing table entries - for (i = 0; i < numTables; i++) { - tableName = tablesNames[i]; - file[offset] = tableName.charCodeAt(0) & 0xFF; - file[offset + 1] = tableName.charCodeAt(1) & 0xFF; - file[offset + 2] = tableName.charCodeAt(2) & 0xFF; - file[offset + 3] = tableName.charCodeAt(3) & 0xFF; - - // checksum - var checksum = 0; - for (j = tableOffsets[i], jj = tableOffsets[i + 1]; j < jj; j += 4) { - var quad = (file[j] << 24) + (file[j + 1] << 16) + - (file[j + 2] << 8) + file[j + 3]; - checksum = (checksum + quad) | 0; + if (useTable) { + potentialTable = { + platformId: platformId, + encodingId: encodingId, + offset: offset + }; + } + if (canBreak) { + break; + } } - writeInt32(file, offset + 4, checksum); - // offset - writeInt32(file, offset + 8, tableOffsets[i]); - // length - writeInt32(file, offset + 12, tables[tableName].length); + if (potentialTable) { + font.pos = start + potentialTable.offset; + } + if (!potentialTable || font.peekByte() === -1) { + warn('Could not find a preferred cmap table.'); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } - offset += OTF_TABLE_ENTRY_SIZE; - } - return file; - }, + var format = font.getUint16(); + var length = font.getUint16(); + var language = font.getUint16(); - addTable: function OpenTypeFileBuilder_addTable(tag, data) { - if (tag in this.tables) { - throw new Error('Table ' + tag + ' already exists'); - } - this.tables[tag] = data; - } - }; + var hasShortCmap = false; + var mappings = []; + var j, glyphId; - return OpenTypeFileBuilder; -})(); + // TODO(mack): refactor this cmap subtable reading logic out + if (format === 0) { + for (j = 0; j < 256; j++) { + var index = font.getByte(); + if (!index) { + continue; + } + mappings.push({ + charCode: j, + glyphId: index + }); + } + hasShortCmap = true; + } else if (format === 4) { + // re-creating the table in format 4 since the encoding + // might be changed + var segCount = (font.getUint16() >> 1); + font.getBytes(6); // skipping range fields + var segIndex, segments = []; + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments.push({ end: font.getUint16() }); + } + font.getUint16(); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments[segIndex].start = font.getUint16(); + } -// Problematic Unicode characters in the fonts that needs to be moved to avoid -// issues when they are painted on the canvas, e.g. complex-script shaping or -// control/whitespace characters. The ranges are listed in pairs: the first item -// is a code of the first problematic code, the second one is the next -// non-problematic code. The ranges must be in sorted order. -var ProblematicCharRanges = new Int32Array([ - // Control characters. - 0x0000, 0x0020, - 0x007F, 0x00A1, - 0x00AD, 0x00AE, - // Chars that is used in complex-script shaping. - 0x0600, 0x0780, - 0x08A0, 0x10A0, - 0x1780, 0x1800, - // General punctuation chars. - 0x2000, 0x2010, - 0x2011, 0x2012, - 0x2028, 0x2030, - 0x205F, 0x2070, - 0x25CC, 0x25CD, - // Chars that is used in complex-script shaping. - 0xAA60, 0xAA80, - // Specials Unicode block. - 0xFFF0, 0x10000 -]); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments[segIndex].delta = font.getUint16(); + } -/** - * 'Font' is the class the outside world should use, it encapsulate all the font - * decoding logics whatever type it is (assuming the font type is supported). - * - * For example to read a Type1 font and to attach it to the document: - * var type1Font = new Font("MyFontName", binaryFile, propertiesObject); - * type1Font.bind(); - */ -var Font = (function FontClosure() { - function Font(name, file, properties) { - var charCode, glyphName, fontChar; + var offsetsCount = 0; + for (segIndex = 0; segIndex < segCount; segIndex++) { + segment = segments[segIndex]; + var rangeOffset = font.getUint16(); + if (!rangeOffset) { + segment.offsetIndex = -1; + continue; + } - this.name = name; - this.loadedName = properties.loadedName; - this.isType3Font = properties.isType3Font; - this.sizes = []; + var offsetIndex = (rangeOffset >> 1) - (segCount - segIndex); + segment.offsetIndex = offsetIndex; + offsetsCount = Math.max(offsetsCount, offsetIndex + + segment.end - segment.start + 1); + } - this.glyphCache = Object.create(null); + var offsets = []; + for (j = 0; j < offsetsCount; j++) { + offsets.push(font.getUint16()); + } - var names = name.split('+'); - names = names.length > 1 ? names[1] : names[0]; - names = names.split(/[-,_]/g)[0]; - this.isSerifFont = !!(properties.flags & FontFlags.Serif); - this.isSymbolicFont = !!(properties.flags & FontFlags.Symbolic); - this.isMonospace = !!(properties.flags & FontFlags.FixedPitch); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segment = segments[segIndex]; + start = segment.start; + var end = segment.end; + var delta = segment.delta; + offsetIndex = segment.offsetIndex; - var type = properties.type; - var subtype = properties.subtype; - this.type = type; + for (j = start; j <= end; j++) { + if (j === 0xFFFF) { + continue; + } - this.fallbackName = (this.isMonospace ? 'monospace' : - (this.isSerifFont ? 'serif' : 'sans-serif')); + glyphId = (offsetIndex < 0 ? + j : offsets[offsetIndex + j - start]); + glyphId = (glyphId + delta) & 0xFFFF; + if (glyphId === 0) { + continue; + } + mappings.push({ + charCode: j, + glyphId: glyphId + }); + } + } + } else if (format === 6) { + // Format 6 is a 2-bytes dense mapping, which means the font data + // lives glue together even if they are pretty far in the unicode + // table. (This looks weird, so I can have missed something), this + // works on Linux but seems to fails on Mac so let's rewrite the + // cmap table to a 3-1-4 style + var firstCode = font.getUint16(); + var entryCount = font.getUint16(); - this.differences = properties.differences; - this.widths = properties.widths; - this.defaultWidth = properties.defaultWidth; - this.composite = properties.composite; - this.wideChars = properties.wideChars; - this.cMap = properties.cMap; - this.ascent = properties.ascent / PDF_GLYPH_SPACE_UNITS; - this.descent = properties.descent / PDF_GLYPH_SPACE_UNITS; - this.fontMatrix = properties.fontMatrix; - this.bbox = properties.bbox; + for (j = 0; j < entryCount; j++) { + glyphId = font.getUint16(); + var charCode = firstCode + j; - this.toUnicode = properties.toUnicode = this.buildToUnicode(properties); + mappings.push({ + charCode: charCode, + glyphId: glyphId + }); + } + } else { + warn('cmap table has unsupported format: ' + format); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } - this.toFontChar = []; + // removing duplicate entries + mappings.sort(function (a, b) { + return a.charCode - b.charCode; + }); + for (i = 1; i < mappings.length; i++) { + if (mappings[i - 1].charCode === mappings[i].charCode) { + mappings.splice(i, 1); + i--; + } + } - if (properties.type === 'Type3') { - for (charCode = 0; charCode < 256; charCode++) { - this.toFontChar[charCode] = (this.differences[charCode] || - properties.defaultEncoding[charCode]); + return { + platformId: potentialTable.platformId, + encodingId: potentialTable.encodingId, + mappings: mappings, + hasShortCmap: hasShortCmap + }; } - this.fontType = FontType.TYPE3; - return; - } - this.cidEncoding = properties.cidEncoding; - this.vertical = properties.vertical; - if (this.vertical) { - this.vmetrics = properties.vmetrics; - this.defaultVMetrics = properties.defaultVMetrics; - } - var glyphsUnicodeMap; - if (!file || file.isEmpty) { - if (file) { - // Some bad PDF generators will include empty font files, - // attempting to recover by assuming that no file exists. - warn('Font file is empty in "' + name + '" (' + this.loadedName + ')'); - } + function sanitizeMetrics(font, header, metrics, numGlyphs) { + if (!header) { + if (metrics) { + metrics.data = null; + } + return; + } - this.missingFile = true; - // The file data is not specified. Trying to fix the font name - // to be used with the canvas.font. - var fontName = name.replace(/[,_]/g, '-'); - var stdFontMap = getStdFontMap(), nonStdFontMap = getNonStdFontMap(); - var isStandardFont = !!stdFontMap[fontName] || - !!(nonStdFontMap[fontName] && stdFontMap[nonStdFontMap[fontName]]); - fontName = stdFontMap[fontName] || nonStdFontMap[fontName] || fontName; + font.pos = (font.start ? font.start : 0) + header.offset; + font.pos += header.length - 2; + var numOfMetrics = font.getUint16(); - this.bold = (fontName.search(/bold/gi) !== -1); - this.italic = ((fontName.search(/oblique/gi) !== -1) || - (fontName.search(/italic/gi) !== -1)); + if (numOfMetrics > numGlyphs) { + info('The numOfMetrics (' + numOfMetrics + ') should not be ' + + 'greater than the numGlyphs (' + numGlyphs + ')'); + // Reduce numOfMetrics if it is greater than numGlyphs + numOfMetrics = numGlyphs; + header.data[34] = (numOfMetrics & 0xff00) >> 8; + header.data[35] = numOfMetrics & 0x00ff; + } - // Use 'name' instead of 'fontName' here because the original - // name ArialBlack for example will be replaced by Helvetica. - this.black = (name.search(/Black/g) !== -1); + var numOfSidebearings = numGlyphs - numOfMetrics; + var numMissing = numOfSidebearings - + ((metrics.length - numOfMetrics * 4) >> 1); - // if at least one width is present, remeasure all chars when exists - this.remeasure = Object.keys(this.widths).length > 0; - if (isStandardFont && type === 'CIDFontType2' && - properties.cidEncoding.indexOf('Identity-') === 0) { - var GlyphMapForStandardFonts = getGlyphMapForStandardFonts(); - // Standard fonts might be embedded as CID font without glyph mapping. - // Building one based on GlyphMapForStandardFonts. - var map = []; - for (charCode in GlyphMapForStandardFonts) { - map[+charCode] = GlyphMapForStandardFonts[charCode]; + if (numMissing > 0) { + // For each missing glyph, we set both the width and lsb to 0 (zero). + // Since we need to add two properties for each glyph, this explains + // the use of |numMissing * 2| when initializing the typed array. + var entries = new Uint8Array(metrics.length + numMissing * 2); + entries.set(metrics.data); + metrics.data = entries; } - if (/ArialBlack/i.test(name)) { - var SupplementalGlyphMapForArialBlack = - getSupplementalGlyphMapForArialBlack(); - for (charCode in SupplementalGlyphMapForArialBlack) { - map[+charCode] = SupplementalGlyphMapForArialBlack[charCode]; - } + } + + function sanitizeGlyph(source, sourceStart, sourceEnd, dest, destStart, + hintsValid) { + if (sourceEnd - sourceStart <= 12) { + // glyph with data less than 12 is invalid one + return 0; } - var isIdentityUnicode = this.toUnicode instanceof IdentityToUnicodeMap; - if (!isIdentityUnicode) { - this.toUnicode.forEach(function(charCode, unicodeCharCode) { - map[+charCode] = unicodeCharCode; - }); + var glyf = source.subarray(sourceStart, sourceEnd); + var contoursCount = (glyf[0] << 8) | glyf[1]; + if (contoursCount & 0x8000) { + // complex glyph, writing as is + dest.set(glyf, destStart); + return glyf.length; } - this.toFontChar = map; - this.toUnicode = new ToUnicodeMap(map); - } else if (/Symbol/i.test(fontName)) { - var symbols = SymbolSetEncoding; - glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode in symbols) { - fontChar = glyphsUnicodeMap[symbols[charCode]]; - if (!fontChar) { - continue; - } - this.toFontChar[charCode] = fontChar; + + var i, j = 10, flagsCount = 0; + for (i = 0; i < contoursCount; i++) { + var endPoint = (glyf[j] << 8) | glyf[j + 1]; + flagsCount = endPoint + 1; + j += 2; } - for (charCode in properties.differences) { - fontChar = glyphsUnicodeMap[properties.differences[charCode]]; - if (!fontChar) { - continue; + // skipping instructions + var instructionsStart = j; + var instructionsLength = (glyf[j] << 8) | glyf[j + 1]; + j += 2 + instructionsLength; + var instructionsEnd = j; + // validating flags + var coordinatesLength = 0; + for (i = 0; i < flagsCount; i++) { + var flag = glyf[j++]; + if (flag & 0xC0) { + // reserved flags must be zero, cleaning up + glyf[j - 1] = flag & 0x3F; + } + var xyLength = ((flag & 2) ? 1 : (flag & 16) ? 0 : 2) + + ((flag & 4) ? 1 : (flag & 32) ? 0 : 2); + coordinatesLength += xyLength; + if (flag & 8) { + var repeat = glyf[j++]; + i += repeat; + coordinatesLength += repeat * xyLength; } - this.toFontChar[charCode] = fontChar; } - } else if (/Dingbats/i.test(fontName)) { - glyphsUnicodeMap = getDingbatsGlyphsUnicode(); - if (/Wingdings/i.test(name)) { - warn('Wingdings font without embedded font file, ' + - 'falling back to the ZapfDingbats encoding.'); + // glyph without coordinates will be rejected + if (coordinatesLength === 0) { + return 0; } - var dingbats = ZapfDingbatsEncoding; - for (charCode in dingbats) { - fontChar = glyphsUnicodeMap[dingbats[charCode]]; - if (!fontChar) { - continue; - } - this.toFontChar[charCode] = fontChar; + var glyphDataLength = j + coordinatesLength; + if (glyphDataLength > glyf.length) { + // not enough data for coordinates + return 0; } - for (charCode in properties.differences) { - fontChar = glyphsUnicodeMap[properties.differences[charCode]]; - if (!fontChar) { - continue; + if (!hintsValid && instructionsLength > 0) { + dest.set(glyf.subarray(0, instructionsStart), destStart); + dest.set([0, 0], destStart + instructionsStart); + dest.set(glyf.subarray(instructionsEnd, glyphDataLength), + destStart + instructionsStart + 2); + glyphDataLength -= instructionsLength; + if (glyf.length - glyphDataLength > 3) { + glyphDataLength = (glyphDataLength + 3) & ~3; } - this.toFontChar[charCode] = fontChar; + return glyphDataLength; } - } else if (isStandardFont) { - this.toFontChar = []; - glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode in properties.defaultEncoding) { - glyphName = (properties.differences[charCode] || - properties.defaultEncoding[charCode]); - this.toFontChar[charCode] = glyphsUnicodeMap[glyphName]; + if (glyf.length - glyphDataLength > 3) { + // truncating and aligning to 4 bytes the long glyph data + glyphDataLength = (glyphDataLength + 3) & ~3; + dest.set(glyf.subarray(0, glyphDataLength), destStart); + return glyphDataLength; } - } else { - var unicodeCharCode, notCidFont = (type.indexOf('CIDFontType') === -1); - glyphsUnicodeMap = getGlyphsUnicode(); - this.toUnicode.forEach(function(charCode, unicodeCharCode) { - if (notCidFont) { - glyphName = (properties.differences[charCode] || - properties.defaultEncoding[charCode]); - unicodeCharCode = (glyphsUnicodeMap[glyphName] || unicodeCharCode); - } - this.toFontChar[charCode] = unicodeCharCode; - }.bind(this)); + // glyph data is fine + dest.set(glyf, destStart); + return glyf.length; } - this.loadedName = fontName.split('-')[0]; - this.loading = false; - this.fontType = getFontType(type, subtype); - return; - } - // Some fonts might use wrong font types for Type1C or CIDFontType0C - if (subtype === 'Type1C' && (type !== 'Type1' && type !== 'MMType1')) { - // Some TrueType fonts by mistake claim Type1C - if (isTrueTypeFile(file)) { - subtype = 'TrueType'; - } else { - type = 'Type1'; - } - } - if (subtype === 'CIDFontType0C' && type !== 'CIDFontType0') { - type = 'CIDFontType0'; - } - if (subtype === 'OpenType') { - type = 'OpenType'; - } - // Some CIDFontType0C fonts by mistake claim CIDFontType0. - if (type === 'CIDFontType0') { - if (isType1File(file)) { - subtype = 'CIDFontType0'; - } else if (isOpenTypeFile(file)) { - // Sometimes the type/subtype can be a complete lie (see issue6782.pdf). - type = subtype = 'OpenType'; - } else { - subtype = 'CIDFontType0C'; - } - } + function sanitizeHead(head, numGlyphs, locaLength) { + var data = head.data; - var data; - switch (type) { - case 'MMType1': - info('MMType1 font (' + name + '), falling back to Type1.'); - /* falls through */ - case 'Type1': - case 'CIDFontType0': - this.mimetype = 'font/opentype'; - - var cff = (subtype === 'Type1C' || subtype === 'CIDFontType0C') ? - new CFFFont(file, properties) : new Type1Font(name, file, properties); - - adjustWidths(properties); - - // Wrap the CFF data inside an OTF font file - data = this.convert(name, cff, properties); - break; - - case 'OpenType': - case 'TrueType': - case 'CIDFontType2': - this.mimetype = 'font/opentype'; - - // Repair the TrueType file. It is can be damaged in the point of - // view of the sanitizer - data = this.checkAndRepair(name, file, properties); - if (this.isOpenType) { - adjustWidths(properties); - - type = 'OpenType'; + // Validate version: + // Should always be 0x00010000 + var version = int32(data[0], data[1], data[2], data[3]); + if (version >> 16 !== 1) { + info('Attempting to fix invalid version in head table: ' + version); + data[0] = 0; + data[1] = 1; + data[2] = 0; + data[3] = 0; } - break; - - default: - error('Font ' + type + ' is not supported'); - break; - } - - this.data = data; - this.fontType = getFontType(type, subtype); - - // Transfer some properties again that could change during font conversion - this.fontMatrix = properties.fontMatrix; - this.widths = properties.widths; - this.defaultWidth = properties.defaultWidth; - this.encoding = properties.baseEncoding; - this.seacMap = properties.seacMap; - - this.loading = true; - } - - Font.getFontID = (function () { - var ID = 1; - return function Font_getFontID() { - return String(ID++); - }; - })(); - - function int16(b0, b1) { - return (b0 << 8) + b1; - } - - function int32(b0, b1, b2, b3) { - return (b0 << 24) + (b1 << 16) + (b2 << 8) + b3; - } - - function string16(value) { - return String.fromCharCode((value >> 8) & 0xff, value & 0xff); - } - - function safeString16(value) { - // clamp value to the 16-bit int range - value = (value > 0x7FFF ? 0x7FFF : (value < -0x8000 ? -0x8000 : value)); - return String.fromCharCode((value >> 8) & 0xff, value & 0xff); - } - - function isTrueTypeFile(file) { - var header = file.peekBytes(4); - return readUint32(header, 0) === 0x00010000; - } - function isOpenTypeFile(file) { - var header = file.peekBytes(4); - return bytesToString(header) === 'OTTO'; - } - - function isType1File(file) { - var header = file.peekBytes(2); - // All Type1 font programs must begin with the comment '%!' (0x25 + 0x21). - if (header[0] === 0x25 && header[1] === 0x21) { - return true; - } - // ... obviously some fonts violate that part of the specification, - // please refer to the comment in |Type1Font| below. - if (header[0] === 0x80 && header[1] === 0x01) { // pfb file header. - return true; - } - return false; - } + var indexToLocFormat = int16(data[50], data[51]); + if (indexToLocFormat < 0 || indexToLocFormat > 1) { + info('Attempting to fix invalid indexToLocFormat in head table: ' + + indexToLocFormat); - /** - * Helper function for |adjustMapping|. - * @return {boolean} - */ - function isProblematicUnicodeLocation(code) { - // Using binary search to find a range start. - var i = 0, j = ProblematicCharRanges.length - 1; - while (i < j) { - var c = (i + j + 1) >> 1; - if (code < ProblematicCharRanges[c]) { - j = c - 1; - } else { - i = c; - } - } - // Even index means code in problematic range. - return !(i & 1); - } + // The value of indexToLocFormat should be 0 if the loca table + // consists of short offsets, and should be 1 if the loca table + // consists of long offsets. + // + // The number of entries in the loca table should be numGlyphs + 1. + // + // Using this information, we can work backwards to deduce if the + // size of each offset in the loca table, and thus figure out the + // appropriate value for indexToLocFormat. - /** - * Rebuilds the char code to glyph ID map by trying to replace the char codes - * with their unicode value. It also moves char codes that are in known - * problematic locations. - * @return {Object} Two properties: - * 'toFontChar' - maps original char codes(the value that will be read - * from commands such as show text) to the char codes that will be used in the - * font that we build - * 'charCodeToGlyphId' - maps the new font char codes to glyph ids - */ - function adjustMapping(charCodeToGlyphId, properties) { - var toUnicode = properties.toUnicode; - var isSymbolic = !!(properties.flags & FontFlags.Symbolic); - var isIdentityUnicode = - properties.toUnicode instanceof IdentityToUnicodeMap; - var newMap = Object.create(null); - var toFontChar = []; - var usedFontCharCodes = []; - var nextAvailableFontCharCode = PRIVATE_USE_OFFSET_START; - for (var originalCharCode in charCodeToGlyphId) { - originalCharCode |= 0; - var glyphId = charCodeToGlyphId[originalCharCode]; - var fontCharCode = originalCharCode; - // First try to map the value to a unicode position if a non identity map - // was created. - if (!isIdentityUnicode && toUnicode.has(originalCharCode)) { - var unicode = toUnicode.get(fontCharCode); - // TODO: Try to map ligatures to the correct spot. - if (unicode.length === 1) { - fontCharCode = unicode.charCodeAt(0); + var numGlyphsPlusOne = numGlyphs + 1; + if (locaLength === numGlyphsPlusOne << 1) { + // 0x0000 indicates the loca table consists of short offsets + data[50] = 0; + data[51] = 0; + } else if (locaLength === numGlyphsPlusOne << 2) { + // 0x0001 indicates the loca table consists of long offsets + data[50] = 0; + data[51] = 1; + } else { + warn('Could not fix indexToLocFormat: ' + indexToLocFormat); + } } } - // Try to move control characters, special characters and already mapped - // characters to the private use area since they will not be drawn by - // canvas if left in their current position. Also, move characters if the - // font was symbolic and there is only an identity unicode map since the - // characters probably aren't in the correct position (fixes an issue - // with firefox and thuluthfont). - if ((usedFontCharCodes[fontCharCode] !== undefined || - isProblematicUnicodeLocation(fontCharCode) || - (isSymbolic && isIdentityUnicode)) && - nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END) { // Room left. - // Loop to try and find a free spot in the private use area. - do { - fontCharCode = nextAvailableFontCharCode++; - if (SKIP_PRIVATE_USE_RANGE_F000_TO_F01F && fontCharCode === 0xF000) { - fontCharCode = 0xF020; - nextAvailableFontCharCode = fontCharCode + 1; + function sanitizeGlyphLocations(loca, glyf, numGlyphs, + isGlyphLocationsLong, hintsValid, + dupFirstEntry) { + var itemSize, itemDecode, itemEncode; + if (isGlyphLocationsLong) { + itemSize = 4; + itemDecode = function fontItemDecodeLong(data, offset) { + return (data[offset] << 24) | (data[offset + 1] << 16) | + (data[offset + 2] << 8) | data[offset + 3]; + }; + itemEncode = function fontItemEncodeLong(data, offset, value) { + data[offset] = (value >>> 24) & 0xFF; + data[offset + 1] = (value >> 16) & 0xFF; + data[offset + 2] = (value >> 8) & 0xFF; + data[offset + 3] = value & 0xFF; + }; + } else { + itemSize = 2; + itemDecode = function fontItemDecode(data, offset) { + return (data[offset] << 9) | (data[offset + 1] << 1); + }; + itemEncode = function fontItemEncode(data, offset, value) { + data[offset] = (value >> 9) & 0xFF; + data[offset + 1] = (value >> 1) & 0xFF; + }; + } + var locaData = loca.data; + var locaDataSize = itemSize * (1 + numGlyphs); + // is loca.data too short or long? + if (locaData.length !== locaDataSize) { + locaData = new Uint8Array(locaDataSize); + locaData.set(loca.data.subarray(0, locaDataSize)); + loca.data = locaData; + } + // removing the invalid glyphs + var oldGlyfData = glyf.data; + var oldGlyfDataLength = oldGlyfData.length; + var newGlyfData = new Uint8Array(oldGlyfDataLength); + var startOffset = itemDecode(locaData, 0); + var writeOffset = 0; + var missingGlyphData = Object.create(null); + itemEncode(locaData, 0, writeOffset); + var i, j; + for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { + var endOffset = itemDecode(locaData, j); + if (endOffset > oldGlyfDataLength && + ((oldGlyfDataLength + 3) & ~3) === endOffset) { + // Aspose breaks fonts by aligning the glyphs to the qword, but not + // the glyf table size, which makes last glyph out of range. + endOffset = oldGlyfDataLength; + } + if (endOffset > oldGlyfDataLength) { + // glyph end offset points outside glyf data, rejecting the glyph + itemEncode(locaData, j, writeOffset); + startOffset = endOffset; + continue; } - } while (usedFontCharCodes[fontCharCode] !== undefined && - nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END); - } + if (startOffset === endOffset) { + missingGlyphData[i] = true; + } - newMap[fontCharCode] = glyphId; - toFontChar[originalCharCode] = fontCharCode; - usedFontCharCodes[fontCharCode] = true; - } - return { - toFontChar: toFontChar, - charCodeToGlyphId: newMap, - nextAvailableFontCharCode: nextAvailableFontCharCode - }; - } + var newLength = sanitizeGlyph(oldGlyfData, startOffset, endOffset, + newGlyfData, writeOffset, hintsValid); + writeOffset += newLength; + itemEncode(locaData, j, writeOffset); + startOffset = endOffset; + } - function getRanges(glyphs, numGlyphs) { - // Array.sort() sorts by characters, not numerically, so convert to an - // array of characters. - var codes = []; - for (var charCode in glyphs) { - // Remove an invalid glyph ID mappings to make OTS happy. - if (glyphs[charCode] >= numGlyphs) { - continue; - } - codes.push({ fontCharCode: charCode | 0, glyphId: glyphs[charCode] }); - } - codes.sort(function fontGetRangesSort(a, b) { - return a.fontCharCode - b.fontCharCode; - }); + if (writeOffset === 0) { + // glyf table cannot be empty -- redoing the glyf and loca tables + // to have single glyph with one point + var simpleGlyph = new Uint8Array( + [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 49, 0]); + for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { + itemEncode(locaData, j, simpleGlyph.length); + } + glyf.data = simpleGlyph; + return missingGlyphData; + } - // Split the sorted codes into ranges. - var ranges = []; - var length = codes.length; - for (var n = 0; n < length; ) { - var start = codes[n].fontCharCode; - var codeIndices = [codes[n].glyphId]; - ++n; - var end = start; - while (n < length && end + 1 === codes[n].fontCharCode) { - codeIndices.push(codes[n].glyphId); - ++end; - ++n; - if (end === 0xFFFF) { - break; + if (dupFirstEntry) { + var firstEntryLength = itemDecode(locaData, itemSize); + if (newGlyfData.length > firstEntryLength + writeOffset) { + glyf.data = newGlyfData.subarray(0, firstEntryLength + writeOffset); + } else { + glyf.data = new Uint8Array(firstEntryLength + writeOffset); + glyf.data.set(newGlyfData.subarray(0, writeOffset)); + } + glyf.data.set(newGlyfData.subarray(0, firstEntryLength), writeOffset); + itemEncode(loca.data, locaData.length - itemSize, + writeOffset + firstEntryLength); + } else { + glyf.data = newGlyfData.subarray(0, writeOffset); } + return missingGlyphData; } - ranges.push([start, end, codeIndices]); - } - return ranges; - } - - function createCmapTable(glyphs, numGlyphs) { - var ranges = getRanges(glyphs, numGlyphs); - var numTables = ranges[ranges.length - 1][1] > 0xFFFF ? 2 : 1; - var cmap = '\x00\x00' + // version - string16(numTables) + // numTables - '\x00\x03' + // platformID - '\x00\x01' + // encodingID - string32(4 + numTables * 8); // start of the table record - - var i, ii, j, jj; - for (i = ranges.length - 1; i >= 0; --i) { - if (ranges[i][0] <= 0xFFFF) { break; } - } - var bmpLength = i + 1; + function readPostScriptTable(post, properties, maxpNumGlyphs) { + var start = (font.start ? font.start : 0) + post.offset; + font.pos = start; - if (ranges[i][0] < 0xFFFF && ranges[i][1] === 0xFFFF) { - ranges[i][1] = 0xFFFE; - } - var trailingRangesCount = ranges[i][1] < 0xFFFF ? 1 : 0; - var segCount = bmpLength + trailingRangesCount; - var searchParams = OpenTypeFileBuilder.getSearchParams(segCount, 2); + var length = post.length, end = start + length; + var version = font.getInt32(); + // skip rest to the tables + font.getBytes(28); - // Fill up the 4 parallel arrays describing the segments. - var startCount = ''; - var endCount = ''; - var idDeltas = ''; - var idRangeOffsets = ''; - var glyphsIds = ''; - var bias = 0; + var glyphNames; + var valid = true; + var i; - var range, start, end, codes; - for (i = 0, ii = bmpLength; i < ii; i++) { - range = ranges[i]; - start = range[0]; - end = range[1]; - startCount += string16(start); - endCount += string16(end); - codes = range[2]; - var contiguous = true; - for (j = 1, jj = codes.length; j < jj; ++j) { - if (codes[j] !== codes[j - 1] + 1) { - contiguous = false; - break; + switch (version) { + case 0x00010000: + glyphNames = MacStandardGlyphOrdering; + break; + case 0x00020000: + var numGlyphs = font.getUint16(); + if (numGlyphs !== maxpNumGlyphs) { + valid = false; + break; + } + var glyphNameIndexes = []; + for (i = 0; i < numGlyphs; ++i) { + var index = font.getUint16(); + if (index >= 32768) { + valid = false; + break; + } + glyphNameIndexes.push(index); + } + if (!valid) { + break; + } + var customNames = []; + var strBuf = []; + while (font.pos < end) { + var stringLength = font.getByte(); + strBuf.length = stringLength; + for (i = 0; i < stringLength; ++i) { + strBuf[i] = String.fromCharCode(font.getByte()); + } + customNames.push(strBuf.join('')); + } + glyphNames = []; + for (i = 0; i < numGlyphs; ++i) { + var j = glyphNameIndexes[i]; + if (j < 258) { + glyphNames.push(MacStandardGlyphOrdering[j]); + continue; + } + glyphNames.push(customNames[j - 258]); + } + break; + case 0x00030000: + break; + default: + warn('Unknown/unsupported post table version ' + version); + valid = false; + if (properties.defaultEncoding) { + glyphNames = properties.defaultEncoding; + } + break; } + properties.glyphNames = glyphNames; + return valid; } - if (!contiguous) { - var offset = (segCount - i) * 2 + bias * 2; - bias += (end - start + 1); - idDeltas += string16(0); - idRangeOffsets += string16(offset); + function readNameTable(nameTable) { + var start = (font.start ? font.start : 0) + nameTable.offset; + font.pos = start; - for (j = 0, jj = codes.length; j < jj; ++j) { - glyphsIds += string16(codes[j]); + var names = [[], []]; + var length = nameTable.length, end = start + length; + var format = font.getUint16(); + var FORMAT_0_HEADER_LENGTH = 6; + if (format !== 0 || length < FORMAT_0_HEADER_LENGTH) { + // unsupported name table format or table "too" small + return names; } - } else { - var startCode = codes[0]; + var numRecords = font.getUint16(); + var stringsStart = font.getUint16(); + var records = []; + var NAME_RECORD_LENGTH = 12; + var i, ii; - idDeltas += string16((startCode - start) & 0xFFFF); - idRangeOffsets += string16(0); + for (i = 0; i < numRecords && + font.pos + NAME_RECORD_LENGTH <= end; i++) { + var r = { + platform: font.getUint16(), + encoding: font.getUint16(), + language: font.getUint16(), + name: font.getUint16(), + length: font.getUint16(), + offset: font.getUint16() + }; + // using only Macintosh and Windows platform/encoding names + if ((r.platform === 1 && r.encoding === 0 && r.language === 0) || + (r.platform === 3 && r.encoding === 1 && r.language === 0x409)) { + records.push(r); + } + } + for (i = 0, ii = records.length; i < ii; i++) { + var record = records[i]; + var pos = start + stringsStart + record.offset; + if (pos + record.length > end) { + continue; // outside of name table, ignoring + } + font.pos = pos; + var nameIndex = record.name; + if (record.encoding) { + // unicode + var str = ''; + for (var j = 0, jj = record.length; j < jj; j += 2) { + str += String.fromCharCode(font.getUint16()); + } + names[1][nameIndex] = str; + } else { + names[0][nameIndex] = bytesToString(font.getBytes(record.length)); + } + } + return names; } - } - if (trailingRangesCount > 0) { - endCount += '\xFF\xFF'; - startCount += '\xFF\xFF'; - idDeltas += '\x00\x01'; - idRangeOffsets += '\x00\x00'; - } + var TTOpsStackDeltas = [ + 0, 0, 0, 0, 0, 0, 0, 0, -2, -2, -2, -2, 0, 0, -2, -5, + -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, -1, -1, + 1, -1, -999, 0, 1, 0, -1, -2, 0, -1, -2, -1, -1, 0, -1, -1, + 0, 0, -999, -999, -1, -1, -1, -1, -2, -999, -2, -2, -999, 0, -2, -2, + 0, 0, -2, 0, -2, 0, 0, 0, -2, -1, -1, 1, 1, 0, 0, -1, + -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, 0, -999, -1, -1, + -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + -2, -999, -999, -999, -999, -999, -1, -1, -2, -2, 0, 0, 0, 0, -1, -1, + -999, -2, -2, 0, 0, -1, -2, -2, 0, 0, 0, -1, -1, -1, -2]; + // 0xC0-DF == -1 and 0xE0-FF == -2 - var format314 = '\x00\x00' + // language - string16(2 * segCount) + - string16(searchParams.range) + - string16(searchParams.entry) + - string16(searchParams.rangeShift) + - endCount + '\x00\x00' + startCount + - idDeltas + idRangeOffsets + glyphsIds; + function sanitizeTTProgram(table, ttContext) { + var data = table.data; + var i = 0, j, n, b, funcId, pc, lastEndf = 0, lastDeff = 0; + var stack = []; + var callstack = []; + var functionsCalled = []; + var tooComplexToFollowFunctions = + ttContext.tooComplexToFollowFunctions; + var inFDEF = false, ifLevel = 0, inELSE = 0; + for (var ii = data.length; i < ii;) { + var op = data[i++]; + // The TrueType instruction set docs can be found at + // https://developer.apple.com/fonts/TTRefMan/RM05/Chap5.html + if (op === 0x40) { // NPUSHB - pushes n bytes + n = data[i++]; + if (inFDEF || inELSE) { + i += n; + } else { + for (j = 0; j < n; j++) { + stack.push(data[i++]); + } + } + } else if (op === 0x41) { // NPUSHW - pushes n words + n = data[i++]; + if (inFDEF || inELSE) { + i += n * 2; + } else { + for (j = 0; j < n; j++) { + b = data[i++]; + stack.push((b << 8) | data[i++]); + } + } + } else if ((op & 0xF8) === 0xB0) { // PUSHB - pushes bytes + n = op - 0xB0 + 1; + if (inFDEF || inELSE) { + i += n; + } else { + for (j = 0; j < n; j++) { + stack.push(data[i++]); + } + } + } else if ((op & 0xF8) === 0xB8) { // PUSHW - pushes words + n = op - 0xB8 + 1; + if (inFDEF || inELSE) { + i += n * 2; + } else { + for (j = 0; j < n; j++) { + b = data[i++]; + stack.push((b << 8) | data[i++]); + } + } + } else if (op === 0x2B && !tooComplexToFollowFunctions) { // CALL + if (!inFDEF && !inELSE) { + // collecting inforamtion about which functions are used + funcId = stack[stack.length - 1]; + ttContext.functionsUsed[funcId] = true; + if (funcId in ttContext.functionsStackDeltas) { + stack.length += ttContext.functionsStackDeltas[funcId]; + } else if (funcId in ttContext.functionsDefined && + functionsCalled.indexOf(funcId) < 0) { + callstack.push({data: data, i: i, stackTop: stack.length - 1}); + functionsCalled.push(funcId); + pc = ttContext.functionsDefined[funcId]; + if (!pc) { + warn('TT: CALL non-existent function'); + ttContext.hintsValid = false; + return; + } + data = pc.data; + i = pc.i; + } + } + } else if (op === 0x2C && !tooComplexToFollowFunctions) { // FDEF + if (inFDEF || inELSE) { + warn('TT: nested FDEFs not allowed'); + tooComplexToFollowFunctions = true; + } + inFDEF = true; + // collecting inforamtion about which functions are defined + lastDeff = i; + funcId = stack.pop(); + ttContext.functionsDefined[funcId] = {data: data, i: i}; + } else if (op === 0x2D) { // ENDF - end of function + if (inFDEF) { + inFDEF = false; + lastEndf = i; + } else { + pc = callstack.pop(); + if (!pc) { + warn('TT: ENDF bad stack'); + ttContext.hintsValid = false; + return; + } + funcId = functionsCalled.pop(); + data = pc.data; + i = pc.i; + ttContext.functionsStackDeltas[funcId] = + stack.length - pc.stackTop; + } + } else if (op === 0x89) { // IDEF - instruction definition + if (inFDEF || inELSE) { + warn('TT: nested IDEFs not allowed'); + tooComplexToFollowFunctions = true; + } + inFDEF = true; + // recording it as a function to track ENDF + lastDeff = i; + } else if (op === 0x58) { // IF + ++ifLevel; + } else if (op === 0x1B) { // ELSE + inELSE = ifLevel; + } else if (op === 0x59) { // EIF + if (inELSE === ifLevel) { + inELSE = 0; + } + --ifLevel; + } else if (op === 0x1C) { // JMPR + if (!inFDEF && !inELSE) { + var offset = stack[stack.length - 1]; + // only jumping forward to prevent infinite loop + if (offset > 0) { + i += offset - 1; + } + } + } + // Adjusting stack not extactly, but just enough to get function id + if (!inFDEF && !inELSE) { + var stackDelta = op <= 0x8E ? TTOpsStackDeltas[op] : + op >= 0xC0 && op <= 0xDF ? -1 : op >= 0xE0 ? -2 : 0; + if (op >= 0x71 && op <= 0x75) { + n = stack.pop(); + if (n === n) { + stackDelta = -n * 2; + } + } + while (stackDelta < 0 && stack.length > 0) { + stack.pop(); + stackDelta++; + } + while (stackDelta > 0) { + stack.push(NaN); // pushing any number into stack + stackDelta--; + } + } + } + ttContext.tooComplexToFollowFunctions = tooComplexToFollowFunctions; + var content = [data]; + if (i > data.length) { + content.push(new Uint8Array(i - data.length)); + } + if (lastDeff > lastEndf) { + warn('TT: complementing a missing function tail'); + // new function definition started, but not finished + // complete function by [CLEAR, ENDF] + content.push(new Uint8Array([0x22, 0x2D])); + } + foldTTTable(table, content); + } - var format31012 = ''; - var header31012 = ''; - if (numTables > 1) { - cmap += '\x00\x03' + // platformID - '\x00\x0A' + // encodingID - string32(4 + numTables * 8 + - 4 + format314.length); // start of the table record - format31012 = ''; - for (i = 0, ii = ranges.length; i < ii; i++) { - range = ranges[i]; - start = range[0]; - codes = range[2]; - var code = codes[0]; - for (j = 1, jj = codes.length; j < jj; ++j) { - if (codes[j] !== codes[j - 1] + 1) { - end = range[0] + j - 1; - format31012 += string32(start) + // startCharCode - string32(end) + // endCharCode - string32(code); // startGlyphID - start = end + 1; - code = codes[j]; + function checkInvalidFunctions(ttContext, maxFunctionDefs) { + if (ttContext.tooComplexToFollowFunctions) { + return; + } + if (ttContext.functionsDefined.length > maxFunctionDefs) { + warn('TT: more functions defined than expected'); + ttContext.hintsValid = false; + return; + } + for (var j = 0, jj = ttContext.functionsUsed.length; j < jj; j++) { + if (j > maxFunctionDefs) { + warn('TT: invalid function id: ' + j); + ttContext.hintsValid = false; + return; + } + if (ttContext.functionsUsed[j] && !ttContext.functionsDefined[j]) { + warn('TT: undefined function: ' + j); + ttContext.hintsValid = false; + return; } } - format31012 += string32(start) + // startCharCode - string32(range[1]) + // endCharCode - string32(code); // startGlyphID } - header31012 = '\x00\x0C' + // format - '\x00\x00' + // reserved - string32(format31012.length + 16) + // length - '\x00\x00\x00\x00' + // language - string32(format31012.length / 12); // nGroups - } - return cmap + '\x00\x04' + // format - string16(format314.length + 4) + // length - format314 + header31012 + format31012; - } + function foldTTTable(table, content) { + if (content.length > 1) { + // concatenating the content items + var newLength = 0; + var j, jj; + for (j = 0, jj = content.length; j < jj; j++) { + newLength += content[j].length; + } + newLength = (newLength + 3) & ~3; + var result = new Uint8Array(newLength); + var pos = 0; + for (j = 0, jj = content.length; j < jj; j++) { + result.set(content[j], pos); + pos += content[j].length; + } + table.data = result; + table.length = newLength; + } + } - function validateOS2Table(os2) { - var stream = new Stream(os2.data); - var version = stream.getUint16(); - // TODO verify all OS/2 tables fields, but currently we validate only those - // that give us issues - stream.getBytes(60); // skipping type, misc sizes, panose, unicode ranges - var selection = stream.getUint16(); - if (version < 4 && (selection & 0x0300)) { - return false; - } - var firstChar = stream.getUint16(); - var lastChar = stream.getUint16(); - if (firstChar > lastChar) { - return false; - } - stream.getBytes(6); // skipping sTypoAscender/Descender/LineGap - var usWinAscent = stream.getUint16(); - if (usWinAscent === 0) { // makes font unreadable by windows - return false; - } + function sanitizeTTPrograms(fpgm, prep, cvt) { + var ttContext = { + functionsDefined: [], + functionsUsed: [], + functionsStackDeltas: [], + tooComplexToFollowFunctions: false, + hintsValid: true + }; + if (fpgm) { + sanitizeTTProgram(fpgm, ttContext); + } + if (prep) { + sanitizeTTProgram(prep, ttContext); + } + if (fpgm) { + checkInvalidFunctions(ttContext, maxFunctionDefs); + } + if (cvt && (cvt.length & 1)) { + var cvtData = new Uint8Array(cvt.length + 1); + cvtData.set(cvt.data); + cvt.data = cvtData; + } + return ttContext.hintsValid; + } - // OS/2 appears to be valid, resetting some fields - os2.data[8] = os2.data[9] = 0; // IE rejects fonts if fsType != 0 - return true; - } + // The following steps modify the original font data, making copy + font = new Stream(new Uint8Array(font.getBytes())); - function createOS2Table(properties, charstrings, override) { - override = override || { - unitsPerEm: 0, - yMax: 0, - yMin: 0, - ascent: 0, - descent: 0 - }; + var VALID_TABLES = ['OS/2', 'cmap', 'head', 'hhea', 'hmtx', 'maxp', + 'name', 'post', 'loca', 'glyf', 'fpgm', 'prep', 'cvt ', 'CFF ']; - var ulUnicodeRange1 = 0; - var ulUnicodeRange2 = 0; - var ulUnicodeRange3 = 0; - var ulUnicodeRange4 = 0; + var header = readOpenTypeHeader(font); + var numTables = header.numTables; + var cff, cffFile; - var firstCharIndex = null; - var lastCharIndex = 0; + var tables = Object.create(null); + tables['OS/2'] = null; + tables['cmap'] = null; + tables['head'] = null; + tables['hhea'] = null; + tables['hmtx'] = null; + tables['maxp'] = null; + tables['name'] = null; + tables['post'] = null; - if (charstrings) { - for (var code in charstrings) { - code |= 0; - if (firstCharIndex > code || !firstCharIndex) { - firstCharIndex = code; - } - if (lastCharIndex < code) { - lastCharIndex = code; + var table; + for (var i = 0; i < numTables; i++) { + table = readTableEntry(font); + if (VALID_TABLES.indexOf(table.tag) < 0) { + continue; // skipping table if it's not a required or optional table } - - var position = getUnicodeRangeFor(code); - if (position < 32) { - ulUnicodeRange1 |= 1 << position; - } else if (position < 64) { - ulUnicodeRange2 |= 1 << position - 32; - } else if (position < 96) { - ulUnicodeRange3 |= 1 << position - 64; - } else if (position < 123) { - ulUnicodeRange4 |= 1 << position - 96; - } else { - error('Unicode ranges Bits > 123 are reserved for internal usage'); + if (table.length === 0) { + continue; // skipping empty tables } + tables[table.tag] = table; } - } else { - // TODO - firstCharIndex = 0; - lastCharIndex = 255; - } - - var bbox = properties.bbox || [0, 0, 0, 0]; - var unitsPerEm = (override.unitsPerEm || - 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]); - // if the font units differ to the PDF glyph space units - // then scale up the values - var scale = (properties.ascentScaled ? 1.0 : - unitsPerEm / PDF_GLYPH_SPACE_UNITS); + var isTrueType = !tables['CFF ']; + if (!isTrueType) { + // OpenType font + if ((header.version === 'OTTO' && properties.type !== 'CIDFontType2') || + !tables['head'] || !tables['hhea'] || !tables['maxp'] || + !tables['post']) { + // no major tables: throwing everything at CFFFont + cffFile = new Stream(tables['CFF '].data); + cff = new CFFFont(cffFile, properties); - var typoAscent = (override.ascent || - Math.round(scale * (properties.ascent || bbox[3]))); - var typoDescent = (override.descent || - Math.round(scale * (properties.descent || bbox[1]))); - if (typoDescent > 0 && properties.descent > 0 && bbox[1] < 0) { - typoDescent = -typoDescent; // fixing incorrect descent - } - var winAscent = override.yMax || typoAscent; - var winDescent = -override.yMin || -typoDescent; + adjustWidths(properties); - return '\x00\x03' + // version - '\x02\x24' + // xAvgCharWidth - '\x01\xF4' + // usWeightClass - '\x00\x05' + // usWidthClass - '\x00\x00' + // fstype (0 to let the font loads via font-face on IE) - '\x02\x8A' + // ySubscriptXSize - '\x02\xBB' + // ySubscriptYSize - '\x00\x00' + // ySubscriptXOffset - '\x00\x8C' + // ySubscriptYOffset - '\x02\x8A' + // ySuperScriptXSize - '\x02\xBB' + // ySuperScriptYSize - '\x00\x00' + // ySuperScriptXOffset - '\x01\xDF' + // ySuperScriptYOffset - '\x00\x31' + // yStrikeOutSize - '\x01\x02' + // yStrikeOutPosition - '\x00\x00' + // sFamilyClass - '\x00\x00\x06' + - String.fromCharCode(properties.fixedPitch ? 0x09 : 0x00) + - '\x00\x00\x00\x00\x00\x00' + // Panose - string32(ulUnicodeRange1) + // ulUnicodeRange1 (Bits 0-31) - string32(ulUnicodeRange2) + // ulUnicodeRange2 (Bits 32-63) - string32(ulUnicodeRange3) + // ulUnicodeRange3 (Bits 64-95) - string32(ulUnicodeRange4) + // ulUnicodeRange4 (Bits 96-127) - '\x2A\x32\x31\x2A' + // achVendID - string16(properties.italicAngle ? 1 : 0) + // fsSelection - string16(firstCharIndex || - properties.firstChar) + // usFirstCharIndex - string16(lastCharIndex || properties.lastChar) + // usLastCharIndex - string16(typoAscent) + // sTypoAscender - string16(typoDescent) + // sTypoDescender - '\x00\x64' + // sTypoLineGap (7%-10% of the unitsPerEM value) - string16(winAscent) + // usWinAscent - string16(winDescent) + // usWinDescent - '\x00\x00\x00\x00' + // ulCodePageRange1 (Bits 0-31) - '\x00\x00\x00\x00' + // ulCodePageRange2 (Bits 32-63) - string16(properties.xHeight) + // sxHeight - string16(properties.capHeight) + // sCapHeight - string16(0) + // usDefaultChar - string16(firstCharIndex || properties.firstChar) + // usBreakChar - '\x00\x03'; // usMaxContext - } + return this.convert(name, cff, properties); + } - function createPostTable(properties) { - var angle = Math.floor(properties.italicAngle * (Math.pow(2, 16))); - return ('\x00\x03\x00\x00' + // Version number - string32(angle) + // italicAngle - '\x00\x00' + // underlinePosition - '\x00\x00' + // underlineThickness - string32(properties.fixedPitch) + // isFixedPitch - '\x00\x00\x00\x00' + // minMemType42 - '\x00\x00\x00\x00' + // maxMemType42 - '\x00\x00\x00\x00' + // minMemType1 - '\x00\x00\x00\x00'); // maxMemType1 - } + delete tables['glyf']; + delete tables['loca']; + delete tables['fpgm']; + delete tables['prep']; + delete tables['cvt ']; + this.isOpenType = true; + } else { + if (!tables['loca']) { + error('Required "loca" table is not found'); + } + if (!tables['glyf']) { + warn('Required "glyf" table is not found -- trying to recover.'); + // Note: We use `sanitizeGlyphLocations` to add dummy glyf data below. + tables['glyf'] = { + tag: 'glyf', + data: new Uint8Array(0), + }; + } + this.isOpenType = false; + } - function createNameTable(name, proto) { - if (!proto) { - proto = [[], []]; // no strings and unicode strings - } + if (!tables['maxp']) { + error('Required "maxp" table is not found'); + } - var strings = [ - proto[0][0] || 'Original licence', // 0.Copyright - proto[0][1] || name, // 1.Font family - proto[0][2] || 'Unknown', // 2.Font subfamily (font weight) - proto[0][3] || 'uniqueID', // 3.Unique ID - proto[0][4] || name, // 4.Full font name - proto[0][5] || 'Version 0.11', // 5.Version - proto[0][6] || '', // 6.Postscript name - proto[0][7] || 'Unknown', // 7.Trademark - proto[0][8] || 'Unknown', // 8.Manufacturer - proto[0][9] || 'Unknown' // 9.Designer - ]; + font.pos = (font.start || 0) + tables['maxp'].offset; + var version = font.getInt32(); + var numGlyphs = font.getUint16(); + var maxFunctionDefs = 0; + if (version >= 0x00010000 && tables['maxp'].length >= 22) { + // maxZones can be invalid + font.pos += 8; + var maxZones = font.getUint16(); + if (maxZones > 2) { // reset to 2 if font has invalid maxZones + tables['maxp'].data[14] = 0; + tables['maxp'].data[15] = 2; + } + font.pos += 4; + maxFunctionDefs = font.getUint16(); + } - // Mac want 1-byte per character strings while Windows want - // 2-bytes per character, so duplicate the names table - var stringsUnicode = []; - var i, ii, j, jj, str; - for (i = 0, ii = strings.length; i < ii; i++) { - str = proto[1][i] || strings[i]; + var dupFirstEntry = false; + if (properties.type === 'CIDFontType2' && properties.toUnicode && + properties.toUnicode.get(0) > '\u0000') { + // oracle's defect (see 3427), duplicating first entry + dupFirstEntry = true; + numGlyphs++; + tables['maxp'].data[4] = numGlyphs >> 8; + tables['maxp'].data[5] = numGlyphs & 255; + } - var strBufUnicode = []; - for (j = 0, jj = str.length; j < jj; j++) { - strBufUnicode.push(string16(str.charCodeAt(j))); + var hintsValid = sanitizeTTPrograms(tables['fpgm'], tables['prep'], + tables['cvt '], maxFunctionDefs); + if (!hintsValid) { + delete tables['fpgm']; + delete tables['prep']; + delete tables['cvt ']; } - stringsUnicode.push(strBufUnicode.join('')); - } - var names = [strings, stringsUnicode]; - var platforms = ['\x00\x01', '\x00\x03']; - var encodings = ['\x00\x00', '\x00\x01']; - var languages = ['\x00\x00', '\x04\x09']; + // Ensure the hmtx table contains the advance width and + // sidebearings information for numGlyphs in the maxp table + sanitizeMetrics(font, tables['hhea'], tables['hmtx'], numGlyphs); - var namesRecordCount = strings.length * platforms.length; - var nameTable = - '\x00\x00' + // format - string16(namesRecordCount) + // Number of names Record - string16(namesRecordCount * 12 + 6); // Storage - - // Build the name records field - var strOffset = 0; - for (i = 0, ii = platforms.length; i < ii; i++) { - var strs = names[i]; - for (j = 0, jj = strs.length; j < jj; j++) { - str = strs[j]; - var nameRecord = - platforms[i] + // platform ID - encodings[i] + // encoding ID - languages[i] + // language ID - string16(j) + // name ID - string16(str.length) + - string16(strOffset); - nameTable += nameRecord; - strOffset += str.length; + if (!tables['head']) { + error('Required "head" table is not found'); } - } - nameTable += strings.join('') + stringsUnicode.join(''); - return nameTable; - } + sanitizeHead(tables['head'], numGlyphs, + isTrueType ? tables['loca'].length : 0); - Font.prototype = { - name: null, - font: null, - mimetype: null, - encoding: null, - get renderer() { - var renderer = FontRendererFactory.create(this); - return shadow(this, 'renderer', renderer); - }, + var missingGlyphs = Object.create(null); + if (isTrueType) { + var isGlyphLocationsLong = int16(tables['head'].data[50], + tables['head'].data[51]); + missingGlyphs = sanitizeGlyphLocations(tables['loca'], tables['glyf'], + numGlyphs, isGlyphLocationsLong, + hintsValid, dupFirstEntry); + } - exportData: function Font_exportData() { - // TODO remove enumerating of the properties, e.g. hardcode exact names. - var data = {}; - for (var i in this) { - if (this.hasOwnProperty(i)) { - data[i] = this[i]; - } + if (!tables['hhea']) { + error('Required "hhea" table is not found'); } - return data; - }, - checkAndRepair: function Font_checkAndRepair(name, font, properties) { - function readTableEntry(file) { - var tag = bytesToString(file.getBytes(4)); + // Sanitizer reduces the glyph advanceWidth to the maxAdvanceWidth + // Sometimes it's 0. That needs to be fixed + if (tables['hhea'].data[10] === 0 && tables['hhea'].data[11] === 0) { + tables['hhea'].data[10] = 0xFF; + tables['hhea'].data[11] = 0xFF; + } - var checksum = file.getInt32(); - var offset = file.getInt32() >>> 0; - var length = file.getInt32() >>> 0; + // Extract some more font properties from the OpenType head and + // hhea tables; yMin and descent value are always negative. + var metricsOverride = { + unitsPerEm: int16(tables['head'].data[18], tables['head'].data[19]), + yMax: int16(tables['head'].data[42], tables['head'].data[43]), + yMin: int16(tables['head'].data[38], tables['head'].data[39]) - 0x10000, + ascent: int16(tables['hhea'].data[4], tables['hhea'].data[5]), + descent: int16(tables['hhea'].data[6], tables['hhea'].data[7]) - 0x10000 + }; - // Read the table associated data - var previousPosition = file.pos; - file.pos = file.start ? file.start : 0; - file.skip(offset); - var data = file.getBytes(length); - file.pos = previousPosition; + // PDF FontDescriptor metrics lie -- using data from actual font. + this.ascent = metricsOverride.ascent / metricsOverride.unitsPerEm; + this.descent = metricsOverride.descent / metricsOverride.unitsPerEm; - if (tag === 'head') { - // clearing checksum adjustment - data[8] = data[9] = data[10] = data[11] = 0; - data[17] |= 0x20; //Set font optimized for cleartype flag + // The 'post' table has glyphs names. + if (tables['post']) { + var valid = readPostScriptTable(tables['post'], properties, numGlyphs); + if (!valid) { + tables['post'] = null; } - - return { - tag: tag, - checksum: checksum, - length: length, - offset: offset, - data: data - }; } - function readOpenTypeHeader(ttf) { - return { - version: bytesToString(ttf.getBytes(4)), - numTables: ttf.getUint16(), - searchRange: ttf.getUint16(), - entrySelector: ttf.getUint16(), - rangeShift: ttf.getUint16() - }; - } + var charCodeToGlyphId = [], charCode; + var toUnicode = properties.toUnicode, widths = properties.widths; + var skipToUnicode = (toUnicode instanceof IdentityToUnicodeMap || + toUnicode.length === 0x10000); - /** - * Read the appropriate subtable from the cmap according to 9.6.6.4 from - * PDF spec - */ - function readCmapTable(cmap, font, isSymbolicFont, hasEncoding) { - if (!cmap) { - warn('No cmap table available.'); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; + // Helper function to try to skip mapping of empty glyphs. + // Note: In some cases, just relying on the glyph data doesn't work, + // hence we also use a few heuristics to fix various PDF files. + function hasGlyph(glyphId, charCode, widthCode) { + if (!missingGlyphs[glyphId]) { + return true; } - var segment; - var start = (font.start ? font.start : 0) + cmap.offset; - font.pos = start; - - var version = font.getUint16(); - var numTables = font.getUint16(); + if (!skipToUnicode && charCode >= 0 && toUnicode.has(charCode)) { + return true; + } + if (widths && widthCode >= 0 && isNum(widths[widthCode])) { + return true; + } + return false; + } - var potentialTable; - var canBreak = false; - // There's an order of preference in terms of which cmap subtable to - // use: - // - non-symbolic fonts the preference is a 3,1 table then a 1,0 table - // - symbolic fonts the preference is a 3,0 table then a 1,0 table - // The following takes advantage of the fact that the tables are sorted - // to work. - for (var i = 0; i < numTables; i++) { - var platformId = font.getUint16(); - var encodingId = font.getUint16(); - var offset = font.getInt32() >>> 0; - var useTable = false; + if (properties.type === 'CIDFontType2') { + var cidToGidMap = properties.cidToGidMap || []; + var isCidToGidMapEmpty = cidToGidMap.length === 0; - if (platformId === 0 && encodingId === 0) { - useTable = true; - // Continue the loop since there still may be a higher priority - // table. - } else if (platformId === 1 && encodingId === 0) { - useTable = true; - // Continue the loop since there still may be a higher priority - // table. - } else if (platformId === 3 && encodingId === 1 && - ((!isSymbolicFont && hasEncoding) || !potentialTable)) { - useTable = true; - if (!isSymbolicFont) { - canBreak = true; - } - } else if (isSymbolicFont && platformId === 3 && encodingId === 0) { - useTable = true; - canBreak = true; + properties.cMap.forEach(function(charCode, cid) { + assert(cid <= 0xffff, 'Max size of CID is 65,535'); + var glyphId = -1; + if (isCidToGidMapEmpty) { + glyphId = charCode; + } else if (cidToGidMap[cid] !== undefined) { + glyphId = cidToGidMap[cid]; } - if (useTable) { - potentialTable = { - platformId: platformId, - encodingId: encodingId, - offset: offset - }; - } - if (canBreak) { - break; + if (glyphId >= 0 && glyphId < numGlyphs && + hasGlyph(glyphId, charCode, cid)) { + charCodeToGlyphId[charCode] = glyphId; } + }); + if (dupFirstEntry) { + charCodeToGlyphId[0] = numGlyphs - 1; } + } else { + // Most of the following logic in this code branch is based on the + // 9.6.6.4 of the PDF spec. + var hasEncoding = + properties.differences.length > 0 || !!properties.baseEncodingName; + var cmapTable = + readCmapTable(tables['cmap'], font, this.isSymbolicFont, hasEncoding); + var cmapPlatformId = cmapTable.platformId; + var cmapEncodingId = cmapTable.encodingId; + var cmapMappings = cmapTable.mappings; + var cmapMappingsLength = cmapMappings.length; - if (potentialTable) { - font.pos = start + potentialTable.offset; - } - if (!potentialTable || font.peekByte() === -1) { - warn('Could not find a preferred cmap table.'); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; - } - - var format = font.getUint16(); - var length = font.getUint16(); - var language = font.getUint16(); - - var hasShortCmap = false; - var mappings = []; - var j, glyphId; - - // TODO(mack): refactor this cmap subtable reading logic out - if (format === 0) { - for (j = 0; j < 256; j++) { - var index = font.getByte(); - if (!index) { - continue; - } - mappings.push({ - charCode: j, - glyphId: index - }); - } - hasShortCmap = true; - } else if (format === 4) { - // re-creating the table in format 4 since the encoding - // might be changed - var segCount = (font.getUint16() >> 1); - font.getBytes(6); // skipping range fields - var segIndex, segments = []; - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments.push({ end: font.getUint16() }); - } - font.getUint16(); - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments[segIndex].start = font.getUint16(); - } + // The spec seems to imply that if the font is symbolic the encoding + // should be ignored, this doesn't appear to work for 'preistabelle.pdf' + // where the the font is symbolic and it has an encoding. + if (hasEncoding && + (cmapPlatformId === 3 && cmapEncodingId === 1 || + cmapPlatformId === 1 && cmapEncodingId === 0) || + (cmapPlatformId === -1 && cmapEncodingId === -1 && // Temporary hack + !!getEncoding(properties.baseEncodingName))) { // Temporary hack + // When no preferred cmap table was found and |baseEncodingName| is + // one of the predefined encodings, we seem to obtain a better + // |charCodeToGlyphId| map from the code below (fixes bug 1057544). + // TODO: Note that this is a hack which should be removed as soon as + // we have proper support for more exotic cmap tables. - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments[segIndex].delta = font.getUint16(); + var baseEncoding = []; + if (properties.baseEncodingName === 'MacRomanEncoding' || + properties.baseEncodingName === 'WinAnsiEncoding') { + baseEncoding = getEncoding(properties.baseEncodingName); } - - var offsetsCount = 0; - for (segIndex = 0; segIndex < segCount; segIndex++) { - segment = segments[segIndex]; - var rangeOffset = font.getUint16(); - if (!rangeOffset) { - segment.offsetIndex = -1; + var glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode = 0; charCode < 256; charCode++) { + var glyphName; + if (this.differences && charCode in this.differences) { + glyphName = this.differences[charCode]; + } else if (charCode in baseEncoding && + baseEncoding[charCode] !== '') { + glyphName = baseEncoding[charCode]; + } else { + glyphName = StandardEncoding[charCode]; + } + if (!glyphName) { continue; } + var unicodeOrCharCode, isUnicode = false; + if (cmapPlatformId === 3 && cmapEncodingId === 1) { + unicodeOrCharCode = glyphsUnicodeMap[glyphName]; + isUnicode = true; + } else if (cmapPlatformId === 1 && cmapEncodingId === 0) { + // TODO: the encoding needs to be updated with mac os table. + unicodeOrCharCode = MacRomanEncoding.indexOf(glyphName); + } - var offsetIndex = (rangeOffset >> 1) - (segCount - segIndex); - segment.offsetIndex = offsetIndex; - offsetsCount = Math.max(offsetsCount, offsetIndex + - segment.end - segment.start + 1); - } - - var offsets = []; - for (j = 0; j < offsetsCount; j++) { - offsets.push(font.getUint16()); - } - - for (segIndex = 0; segIndex < segCount; segIndex++) { - segment = segments[segIndex]; - start = segment.start; - var end = segment.end; - var delta = segment.delta; - offsetIndex = segment.offsetIndex; - - for (j = start; j <= end; j++) { - if (j === 0xFFFF) { + var found = false; + for (i = 0; i < cmapMappingsLength; ++i) { + if (cmapMappings[i].charCode !== unicodeOrCharCode) { continue; } - - glyphId = (offsetIndex < 0 ? - j : offsets[offsetIndex + j - start]); - glyphId = (glyphId + delta) & 0xFFFF; - if (glyphId === 0) { - continue; + var code = isUnicode ? charCode : unicodeOrCharCode; + if (hasGlyph(cmapMappings[i].glyphId, code, -1)) { + charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; + found = true; + break; } - mappings.push({ - charCode: j, - glyphId: glyphId - }); + } + if (!found && properties.glyphNames) { + // Try to map using the post table. + var glyphId = properties.glyphNames.indexOf(glyphName); + if (glyphId > 0 && hasGlyph(glyphId, -1, -1)) { + charCodeToGlyphId[charCode] = glyphId; + found = true; + } + } + if (!found) { + charCodeToGlyphId[charCode] = 0; // notdef } } - } else if (format === 6) { - // Format 6 is a 2-bytes dense mapping, which means the font data - // lives glue together even if they are pretty far in the unicode - // table. (This looks weird, so I can have missed something), this - // works on Linux but seems to fails on Mac so let's rewrite the - // cmap table to a 3-1-4 style - var firstCode = font.getUint16(); - var entryCount = font.getUint16(); - - for (j = 0; j < entryCount; j++) { - glyphId = font.getUint16(); - var charCode = firstCode + j; - - mappings.push({ - charCode: charCode, - glyphId: glyphId - }); + } else if (cmapPlatformId === 0 && cmapEncodingId === 0) { + // Default Unicode semantics, use the charcodes as is. + for (i = 0; i < cmapMappingsLength; ++i) { + charCodeToGlyphId[cmapMappings[i].charCode] = + cmapMappings[i].glyphId; } } else { - warn('cmap table has unsupported format: ' + format); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; - } - - // removing duplicate entries - mappings.sort(function (a, b) { - return a.charCode - b.charCode; - }); - for (i = 1; i < mappings.length; i++) { - if (mappings[i - 1].charCode === mappings[i].charCode) { - mappings.splice(i, 1); - i--; + // For (3, 0) cmap tables: + // The charcode key being stored in charCodeToGlyphId is the lower + // byte of the two-byte charcodes of the cmap table since according to + // the spec: 'each byte from the string shall be prepended with the + // high byte of the range [of charcodes in the cmap table], to form + // a two-byte character, which shall be used to select the + // associated glyph description from the subtable'. + // + // For (1, 0) cmap tables: + // 'single bytes from the string shall be used to look up the + // associated glyph descriptions from the subtable'. This means + // charcodes in the cmap will be single bytes, so no-op since + // glyph.charCode & 0xFF === glyph.charCode + for (i = 0; i < cmapMappingsLength; ++i) { + charCode = cmapMappings[i].charCode & 0xFF; + charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; } } - - return { - platformId: potentialTable.platformId, - encodingId: potentialTable.encodingId, - mappings: mappings, - hasShortCmap: hasShortCmap - }; } - function sanitizeMetrics(font, header, metrics, numGlyphs) { - if (!header) { - if (metrics) { - metrics.data = null; - } - return; - } + if (charCodeToGlyphId.length === 0) { + // defines at least one glyph + charCodeToGlyphId[0] = 0; + } - font.pos = (font.start ? font.start : 0) + header.offset; - font.pos += header.length - 2; - var numOfMetrics = font.getUint16(); + // Converting glyphs and ids into font's cmap table + var newMapping = adjustMapping(charCodeToGlyphId, properties); + this.toFontChar = newMapping.toFontChar; + tables['cmap'] = { + tag: 'cmap', + data: createCmapTable(newMapping.charCodeToGlyphId, numGlyphs) + }; - if (numOfMetrics > numGlyphs) { - info('The numOfMetrics (' + numOfMetrics + ') should not be ' + - 'greater than the numGlyphs (' + numGlyphs + ')'); - // Reduce numOfMetrics if it is greater than numGlyphs - numOfMetrics = numGlyphs; - header.data[34] = (numOfMetrics & 0xff00) >> 8; - header.data[35] = numOfMetrics & 0x00ff; - } + if (!tables['OS/2'] || !validateOS2Table(tables['OS/2'])) { + tables['OS/2'] = { + tag: 'OS/2', + data: createOS2Table(properties, newMapping.charCodeToGlyphId, + metricsOverride) + }; + } - var numOfSidebearings = numGlyphs - numOfMetrics; - var numMissing = numOfSidebearings - - ((metrics.length - numOfMetrics * 4) >> 1); + // Rewrite the 'post' table if needed + if (!tables['post']) { + tables['post'] = { + tag: 'post', + data: createPostTable(properties) + }; + } - if (numMissing > 0) { - // For each missing glyph, we set both the width and lsb to 0 (zero). - // Since we need to add two properties for each glyph, this explains - // the use of |numMissing * 2| when initializing the typed array. - var entries = new Uint8Array(metrics.length + numMissing * 2); - entries.set(metrics.data); - metrics.data = entries; + if (!isTrueType) { + try { + // Trying to repair CFF file + cffFile = new Stream(tables['CFF '].data); + var parser = new CFFParser(cffFile, properties); + cff = parser.parse(); + var compiler = new CFFCompiler(cff); + tables['CFF '].data = compiler.compile(); + } catch (e) { + warn('Failed to compile font ' + properties.loadedName); } } - function sanitizeGlyph(source, sourceStart, sourceEnd, dest, destStart, - hintsValid) { - if (sourceEnd - sourceStart <= 12) { - // glyph with data less than 12 is invalid one - return 0; - } - var glyf = source.subarray(sourceStart, sourceEnd); - var contoursCount = (glyf[0] << 8) | glyf[1]; - if (contoursCount & 0x8000) { - // complex glyph, writing as is - dest.set(glyf, destStart); - return glyf.length; - } + // Re-creating 'name' table + if (!tables['name']) { + tables['name'] = { + tag: 'name', + data: createNameTable(this.name) + }; + } else { + // ... using existing 'name' table as prototype + var namePrototype = readNameTable(tables['name']); + tables['name'].data = createNameTable(name, namePrototype); + } - var i, j = 10, flagsCount = 0; - for (i = 0; i < contoursCount; i++) { - var endPoint = (glyf[j] << 8) | glyf[j + 1]; - flagsCount = endPoint + 1; - j += 2; - } - // skipping instructions - var instructionsStart = j; - var instructionsLength = (glyf[j] << 8) | glyf[j + 1]; - j += 2 + instructionsLength; - var instructionsEnd = j; - // validating flags - var coordinatesLength = 0; - for (i = 0; i < flagsCount; i++) { - var flag = glyf[j++]; - if (flag & 0xC0) { - // reserved flags must be zero, cleaning up - glyf[j - 1] = flag & 0x3F; - } - var xyLength = ((flag & 2) ? 1 : (flag & 16) ? 0 : 2) + - ((flag & 4) ? 1 : (flag & 32) ? 0 : 2); - coordinatesLength += xyLength; - if (flag & 8) { - var repeat = glyf[j++]; - i += repeat; - coordinatesLength += repeat * xyLength; - } - } - // glyph without coordinates will be rejected - if (coordinatesLength === 0) { - return 0; - } - var glyphDataLength = j + coordinatesLength; - if (glyphDataLength > glyf.length) { - // not enough data for coordinates - return 0; - } - if (!hintsValid && instructionsLength > 0) { - dest.set(glyf.subarray(0, instructionsStart), destStart); - dest.set([0, 0], destStart + instructionsStart); - dest.set(glyf.subarray(instructionsEnd, glyphDataLength), - destStart + instructionsStart + 2); - glyphDataLength -= instructionsLength; - if (glyf.length - glyphDataLength > 3) { - glyphDataLength = (glyphDataLength + 3) & ~3; - } - return glyphDataLength; - } - if (glyf.length - glyphDataLength > 3) { - // truncating and aligning to 4 bytes the long glyph data - glyphDataLength = (glyphDataLength + 3) & ~3; - dest.set(glyf.subarray(0, glyphDataLength), destStart); - return glyphDataLength; - } - // glyph data is fine - dest.set(glyf, destStart); - return glyf.length; + var builder = new OpenTypeFileBuilder(header.version); + for (var tableTag in tables) { + builder.addTable(tableTag, tables[tableTag].data); } + return builder.toArray(); + }, - function sanitizeHead(head, numGlyphs, locaLength) { - var data = head.data; - - // Validate version: - // Should always be 0x00010000 - var version = int32(data[0], data[1], data[2], data[3]); - if (version >> 16 !== 1) { - info('Attempting to fix invalid version in head table: ' + version); - data[0] = 0; - data[1] = 1; - data[2] = 0; - data[3] = 0; - } - - var indexToLocFormat = int16(data[50], data[51]); - if (indexToLocFormat < 0 || indexToLocFormat > 1) { - info('Attempting to fix invalid indexToLocFormat in head table: ' + - indexToLocFormat); + convert: function Font_convert(fontName, font, properties) { + // TODO: Check the charstring widths to determine this. + properties.fixedPitch = false; - // The value of indexToLocFormat should be 0 if the loca table - // consists of short offsets, and should be 1 if the loca table - // consists of long offsets. - // - // The number of entries in the loca table should be numGlyphs + 1. - // - // Using this information, we can work backwards to deduce if the - // size of each offset in the loca table, and thus figure out the - // appropriate value for indexToLocFormat. + var mapping = font.getGlyphMapping(properties); + var newMapping = adjustMapping(mapping, properties); + this.toFontChar = newMapping.toFontChar; + var numGlyphs = font.numGlyphs; - var numGlyphsPlusOne = numGlyphs + 1; - if (locaLength === numGlyphsPlusOne << 1) { - // 0x0000 indicates the loca table consists of short offsets - data[50] = 0; - data[51] = 0; - } else if (locaLength === numGlyphsPlusOne << 2) { - // 0x0001 indicates the loca table consists of long offsets - data[50] = 0; - data[51] = 1; - } else { - warn('Could not fix indexToLocFormat: ' + indexToLocFormat); + function getCharCodes(charCodeToGlyphId, glyphId) { + var charCodes = null; + for (var charCode in charCodeToGlyphId) { + if (glyphId === charCodeToGlyphId[charCode]) { + if (!charCodes) { + charCodes = []; + } + charCodes.push(charCode | 0); } } + return charCodes; } - function sanitizeGlyphLocations(loca, glyf, numGlyphs, - isGlyphLocationsLong, hintsValid, - dupFirstEntry) { - var itemSize, itemDecode, itemEncode; - if (isGlyphLocationsLong) { - itemSize = 4; - itemDecode = function fontItemDecodeLong(data, offset) { - return (data[offset] << 24) | (data[offset + 1] << 16) | - (data[offset + 2] << 8) | data[offset + 3]; - }; - itemEncode = function fontItemEncodeLong(data, offset, value) { - data[offset] = (value >>> 24) & 0xFF; - data[offset + 1] = (value >> 16) & 0xFF; - data[offset + 2] = (value >> 8) & 0xFF; - data[offset + 3] = value & 0xFF; - }; - } else { - itemSize = 2; - itemDecode = function fontItemDecode(data, offset) { - return (data[offset] << 9) | (data[offset + 1] << 1); - }; - itemEncode = function fontItemEncode(data, offset, value) { - data[offset] = (value >> 9) & 0xFF; - data[offset + 1] = (value >> 1) & 0xFF; - }; - } - var locaData = loca.data; - var locaDataSize = itemSize * (1 + numGlyphs); - // is loca.data too short or long? - if (locaData.length !== locaDataSize) { - locaData = new Uint8Array(locaDataSize); - locaData.set(loca.data.subarray(0, locaDataSize)); - loca.data = locaData; - } - // removing the invalid glyphs - var oldGlyfData = glyf.data; - var oldGlyfDataLength = oldGlyfData.length; - var newGlyfData = new Uint8Array(oldGlyfDataLength); - var startOffset = itemDecode(locaData, 0); - var writeOffset = 0; - var missingGlyphData = Object.create(null); - itemEncode(locaData, 0, writeOffset); - var i, j; - for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { - var endOffset = itemDecode(locaData, j); - if (endOffset > oldGlyfDataLength && - ((oldGlyfDataLength + 3) & ~3) === endOffset) { - // Aspose breaks fonts by aligning the glyphs to the qword, but not - // the glyf table size, which makes last glyph out of range. - endOffset = oldGlyfDataLength; - } - if (endOffset > oldGlyfDataLength) { - // glyph end offset points outside glyf data, rejecting the glyph - itemEncode(locaData, j, writeOffset); - startOffset = endOffset; - continue; - } - - if (startOffset === endOffset) { - missingGlyphData[i] = true; + function createCharCode(charCodeToGlyphId, glyphId) { + for (var charCode in charCodeToGlyphId) { + if (glyphId === charCodeToGlyphId[charCode]) { + return charCode | 0; } - - var newLength = sanitizeGlyph(oldGlyfData, startOffset, endOffset, - newGlyfData, writeOffset, hintsValid); - writeOffset += newLength; - itemEncode(locaData, j, writeOffset); - startOffset = endOffset; } + newMapping.charCodeToGlyphId[newMapping.nextAvailableFontCharCode] = + glyphId; + return newMapping.nextAvailableFontCharCode++; + } - if (writeOffset === 0) { - // glyf table cannot be empty -- redoing the glyf and loca tables - // to have single glyph with one point - var simpleGlyph = new Uint8Array( - [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 49, 0]); - for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { - itemEncode(locaData, j, simpleGlyph.length); + var seacs = font.seacs; + if (SEAC_ANALYSIS_ENABLED && seacs && seacs.length) { + var matrix = properties.fontMatrix || FONT_IDENTITY_MATRIX; + var charset = font.getCharset(); + var seacMap = Object.create(null); + for (var glyphId in seacs) { + glyphId |= 0; + var seac = seacs[glyphId]; + var baseGlyphName = StandardEncoding[seac[2]]; + var accentGlyphName = StandardEncoding[seac[3]]; + var baseGlyphId = charset.indexOf(baseGlyphName); + var accentGlyphId = charset.indexOf(accentGlyphName); + if (baseGlyphId < 0 || accentGlyphId < 0) { + continue; } - glyf.data = simpleGlyph; - return missingGlyphData; - } + var accentOffset = { + x: seac[0] * matrix[0] + seac[1] * matrix[2] + matrix[4], + y: seac[0] * matrix[1] + seac[1] * matrix[3] + matrix[5] + }; - if (dupFirstEntry) { - var firstEntryLength = itemDecode(locaData, itemSize); - if (newGlyfData.length > firstEntryLength + writeOffset) { - glyf.data = newGlyfData.subarray(0, firstEntryLength + writeOffset); - } else { - glyf.data = new Uint8Array(firstEntryLength + writeOffset); - glyf.data.set(newGlyfData.subarray(0, writeOffset)); + var charCodes = getCharCodes(mapping, glyphId); + if (!charCodes) { + // There's no point in mapping it if the char code was never mapped + // to begin with. + continue; + } + for (var i = 0, ii = charCodes.length; i < ii; i++) { + var charCode = charCodes[i]; + // Find a fontCharCode that maps to the base and accent glyphs. + // If one doesn't exists, create it. + var charCodeToGlyphId = newMapping.charCodeToGlyphId; + var baseFontCharCode = createCharCode(charCodeToGlyphId, + baseGlyphId); + var accentFontCharCode = createCharCode(charCodeToGlyphId, + accentGlyphId); + seacMap[charCode] = { + baseFontCharCode: baseFontCharCode, + accentFontCharCode: accentFontCharCode, + accentOffset: accentOffset + }; } - glyf.data.set(newGlyfData.subarray(0, firstEntryLength), writeOffset); - itemEncode(loca.data, locaData.length - itemSize, - writeOffset + firstEntryLength); - } else { - glyf.data = newGlyfData.subarray(0, writeOffset); } - return missingGlyphData; + properties.seacMap = seacMap; } - function readPostScriptTable(post, properties, maxpNumGlyphs) { - var start = (font.start ? font.start : 0) + post.offset; - font.pos = start; + var unitsPerEm = 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]; - var length = post.length, end = start + length; - var version = font.getInt32(); - // skip rest to the tables - font.getBytes(28); + var builder = new OpenTypeFileBuilder('\x4F\x54\x54\x4F'); + // PostScript Font Program + builder.addTable('CFF ', font.data); + // OS/2 and Windows Specific metrics + builder.addTable('OS/2', createOS2Table(properties, + newMapping.charCodeToGlyphId)); + // Character to glyphs mapping + builder.addTable('cmap', createCmapTable(newMapping.charCodeToGlyphId, + numGlyphs)); + // Font header + builder.addTable('head', + '\x00\x01\x00\x00' + // Version number + '\x00\x00\x10\x00' + // fontRevision + '\x00\x00\x00\x00' + // checksumAdjustement + '\x5F\x0F\x3C\xF5' + // magicNumber + '\x00\x00' + // Flags + safeString16(unitsPerEm) + // unitsPerEM + '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // creation date + '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // modifification date + '\x00\x00' + // xMin + safeString16(properties.descent) + // yMin + '\x0F\xFF' + // xMax + safeString16(properties.ascent) + // yMax + string16(properties.italicAngle ? 2 : 0) + // macStyle + '\x00\x11' + // lowestRecPPEM + '\x00\x00' + // fontDirectionHint + '\x00\x00' + // indexToLocFormat + '\x00\x00'); // glyphDataFormat - var glyphNames; - var valid = true; - var i; + // Horizontal header + builder.addTable('hhea', + '\x00\x01\x00\x00' + // Version number + safeString16(properties.ascent) + // Typographic Ascent + safeString16(properties.descent) + // Typographic Descent + '\x00\x00' + // Line Gap + '\xFF\xFF' + // advanceWidthMax + '\x00\x00' + // minLeftSidebearing + '\x00\x00' + // minRightSidebearing + '\x00\x00' + // xMaxExtent + safeString16(properties.capHeight) + // caretSlopeRise + safeString16(Math.tan(properties.italicAngle) * + properties.xHeight) + // caretSlopeRun + '\x00\x00' + // caretOffset + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // metricDataFormat + string16(numGlyphs)); // Number of HMetrics - switch (version) { - case 0x00010000: - glyphNames = MacStandardGlyphOrdering; - break; - case 0x00020000: - var numGlyphs = font.getUint16(); - if (numGlyphs !== maxpNumGlyphs) { - valid = false; - break; - } - var glyphNameIndexes = []; - for (i = 0; i < numGlyphs; ++i) { - var index = font.getUint16(); - if (index >= 32768) { - valid = false; - break; - } - glyphNameIndexes.push(index); - } - if (!valid) { - break; - } - var customNames = []; - var strBuf = []; - while (font.pos < end) { - var stringLength = font.getByte(); - strBuf.length = stringLength; - for (i = 0; i < stringLength; ++i) { - strBuf[i] = String.fromCharCode(font.getByte()); - } - customNames.push(strBuf.join('')); - } - glyphNames = []; - for (i = 0; i < numGlyphs; ++i) { - var j = glyphNameIndexes[i]; - if (j < 258) { - glyphNames.push(MacStandardGlyphOrdering[j]); - continue; - } - glyphNames.push(customNames[j - 258]); - } - break; - case 0x00030000: - break; - default: - warn('Unknown/unsupported post table version ' + version); - valid = false; - if (properties.defaultEncoding) { - glyphNames = properties.defaultEncoding; + // Horizontal metrics + builder.addTable('hmtx', (function fontFieldsHmtx() { + var charstrings = font.charstrings; + var cffWidths = font.cff ? font.cff.widths : null; + var hmtx = '\x00\x00\x00\x00'; // Fake .notdef + for (var i = 1, ii = numGlyphs; i < ii; i++) { + var width = 0; + if (charstrings) { + var charstring = charstrings[i - 1]; + width = 'width' in charstring ? charstring.width : 0; + } else if (cffWidths) { + width = Math.ceil(cffWidths[i] || 0); } - break; - } - properties.glyphNames = glyphNames; - return valid; - } + hmtx += string16(width) + string16(0); + } + return hmtx; + })()); - function readNameTable(nameTable) { - var start = (font.start ? font.start : 0) + nameTable.offset; - font.pos = start; + // Maximum profile + builder.addTable('maxp', + '\x00\x00\x50\x00' + // Version number + string16(numGlyphs)); // Num of glyphs - var names = [[], []]; - var length = nameTable.length, end = start + length; - var format = font.getUint16(); - var FORMAT_0_HEADER_LENGTH = 6; - if (format !== 0 || length < FORMAT_0_HEADER_LENGTH) { - // unsupported name table format or table "too" small - return names; - } - var numRecords = font.getUint16(); - var stringsStart = font.getUint16(); - var records = []; - var NAME_RECORD_LENGTH = 12; - var i, ii; + // Naming tables + builder.addTable('name', createNameTable(fontName)); - for (i = 0; i < numRecords && - font.pos + NAME_RECORD_LENGTH <= end; i++) { - var r = { - platform: font.getUint16(), - encoding: font.getUint16(), - language: font.getUint16(), - name: font.getUint16(), - length: font.getUint16(), - offset: font.getUint16() - }; - // using only Macintosh and Windows platform/encoding names - if ((r.platform === 1 && r.encoding === 0 && r.language === 0) || - (r.platform === 3 && r.encoding === 1 && r.language === 0x409)) { - records.push(r); - } + // PostScript informations + builder.addTable('post', createPostTable(properties)); + + return builder.toArray(); + }, + + /** + * Builds a char code to unicode map based on section 9.10 of the spec. + * @param {Object} properties Font properties object. + * @return {Object} A ToUnicodeMap object. + */ + buildToUnicode: function Font_buildToUnicode(properties) { + // Section 9.10.2 Mapping Character Codes to Unicode Values + if (properties.toUnicode && properties.toUnicode.length !== 0) { + return properties.toUnicode; + } + // According to the spec if the font is a simple font we should only map + // to unicode if the base encoding is MacRoman, MacExpert, or WinAnsi or + // the differences array only contains adobe standard or symbol set names, + // in pratice it seems better to always try to create a toUnicode + // map based of the default encoding. + var toUnicode, charcode; + if (!properties.composite /* is simple font */) { + toUnicode = []; + var encoding = properties.defaultEncoding.slice(); + var baseEncodingName = properties.baseEncodingName; + // Merge in the differences array. + var differences = properties.differences; + for (charcode in differences) { + encoding[charcode] = differences[charcode]; } - for (i = 0, ii = records.length; i < ii; i++) { - var record = records[i]; - var pos = start + stringsStart + record.offset; - if (pos + record.length > end) { - continue; // outside of name table, ignoring - } - font.pos = pos; - var nameIndex = record.name; - if (record.encoding) { - // unicode - var str = ''; - for (var j = 0, jj = record.length; j < jj; j += 2) { - str += String.fromCharCode(font.getUint16()); + var glyphsUnicodeMap = getGlyphsUnicode(); + for (charcode in encoding) { + // a) Map the character code to a character name. + var glyphName = encoding[charcode]; + // b) Look up the character name in the Adobe Glyph List (see the + // Bibliography) to obtain the corresponding Unicode value. + if (glyphName === '') { + continue; + } else if (glyphsUnicodeMap[glyphName] === undefined) { + // (undocumented) c) Few heuristics to recognize unknown glyphs + // NOTE: Adobe Reader does not do this step, but OSX Preview does + var code = 0; + switch (glyphName[0]) { + case 'G': // Gxx glyph + if (glyphName.length === 3) { + code = parseInt(glyphName.substr(1), 16); + } + break; + case 'g': // g00xx glyph + if (glyphName.length === 5) { + code = parseInt(glyphName.substr(1), 16); + } + break; + case 'C': // Cddd glyph + case 'c': // cddd glyph + if (glyphName.length >= 3) { + code = +glyphName.substr(1); + } + break; } - names[1][nameIndex] = str; - } else { - names[0][nameIndex] = bytesToString(font.getBytes(record.length)); + if (code) { + // If |baseEncodingName| is one the predefined encodings, + // and |code| equals |charcode|, using the glyph defined in the + // baseEncoding seems to yield a better |toUnicode| mapping + // (fixes issue 5070). + if (baseEncodingName && code === +charcode) { + var baseEncoding = getEncoding(baseEncodingName); + if (baseEncoding && (glyphName = baseEncoding[charcode])) { + toUnicode[charcode] = + String.fromCharCode(glyphsUnicodeMap[glyphName]); + continue; + } + } + toUnicode[charcode] = String.fromCharCode(code); + } + continue; } + toUnicode[charcode] = + String.fromCharCode(glyphsUnicodeMap[glyphName]); } - return names; + return new ToUnicodeMap(toUnicode); } - - var TTOpsStackDeltas = [ - 0, 0, 0, 0, 0, 0, 0, 0, -2, -2, -2, -2, 0, 0, -2, -5, - -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, -1, -1, - 1, -1, -999, 0, 1, 0, -1, -2, 0, -1, -2, -1, -1, 0, -1, -1, - 0, 0, -999, -999, -1, -1, -1, -1, -2, -999, -2, -2, -999, 0, -2, -2, - 0, 0, -2, 0, -2, 0, 0, 0, -2, -1, -1, 1, 1, 0, 0, -1, - -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, 0, -999, -1, -1, - -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - -2, -999, -999, -999, -999, -999, -1, -1, -2, -2, 0, 0, 0, 0, -1, -1, - -999, -2, -2, 0, 0, -1, -2, -2, 0, 0, 0, -1, -1, -1, -2]; - // 0xC0-DF == -1 and 0xE0-FF == -2 - - function sanitizeTTProgram(table, ttContext) { - var data = table.data; - var i = 0, j, n, b, funcId, pc, lastEndf = 0, lastDeff = 0; - var stack = []; - var callstack = []; - var functionsCalled = []; - var tooComplexToFollowFunctions = - ttContext.tooComplexToFollowFunctions; - var inFDEF = false, ifLevel = 0, inELSE = 0; - for (var ii = data.length; i < ii;) { - var op = data[i++]; - // The TrueType instruction set docs can be found at - // https://developer.apple.com/fonts/TTRefMan/RM05/Chap5.html - if (op === 0x40) { // NPUSHB - pushes n bytes - n = data[i++]; - if (inFDEF || inELSE) { - i += n; - } else { - for (j = 0; j < n; j++) { - stack.push(data[i++]); - } - } - } else if (op === 0x41) { // NPUSHW - pushes n words - n = data[i++]; - if (inFDEF || inELSE) { - i += n * 2; - } else { - for (j = 0; j < n; j++) { - b = data[i++]; - stack.push((b << 8) | data[i++]); - } - } - } else if ((op & 0xF8) === 0xB0) { // PUSHB - pushes bytes - n = op - 0xB0 + 1; - if (inFDEF || inELSE) { - i += n; - } else { - for (j = 0; j < n; j++) { - stack.push(data[i++]); - } - } - } else if ((op & 0xF8) === 0xB8) { // PUSHW - pushes words - n = op - 0xB8 + 1; - if (inFDEF || inELSE) { - i += n * 2; - } else { - for (j = 0; j < n; j++) { - b = data[i++]; - stack.push((b << 8) | data[i++]); - } - } - } else if (op === 0x2B && !tooComplexToFollowFunctions) { // CALL - if (!inFDEF && !inELSE) { - // collecting inforamtion about which functions are used - funcId = stack[stack.length - 1]; - ttContext.functionsUsed[funcId] = true; - if (funcId in ttContext.functionsStackDeltas) { - stack.length += ttContext.functionsStackDeltas[funcId]; - } else if (funcId in ttContext.functionsDefined && - functionsCalled.indexOf(funcId) < 0) { - callstack.push({data: data, i: i, stackTop: stack.length - 1}); - functionsCalled.push(funcId); - pc = ttContext.functionsDefined[funcId]; - if (!pc) { - warn('TT: CALL non-existent function'); - ttContext.hintsValid = false; - return; - } - data = pc.data; - i = pc.i; - } - } - } else if (op === 0x2C && !tooComplexToFollowFunctions) { // FDEF - if (inFDEF || inELSE) { - warn('TT: nested FDEFs not allowed'); - tooComplexToFollowFunctions = true; - } - inFDEF = true; - // collecting inforamtion about which functions are defined - lastDeff = i; - funcId = stack.pop(); - ttContext.functionsDefined[funcId] = {data: data, i: i}; - } else if (op === 0x2D) { // ENDF - end of function - if (inFDEF) { - inFDEF = false; - lastEndf = i; - } else { - pc = callstack.pop(); - if (!pc) { - warn('TT: ENDF bad stack'); - ttContext.hintsValid = false; - return; - } - funcId = functionsCalled.pop(); - data = pc.data; - i = pc.i; - ttContext.functionsStackDeltas[funcId] = - stack.length - pc.stackTop; - } - } else if (op === 0x89) { // IDEF - instruction definition - if (inFDEF || inELSE) { - warn('TT: nested IDEFs not allowed'); - tooComplexToFollowFunctions = true; - } - inFDEF = true; - // recording it as a function to track ENDF - lastDeff = i; - } else if (op === 0x58) { // IF - ++ifLevel; - } else if (op === 0x1B) { // ELSE - inELSE = ifLevel; - } else if (op === 0x59) { // EIF - if (inELSE === ifLevel) { - inELSE = 0; - } - --ifLevel; - } else if (op === 0x1C) { // JMPR - if (!inFDEF && !inELSE) { - var offset = stack[stack.length - 1]; - // only jumping forward to prevent infinite loop - if (offset > 0) { - i += offset - 1; - } - } - } - // Adjusting stack not extactly, but just enough to get function id - if (!inFDEF && !inELSE) { - var stackDelta = op <= 0x8E ? TTOpsStackDeltas[op] : - op >= 0xC0 && op <= 0xDF ? -1 : op >= 0xE0 ? -2 : 0; - if (op >= 0x71 && op <= 0x75) { - n = stack.pop(); - if (n === n) { - stackDelta = -n * 2; - } - } - while (stackDelta < 0 && stack.length > 0) { - stack.pop(); - stackDelta++; - } - while (stackDelta > 0) { - stack.push(NaN); // pushing any number into stack - stackDelta--; - } + // If the font is a composite font that uses one of the predefined CMaps + // listed in Table 118 (except Identity–H and Identity–V) or whose + // descendant CIDFont uses the Adobe-GB1, Adobe-CNS1, Adobe-Japan1, or + // Adobe-Korea1 character collection: + if (properties.composite && ( + (properties.cMap.builtInCMap && + !(properties.cMap instanceof IdentityCMap)) || + (properties.cidSystemInfo.registry === 'Adobe' && + (properties.cidSystemInfo.ordering === 'GB1' || + properties.cidSystemInfo.ordering === 'CNS1' || + properties.cidSystemInfo.ordering === 'Japan1' || + properties.cidSystemInfo.ordering === 'Korea1')))) { + // Then: + // a) Map the character code to a character identifier (CID) according + // to the font’s CMap. + // b) Obtain the registry and ordering of the character collection used + // by the font’s CMap (for example, Adobe and Japan1) from its + // CIDSystemInfo dictionary. + var registry = properties.cidSystemInfo.registry; + var ordering = properties.cidSystemInfo.ordering; + // c) Construct a second CMap name by concatenating the registry and + // ordering obtained in step (b) in the format registry–ordering–UCS2 + // (for example, Adobe–Japan1–UCS2). + var ucs2CMapName = new Name(registry + '-' + ordering + '-UCS2'); + // d) Obtain the CMap with the name constructed in step (c) (available + // from the ASN Web site; see the Bibliography). + var ucs2CMap = CMapFactory.create(ucs2CMapName, + { url: PDFJS.cMapUrl, packed: PDFJS.cMapPacked }, null); + var cMap = properties.cMap; + toUnicode = []; + cMap.forEach(function(charcode, cid) { + assert(cid <= 0xffff, 'Max size of CID is 65,535'); + // e) Map the CID obtained in step (a) according to the CMap obtained + // in step (d), producing a Unicode value. + var ucs2 = ucs2CMap.lookup(cid); + if (ucs2) { + toUnicode[charcode] = + String.fromCharCode((ucs2.charCodeAt(0) << 8) + + ucs2.charCodeAt(1)); } - } - ttContext.tooComplexToFollowFunctions = tooComplexToFollowFunctions; - var content = [data]; - if (i > data.length) { - content.push(new Uint8Array(i - data.length)); - } - if (lastDeff > lastEndf) { - warn('TT: complementing a missing function tail'); - // new function definition started, but not finished - // complete function by [CLEAR, ENDF] - content.push(new Uint8Array([0x22, 0x2D])); - } - foldTTTable(table, content); + }); + return new ToUnicodeMap(toUnicode); } - function checkInvalidFunctions(ttContext, maxFunctionDefs) { - if (ttContext.tooComplexToFollowFunctions) { - return; - } - if (ttContext.functionsDefined.length > maxFunctionDefs) { - warn('TT: more functions defined than expected'); - ttContext.hintsValid = false; - return; - } - for (var j = 0, jj = ttContext.functionsUsed.length; j < jj; j++) { - if (j > maxFunctionDefs) { - warn('TT: invalid function id: ' + j); - ttContext.hintsValid = false; - return; - } - if (ttContext.functionsUsed[j] && !ttContext.functionsDefined[j]) { - warn('TT: undefined function: ' + j); - ttContext.hintsValid = false; - return; - } - } - } + // The viewer's choice, just use an identity map. + return new IdentityToUnicodeMap(properties.firstChar, + properties.lastChar); + }, - function foldTTTable(table, content) { - if (content.length > 1) { - // concatenating the content items - var newLength = 0; - var j, jj; - for (j = 0, jj = content.length; j < jj; j++) { - newLength += content[j].length; - } - newLength = (newLength + 3) & ~3; - var result = new Uint8Array(newLength); - var pos = 0; - for (j = 0, jj = content.length; j < jj; j++) { - result.set(content[j], pos); - pos += content[j].length; - } - table.data = result; - table.length = newLength; - } + get spaceWidth() { + if ('_shadowWidth' in this) { + return this._shadowWidth; } - function sanitizeTTPrograms(fpgm, prep, cvt) { - var ttContext = { - functionsDefined: [], - functionsUsed: [], - functionsStackDeltas: [], - tooComplexToFollowFunctions: false, - hintsValid: true - }; - if (fpgm) { - sanitizeTTProgram(fpgm, ttContext); - } - if (prep) { - sanitizeTTProgram(prep, ttContext); + // trying to estimate space character width + var possibleSpaceReplacements = ['space', 'minus', 'one', 'i']; + var width; + for (var i = 0, ii = possibleSpaceReplacements.length; i < ii; i++) { + var glyphName = possibleSpaceReplacements[i]; + // if possible, getting width by glyph name + if (glyphName in this.widths) { + width = this.widths[glyphName]; + break; } - if (fpgm) { - checkInvalidFunctions(ttContext, maxFunctionDefs); + var glyphsUnicodeMap = getGlyphsUnicode(); + var glyphUnicode = glyphsUnicodeMap[glyphName]; + // finding the charcode via unicodeToCID map + var charcode = 0; + if (this.composite) { + if (this.cMap.contains(glyphUnicode)) { + charcode = this.cMap.lookup(glyphUnicode); + } } - if (cvt && (cvt.length & 1)) { - var cvtData = new Uint8Array(cvt.length + 1); - cvtData.set(cvt.data); - cvt.data = cvtData; + // ... via toUnicode map + if (!charcode && this.toUnicode) { + charcode = this.toUnicode.charCodeOf(glyphUnicode); } - return ttContext.hintsValid; - } - - // The following steps modify the original font data, making copy - font = new Stream(new Uint8Array(font.getBytes())); - - var VALID_TABLES = ['OS/2', 'cmap', 'head', 'hhea', 'hmtx', 'maxp', - 'name', 'post', 'loca', 'glyf', 'fpgm', 'prep', 'cvt ', 'CFF ']; - - var header = readOpenTypeHeader(font); - var numTables = header.numTables; - var cff, cffFile; - - var tables = Object.create(null); - tables['OS/2'] = null; - tables['cmap'] = null; - tables['head'] = null; - tables['hhea'] = null; - tables['hmtx'] = null; - tables['maxp'] = null; - tables['name'] = null; - tables['post'] = null; - - var table; - for (var i = 0; i < numTables; i++) { - table = readTableEntry(font); - if (VALID_TABLES.indexOf(table.tag) < 0) { - continue; // skipping table if it's not a required or optional table + // setting it to unicode if negative or undefined + if (charcode <= 0) { + charcode = glyphUnicode; } - if (table.length === 0) { - continue; // skipping empty tables + // trying to get width via charcode + width = this.widths[charcode]; + if (width) { + break; // the non-zero width found } - tables[table.tag] = table; } + width = width || this.defaultWidth; + // Do not shadow the property here. See discussion: + // https://github.com/mozilla/pdf.js/pull/2127#discussion_r1662280 + this._shadowWidth = width; + return width; + }, - var isTrueType = !tables['CFF ']; - if (!isTrueType) { - // OpenType font - if ((header.version === 'OTTO' && properties.type !== 'CIDFontType2') || - !tables['head'] || !tables['hhea'] || !tables['maxp'] || - !tables['post']) { - // no major tables: throwing everything at CFFFont - cffFile = new Stream(tables['CFF '].data); - cff = new CFFFont(cffFile, properties); - - adjustWidths(properties); - - return this.convert(name, cff, properties); - } + charToGlyph: function Font_charToGlyph(charcode, isSpace) { + var fontCharCode, width, operatorListId; - delete tables['glyf']; - delete tables['loca']; - delete tables['fpgm']; - delete tables['prep']; - delete tables['cvt ']; - this.isOpenType = true; - } else { - if (!tables['loca']) { - error('Required "loca" table is not found'); - } - if (!tables['glyf']) { - warn('Required "glyf" table is not found -- trying to recover.'); - // Note: We use `sanitizeGlyphLocations` to add dummy glyf data below. - tables['glyf'] = { - tag: 'glyf', - data: new Uint8Array(0), - }; - } - this.isOpenType = false; + var widthCode = charcode; + if (this.cMap && this.cMap.contains(charcode)) { + widthCode = this.cMap.lookup(charcode); } + width = this.widths[widthCode]; + width = isNum(width) ? width : this.defaultWidth; + var vmetric = this.vmetrics && this.vmetrics[widthCode]; - if (!tables['maxp']) { - error('Required "maxp" table is not found'); + var unicode = this.toUnicode.get(charcode) || charcode; + if (typeof unicode === 'number') { + unicode = String.fromCharCode(unicode); } - font.pos = (font.start || 0) + tables['maxp'].offset; - var version = font.getInt32(); - var numGlyphs = font.getUint16(); - var maxFunctionDefs = 0; - if (version >= 0x00010000 && tables['maxp'].length >= 22) { - // maxZones can be invalid - font.pos += 8; - var maxZones = font.getUint16(); - if (maxZones > 2) { // reset to 2 if font has invalid maxZones - tables['maxp'].data[14] = 0; - tables['maxp'].data[15] = 2; - } - font.pos += 4; - maxFunctionDefs = font.getUint16(); + // First try the toFontChar map, if it's not there then try falling + // back to the char code. + fontCharCode = this.toFontChar[charcode] || charcode; + if (this.missingFile) { + fontCharCode = mapSpecialUnicodeValues(fontCharCode); } - var dupFirstEntry = false; - if (properties.type === 'CIDFontType2' && properties.toUnicode && - properties.toUnicode.get(0) > '\u0000') { - // oracle's defect (see 3427), duplicating first entry - dupFirstEntry = true; - numGlyphs++; - tables['maxp'].data[4] = numGlyphs >> 8; - tables['maxp'].data[5] = numGlyphs & 255; + if (this.isType3Font) { + // Font char code in this case is actually a glyph name. + operatorListId = fontCharCode; } - var hintsValid = sanitizeTTPrograms(tables['fpgm'], tables['prep'], - tables['cvt '], maxFunctionDefs); - if (!hintsValid) { - delete tables['fpgm']; - delete tables['prep']; - delete tables['cvt ']; + var accent = null; + if (this.seacMap && this.seacMap[charcode]) { + var seac = this.seacMap[charcode]; + fontCharCode = seac.baseFontCharCode; + accent = { + fontChar: String.fromCharCode(seac.accentFontCharCode), + offset: seac.accentOffset + }; } - // Ensure the hmtx table contains the advance width and - // sidebearings information for numGlyphs in the maxp table - sanitizeMetrics(font, tables['hhea'], tables['hmtx'], numGlyphs); + var fontChar = String.fromCharCode(fontCharCode); - if (!tables['head']) { - error('Required "head" table is not found'); + var glyph = this.glyphCache[charcode]; + if (!glyph || + !glyph.matchesForCache(fontChar, unicode, accent, width, vmetric, + operatorListId, isSpace)) { + glyph = new Glyph(fontChar, unicode, accent, width, vmetric, + operatorListId, isSpace); + this.glyphCache[charcode] = glyph; } + return glyph; + }, - sanitizeHead(tables['head'], numGlyphs, - isTrueType ? tables['loca'].length : 0); - - var missingGlyphs = Object.create(null); - if (isTrueType) { - var isGlyphLocationsLong = int16(tables['head'].data[50], - tables['head'].data[51]); - missingGlyphs = sanitizeGlyphLocations(tables['loca'], tables['glyf'], - numGlyphs, isGlyphLocationsLong, - hintsValid, dupFirstEntry); - } + charsToGlyphs: function Font_charsToGlyphs(chars) { + var charsCache = this.charsCache; + var glyphs, glyph, charcode; - if (!tables['hhea']) { - error('Required "hhea" table is not found'); + // if we translated this string before, just grab it from the cache + if (charsCache) { + glyphs = charsCache[chars]; + if (glyphs) { + return glyphs; + } } - // Sanitizer reduces the glyph advanceWidth to the maxAdvanceWidth - // Sometimes it's 0. That needs to be fixed - if (tables['hhea'].data[10] === 0 && tables['hhea'].data[11] === 0) { - tables['hhea'].data[10] = 0xFF; - tables['hhea'].data[11] = 0xFF; + // lazily create the translation cache + if (!charsCache) { + charsCache = this.charsCache = Object.create(null); } - // Extract some more font properties from the OpenType head and - // hhea tables; yMin and descent value are always negative. - var metricsOverride = { - unitsPerEm: int16(tables['head'].data[18], tables['head'].data[19]), - yMax: int16(tables['head'].data[42], tables['head'].data[43]), - yMin: int16(tables['head'].data[38], tables['head'].data[39]) - 0x10000, - ascent: int16(tables['hhea'].data[4], tables['hhea'].data[5]), - descent: int16(tables['hhea'].data[6], tables['hhea'].data[7]) - 0x10000 - }; - - // PDF FontDescriptor metrics lie -- using data from actual font. - this.ascent = metricsOverride.ascent / metricsOverride.unitsPerEm; - this.descent = metricsOverride.descent / metricsOverride.unitsPerEm; + glyphs = []; + var charsCacheKey = chars; + var i = 0, ii; - // The 'post' table has glyphs names. - if (tables['post']) { - var valid = readPostScriptTable(tables['post'], properties, numGlyphs); - if (!valid) { - tables['post'] = null; + if (this.cMap) { + // composite fonts have multi-byte strings convert the string from + // single-byte to multi-byte + var c = Object.create(null); + while (i < chars.length) { + this.cMap.readCharCode(chars, i, c); + charcode = c.charcode; + var length = c.length; + i += length; + // Space is char with code 0x20 and length 1 in multiple-byte codes. + var isSpace = length === 1 && chars.charCodeAt(i - 1) === 0x20; + glyph = this.charToGlyph(charcode, isSpace); + glyphs.push(glyph); + } + } else { + for (i = 0, ii = chars.length; i < ii; ++i) { + charcode = chars.charCodeAt(i); + glyph = this.charToGlyph(charcode, charcode === 0x20); + glyphs.push(glyph); } } - var charCodeToGlyphId = [], charCode; - var toUnicode = properties.toUnicode, widths = properties.widths; - var skipToUnicode = (toUnicode instanceof IdentityToUnicodeMap || - toUnicode.length === 0x10000); + // Enter the translated string into the cache + return (charsCache[charsCacheKey] = glyphs); + } + }; - // Helper function to try to skip mapping of empty glyphs. - // Note: In some cases, just relying on the glyph data doesn't work, - // hence we also use a few heuristics to fix various PDF files. - function hasGlyph(glyphId, charCode, widthCode) { - if (!missingGlyphs[glyphId]) { - return true; - } - if (!skipToUnicode && charCode >= 0 && toUnicode.has(charCode)) { - return true; - } - if (widths && widthCode >= 0 && isNum(widths[widthCode])) { - return true; - } - return false; - } + return Font; +})(); - if (properties.type === 'CIDFontType2') { - var cidToGidMap = properties.cidToGidMap || []; - var isCidToGidMapEmpty = cidToGidMap.length === 0; +var ErrorFont = (function ErrorFontClosure() { + function ErrorFont(error) { + this.error = error; + this.loadedName = 'g_font_error'; + this.loading = false; + } - properties.cMap.forEach(function(charCode, cid) { - assert(cid <= 0xffff, 'Max size of CID is 65,535'); - var glyphId = -1; - if (isCidToGidMapEmpty) { - glyphId = charCode; - } else if (cidToGidMap[cid] !== undefined) { - glyphId = cidToGidMap[cid]; - } + ErrorFont.prototype = { + charsToGlyphs: function ErrorFont_charsToGlyphs() { + return []; + }, + exportData: function ErrorFont_exportData() { + return {error: this.error}; + } + }; - if (glyphId >= 0 && glyphId < numGlyphs && - hasGlyph(glyphId, charCode, cid)) { - charCodeToGlyphId[charCode] = glyphId; - } - }); - if (dupFirstEntry) { - charCodeToGlyphId[0] = numGlyphs - 1; - } + return ErrorFont; +})(); + +/** + * Shared logic for building a char code to glyph id mapping for Type1 and + * simple CFF fonts. See section 9.6.6.2 of the spec. + * @param {Object} properties Font properties object. + * @param {Object} builtInEncoding The encoding contained within the actual font + * data. + * @param {Array} Array of glyph names where the index is the glyph ID. + * @returns {Object} A char code to glyph ID map. + */ +function type1FontGlyphMapping(properties, builtInEncoding, glyphNames) { + var charCodeToGlyphId = Object.create(null); + var glyphId, charCode, baseEncoding; + + if (properties.baseEncodingName) { + // If a valid base encoding name was used, the mapping is initialized with + // that. + baseEncoding = getEncoding(properties.baseEncodingName); + for (charCode = 0; charCode < baseEncoding.length; charCode++) { + glyphId = glyphNames.indexOf(baseEncoding[charCode]); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; } else { - // Most of the following logic in this code branch is based on the - // 9.6.6.4 of the PDF spec. - var hasEncoding = - properties.differences.length > 0 || !!properties.baseEncodingName; - var cmapTable = - readCmapTable(tables['cmap'], font, this.isSymbolicFont, hasEncoding); - var cmapPlatformId = cmapTable.platformId; - var cmapEncodingId = cmapTable.encodingId; - var cmapMappings = cmapTable.mappings; - var cmapMappingsLength = cmapMappings.length; + charCodeToGlyphId[charCode] = 0; // notdef + } + } + } else if (!!(properties.flags & FontFlags.Symbolic)) { + // For a symbolic font the encoding should be the fonts built-in + // encoding. + for (charCode in builtInEncoding) { + charCodeToGlyphId[charCode] = builtInEncoding[charCode]; + } + } else { + // For non-symbolic fonts that don't have a base encoding the standard + // encoding should be used. + baseEncoding = StandardEncoding; + for (charCode = 0; charCode < baseEncoding.length; charCode++) { + glyphId = glyphNames.indexOf(baseEncoding[charCode]); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; + } else { + charCodeToGlyphId[charCode] = 0; // notdef + } + } + } - // The spec seems to imply that if the font is symbolic the encoding - // should be ignored, this doesn't appear to work for 'preistabelle.pdf' - // where the the font is symbolic and it has an encoding. - if (hasEncoding && - (cmapPlatformId === 3 && cmapEncodingId === 1 || - cmapPlatformId === 1 && cmapEncodingId === 0) || - (cmapPlatformId === -1 && cmapEncodingId === -1 && // Temporary hack - !!getEncoding(properties.baseEncodingName))) { // Temporary hack - // When no preferred cmap table was found and |baseEncodingName| is - // one of the predefined encodings, we seem to obtain a better - // |charCodeToGlyphId| map from the code below (fixes bug 1057544). - // TODO: Note that this is a hack which should be removed as soon as - // we have proper support for more exotic cmap tables. + // Lastly, merge in the differences. + var differences = properties.differences; + if (differences) { + for (charCode in differences) { + var glyphName = differences[charCode]; + glyphId = glyphNames.indexOf(glyphName); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; + } else { + charCodeToGlyphId[charCode] = 0; // notdef + } + } + } + return charCodeToGlyphId; +} - var baseEncoding = []; - if (properties.baseEncodingName === 'MacRomanEncoding' || - properties.baseEncodingName === 'WinAnsiEncoding') { - baseEncoding = getEncoding(properties.baseEncodingName); - } - var glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode = 0; charCode < 256; charCode++) { - var glyphName; - if (this.differences && charCode in this.differences) { - glyphName = this.differences[charCode]; - } else if (charCode in baseEncoding && - baseEncoding[charCode] !== '') { - glyphName = baseEncoding[charCode]; - } else { - glyphName = StandardEncoding[charCode]; - } - if (!glyphName) { - continue; - } - var unicodeOrCharCode, isUnicode = false; - if (cmapPlatformId === 3 && cmapEncodingId === 1) { - unicodeOrCharCode = glyphsUnicodeMap[glyphName]; - isUnicode = true; - } else if (cmapPlatformId === 1 && cmapEncodingId === 0) { - // TODO: the encoding needs to be updated with mac os table. - unicodeOrCharCode = MacRomanEncoding.indexOf(glyphName); - } +/* + * CharStrings are encoded following the the CharString Encoding sequence + * describe in Chapter 6 of the "Adobe Type1 Font Format" specification. + * The value in a byte indicates a command, a number, or subsequent bytes + * that are to be interpreted in a special way. + * + * CharString Number Encoding: + * A CharString byte containing the values from 32 through 255 inclusive + * indicate an integer. These values are decoded in four ranges. + * + * 1. A CharString byte containing a value, v, between 32 and 246 inclusive, + * indicate the integer v - 139. Thus, the integer values from -107 through + * 107 inclusive may be encoded in single byte. + * + * 2. A CharString byte containing a value, v, between 247 and 250 inclusive, + * indicates an integer involving the next byte, w, according to the formula: + * [(v - 247) x 256] + w + 108 + * + * 3. A CharString byte containing a value, v, between 251 and 254 inclusive, + * indicates an integer involving the next byte, w, according to the formula: + * -[(v - 251) * 256] - w - 108 + * + * 4. A CharString containing the value 255 indicates that the next 4 bytes + * are a two complement signed integer. The first of these bytes contains the + * highest order bits, the second byte contains the next higher order bits + * and the fourth byte contain the lowest order bits. + * + * + * CharString Command Encoding: + * CharStrings commands are encoded in 1 or 2 bytes. + * + * Single byte commands are encoded in 1 byte that contains a value between + * 0 and 31 inclusive. + * If a command byte contains the value 12, then the value in the next byte + * indicates a command. This "escape" mechanism allows many extra commands + * to be encoded and this encoding technique helps to minimize the length of + * the charStrings. + */ +var Type1CharString = (function Type1CharStringClosure() { + var COMMAND_MAP = { + 'hstem': [1], + 'vstem': [3], + 'vmoveto': [4], + 'rlineto': [5], + 'hlineto': [6], + 'vlineto': [7], + 'rrcurveto': [8], + 'callsubr': [10], + 'flex': [12, 35], + 'drop' : [12, 18], + 'endchar': [14], + 'rmoveto': [21], + 'hmoveto': [22], + 'vhcurveto': [30], + 'hvcurveto': [31] + }; - var found = false; - for (i = 0; i < cmapMappingsLength; ++i) { - if (cmapMappings[i].charCode !== unicodeOrCharCode) { - continue; + function Type1CharString() { + this.width = 0; + this.lsb = 0; + this.flexing = false; + this.output = []; + this.stack = []; + } + + Type1CharString.prototype = { + convert: function Type1CharString_convert(encoded, subrs) { + var count = encoded.length; + var error = false; + var wx, sbx, subrNumber; + for (var i = 0; i < count; i++) { + var value = encoded[i]; + if (value < 32) { + if (value === 12) { + value = (value << 8) + encoded[++i]; + } + switch (value) { + case 1: // hstem + if (!HINTING_ENABLED) { + this.stack = []; + break; } - var code = isUnicode ? charCode : unicodeOrCharCode; - if (hasGlyph(cmapMappings[i].glyphId, code, -1)) { - charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; - found = true; + error = this.executeCommand(2, COMMAND_MAP.hstem); + break; + case 3: // vstem + if (!HINTING_ENABLED) { + this.stack = []; break; } - } - if (!found && properties.glyphNames) { - // Try to map using the post table. - var glyphId = properties.glyphNames.indexOf(glyphName); - if (glyphId > 0 && hasGlyph(glyphId, -1, -1)) { - charCodeToGlyphId[charCode] = glyphId; - found = true; + error = this.executeCommand(2, COMMAND_MAP.vstem); + break; + case 4: // vmoveto + if (this.flexing) { + if (this.stack.length < 1) { + error = true; + break; + } + // Add the dx for flex and but also swap the values so they are + // the right order. + var dy = this.stack.pop(); + this.stack.push(0, dy); + break; } - } - if (!found) { - charCodeToGlyphId[charCode] = 0; // notdef - } + error = this.executeCommand(1, COMMAND_MAP.vmoveto); + break; + case 5: // rlineto + error = this.executeCommand(2, COMMAND_MAP.rlineto); + break; + case 6: // hlineto + error = this.executeCommand(1, COMMAND_MAP.hlineto); + break; + case 7: // vlineto + error = this.executeCommand(1, COMMAND_MAP.vlineto); + break; + case 8: // rrcurveto + error = this.executeCommand(6, COMMAND_MAP.rrcurveto); + break; + case 9: // closepath + // closepath is a Type1 command that does not take argument and is + // useless in Type2 and it can simply be ignored. + this.stack = []; + break; + case 10: // callsubr + if (this.stack.length < 1) { + error = true; + break; + } + subrNumber = this.stack.pop(); + error = this.convert(subrs[subrNumber], subrs); + break; + case 11: // return + return error; + case 13: // hsbw + if (this.stack.length < 2) { + error = true; + break; + } + // To convert to type2 we have to move the width value to the + // first part of the charstring and then use hmoveto with lsb. + wx = this.stack.pop(); + sbx = this.stack.pop(); + this.lsb = sbx; + this.width = wx; + this.stack.push(wx, sbx); + error = this.executeCommand(2, COMMAND_MAP.hmoveto); + break; + case 14: // endchar + this.output.push(COMMAND_MAP.endchar[0]); + break; + case 21: // rmoveto + if (this.flexing) { + break; + } + error = this.executeCommand(2, COMMAND_MAP.rmoveto); + break; + case 22: // hmoveto + if (this.flexing) { + // Add the dy for flex. + this.stack.push(0); + break; + } + error = this.executeCommand(1, COMMAND_MAP.hmoveto); + break; + case 30: // vhcurveto + error = this.executeCommand(4, COMMAND_MAP.vhcurveto); + break; + case 31: // hvcurveto + error = this.executeCommand(4, COMMAND_MAP.hvcurveto); + break; + case (12 << 8) + 0: // dotsection + // dotsection is a Type1 command to specify some hinting feature + // for dots that do not take a parameter and it can safely be + // ignored for Type2. + this.stack = []; + break; + case (12 << 8) + 1: // vstem3 + if (!HINTING_ENABLED) { + this.stack = []; + break; + } + // [vh]stem3 are Type1 only and Type2 supports [vh]stem with + // multiple parameters, so instead of returning [vh]stem3 take a + // shortcut and return [vhstem] instead. + error = this.executeCommand(2, COMMAND_MAP.vstem); + break; + case (12 << 8) + 2: // hstem3 + if (!HINTING_ENABLED) { + this.stack = []; + break; + } + // See vstem3. + error = this.executeCommand(2, COMMAND_MAP.hstem); + break; + case (12 << 8) + 6: // seac + // seac is like type 2's special endchar but it doesn't use the + // first argument asb, so remove it. + if (SEAC_ANALYSIS_ENABLED) { + this.seac = this.stack.splice(-4, 4); + error = this.executeCommand(0, COMMAND_MAP.endchar); + } else { + error = this.executeCommand(4, COMMAND_MAP.endchar); + } + break; + case (12 << 8) + 7: // sbw + if (this.stack.length < 4) { + error = true; + break; + } + // To convert to type2 we have to move the width value to the + // first part of the charstring and then use rmoveto with + // (dx, dy). The height argument will not be used for vmtx and + // vhea tables reconstruction -- ignoring it. + var wy = this.stack.pop(); + wx = this.stack.pop(); + var sby = this.stack.pop(); + sbx = this.stack.pop(); + this.lsb = sbx; + this.width = wx; + this.stack.push(wx, sbx, sby); + error = this.executeCommand(3, COMMAND_MAP.rmoveto); + break; + case (12 << 8) + 12: // div + if (this.stack.length < 2) { + error = true; + break; + } + var num2 = this.stack.pop(); + var num1 = this.stack.pop(); + this.stack.push(num1 / num2); + break; + case (12 << 8) + 16: // callothersubr + if (this.stack.length < 2) { + error = true; + break; + } + subrNumber = this.stack.pop(); + var numArgs = this.stack.pop(); + if (subrNumber === 0 && numArgs === 3) { + var flexArgs = this.stack.splice(this.stack.length - 17, 17); + this.stack.push( + flexArgs[2] + flexArgs[0], // bcp1x + rpx + flexArgs[3] + flexArgs[1], // bcp1y + rpy + flexArgs[4], // bcp2x + flexArgs[5], // bcp2y + flexArgs[6], // p2x + flexArgs[7], // p2y + flexArgs[8], // bcp3x + flexArgs[9], // bcp3y + flexArgs[10], // bcp4x + flexArgs[11], // bcp4y + flexArgs[12], // p3x + flexArgs[13], // p3y + flexArgs[14] // flexDepth + // 15 = finalx unused by flex + // 16 = finaly unused by flex + ); + error = this.executeCommand(13, COMMAND_MAP.flex, true); + this.flexing = false; + this.stack.push(flexArgs[15], flexArgs[16]); + } else if (subrNumber === 1 && numArgs === 0) { + this.flexing = true; + } + break; + case (12 << 8) + 17: // pop + // Ignore this since it is only used with othersubr. + break; + case (12 << 8) + 33: // setcurrentpoint + // Ignore for now. + this.stack = []; + break; + default: + warn('Unknown type 1 charstring command of "' + value + '"'); + break; } - } else if (cmapPlatformId === 0 && cmapEncodingId === 0) { - // Default Unicode semantics, use the charcodes as is. - for (i = 0; i < cmapMappingsLength; ++i) { - charCodeToGlyphId[cmapMappings[i].charCode] = - cmapMappings[i].glyphId; + if (error) { + break; } + continue; + } else if (value <= 246) { + value = value - 139; + } else if (value <= 250) { + value = ((value - 247) * 256) + encoded[++i] + 108; + } else if (value <= 254) { + value = -((value - 251) * 256) - encoded[++i] - 108; } else { - // For (3, 0) cmap tables: - // The charcode key being stored in charCodeToGlyphId is the lower - // byte of the two-byte charcodes of the cmap table since according to - // the spec: 'each byte from the string shall be prepended with the - // high byte of the range [of charcodes in the cmap table], to form - // a two-byte character, which shall be used to select the - // associated glyph description from the subtable'. - // - // For (1, 0) cmap tables: - // 'single bytes from the string shall be used to look up the - // associated glyph descriptions from the subtable'. This means - // charcodes in the cmap will be single bytes, so no-op since - // glyph.charCode & 0xFF === glyph.charCode - for (i = 0; i < cmapMappingsLength; ++i) { - charCode = cmapMappings[i].charCode & 0xFF; - charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; - } + value = (encoded[++i] & 0xff) << 24 | (encoded[++i] & 0xff) << 16 | + (encoded[++i] & 0xff) << 8 | (encoded[++i] & 0xff) << 0; } + this.stack.push(value); } + return error; + }, - if (charCodeToGlyphId.length === 0) { - // defines at least one glyph - charCodeToGlyphId[0] = 0; + executeCommand: function(howManyArgs, command, keepStack) { + var stackLength = this.stack.length; + if (howManyArgs > stackLength) { + return true; } - - // Converting glyphs and ids into font's cmap table - var newMapping = adjustMapping(charCodeToGlyphId, properties); - this.toFontChar = newMapping.toFontChar; - tables['cmap'] = { - tag: 'cmap', - data: createCmapTable(newMapping.charCodeToGlyphId, numGlyphs) - }; - - if (!tables['OS/2'] || !validateOS2Table(tables['OS/2'])) { - tables['OS/2'] = { - tag: 'OS/2', - data: createOS2Table(properties, newMapping.charCodeToGlyphId, - metricsOverride) - }; - } - - // Rewrite the 'post' table if needed - if (!tables['post']) { - tables['post'] = { - tag: 'post', - data: createPostTable(properties) - }; - } - - if (!isTrueType) { - try { - // Trying to repair CFF file - cffFile = new Stream(tables['CFF '].data); - var parser = new CFFParser(cffFile, properties); - cff = parser.parse(); - var compiler = new CFFCompiler(cff); - tables['CFF '].data = compiler.compile(); - } catch (e) { - warn('Failed to compile font ' + properties.loadedName); + var start = stackLength - howManyArgs; + for (var i = start; i < stackLength; i++) { + var value = this.stack[i]; + if (value === (value | 0)) { // int + this.output.push(28, (value >> 8) & 0xff, value & 0xff); + } else { // fixed point + value = (65536 * value) | 0; + this.output.push(255, + (value >> 24) & 0xFF, + (value >> 16) & 0xFF, + (value >> 8) & 0xFF, + value & 0xFF); } } - - // Re-creating 'name' table - if (!tables['name']) { - tables['name'] = { - tag: 'name', - data: createNameTable(this.name) - }; + this.output.push.apply(this.output, command); + if (keepStack) { + this.stack.splice(start, howManyArgs); } else { - // ... using existing 'name' table as prototype - var namePrototype = readNameTable(tables['name']); - tables['name'].data = createNameTable(name, namePrototype); + this.stack.length = 0; } + return false; + } + }; - var builder = new OpenTypeFileBuilder(header.version); - for (var tableTag in tables) { - builder.addTable(tableTag, tables[tableTag].data); - } - return builder.toArray(); - }, + return Type1CharString; +})(); - convert: function Font_convert(fontName, font, properties) { - // TODO: Check the charstring widths to determine this. - properties.fixedPitch = false; +/* + * Type1Parser encapsulate the needed code for parsing a Type1 font + * program. Some of its logic depends on the Type2 charstrings + * structure. + * Note: this doesn't really parse the font since that would require evaluation + * of PostScript, but it is possible in most cases to extract what we need + * without a full parse. + */ +var Type1Parser = (function Type1ParserClosure() { + /* + * Decrypt a Sequence of Ciphertext Bytes to Produce the Original Sequence + * of Plaintext Bytes. The function took a key as a parameter which can be + * for decrypting the eexec block of for decoding charStrings. + */ + var EEXEC_ENCRYPT_KEY = 55665; + var CHAR_STRS_ENCRYPT_KEY = 4330; - var mapping = font.getGlyphMapping(properties); - var newMapping = adjustMapping(mapping, properties); - this.toFontChar = newMapping.toFontChar; - var numGlyphs = font.numGlyphs; + function isHexDigit(code) { + return code >= 48 && code <= 57 || // '0'-'9' + code >= 65 && code <= 70 || // 'A'-'F' + code >= 97 && code <= 102; // 'a'-'f' + } - function getCharCodes(charCodeToGlyphId, glyphId) { - var charCodes = null; - for (var charCode in charCodeToGlyphId) { - if (glyphId === charCodeToGlyphId[charCode]) { - if (!charCodes) { - charCodes = []; - } - charCodes.push(charCode | 0); - } - } - return charCodes; - } + function decrypt(data, key, discardNumber) { + if (discardNumber >= data.length) { + return new Uint8Array(0); + } + var r = key | 0, c1 = 52845, c2 = 22719, i, j; + for (i = 0; i < discardNumber; i++) { + r = ((data[i] + r) * c1 + c2) & ((1 << 16) - 1); + } + var count = data.length - discardNumber; + var decrypted = new Uint8Array(count); + for (i = discardNumber, j = 0; j < count; i++, j++) { + var value = data[i]; + decrypted[j] = value ^ (r >> 8); + r = ((value + r) * c1 + c2) & ((1 << 16) - 1); + } + return decrypted; + } - function createCharCode(charCodeToGlyphId, glyphId) { - for (var charCode in charCodeToGlyphId) { - if (glyphId === charCodeToGlyphId[charCode]) { - return charCode | 0; - } - } - newMapping.charCodeToGlyphId[newMapping.nextAvailableFontCharCode] = - glyphId; - return newMapping.nextAvailableFontCharCode++; + function decryptAscii(data, key, discardNumber) { + var r = key | 0, c1 = 52845, c2 = 22719; + var count = data.length, maybeLength = count >>> 1; + var decrypted = new Uint8Array(maybeLength); + var i, j; + for (i = 0, j = 0; i < count; i++) { + var digit1 = data[i]; + if (!isHexDigit(digit1)) { + continue; } - - var seacs = font.seacs; - if (SEAC_ANALYSIS_ENABLED && seacs && seacs.length) { - var matrix = properties.fontMatrix || FONT_IDENTITY_MATRIX; - var charset = font.getCharset(); - var seacMap = Object.create(null); - for (var glyphId in seacs) { - glyphId |= 0; - var seac = seacs[glyphId]; - var baseGlyphName = StandardEncoding[seac[2]]; - var accentGlyphName = StandardEncoding[seac[3]]; - var baseGlyphId = charset.indexOf(baseGlyphName); - var accentGlyphId = charset.indexOf(accentGlyphName); - if (baseGlyphId < 0 || accentGlyphId < 0) { - continue; - } - var accentOffset = { - x: seac[0] * matrix[0] + seac[1] * matrix[2] + matrix[4], - y: seac[0] * matrix[1] + seac[1] * matrix[3] + matrix[5] - }; - - var charCodes = getCharCodes(mapping, glyphId); - if (!charCodes) { - // There's no point in mapping it if the char code was never mapped - // to begin with. - continue; - } - for (var i = 0, ii = charCodes.length; i < ii; i++) { - var charCode = charCodes[i]; - // Find a fontCharCode that maps to the base and accent glyphs. - // If one doesn't exists, create it. - var charCodeToGlyphId = newMapping.charCodeToGlyphId; - var baseFontCharCode = createCharCode(charCodeToGlyphId, - baseGlyphId); - var accentFontCharCode = createCharCode(charCodeToGlyphId, - accentGlyphId); - seacMap[charCode] = { - baseFontCharCode: baseFontCharCode, - accentFontCharCode: accentFontCharCode, - accentOffset: accentOffset - }; - } - } - properties.seacMap = seacMap; + i++; + var digit2; + while (i < count && !isHexDigit(digit2 = data[i])) { + i++; + } + if (i < count) { + var value = parseInt(String.fromCharCode(digit1, digit2), 16); + decrypted[j++] = value ^ (r >> 8); + r = ((value + r) * c1 + c2) & ((1 << 16) - 1); } + } + return Array.prototype.slice.call(decrypted, discardNumber, j); + } - var unitsPerEm = 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]; + function isSpecial(c) { + return c === 0x2F || // '/' + c === 0x5B || c === 0x5D || // '[', ']' + c === 0x7B || c === 0x7D || // '{', '}' + c === 0x28 || c === 0x29; // '(', ')' + } - var builder = new OpenTypeFileBuilder('\x4F\x54\x54\x4F'); - // PostScript Font Program - builder.addTable('CFF ', font.data); - // OS/2 and Windows Specific metrics - builder.addTable('OS/2', createOS2Table(properties, - newMapping.charCodeToGlyphId)); - // Character to glyphs mapping - builder.addTable('cmap', createCmapTable(newMapping.charCodeToGlyphId, - numGlyphs)); - // Font header - builder.addTable('head', - '\x00\x01\x00\x00' + // Version number - '\x00\x00\x10\x00' + // fontRevision - '\x00\x00\x00\x00' + // checksumAdjustement - '\x5F\x0F\x3C\xF5' + // magicNumber - '\x00\x00' + // Flags - safeString16(unitsPerEm) + // unitsPerEM - '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // creation date - '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // modifification date - '\x00\x00' + // xMin - safeString16(properties.descent) + // yMin - '\x0F\xFF' + // xMax - safeString16(properties.ascent) + // yMax - string16(properties.italicAngle ? 2 : 0) + // macStyle - '\x00\x11' + // lowestRecPPEM - '\x00\x00' + // fontDirectionHint - '\x00\x00' + // indexToLocFormat - '\x00\x00'); // glyphDataFormat + function Type1Parser(stream, encrypted) { + if (encrypted) { + var data = stream.getBytes(); + var isBinary = !(isHexDigit(data[0]) && isHexDigit(data[1]) && + isHexDigit(data[2]) && isHexDigit(data[3])); + stream = new Stream(isBinary ? decrypt(data, EEXEC_ENCRYPT_KEY, 4) : + decryptAscii(data, EEXEC_ENCRYPT_KEY, 4)); + } + this.stream = stream; + this.nextChar(); + } - // Horizontal header - builder.addTable('hhea', - '\x00\x01\x00\x00' + // Version number - safeString16(properties.ascent) + // Typographic Ascent - safeString16(properties.descent) + // Typographic Descent - '\x00\x00' + // Line Gap - '\xFF\xFF' + // advanceWidthMax - '\x00\x00' + // minLeftSidebearing - '\x00\x00' + // minRightSidebearing - '\x00\x00' + // xMaxExtent - safeString16(properties.capHeight) + // caretSlopeRise - safeString16(Math.tan(properties.italicAngle) * - properties.xHeight) + // caretSlopeRun - '\x00\x00' + // caretOffset - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // metricDataFormat - string16(numGlyphs)); // Number of HMetrics + Type1Parser.prototype = { + readNumberArray: function Type1Parser_readNumberArray() { + this.getToken(); // read '[' or '{' (arrays can start with either) + var array = []; + while (true) { + var token = this.getToken(); + if (token === null || token === ']' || token === '}') { + break; + } + array.push(parseFloat(token || 0)); + } + return array; + }, - // Horizontal metrics - builder.addTable('hmtx', (function fontFieldsHmtx() { - var charstrings = font.charstrings; - var cffWidths = font.cff ? font.cff.widths : null; - var hmtx = '\x00\x00\x00\x00'; // Fake .notdef - for (var i = 1, ii = numGlyphs; i < ii; i++) { - var width = 0; - if (charstrings) { - var charstring = charstrings[i - 1]; - width = 'width' in charstring ? charstring.width : 0; - } else if (cffWidths) { - width = Math.ceil(cffWidths[i] || 0); - } - hmtx += string16(width) + string16(0); - } - return hmtx; - })()); + readNumber: function Type1Parser_readNumber() { + var token = this.getToken(); + return parseFloat(token || 0); + }, - // Maximum profile - builder.addTable('maxp', - '\x00\x00\x50\x00' + // Version number - string16(numGlyphs)); // Num of glyphs + readInt: function Type1Parser_readInt() { + // Use '| 0' to prevent setting a double into length such as the double + // does not flow into the loop variable. + var token = this.getToken(); + return parseInt(token || 0, 10) | 0; + }, - // Naming tables - builder.addTable('name', createNameTable(fontName)); + readBoolean: function Type1Parser_readBoolean() { + var token = this.getToken(); - // PostScript informations - builder.addTable('post', createPostTable(properties)); + // Use 1 and 0 since that's what type2 charstrings use. + return token === 'true' ? 1 : 0; + }, - return builder.toArray(); + nextChar : function Type1_nextChar() { + return (this.currentChar = this.stream.getByte()); }, - /** - * Builds a char code to unicode map based on section 9.10 of the spec. - * @param {Object} properties Font properties object. - * @return {Object} A ToUnicodeMap object. - */ - buildToUnicode: function Font_buildToUnicode(properties) { - // Section 9.10.2 Mapping Character Codes to Unicode Values - if (properties.toUnicode && properties.toUnicode.length !== 0) { - return properties.toUnicode; - } - // According to the spec if the font is a simple font we should only map - // to unicode if the base encoding is MacRoman, MacExpert, or WinAnsi or - // the differences array only contains adobe standard or symbol set names, - // in pratice it seems better to always try to create a toUnicode - // map based of the default encoding. - var toUnicode, charcode; - if (!properties.composite /* is simple font */) { - toUnicode = []; - var encoding = properties.defaultEncoding.slice(); - var baseEncodingName = properties.baseEncodingName; - // Merge in the differences array. - var differences = properties.differences; - for (charcode in differences) { - encoding[charcode] = differences[charcode]; + getToken: function Type1Parser_getToken() { + // Eat whitespace and comments. + var comment = false; + var ch = this.currentChar; + while (true) { + if (ch === -1) { + return null; } - var glyphsUnicodeMap = getGlyphsUnicode(); - for (charcode in encoding) { - // a) Map the character code to a character name. - var glyphName = encoding[charcode]; - // b) Look up the character name in the Adobe Glyph List (see the - // Bibliography) to obtain the corresponding Unicode value. - if (glyphName === '') { - continue; - } else if (glyphsUnicodeMap[glyphName] === undefined) { - // (undocumented) c) Few heuristics to recognize unknown glyphs - // NOTE: Adobe Reader does not do this step, but OSX Preview does - var code = 0; - switch (glyphName[0]) { - case 'G': // Gxx glyph - if (glyphName.length === 3) { - code = parseInt(glyphName.substr(1), 16); - } - break; - case 'g': // g00xx glyph - if (glyphName.length === 5) { - code = parseInt(glyphName.substr(1), 16); - } - break; - case 'C': // Cddd glyph - case 'c': // cddd glyph - if (glyphName.length >= 3) { - code = +glyphName.substr(1); - } - break; - } - if (code) { - // If |baseEncodingName| is one the predefined encodings, - // and |code| equals |charcode|, using the glyph defined in the - // baseEncoding seems to yield a better |toUnicode| mapping - // (fixes issue 5070). - if (baseEncodingName && code === +charcode) { - var baseEncoding = getEncoding(baseEncodingName); - if (baseEncoding && (glyphName = baseEncoding[charcode])) { - toUnicode[charcode] = - String.fromCharCode(glyphsUnicodeMap[glyphName]); - continue; - } - } - toUnicode[charcode] = String.fromCharCode(code); - } - continue; + + if (comment) { + if (ch === 0x0A || ch === 0x0D) { + comment = false; } - toUnicode[charcode] = - String.fromCharCode(glyphsUnicodeMap[glyphName]); + } else if (ch === 0x25) { // '%' + comment = true; + } else if (!Lexer.isSpace(ch)) { + break; } - return new ToUnicodeMap(toUnicode); + ch = this.nextChar(); } - // If the font is a composite font that uses one of the predefined CMaps - // listed in Table 118 (except Identity–H and Identity–V) or whose - // descendant CIDFont uses the Adobe-GB1, Adobe-CNS1, Adobe-Japan1, or - // Adobe-Korea1 character collection: - if (properties.composite && ( - (properties.cMap.builtInCMap && - !(properties.cMap instanceof IdentityCMap)) || - (properties.cidSystemInfo.registry === 'Adobe' && - (properties.cidSystemInfo.ordering === 'GB1' || - properties.cidSystemInfo.ordering === 'CNS1' || - properties.cidSystemInfo.ordering === 'Japan1' || - properties.cidSystemInfo.ordering === 'Korea1')))) { - // Then: - // a) Map the character code to a character identifier (CID) according - // to the font’s CMap. - // b) Obtain the registry and ordering of the character collection used - // by the font’s CMap (for example, Adobe and Japan1) from its - // CIDSystemInfo dictionary. - var registry = properties.cidSystemInfo.registry; - var ordering = properties.cidSystemInfo.ordering; - // c) Construct a second CMap name by concatenating the registry and - // ordering obtained in step (b) in the format registry–ordering–UCS2 - // (for example, Adobe–Japan1–UCS2). - var ucs2CMapName = new Name(registry + '-' + ordering + '-UCS2'); - // d) Obtain the CMap with the name constructed in step (c) (available - // from the ASN Web site; see the Bibliography). - var ucs2CMap = CMapFactory.create(ucs2CMapName, - { url: PDFJS.cMapUrl, packed: PDFJS.cMapPacked }, null); - var cMap = properties.cMap; - toUnicode = []; - cMap.forEach(function(charcode, cid) { - assert(cid <= 0xffff, 'Max size of CID is 65,535'); - // e) Map the CID obtained in step (a) according to the CMap obtained - // in step (d), producing a Unicode value. - var ucs2 = ucs2CMap.lookup(cid); - if (ucs2) { - toUnicode[charcode] = - String.fromCharCode((ucs2.charCodeAt(0) << 8) + - ucs2.charCodeAt(1)); - } - }); - return new ToUnicodeMap(toUnicode); + if (isSpecial(ch)) { + this.nextChar(); + return String.fromCharCode(ch); } - - // The viewer's choice, just use an identity map. - return new IdentityToUnicodeMap(properties.firstChar, - properties.lastChar); + var token = ''; + do { + token += String.fromCharCode(ch); + ch = this.nextChar(); + } while (ch >= 0 && !Lexer.isSpace(ch) && !isSpecial(ch)); + return token; }, - get spaceWidth() { - if ('_shadowWidth' in this) { - return this._shadowWidth; - } + /* + * Returns an object containing a Subrs array and a CharStrings + * array extracted from and eexec encrypted block of data + */ + extractFontProgram: function Type1Parser_extractFontProgram() { + var stream = this.stream; - // trying to estimate space character width - var possibleSpaceReplacements = ['space', 'minus', 'one', 'i']; - var width; - for (var i = 0, ii = possibleSpaceReplacements.length; i < ii; i++) { - var glyphName = possibleSpaceReplacements[i]; - // if possible, getting width by glyph name - if (glyphName in this.widths) { - width = this.widths[glyphName]; - break; - } - var glyphsUnicodeMap = getGlyphsUnicode(); - var glyphUnicode = glyphsUnicodeMap[glyphName]; - // finding the charcode via unicodeToCID map - var charcode = 0; - if (this.composite) { - if (this.cMap.contains(glyphUnicode)) { - charcode = this.cMap.lookup(glyphUnicode); - } - } - // ... via toUnicode map - if (!charcode && this.toUnicode) { - charcode = this.toUnicode.charCodeOf(glyphUnicode); - } - // setting it to unicode if negative or undefined - if (charcode <= 0) { - charcode = glyphUnicode; + var subrs = [], charstrings = []; + var privateData = Object.create(null); + privateData['lenIV'] = 4; + var program = { + subrs: [], + charstrings: [], + properties: { + 'privateData': privateData } - // trying to get width via charcode - width = this.widths[charcode]; - if (width) { - break; // the non-zero width found + }; + var token, length, data, lenIV, encoded; + while ((token = this.getToken()) !== null) { + if (token !== '/') { + continue; } - } - width = width || this.defaultWidth; - // Do not shadow the property here. See discussion: - // https://github.com/mozilla/pdf.js/pull/2127#discussion_r1662280 - this._shadowWidth = width; - return width; - }, - - charToGlyph: function Font_charToGlyph(charcode, isSpace) { - var fontCharCode, width, operatorListId; - - var widthCode = charcode; - if (this.cMap && this.cMap.contains(charcode)) { - widthCode = this.cMap.lookup(charcode); - } - width = this.widths[widthCode]; - width = isNum(width) ? width : this.defaultWidth; - var vmetric = this.vmetrics && this.vmetrics[widthCode]; - - var unicode = this.toUnicode.get(charcode) || charcode; - if (typeof unicode === 'number') { - unicode = String.fromCharCode(unicode); - } - - // First try the toFontChar map, if it's not there then try falling - // back to the char code. - fontCharCode = this.toFontChar[charcode] || charcode; - if (this.missingFile) { - fontCharCode = mapSpecialUnicodeValues(fontCharCode); - } + token = this.getToken(); + switch (token) { + case 'CharStrings': + // The number immediately following CharStrings must be greater or + // equal to the number of CharStrings. + this.getToken(); + this.getToken(); // read in 'dict' + this.getToken(); // read in 'dup' + this.getToken(); // read in 'begin' + while(true) { + token = this.getToken(); + if (token === null || token === 'end') { + break; + } - if (this.isType3Font) { - // Font char code in this case is actually a glyph name. - operatorListId = fontCharCode; + if (token !== '/') { + continue; + } + var glyph = this.getToken(); + length = this.readInt(); + this.getToken(); // read in 'RD' or '-|' + data = stream.makeSubStream(stream.pos, length); + lenIV = program.properties.privateData['lenIV']; + encoded = decrypt(data.getBytes(), CHAR_STRS_ENCRYPT_KEY, lenIV); + // Skip past the required space and binary data. + stream.skip(length); + this.nextChar(); + token = this.getToken(); // read in 'ND' or '|-' + if (token === 'noaccess') { + this.getToken(); // read in 'def' + } + charstrings.push({ + glyph: glyph, + encoded: encoded + }); + } + break; + case 'Subrs': + var num = this.readInt(); + this.getToken(); // read in 'array' + while ((token = this.getToken()) === 'dup') { + var index = this.readInt(); + length = this.readInt(); + this.getToken(); // read in 'RD' or '-|' + data = stream.makeSubStream(stream.pos, length); + lenIV = program.properties.privateData['lenIV']; + encoded = decrypt(data.getBytes(), CHAR_STRS_ENCRYPT_KEY, lenIV); + // Skip past the required space and binary data. + stream.skip(length); + this.nextChar(); + token = this.getToken(); // read in 'NP' or '|' + if (token === 'noaccess') { + this.getToken(); // read in 'put' + } + subrs[index] = encoded; + } + break; + case 'BlueValues': + case 'OtherBlues': + case 'FamilyBlues': + case 'FamilyOtherBlues': + var blueArray = this.readNumberArray(); + // *Blue* values may contain invalid data: disables reading of + // those values when hinting is disabled. + if (blueArray.length > 0 && (blueArray.length % 2) === 0 && + HINTING_ENABLED) { + program.properties.privateData[token] = blueArray; + } + break; + case 'StemSnapH': + case 'StemSnapV': + program.properties.privateData[token] = this.readNumberArray(); + break; + case 'StdHW': + case 'StdVW': + program.properties.privateData[token] = + this.readNumberArray()[0]; + break; + case 'BlueShift': + case 'lenIV': + case 'BlueFuzz': + case 'BlueScale': + case 'LanguageGroup': + case 'ExpansionFactor': + program.properties.privateData[token] = this.readNumber(); + break; + case 'ForceBold': + program.properties.privateData[token] = this.readBoolean(); + break; + } } - var accent = null; - if (this.seacMap && this.seacMap[charcode]) { - var seac = this.seacMap[charcode]; - fontCharCode = seac.baseFontCharCode; - accent = { - fontChar: String.fromCharCode(seac.accentFontCharCode), - offset: seac.accentOffset - }; + for (var i = 0; i < charstrings.length; i++) { + glyph = charstrings[i].glyph; + encoded = charstrings[i].encoded; + var charString = new Type1CharString(); + var error = charString.convert(encoded, subrs); + var output = charString.output; + if (error) { + // It seems when FreeType encounters an error while evaluating a glyph + // that it completely ignores the glyph so we'll mimic that behaviour + // here and put an endchar to make the validator happy. + output = [14]; + } + program.charstrings.push({ + glyphName: glyph, + charstring: output, + width: charString.width, + lsb: charString.lsb, + seac: charString.seac + }); } - var fontChar = String.fromCharCode(fontCharCode); - - var glyph = this.glyphCache[charcode]; - if (!glyph || - !glyph.matchesForCache(fontChar, unicode, accent, width, vmetric, - operatorListId, isSpace)) { - glyph = new Glyph(fontChar, unicode, accent, width, vmetric, - operatorListId, isSpace); - this.glyphCache[charcode] = glyph; - } - return glyph; + return program; }, - charsToGlyphs: function Font_charsToGlyphs(chars) { - var charsCache = this.charsCache; - var glyphs, glyph, charcode; - - // if we translated this string before, just grab it from the cache - if (charsCache) { - glyphs = charsCache[chars]; - if (glyphs) { - return glyphs; + extractFontHeader: function Type1Parser_extractFontHeader(properties) { + var token; + while ((token = this.getToken()) !== null) { + if (token !== '/') { + continue; } - } - - // lazily create the translation cache - if (!charsCache) { - charsCache = this.charsCache = Object.create(null); - } - - glyphs = []; - var charsCacheKey = chars; - var i = 0, ii; + token = this.getToken(); + switch (token) { + case 'FontMatrix': + var matrix = this.readNumberArray(); + properties.fontMatrix = matrix; + break; + case 'Encoding': + var encodingArg = this.getToken(); + var encoding; + if (!/^\d+$/.test(encodingArg)) { + // encoding name is specified + encoding = getEncoding(encodingArg); + } else { + encoding = []; + var size = parseInt(encodingArg, 10) | 0; + this.getToken(); // read in 'array' - if (this.cMap) { - // composite fonts have multi-byte strings convert the string from - // single-byte to multi-byte - var c = Object.create(null); - while (i < chars.length) { - this.cMap.readCharCode(chars, i, c); - charcode = c.charcode; - var length = c.length; - i += length; - // Space is char with code 0x20 and length 1 in multiple-byte codes. - var isSpace = length === 1 && chars.charCodeAt(i - 1) === 0x20; - glyph = this.charToGlyph(charcode, isSpace); - glyphs.push(glyph); - } - } else { - for (i = 0, ii = chars.length; i < ii; ++i) { - charcode = chars.charCodeAt(i); - glyph = this.charToGlyph(charcode, charcode === 0x20); - glyphs.push(glyph); + for (var j = 0; j < size; j++) { + token = this.getToken(); + // skipping till first dup or def (e.g. ignoring for statement) + while (token !== 'dup' && token !== 'def') { + token = this.getToken(); + if (token === null) { + return; // invalid header + } + } + if (token === 'def') { + break; // read all array data + } + var index = this.readInt(); + this.getToken(); // read in '/' + var glyph = this.getToken(); + encoding[index] = glyph; + this.getToken(); // read the in 'put' + } + } + properties.builtInEncoding = encoding; + break; + case 'FontBBox': + var fontBBox = this.readNumberArray(); + // adjusting ascent/descent + properties.ascent = fontBBox[3]; + properties.descent = fontBBox[1]; + properties.ascentScaled = true; + break; } } - - // Enter the translated string into the cache - return (charsCache[charsCacheKey] = glyphs); } }; - return Font; + return Type1Parser; })(); -var ErrorFont = (function ErrorFontClosure() { - function ErrorFont(error) { - this.error = error; - this.loadedName = 'g_font_error'; - this.loading = false; +/** + * The CFF class takes a Type1 file and wrap it into a + * 'Compact Font Format' which itself embed Type2 charstrings. + */ +var CFFStandardStrings = [ + '.notdef', 'space', 'exclam', 'quotedbl', 'numbersign', 'dollar', 'percent', + 'ampersand', 'quoteright', 'parenleft', 'parenright', 'asterisk', 'plus', + 'comma', 'hyphen', 'period', 'slash', 'zero', 'one', 'two', 'three', 'four', + 'five', 'six', 'seven', 'eight', 'nine', 'colon', 'semicolon', 'less', + 'equal', 'greater', 'question', 'at', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', + 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', + 'X', 'Y', 'Z', 'bracketleft', 'backslash', 'bracketright', 'asciicircum', + 'underscore', 'quoteleft', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', + 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', + 'z', 'braceleft', 'bar', 'braceright', 'asciitilde', 'exclamdown', 'cent', + 'sterling', 'fraction', 'yen', 'florin', 'section', 'currency', + 'quotesingle', 'quotedblleft', 'guillemotleft', 'guilsinglleft', + 'guilsinglright', 'fi', 'fl', 'endash', 'dagger', 'daggerdbl', + 'periodcentered', 'paragraph', 'bullet', 'quotesinglbase', 'quotedblbase', + 'quotedblright', 'guillemotright', 'ellipsis', 'perthousand', 'questiondown', + 'grave', 'acute', 'circumflex', 'tilde', 'macron', 'breve', 'dotaccent', + 'dieresis', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', 'emdash', + 'AE', 'ordfeminine', 'Lslash', 'Oslash', 'OE', 'ordmasculine', 'ae', + 'dotlessi', 'lslash', 'oslash', 'oe', 'germandbls', 'onesuperior', + 'logicalnot', 'mu', 'trademark', 'Eth', 'onehalf', 'plusminus', 'Thorn', + 'onequarter', 'divide', 'brokenbar', 'degree', 'thorn', 'threequarters', + 'twosuperior', 'registered', 'minus', 'eth', 'multiply', 'threesuperior', + 'copyright', 'Aacute', 'Acircumflex', 'Adieresis', 'Agrave', 'Aring', + 'Atilde', 'Ccedilla', 'Eacute', 'Ecircumflex', 'Edieresis', 'Egrave', + 'Iacute', 'Icircumflex', 'Idieresis', 'Igrave', 'Ntilde', 'Oacute', + 'Ocircumflex', 'Odieresis', 'Ograve', 'Otilde', 'Scaron', 'Uacute', + 'Ucircumflex', 'Udieresis', 'Ugrave', 'Yacute', 'Ydieresis', 'Zcaron', + 'aacute', 'acircumflex', 'adieresis', 'agrave', 'aring', 'atilde', + 'ccedilla', 'eacute', 'ecircumflex', 'edieresis', 'egrave', 'iacute', + 'icircumflex', 'idieresis', 'igrave', 'ntilde', 'oacute', 'ocircumflex', + 'odieresis', 'ograve', 'otilde', 'scaron', 'uacute', 'ucircumflex', + 'udieresis', 'ugrave', 'yacute', 'ydieresis', 'zcaron', 'exclamsmall', + 'Hungarumlautsmall', 'dollaroldstyle', 'dollarsuperior', 'ampersandsmall', + 'Acutesmall', 'parenleftsuperior', 'parenrightsuperior', 'twodotenleader', + 'onedotenleader', 'zerooldstyle', 'oneoldstyle', 'twooldstyle', + 'threeoldstyle', 'fouroldstyle', 'fiveoldstyle', 'sixoldstyle', + 'sevenoldstyle', 'eightoldstyle', 'nineoldstyle', 'commasuperior', + 'threequartersemdash', 'periodsuperior', 'questionsmall', 'asuperior', + 'bsuperior', 'centsuperior', 'dsuperior', 'esuperior', 'isuperior', + 'lsuperior', 'msuperior', 'nsuperior', 'osuperior', 'rsuperior', 'ssuperior', + 'tsuperior', 'ff', 'ffi', 'ffl', 'parenleftinferior', 'parenrightinferior', + 'Circumflexsmall', 'hyphensuperior', 'Gravesmall', 'Asmall', 'Bsmall', + 'Csmall', 'Dsmall', 'Esmall', 'Fsmall', 'Gsmall', 'Hsmall', 'Ismall', + 'Jsmall', 'Ksmall', 'Lsmall', 'Msmall', 'Nsmall', 'Osmall', 'Psmall', + 'Qsmall', 'Rsmall', 'Ssmall', 'Tsmall', 'Usmall', 'Vsmall', 'Wsmall', + 'Xsmall', 'Ysmall', 'Zsmall', 'colonmonetary', 'onefitted', 'rupiah', + 'Tildesmall', 'exclamdownsmall', 'centoldstyle', 'Lslashsmall', + 'Scaronsmall', 'Zcaronsmall', 'Dieresissmall', 'Brevesmall', 'Caronsmall', + 'Dotaccentsmall', 'Macronsmall', 'figuredash', 'hypheninferior', + 'Ogoneksmall', 'Ringsmall', 'Cedillasmall', 'questiondownsmall', 'oneeighth', + 'threeeighths', 'fiveeighths', 'seveneighths', 'onethird', 'twothirds', + 'zerosuperior', 'foursuperior', 'fivesuperior', 'sixsuperior', + 'sevensuperior', 'eightsuperior', 'ninesuperior', 'zeroinferior', + 'oneinferior', 'twoinferior', 'threeinferior', 'fourinferior', + 'fiveinferior', 'sixinferior', 'seveninferior', 'eightinferior', + 'nineinferior', 'centinferior', 'dollarinferior', 'periodinferior', + 'commainferior', 'Agravesmall', 'Aacutesmall', 'Acircumflexsmall', + 'Atildesmall', 'Adieresissmall', 'Aringsmall', 'AEsmall', 'Ccedillasmall', + 'Egravesmall', 'Eacutesmall', 'Ecircumflexsmall', 'Edieresissmall', + 'Igravesmall', 'Iacutesmall', 'Icircumflexsmall', 'Idieresissmall', + 'Ethsmall', 'Ntildesmall', 'Ogravesmall', 'Oacutesmall', 'Ocircumflexsmall', + 'Otildesmall', 'Odieresissmall', 'OEsmall', 'Oslashsmall', 'Ugravesmall', + 'Uacutesmall', 'Ucircumflexsmall', 'Udieresissmall', 'Yacutesmall', + 'Thornsmall', 'Ydieresissmall', '001.000', '001.001', '001.002', '001.003', + 'Black', 'Bold', 'Book', 'Light', 'Medium', 'Regular', 'Roman', 'Semibold' +]; + +// Type1Font is also a CIDFontType0. +var Type1Font = function Type1Font(name, file, properties) { + // Some bad generators embed pfb file as is, we have to strip 6-byte headers. + // Also, length1 and length2 might be off by 6 bytes as well. + // http://www.math.ubc.ca/~cass/piscript/type1.pdf + var PFB_HEADER_SIZE = 6; + var headerBlockLength = properties.length1; + var eexecBlockLength = properties.length2; + var pfbHeader = file.peekBytes(PFB_HEADER_SIZE); + var pfbHeaderPresent = pfbHeader[0] === 0x80 && pfbHeader[1] === 0x01; + if (pfbHeaderPresent) { + file.skip(PFB_HEADER_SIZE); + headerBlockLength = (pfbHeader[5] << 24) | (pfbHeader[4] << 16) | + (pfbHeader[3] << 8) | pfbHeader[2]; } - ErrorFont.prototype = { - charsToGlyphs: function ErrorFont_charsToGlyphs() { - return []; - }, - exportData: function ErrorFont_exportData() { - return {error: this.error}; - } - }; + // Get the data block containing glyphs and subrs informations + var headerBlock = new Stream(file.getBytes(headerBlockLength)); + var headerBlockParser = new Type1Parser(headerBlock); + headerBlockParser.extractFontHeader(properties); - return ErrorFont; -})(); + if (pfbHeaderPresent) { + pfbHeader = file.getBytes(PFB_HEADER_SIZE); + eexecBlockLength = (pfbHeader[5] << 24) | (pfbHeader[4] << 16) | + (pfbHeader[3] << 8) | pfbHeader[2]; + } -/** - * Shared logic for building a char code to glyph id mapping for Type1 and - * simple CFF fonts. See section 9.6.6.2 of the spec. - * @param {Object} properties Font properties object. - * @param {Object} builtInEncoding The encoding contained within the actual font - * data. - * @param {Array} Array of glyph names where the index is the glyph ID. - * @returns {Object} A char code to glyph ID map. - */ -function type1FontGlyphMapping(properties, builtInEncoding, glyphNames) { - var charCodeToGlyphId = Object.create(null); - var glyphId, charCode, baseEncoding; + // Decrypt the data blocks and retrieve it's content + var eexecBlock = new Stream(file.getBytes(eexecBlockLength)); + var eexecBlockParser = new Type1Parser(eexecBlock, true); + var data = eexecBlockParser.extractFontProgram(); + for (var info in data.properties) { + properties[info] = data.properties[info]; + } - if (properties.baseEncodingName) { - // If a valid base encoding name was used, the mapping is initialized with - // that. - baseEncoding = getEncoding(properties.baseEncodingName); - for (charCode = 0; charCode < baseEncoding.length; charCode++) { - glyphId = glyphNames.indexOf(baseEncoding[charCode]); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef - } + var charstrings = data.charstrings; + var type2Charstrings = this.getType2Charstrings(charstrings); + var subrs = this.getType2Subrs(data.subrs); + + this.charstrings = charstrings; + this.data = this.wrap(name, type2Charstrings, this.charstrings, + subrs, properties); + this.seacs = this.getSeacs(data.charstrings); +}; + +Type1Font.prototype = { + get numGlyphs() { + return this.charstrings.length + 1; + }, + + getCharset: function Type1Font_getCharset() { + var charset = ['.notdef']; + var charstrings = this.charstrings; + for (var glyphId = 0; glyphId < charstrings.length; glyphId++) { + charset.push(charstrings[glyphId].glyphName); } - } else if (!!(properties.flags & FontFlags.Symbolic)) { - // For a symbolic font the encoding should be the fonts built-in - // encoding. - for (charCode in builtInEncoding) { - charCodeToGlyphId[charCode] = builtInEncoding[charCode]; + return charset; + }, + + getGlyphMapping: function Type1Font_getGlyphMapping(properties) { + var charstrings = this.charstrings; + var glyphNames = ['.notdef'], glyphId; + for (glyphId = 0; glyphId < charstrings.length; glyphId++) { + glyphNames.push(charstrings[glyphId].glyphName); } - } else { - // For non-symbolic fonts that don't have a base encoding the standard - // encoding should be used. - baseEncoding = StandardEncoding; - for (charCode = 0; charCode < baseEncoding.length; charCode++) { - glyphId = glyphNames.indexOf(baseEncoding[charCode]); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef + var encoding = properties.builtInEncoding; + if (encoding) { + var builtInEncoding = Object.create(null); + for (var charCode in encoding) { + glyphId = glyphNames.indexOf(encoding[charCode]); + if (glyphId >= 0) { + builtInEncoding[charCode] = glyphId; + } } } - } - // Lastly, merge in the differences. - var differences = properties.differences; - if (differences) { - for (charCode in differences) { - var glyphName = differences[charCode]; - glyphId = glyphNames.indexOf(glyphName); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef + return type1FontGlyphMapping(properties, builtInEncoding, glyphNames); + }, + + getSeacs: function Type1Font_getSeacs(charstrings) { + var i, ii; + var seacMap = []; + for (i = 0, ii = charstrings.length; i < ii; i++) { + var charstring = charstrings[i]; + if (charstring.seac) { + // Offset by 1 for .notdef + seacMap[i + 1] = charstring.seac; } } - } - return charCodeToGlyphId; -} + return seacMap; + }, -/* - * CharStrings are encoded following the the CharString Encoding sequence - * describe in Chapter 6 of the "Adobe Type1 Font Format" specification. - * The value in a byte indicates a command, a number, or subsequent bytes - * that are to be interpreted in a special way. - * - * CharString Number Encoding: - * A CharString byte containing the values from 32 through 255 inclusive - * indicate an integer. These values are decoded in four ranges. - * - * 1. A CharString byte containing a value, v, between 32 and 246 inclusive, - * indicate the integer v - 139. Thus, the integer values from -107 through - * 107 inclusive may be encoded in single byte. - * - * 2. A CharString byte containing a value, v, between 247 and 250 inclusive, - * indicates an integer involving the next byte, w, according to the formula: - * [(v - 247) x 256] + w + 108 - * - * 3. A CharString byte containing a value, v, between 251 and 254 inclusive, - * indicates an integer involving the next byte, w, according to the formula: - * -[(v - 251) * 256] - w - 108 - * - * 4. A CharString containing the value 255 indicates that the next 4 bytes - * are a two complement signed integer. The first of these bytes contains the - * highest order bits, the second byte contains the next higher order bits - * and the fourth byte contain the lowest order bits. - * - * - * CharString Command Encoding: - * CharStrings commands are encoded in 1 or 2 bytes. - * - * Single byte commands are encoded in 1 byte that contains a value between - * 0 and 31 inclusive. - * If a command byte contains the value 12, then the value in the next byte - * indicates a command. This "escape" mechanism allows many extra commands - * to be encoded and this encoding technique helps to minimize the length of - * the charStrings. - */ -var Type1CharString = (function Type1CharStringClosure() { - var COMMAND_MAP = { - 'hstem': [1], - 'vstem': [3], - 'vmoveto': [4], - 'rlineto': [5], - 'hlineto': [6], - 'vlineto': [7], - 'rrcurveto': [8], - 'callsubr': [10], - 'flex': [12, 35], - 'drop' : [12, 18], - 'endchar': [14], - 'rmoveto': [21], - 'hmoveto': [22], - 'vhcurveto': [30], - 'hvcurveto': [31] - }; + getType2Charstrings: function Type1Font_getType2Charstrings( + type1Charstrings) { + var type2Charstrings = []; + for (var i = 0, ii = type1Charstrings.length; i < ii; i++) { + type2Charstrings.push(type1Charstrings[i].charstring); + } + return type2Charstrings; + }, - function Type1CharString() { - this.width = 0; - this.lsb = 0; - this.flexing = false; - this.output = []; - this.stack = []; - } + getType2Subrs: function Type1Font_getType2Subrs(type1Subrs) { + var bias = 0; + var count = type1Subrs.length; + if (count < 1133) { + bias = 107; + } else if (count < 33769) { + bias = 1131; + } else { + bias = 32768; + } - Type1CharString.prototype = { - convert: function Type1CharString_convert(encoded, subrs) { - var count = encoded.length; - var error = false; - var wx, sbx, subrNumber; - for (var i = 0; i < count; i++) { - var value = encoded[i]; - if (value < 32) { - if (value === 12) { - value = (value << 8) + encoded[++i]; - } - switch (value) { - case 1: // hstem - if (!HINTING_ENABLED) { - this.stack = []; - break; - } - error = this.executeCommand(2, COMMAND_MAP.hstem); - break; - case 3: // vstem - if (!HINTING_ENABLED) { - this.stack = []; - break; - } - error = this.executeCommand(2, COMMAND_MAP.vstem); - break; - case 4: // vmoveto - if (this.flexing) { - if (this.stack.length < 1) { - error = true; - break; - } - // Add the dx for flex and but also swap the values so they are - // the right order. - var dy = this.stack.pop(); - this.stack.push(0, dy); - break; - } - error = this.executeCommand(1, COMMAND_MAP.vmoveto); - break; - case 5: // rlineto - error = this.executeCommand(2, COMMAND_MAP.rlineto); - break; - case 6: // hlineto - error = this.executeCommand(1, COMMAND_MAP.hlineto); - break; - case 7: // vlineto - error = this.executeCommand(1, COMMAND_MAP.vlineto); - break; - case 8: // rrcurveto - error = this.executeCommand(6, COMMAND_MAP.rrcurveto); - break; - case 9: // closepath - // closepath is a Type1 command that does not take argument and is - // useless in Type2 and it can simply be ignored. - this.stack = []; - break; - case 10: // callsubr - if (this.stack.length < 1) { - error = true; - break; - } - subrNumber = this.stack.pop(); - error = this.convert(subrs[subrNumber], subrs); - break; - case 11: // return - return error; - case 13: // hsbw - if (this.stack.length < 2) { - error = true; - break; - } - // To convert to type2 we have to move the width value to the - // first part of the charstring and then use hmoveto with lsb. - wx = this.stack.pop(); - sbx = this.stack.pop(); - this.lsb = sbx; - this.width = wx; - this.stack.push(wx, sbx); - error = this.executeCommand(2, COMMAND_MAP.hmoveto); - break; - case 14: // endchar - this.output.push(COMMAND_MAP.endchar[0]); - break; - case 21: // rmoveto - if (this.flexing) { - break; - } - error = this.executeCommand(2, COMMAND_MAP.rmoveto); - break; - case 22: // hmoveto - if (this.flexing) { - // Add the dy for flex. - this.stack.push(0); - break; - } - error = this.executeCommand(1, COMMAND_MAP.hmoveto); - break; - case 30: // vhcurveto - error = this.executeCommand(4, COMMAND_MAP.vhcurveto); - break; - case 31: // hvcurveto - error = this.executeCommand(4, COMMAND_MAP.hvcurveto); - break; - case (12 << 8) + 0: // dotsection - // dotsection is a Type1 command to specify some hinting feature - // for dots that do not take a parameter and it can safely be - // ignored for Type2. - this.stack = []; - break; - case (12 << 8) + 1: // vstem3 - if (!HINTING_ENABLED) { - this.stack = []; - break; - } - // [vh]stem3 are Type1 only and Type2 supports [vh]stem with - // multiple parameters, so instead of returning [vh]stem3 take a - // shortcut and return [vhstem] instead. - error = this.executeCommand(2, COMMAND_MAP.vstem); - break; - case (12 << 8) + 2: // hstem3 - if (!HINTING_ENABLED) { - this.stack = []; - break; - } - // See vstem3. - error = this.executeCommand(2, COMMAND_MAP.hstem); - break; - case (12 << 8) + 6: // seac - // seac is like type 2's special endchar but it doesn't use the - // first argument asb, so remove it. - if (SEAC_ANALYSIS_ENABLED) { - this.seac = this.stack.splice(-4, 4); - error = this.executeCommand(0, COMMAND_MAP.endchar); - } else { - error = this.executeCommand(4, COMMAND_MAP.endchar); - } - break; - case (12 << 8) + 7: // sbw - if (this.stack.length < 4) { - error = true; - break; - } - // To convert to type2 we have to move the width value to the - // first part of the charstring and then use rmoveto with - // (dx, dy). The height argument will not be used for vmtx and - // vhea tables reconstruction -- ignoring it. - var wy = this.stack.pop(); - wx = this.stack.pop(); - var sby = this.stack.pop(); - sbx = this.stack.pop(); - this.lsb = sbx; - this.width = wx; - this.stack.push(wx, sbx, sby); - error = this.executeCommand(3, COMMAND_MAP.rmoveto); - break; - case (12 << 8) + 12: // div - if (this.stack.length < 2) { - error = true; - break; - } - var num2 = this.stack.pop(); - var num1 = this.stack.pop(); - this.stack.push(num1 / num2); - break; - case (12 << 8) + 16: // callothersubr - if (this.stack.length < 2) { - error = true; - break; - } - subrNumber = this.stack.pop(); - var numArgs = this.stack.pop(); - if (subrNumber === 0 && numArgs === 3) { - var flexArgs = this.stack.splice(this.stack.length - 17, 17); - this.stack.push( - flexArgs[2] + flexArgs[0], // bcp1x + rpx - flexArgs[3] + flexArgs[1], // bcp1y + rpy - flexArgs[4], // bcp2x - flexArgs[5], // bcp2y - flexArgs[6], // p2x - flexArgs[7], // p2y - flexArgs[8], // bcp3x - flexArgs[9], // bcp3y - flexArgs[10], // bcp4x - flexArgs[11], // bcp4y - flexArgs[12], // p3x - flexArgs[13], // p3y - flexArgs[14] // flexDepth - // 15 = finalx unused by flex - // 16 = finaly unused by flex - ); - error = this.executeCommand(13, COMMAND_MAP.flex, true); - this.flexing = false; - this.stack.push(flexArgs[15], flexArgs[16]); - } else if (subrNumber === 1 && numArgs === 0) { - this.flexing = true; - } - break; - case (12 << 8) + 17: // pop - // Ignore this since it is only used with othersubr. - break; - case (12 << 8) + 33: // setcurrentpoint - // Ignore for now. - this.stack = []; - break; - default: - warn('Unknown type 1 charstring command of "' + value + '"'); - break; - } - if (error) { - break; - } - continue; - } else if (value <= 246) { - value = value - 139; - } else if (value <= 250) { - value = ((value - 247) * 256) + encoded[++i] + 108; - } else if (value <= 254) { - value = -((value - 251) * 256) - encoded[++i] - 108; - } else { - value = (encoded[++i] & 0xff) << 24 | (encoded[++i] & 0xff) << 16 | - (encoded[++i] & 0xff) << 8 | (encoded[++i] & 0xff) << 0; - } - this.stack.push(value); - } - return error; - }, + // Add a bunch of empty subrs to deal with the Type2 bias + var type2Subrs = []; + var i; + for (i = 0; i < bias; i++) { + type2Subrs.push([0x0B]); + } - executeCommand: function(howManyArgs, command, keepStack) { - var stackLength = this.stack.length; - if (howManyArgs > stackLength) { - return true; - } - var start = stackLength - howManyArgs; - for (var i = start; i < stackLength; i++) { - var value = this.stack[i]; - if (value === (value | 0)) { // int - this.output.push(28, (value >> 8) & 0xff, value & 0xff); - } else { // fixed point - value = (65536 * value) | 0; - this.output.push(255, - (value >> 24) & 0xFF, - (value >> 16) & 0xFF, - (value >> 8) & 0xFF, - value & 0xFF); - } - } - this.output.push.apply(this.output, command); - if (keepStack) { - this.stack.splice(start, howManyArgs); - } else { - this.stack.length = 0; - } - return false; + for (i = 0; i < count; i++) { + type2Subrs.push(type1Subrs[i]); } - }; - return Type1CharString; -})(); + return type2Subrs; + }, -/* - * Type1Parser encapsulate the needed code for parsing a Type1 font - * program. Some of its logic depends on the Type2 charstrings - * structure. - * Note: this doesn't really parse the font since that would require evaluation - * of PostScript, but it is possible in most cases to extract what we need - * without a full parse. - */ -var Type1Parser = (function Type1ParserClosure() { - /* - * Decrypt a Sequence of Ciphertext Bytes to Produce the Original Sequence - * of Plaintext Bytes. The function took a key as a parameter which can be - * for decrypting the eexec block of for decoding charStrings. - */ - var EEXEC_ENCRYPT_KEY = 55665; - var CHAR_STRS_ENCRYPT_KEY = 4330; + wrap: function Type1Font_wrap(name, glyphs, charstrings, subrs, properties) { + var cff = new CFF(); + cff.header = new CFFHeader(1, 0, 4, 4); - function isHexDigit(code) { - return code >= 48 && code <= 57 || // '0'-'9' - code >= 65 && code <= 70 || // 'A'-'F' - code >= 97 && code <= 102; // 'a'-'f' - } + cff.names = [name]; - function decrypt(data, key, discardNumber) { - if (discardNumber >= data.length) { - return new Uint8Array(0); - } - var r = key | 0, c1 = 52845, c2 = 22719, i, j; - for (i = 0; i < discardNumber; i++) { - r = ((data[i] + r) * c1 + c2) & ((1 << 16) - 1); + var topDict = new CFFTopDict(); + // CFF strings IDs 0...390 are predefined names, so refering + // to entries in our own String INDEX starts at SID 391. + topDict.setByName('version', 391); + topDict.setByName('Notice', 392); + topDict.setByName('FullName', 393); + topDict.setByName('FamilyName', 394); + topDict.setByName('Weight', 395); + topDict.setByName('Encoding', null); // placeholder + topDict.setByName('FontMatrix', properties.fontMatrix); + topDict.setByName('FontBBox', properties.bbox); + topDict.setByName('charset', null); // placeholder + topDict.setByName('CharStrings', null); // placeholder + topDict.setByName('Private', null); // placeholder + cff.topDict = topDict; + + var strings = new CFFStrings(); + strings.add('Version 0.11'); // Version + strings.add('See original notice'); // Notice + strings.add(name); // FullName + strings.add(name); // FamilyName + strings.add('Medium'); // Weight + cff.strings = strings; + + cff.globalSubrIndex = new CFFIndex(); + + var count = glyphs.length; + var charsetArray = [0]; + var i, ii; + for (i = 0; i < count; i++) { + var index = CFFStandardStrings.indexOf(charstrings[i].glyphName); + // TODO: Insert the string and correctly map it. Previously it was + // thought mapping names that aren't in the standard strings to .notdef + // was fine, however in issue818 when mapping them all to .notdef the + // adieresis glyph no longer worked. + if (index === -1) { + index = 0; + } + charsetArray.push((index >> 8) & 0xff, index & 0xff); } - var count = data.length - discardNumber; - var decrypted = new Uint8Array(count); - for (i = discardNumber, j = 0; j < count; i++, j++) { - var value = data[i]; - decrypted[j] = value ^ (r >> 8); - r = ((value + r) * c1 + c2) & ((1 << 16) - 1); + cff.charset = new CFFCharset(false, 0, [], charsetArray); + + var charStringsIndex = new CFFIndex(); + charStringsIndex.add([0x8B, 0x0E]); // .notdef + for (i = 0; i < count; i++) { + charStringsIndex.add(glyphs[i]); } - return decrypted; - } + cff.charStrings = charStringsIndex; - function decryptAscii(data, key, discardNumber) { - var r = key | 0, c1 = 52845, c2 = 22719; - var count = data.length, maybeLength = count >>> 1; - var decrypted = new Uint8Array(maybeLength); - var i, j; - for (i = 0, j = 0; i < count; i++) { - var digit1 = data[i]; - if (!isHexDigit(digit1)) { + var privateDict = new CFFPrivateDict(); + privateDict.setByName('Subrs', null); // placeholder + var fields = [ + 'BlueValues', + 'OtherBlues', + 'FamilyBlues', + 'FamilyOtherBlues', + 'StemSnapH', + 'StemSnapV', + 'BlueShift', + 'BlueFuzz', + 'BlueScale', + 'LanguageGroup', + 'ExpansionFactor', + 'ForceBold', + 'StdHW', + 'StdVW' + ]; + for (i = 0, ii = fields.length; i < ii; i++) { + var field = fields[i]; + if (!(field in properties.privateData)) { continue; } - i++; - var digit2; - while (i < count && !isHexDigit(digit2 = data[i])) { - i++; - } - if (i < count) { - var value = parseInt(String.fromCharCode(digit1, digit2), 16); - decrypted[j++] = value ^ (r >> 8); - r = ((value + r) * c1 + c2) & ((1 << 16) - 1); + var value = properties.privateData[field]; + if (isArray(value)) { + // All of the private dictionary array data in CFF must be stored as + // "delta-encoded" numbers. + for (var j = value.length - 1; j > 0; j--) { + value[j] -= value[j - 1]; // ... difference from previous value + } } + privateDict.setByName(field, value); } - return Array.prototype.slice.call(decrypted, discardNumber, j); - } + cff.topDict.privateDict = privateDict; - function isSpecial(c) { - return c === 0x2F || // '/' - c === 0x5B || c === 0x5D || // '[', ']' - c === 0x7B || c === 0x7D || // '{', '}' - c === 0x28 || c === 0x29; // '(', ')' + var subrIndex = new CFFIndex(); + for (i = 0, ii = subrs.length; i < ii; i++) { + subrIndex.add(subrs[i]); + } + privateDict.subrsIndex = subrIndex; + + var compiler = new CFFCompiler(cff); + return compiler.compile(); } +}; - function Type1Parser(stream, encrypted) { - if (encrypted) { - var data = stream.getBytes(); - var isBinary = !(isHexDigit(data[0]) && isHexDigit(data[1]) && - isHexDigit(data[2]) && isHexDigit(data[3])); - stream = new Stream(isBinary ? decrypt(data, EEXEC_ENCRYPT_KEY, 4) : - decryptAscii(data, EEXEC_ENCRYPT_KEY, 4)); +var CFFFont = (function CFFFontClosure() { + function CFFFont(file, properties) { + this.properties = properties; + + var parser = new CFFParser(file, properties); + this.cff = parser.parse(); + var compiler = new CFFCompiler(this.cff); + this.seacs = this.cff.seacs; + try { + this.data = compiler.compile(); + } catch (e) { + warn('Failed to compile font ' + properties.loadedName); + // There may have just been an issue with the compiler, set the data + // anyway and hope the font loaded. + this.data = file; } - this.stream = stream; - this.nextChar(); } - Type1Parser.prototype = { - readNumberArray: function Type1Parser_readNumberArray() { - this.getToken(); // read '[' or '{' (arrays can start with either) - var array = []; - while (true) { - var token = this.getToken(); - if (token === null || token === ']' || token === '}') { - break; - } - array.push(parseFloat(token || 0)); - } - return array; - }, - - readNumber: function Type1Parser_readNumber() { - var token = this.getToken(); - return parseFloat(token || 0); + CFFFont.prototype = { + get numGlyphs() { + return this.cff.charStrings.count; }, - - readInt: function Type1Parser_readInt() { - // Use '| 0' to prevent setting a double into length such as the double - // does not flow into the loop variable. - var token = this.getToken(); - return parseInt(token || 0, 10) | 0; + getCharset: function CFFFont_getCharset() { + return this.cff.charset.charset; }, + getGlyphMapping: function CFFFont_getGlyphMapping() { + var cff = this.cff; + var properties = this.properties; + var charsets = cff.charset.charset; + var charCodeToGlyphId; + var glyphId; - readBoolean: function Type1Parser_readBoolean() { - var token = this.getToken(); - - // Use 1 and 0 since that's what type2 charstrings use. - return token === 'true' ? 1 : 0; - }, + if (properties.composite) { + charCodeToGlyphId = Object.create(null); + if (cff.isCIDFont) { + // If the font is actually a CID font then we should use the charset + // to map CIDs to GIDs. + for (glyphId = 0; glyphId < charsets.length; glyphId++) { + var cid = charsets[glyphId]; + var charCode = properties.cMap.charCodeOf(cid); + charCodeToGlyphId[charCode] = glyphId; + } + } else { + // If it is NOT actually a CID font then CIDs should be mapped + // directly to GIDs. + for (glyphId = 0; glyphId < cff.charStrings.count; glyphId++) { + charCodeToGlyphId[glyphId] = glyphId; + } + } + return charCodeToGlyphId; + } - nextChar : function Type1_nextChar() { - return (this.currentChar = this.stream.getByte()); - }, + var encoding = cff.encoding ? cff.encoding.encoding : null; + charCodeToGlyphId = type1FontGlyphMapping(properties, encoding, charsets); + return charCodeToGlyphId; + } + }; - getToken: function Type1Parser_getToken() { - // Eat whitespace and comments. - var comment = false; - var ch = this.currentChar; - while (true) { - if (ch === -1) { - return null; - } + return CFFFont; +})(); - if (comment) { - if (ch === 0x0A || ch === 0x0D) { - comment = false; - } - } else if (ch === 0x25) { // '%' - comment = true; - } else if (!Lexer.isSpace(ch)) { - break; - } - ch = this.nextChar(); +var CFFParser = (function CFFParserClosure() { + var CharstringValidationData = [ + null, + { id: 'hstem', min: 2, stackClearing: true, stem: true }, + null, + { id: 'vstem', min: 2, stackClearing: true, stem: true }, + { id: 'vmoveto', min: 1, stackClearing: true }, + { id: 'rlineto', min: 2, resetStack: true }, + { id: 'hlineto', min: 1, resetStack: true }, + { id: 'vlineto', min: 1, resetStack: true }, + { id: 'rrcurveto', min: 6, resetStack: true }, + null, + { id: 'callsubr', min: 1, undefStack: true }, + { id: 'return', min: 0, undefStack: true }, + null, // 12 + null, + { id: 'endchar', min: 0, stackClearing: true }, + null, + null, + null, + { id: 'hstemhm', min: 2, stackClearing: true, stem: true }, + { id: 'hintmask', min: 0, stackClearing: true }, + { id: 'cntrmask', min: 0, stackClearing: true }, + { id: 'rmoveto', min: 2, stackClearing: true }, + { id: 'hmoveto', min: 1, stackClearing: true }, + { id: 'vstemhm', min: 2, stackClearing: true, stem: true }, + { id: 'rcurveline', min: 8, resetStack: true }, + { id: 'rlinecurve', min: 8, resetStack: true }, + { id: 'vvcurveto', min: 4, resetStack: true }, + { id: 'hhcurveto', min: 4, resetStack: true }, + null, // shortint + { id: 'callgsubr', min: 1, undefStack: true }, + { id: 'vhcurveto', min: 4, resetStack: true }, + { id: 'hvcurveto', min: 4, resetStack: true } + ]; + var CharstringValidationData12 = [ + null, + null, + null, + { id: 'and', min: 2, stackDelta: -1 }, + { id: 'or', min: 2, stackDelta: -1 }, + { id: 'not', min: 1, stackDelta: 0 }, + null, + null, + null, + { id: 'abs', min: 1, stackDelta: 0 }, + { id: 'add', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] + stack[index - 1]; } - if (isSpecial(ch)) { - this.nextChar(); - return String.fromCharCode(ch); + }, + { id: 'sub', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] - stack[index - 1]; + } + }, + { id: 'div', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] / stack[index - 1]; + } + }, + null, + { id: 'neg', min: 1, stackDelta: 0, + stackFn: function stack_div(stack, index) { + stack[index - 1] = -stack[index - 1]; + } + }, + { id: 'eq', min: 2, stackDelta: -1 }, + null, + null, + { id: 'drop', min: 1, stackDelta: -1 }, + null, + { id: 'put', min: 2, stackDelta: -2 }, + { id: 'get', min: 1, stackDelta: 0 }, + { id: 'ifelse', min: 4, stackDelta: -3 }, + { id: 'random', min: 0, stackDelta: 1 }, + { id: 'mul', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] * stack[index - 1]; } - var token = ''; - do { - token += String.fromCharCode(ch); - ch = this.nextChar(); - } while (ch >= 0 && !Lexer.isSpace(ch) && !isSpecial(ch)); - return token; }, + null, + { id: 'sqrt', min: 1, stackDelta: 0 }, + { id: 'dup', min: 1, stackDelta: 1 }, + { id: 'exch', min: 2, stackDelta: 0 }, + { id: 'index', min: 2, stackDelta: 0 }, + { id: 'roll', min: 3, stackDelta: -2 }, + null, + null, + null, + { id: 'hflex', min: 7, resetStack: true }, + { id: 'flex', min: 13, resetStack: true }, + { id: 'hflex1', min: 9, resetStack: true }, + { id: 'flex1', min: 11, resetStack: true } + ]; - /* - * Returns an object containing a Subrs array and a CharStrings - * array extracted from and eexec encrypted block of data - */ - extractFontProgram: function Type1Parser_extractFontProgram() { - var stream = this.stream; + function CFFParser(file, properties) { + this.bytes = file.getBytes(); + this.properties = properties; + } + CFFParser.prototype = { + parse: function CFFParser_parse() { + var properties = this.properties; + var cff = new CFF(); + this.cff = cff; - var subrs = [], charstrings = []; - var privateData = Object.create(null); - privateData['lenIV'] = 4; - var program = { - subrs: [], - charstrings: [], - properties: { - 'privateData': privateData - } - }; - var token, length, data, lenIV, encoded; - while ((token = this.getToken()) !== null) { - if (token !== '/') { - continue; - } - token = this.getToken(); - switch (token) { - case 'CharStrings': - // The number immediately following CharStrings must be greater or - // equal to the number of CharStrings. - this.getToken(); - this.getToken(); // read in 'dict' - this.getToken(); // read in 'dup' - this.getToken(); // read in 'begin' - while(true) { - token = this.getToken(); - if (token === null || token === 'end') { - break; - } + // The first five sections must be in order, all the others are reached + // via offsets contained in one of the below. + var header = this.parseHeader(); + var nameIndex = this.parseIndex(header.endPos); + var topDictIndex = this.parseIndex(nameIndex.endPos); + var stringIndex = this.parseIndex(topDictIndex.endPos); + var globalSubrIndex = this.parseIndex(stringIndex.endPos); - if (token !== '/') { - continue; - } - var glyph = this.getToken(); - length = this.readInt(); - this.getToken(); // read in 'RD' or '-|' - data = stream.makeSubStream(stream.pos, length); - lenIV = program.properties.privateData['lenIV']; - encoded = decrypt(data.getBytes(), CHAR_STRS_ENCRYPT_KEY, lenIV); - // Skip past the required space and binary data. - stream.skip(length); - this.nextChar(); - token = this.getToken(); // read in 'ND' or '|-' - if (token === 'noaccess') { - this.getToken(); // read in 'def' - } - charstrings.push({ - glyph: glyph, - encoded: encoded - }); - } - break; - case 'Subrs': - var num = this.readInt(); - this.getToken(); // read in 'array' - while ((token = this.getToken()) === 'dup') { - var index = this.readInt(); - length = this.readInt(); - this.getToken(); // read in 'RD' or '-|' - data = stream.makeSubStream(stream.pos, length); - lenIV = program.properties.privateData['lenIV']; - encoded = decrypt(data.getBytes(), CHAR_STRS_ENCRYPT_KEY, lenIV); - // Skip past the required space and binary data. - stream.skip(length); - this.nextChar(); - token = this.getToken(); // read in 'NP' or '|' - if (token === 'noaccess') { - this.getToken(); // read in 'put' - } - subrs[index] = encoded; - } - break; - case 'BlueValues': - case 'OtherBlues': - case 'FamilyBlues': - case 'FamilyOtherBlues': - var blueArray = this.readNumberArray(); - // *Blue* values may contain invalid data: disables reading of - // those values when hinting is disabled. - if (blueArray.length > 0 && (blueArray.length % 2) === 0 && - HINTING_ENABLED) { - program.properties.privateData[token] = blueArray; - } - break; - case 'StemSnapH': - case 'StemSnapV': - program.properties.privateData[token] = this.readNumberArray(); - break; - case 'StdHW': - case 'StdVW': - program.properties.privateData[token] = - this.readNumberArray()[0]; - break; - case 'BlueShift': - case 'lenIV': - case 'BlueFuzz': - case 'BlueScale': - case 'LanguageGroup': - case 'ExpansionFactor': - program.properties.privateData[token] = this.readNumber(); - break; - case 'ForceBold': - program.properties.privateData[token] = this.readBoolean(); - break; - } + var topDictParsed = this.parseDict(topDictIndex.obj.get(0)); + var topDict = this.createDict(CFFTopDict, topDictParsed, cff.strings); + + cff.header = header.obj; + cff.names = this.parseNameIndex(nameIndex.obj); + cff.strings = this.parseStringIndex(stringIndex.obj); + cff.topDict = topDict; + cff.globalSubrIndex = globalSubrIndex.obj; + + this.parsePrivateDict(cff.topDict); + + cff.isCIDFont = topDict.hasName('ROS'); + + var charStringOffset = topDict.getByName('CharStrings'); + var charStringIndex = this.parseIndex(charStringOffset).obj; + + var fontMatrix = topDict.getByName('FontMatrix'); + if (fontMatrix) { + properties.fontMatrix = fontMatrix; } - for (var i = 0; i < charstrings.length; i++) { - glyph = charstrings[i].glyph; - encoded = charstrings[i].encoded; - var charString = new Type1CharString(); - var error = charString.convert(encoded, subrs); - var output = charString.output; - if (error) { - // It seems when FreeType encounters an error while evaluating a glyph - // that it completely ignores the glyph so we'll mimic that behaviour - // here and put an endchar to make the validator happy. - output = [14]; + var fontBBox = topDict.getByName('FontBBox'); + if (fontBBox) { + // adjusting ascent/descent + properties.ascent = fontBBox[3]; + properties.descent = fontBBox[1]; + properties.ascentScaled = true; + } + + var charset, encoding; + if (cff.isCIDFont) { + var fdArrayIndex = this.parseIndex(topDict.getByName('FDArray')).obj; + for (var i = 0, ii = fdArrayIndex.count; i < ii; ++i) { + var dictRaw = fdArrayIndex.get(i); + var fontDict = this.createDict(CFFTopDict, this.parseDict(dictRaw), + cff.strings); + this.parsePrivateDict(fontDict); + cff.fdArray.push(fontDict); } - program.charstrings.push({ - glyphName: glyph, - charstring: output, - width: charString.width, - lsb: charString.lsb, - seac: charString.seac - }); + // cid fonts don't have an encoding + encoding = null; + charset = this.parseCharsets(topDict.getByName('charset'), + charStringIndex.count, cff.strings, true); + cff.fdSelect = this.parseFDSelect(topDict.getByName('FDSelect'), + charStringIndex.count); + } else { + charset = this.parseCharsets(topDict.getByName('charset'), + charStringIndex.count, cff.strings, false); + encoding = this.parseEncoding(topDict.getByName('Encoding'), + properties, + cff.strings, charset.charset); } - return program; + cff.charset = charset; + cff.encoding = encoding; + + var charStringsAndSeacs = this.parseCharStrings( + charStringIndex, + topDict.privateDict.subrsIndex, + globalSubrIndex.obj, + cff.fdSelect, + cff.fdArray); + cff.charStrings = charStringsAndSeacs.charStrings; + cff.seacs = charStringsAndSeacs.seacs; + cff.widths = charStringsAndSeacs.widths; + + return cff; }, + parseHeader: function CFFParser_parseHeader() { + var bytes = this.bytes; + var bytesLength = bytes.length; + var offset = 0; - extractFontHeader: function Type1Parser_extractFontHeader(properties) { - var token; - while ((token = this.getToken()) !== null) { - if (token !== '/') { - continue; + // Prevent an infinite loop, by checking that the offset is within the + // bounds of the bytes array. Necessary in empty, or invalid, font files. + while (offset < bytesLength && bytes[offset] !== 1) { + ++offset; + } + if (offset >= bytesLength) { + error('Invalid CFF header'); + } else if (offset !== 0) { + info('cff data is shifted'); + bytes = bytes.subarray(offset); + this.bytes = bytes; + } + var major = bytes[0]; + var minor = bytes[1]; + var hdrSize = bytes[2]; + var offSize = bytes[3]; + var header = new CFFHeader(major, minor, hdrSize, offSize); + return { obj: header, endPos: hdrSize }; + }, + parseDict: function CFFParser_parseDict(dict) { + var pos = 0; + + function parseOperand() { + var value = dict[pos++]; + if (value === 30) { + return parseFloatOperand(pos); + } else if (value === 28) { + value = dict[pos++]; + value = ((value << 24) | (dict[pos++] << 16)) >> 16; + return value; + } else if (value === 29) { + value = dict[pos++]; + value = (value << 8) | dict[pos++]; + value = (value << 8) | dict[pos++]; + value = (value << 8) | dict[pos++]; + return value; + } else if (value >= 32 && value <= 246) { + return value - 139; + } else if (value >= 247 && value <= 250) { + return ((value - 247) * 256) + dict[pos++] + 108; + } else if (value >= 251 && value <= 254) { + return -((value - 251) * 256) - dict[pos++] - 108; + } else { + error('255 is not a valid DICT command'); } - token = this.getToken(); - switch (token) { - case 'FontMatrix': - var matrix = this.readNumberArray(); - properties.fontMatrix = matrix; - break; - case 'Encoding': - var encodingArg = this.getToken(); - var encoding; - if (!/^\d+$/.test(encodingArg)) { - // encoding name is specified - encoding = getEncoding(encodingArg); - } else { - encoding = []; - var size = parseInt(encodingArg, 10) | 0; - this.getToken(); // read in 'array' + return -1; + } - for (var j = 0; j < size; j++) { - token = this.getToken(); - // skipping till first dup or def (e.g. ignoring for statement) - while (token !== 'dup' && token !== 'def') { - token = this.getToken(); - if (token === null) { - return; // invalid header - } - } - if (token === 'def') { - break; // read all array data - } - var index = this.readInt(); - this.getToken(); // read in '/' - var glyph = this.getToken(); - encoding[index] = glyph; - this.getToken(); // read the in 'put' - } - } - properties.builtInEncoding = encoding; + function parseFloatOperand() { + var str = ''; + var eof = 15; + var lookup = ['0', '1', '2', '3', '4', '5', '6', '7', '8', + '9', '.', 'E', 'E-', null, '-']; + var length = dict.length; + while (pos < length) { + var b = dict[pos++]; + var b1 = b >> 4; + var b2 = b & 15; + + if (b1 === eof) { break; - case 'FontBBox': - var fontBBox = this.readNumberArray(); - // adjusting ascent/descent - properties.ascent = fontBBox[3]; - properties.descent = fontBBox[1]; - properties.ascentScaled = true; + } + str += lookup[b1]; + + if (b2 === eof) { break; + } + str += lookup[b2]; } + return parseFloat(str); } - } - }; - return Type1Parser; -})(); + var operands = []; + var entries = []; -/** - * The CFF class takes a Type1 file and wrap it into a - * 'Compact Font Format' which itself embed Type2 charstrings. - */ -var CFFStandardStrings = [ - '.notdef', 'space', 'exclam', 'quotedbl', 'numbersign', 'dollar', 'percent', - 'ampersand', 'quoteright', 'parenleft', 'parenright', 'asterisk', 'plus', - 'comma', 'hyphen', 'period', 'slash', 'zero', 'one', 'two', 'three', 'four', - 'five', 'six', 'seven', 'eight', 'nine', 'colon', 'semicolon', 'less', - 'equal', 'greater', 'question', 'at', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', - 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', - 'X', 'Y', 'Z', 'bracketleft', 'backslash', 'bracketright', 'asciicircum', - 'underscore', 'quoteleft', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', - 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', - 'z', 'braceleft', 'bar', 'braceright', 'asciitilde', 'exclamdown', 'cent', - 'sterling', 'fraction', 'yen', 'florin', 'section', 'currency', - 'quotesingle', 'quotedblleft', 'guillemotleft', 'guilsinglleft', - 'guilsinglright', 'fi', 'fl', 'endash', 'dagger', 'daggerdbl', - 'periodcentered', 'paragraph', 'bullet', 'quotesinglbase', 'quotedblbase', - 'quotedblright', 'guillemotright', 'ellipsis', 'perthousand', 'questiondown', - 'grave', 'acute', 'circumflex', 'tilde', 'macron', 'breve', 'dotaccent', - 'dieresis', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', 'emdash', - 'AE', 'ordfeminine', 'Lslash', 'Oslash', 'OE', 'ordmasculine', 'ae', - 'dotlessi', 'lslash', 'oslash', 'oe', 'germandbls', 'onesuperior', - 'logicalnot', 'mu', 'trademark', 'Eth', 'onehalf', 'plusminus', 'Thorn', - 'onequarter', 'divide', 'brokenbar', 'degree', 'thorn', 'threequarters', - 'twosuperior', 'registered', 'minus', 'eth', 'multiply', 'threesuperior', - 'copyright', 'Aacute', 'Acircumflex', 'Adieresis', 'Agrave', 'Aring', - 'Atilde', 'Ccedilla', 'Eacute', 'Ecircumflex', 'Edieresis', 'Egrave', - 'Iacute', 'Icircumflex', 'Idieresis', 'Igrave', 'Ntilde', 'Oacute', - 'Ocircumflex', 'Odieresis', 'Ograve', 'Otilde', 'Scaron', 'Uacute', - 'Ucircumflex', 'Udieresis', 'Ugrave', 'Yacute', 'Ydieresis', 'Zcaron', - 'aacute', 'acircumflex', 'adieresis', 'agrave', 'aring', 'atilde', - 'ccedilla', 'eacute', 'ecircumflex', 'edieresis', 'egrave', 'iacute', - 'icircumflex', 'idieresis', 'igrave', 'ntilde', 'oacute', 'ocircumflex', - 'odieresis', 'ograve', 'otilde', 'scaron', 'uacute', 'ucircumflex', - 'udieresis', 'ugrave', 'yacute', 'ydieresis', 'zcaron', 'exclamsmall', - 'Hungarumlautsmall', 'dollaroldstyle', 'dollarsuperior', 'ampersandsmall', - 'Acutesmall', 'parenleftsuperior', 'parenrightsuperior', 'twodotenleader', - 'onedotenleader', 'zerooldstyle', 'oneoldstyle', 'twooldstyle', - 'threeoldstyle', 'fouroldstyle', 'fiveoldstyle', 'sixoldstyle', - 'sevenoldstyle', 'eightoldstyle', 'nineoldstyle', 'commasuperior', - 'threequartersemdash', 'periodsuperior', 'questionsmall', 'asuperior', - 'bsuperior', 'centsuperior', 'dsuperior', 'esuperior', 'isuperior', - 'lsuperior', 'msuperior', 'nsuperior', 'osuperior', 'rsuperior', 'ssuperior', - 'tsuperior', 'ff', 'ffi', 'ffl', 'parenleftinferior', 'parenrightinferior', - 'Circumflexsmall', 'hyphensuperior', 'Gravesmall', 'Asmall', 'Bsmall', - 'Csmall', 'Dsmall', 'Esmall', 'Fsmall', 'Gsmall', 'Hsmall', 'Ismall', - 'Jsmall', 'Ksmall', 'Lsmall', 'Msmall', 'Nsmall', 'Osmall', 'Psmall', - 'Qsmall', 'Rsmall', 'Ssmall', 'Tsmall', 'Usmall', 'Vsmall', 'Wsmall', - 'Xsmall', 'Ysmall', 'Zsmall', 'colonmonetary', 'onefitted', 'rupiah', - 'Tildesmall', 'exclamdownsmall', 'centoldstyle', 'Lslashsmall', - 'Scaronsmall', 'Zcaronsmall', 'Dieresissmall', 'Brevesmall', 'Caronsmall', - 'Dotaccentsmall', 'Macronsmall', 'figuredash', 'hypheninferior', - 'Ogoneksmall', 'Ringsmall', 'Cedillasmall', 'questiondownsmall', 'oneeighth', - 'threeeighths', 'fiveeighths', 'seveneighths', 'onethird', 'twothirds', - 'zerosuperior', 'foursuperior', 'fivesuperior', 'sixsuperior', - 'sevensuperior', 'eightsuperior', 'ninesuperior', 'zeroinferior', - 'oneinferior', 'twoinferior', 'threeinferior', 'fourinferior', - 'fiveinferior', 'sixinferior', 'seveninferior', 'eightinferior', - 'nineinferior', 'centinferior', 'dollarinferior', 'periodinferior', - 'commainferior', 'Agravesmall', 'Aacutesmall', 'Acircumflexsmall', - 'Atildesmall', 'Adieresissmall', 'Aringsmall', 'AEsmall', 'Ccedillasmall', - 'Egravesmall', 'Eacutesmall', 'Ecircumflexsmall', 'Edieresissmall', - 'Igravesmall', 'Iacutesmall', 'Icircumflexsmall', 'Idieresissmall', - 'Ethsmall', 'Ntildesmall', 'Ogravesmall', 'Oacutesmall', 'Ocircumflexsmall', - 'Otildesmall', 'Odieresissmall', 'OEsmall', 'Oslashsmall', 'Ugravesmall', - 'Uacutesmall', 'Ucircumflexsmall', 'Udieresissmall', 'Yacutesmall', - 'Thornsmall', 'Ydieresissmall', '001.000', '001.001', '001.002', '001.003', - 'Black', 'Bold', 'Book', 'Light', 'Medium', 'Regular', 'Roman', 'Semibold' -]; - -// Type1Font is also a CIDFontType0. -var Type1Font = function Type1Font(name, file, properties) { - // Some bad generators embed pfb file as is, we have to strip 6-byte headers. - // Also, length1 and length2 might be off by 6 bytes as well. - // http://www.math.ubc.ca/~cass/piscript/type1.pdf - var PFB_HEADER_SIZE = 6; - var headerBlockLength = properties.length1; - var eexecBlockLength = properties.length2; - var pfbHeader = file.peekBytes(PFB_HEADER_SIZE); - var pfbHeaderPresent = pfbHeader[0] === 0x80 && pfbHeader[1] === 0x01; - if (pfbHeaderPresent) { - file.skip(PFB_HEADER_SIZE); - headerBlockLength = (pfbHeader[5] << 24) | (pfbHeader[4] << 16) | - (pfbHeader[3] << 8) | pfbHeader[2]; - } - - // Get the data block containing glyphs and subrs informations - var headerBlock = new Stream(file.getBytes(headerBlockLength)); - var headerBlockParser = new Type1Parser(headerBlock); - headerBlockParser.extractFontHeader(properties); - - if (pfbHeaderPresent) { - pfbHeader = file.getBytes(PFB_HEADER_SIZE); - eexecBlockLength = (pfbHeader[5] << 24) | (pfbHeader[4] << 16) | - (pfbHeader[3] << 8) | pfbHeader[2]; - } - - // Decrypt the data blocks and retrieve it's content - var eexecBlock = new Stream(file.getBytes(eexecBlockLength)); - var eexecBlockParser = new Type1Parser(eexecBlock, true); - var data = eexecBlockParser.extractFontProgram(); - for (var info in data.properties) { - properties[info] = data.properties[info]; - } - - var charstrings = data.charstrings; - var type2Charstrings = this.getType2Charstrings(charstrings); - var subrs = this.getType2Subrs(data.subrs); - - this.charstrings = charstrings; - this.data = this.wrap(name, type2Charstrings, this.charstrings, - subrs, properties); - this.seacs = this.getSeacs(data.charstrings); -}; - -Type1Font.prototype = { - get numGlyphs() { - return this.charstrings.length + 1; - }, - - getCharset: function Type1Font_getCharset() { - var charset = ['.notdef']; - var charstrings = this.charstrings; - for (var glyphId = 0; glyphId < charstrings.length; glyphId++) { - charset.push(charstrings[glyphId].glyphName); - } - return charset; - }, - - getGlyphMapping: function Type1Font_getGlyphMapping(properties) { - var charstrings = this.charstrings; - var glyphNames = ['.notdef'], glyphId; - for (glyphId = 0; glyphId < charstrings.length; glyphId++) { - glyphNames.push(charstrings[glyphId].glyphName); - } - var encoding = properties.builtInEncoding; - if (encoding) { - var builtInEncoding = Object.create(null); - for (var charCode in encoding) { - glyphId = glyphNames.indexOf(encoding[charCode]); - if (glyphId >= 0) { - builtInEncoding[charCode] = glyphId; + pos = 0; + var end = dict.length; + while (pos < end) { + var b = dict[pos]; + if (b <= 21) { + if (b === 12) { + b = (b << 8) | dict[++pos]; + } + entries.push([b, operands]); + operands = []; + ++pos; + } else { + operands.push(parseOperand()); } } - } + return entries; + }, + parseIndex: function CFFParser_parseIndex(pos) { + var cffIndex = new CFFIndex(); + var bytes = this.bytes; + var count = (bytes[pos++] << 8) | bytes[pos++]; + var offsets = []; + var end = pos; + var i, ii; - return type1FontGlyphMapping(properties, builtInEncoding, glyphNames); - }, + if (count !== 0) { + var offsetSize = bytes[pos++]; + // add 1 for offset to determine size of last object + var startPos = pos + ((count + 1) * offsetSize) - 1; - getSeacs: function Type1Font_getSeacs(charstrings) { - var i, ii; - var seacMap = []; - for (i = 0, ii = charstrings.length; i < ii; i++) { - var charstring = charstrings[i]; - if (charstring.seac) { - // Offset by 1 for .notdef - seacMap[i + 1] = charstring.seac; + for (i = 0, ii = count + 1; i < ii; ++i) { + var offset = 0; + for (var j = 0; j < offsetSize; ++j) { + offset <<= 8; + offset += bytes[pos++]; + } + offsets.push(startPos + offset); + } + end = offsets[count]; } - } - return seacMap; - }, - - getType2Charstrings: function Type1Font_getType2Charstrings( - type1Charstrings) { - var type2Charstrings = []; - for (var i = 0, ii = type1Charstrings.length; i < ii; i++) { - type2Charstrings.push(type1Charstrings[i].charstring); - } - return type2Charstrings; - }, - - getType2Subrs: function Type1Font_getType2Subrs(type1Subrs) { - var bias = 0; - var count = type1Subrs.length; - if (count < 1133) { - bias = 107; - } else if (count < 33769) { - bias = 1131; - } else { - bias = 32768; - } - - // Add a bunch of empty subrs to deal with the Type2 bias - var type2Subrs = []; - var i; - for (i = 0; i < bias; i++) { - type2Subrs.push([0x0B]); - } - - for (i = 0; i < count; i++) { - type2Subrs.push(type1Subrs[i]); - } + for (i = 0, ii = offsets.length - 1; i < ii; ++i) { + var offsetStart = offsets[i]; + var offsetEnd = offsets[i + 1]; + cffIndex.add(bytes.subarray(offsetStart, offsetEnd)); + } + return {obj: cffIndex, endPos: end}; + }, + parseNameIndex: function CFFParser_parseNameIndex(index) { + var names = []; + for (var i = 0, ii = index.count; i < ii; ++i) { + var name = index.get(i); + // OTS doesn't allow names to be over 127 characters. + var length = Math.min(name.length, 127); + var data = []; + // OTS also only permits certain characters in the name. + for (var j = 0; j < length; ++j) { + var c = name[j]; + if (j === 0 && c === 0) { + data[j] = c; + continue; + } + if ((c < 33 || c > 126) || c === 91 /* [ */ || c === 93 /* ] */ || + c === 40 /* ( */ || c === 41 /* ) */ || c === 123 /* { */ || + c === 125 /* } */ || c === 60 /* < */ || c === 62 /* > */ || + c === 47 /* / */ || c === 37 /* % */ || c === 35 /* # */) { + data[j] = 95; + continue; + } + data[j] = c; + } + names.push(bytesToString(data)); + } + return names; + }, + parseStringIndex: function CFFParser_parseStringIndex(index) { + var strings = new CFFStrings(); + for (var i = 0, ii = index.count; i < ii; ++i) { + var data = index.get(i); + strings.add(bytesToString(data)); + } + return strings; + }, + createDict: function CFFParser_createDict(Type, dict, strings) { + var cffDict = new Type(strings); + for (var i = 0, ii = dict.length; i < ii; ++i) { + var pair = dict[i]; + var key = pair[0]; + var value = pair[1]; + cffDict.setByKey(key, value); + } + return cffDict; + }, + parseCharString: function CFFParser_parseCharString(state, data, + localSubrIndex, + globalSubrIndex) { + if (state.callDepth > MAX_SUBR_NESTING) { + return false; + } + var stackSize = state.stackSize; + var stack = state.stack; - return type2Subrs; - }, + var length = data.length; - wrap: function Type1Font_wrap(name, glyphs, charstrings, subrs, properties) { - var cff = new CFF(); - cff.header = new CFFHeader(1, 0, 4, 4); + for (var j = 0; j < length;) { + var value = data[j++]; + var validationCommand = null; + if (value === 12) { + var q = data[j++]; + if (q === 0) { + // The CFF specification state that the 'dotsection' command + // (12, 0) is deprecated and treated as a no-op, but all Type2 + // charstrings processors should support them. Unfortunately + // the font sanitizer don't. As a workaround the sequence (12, 0) + // is replaced by a useless (0, hmoveto). + data[j - 2] = 139; + data[j - 1] = 22; + stackSize = 0; + } else { + validationCommand = CharstringValidationData12[q]; + } + } else if (value === 28) { // number (16 bit) + stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16)) >> 16; + j += 2; + stackSize++; + } else if (value === 14) { + if (stackSize >= 4) { + stackSize -= 4; + if (SEAC_ANALYSIS_ENABLED) { + state.seac = stack.slice(stackSize, stackSize + 4); + return false; + } + } + validationCommand = CharstringValidationData[value]; + } else if (value >= 32 && value <= 246) { // number + stack[stackSize] = value - 139; + stackSize++; + } else if (value >= 247 && value <= 254) { // number (+1 bytes) + stack[stackSize] = (value < 251 ? + ((value - 247) << 8) + data[j] + 108 : + -((value - 251) << 8) - data[j] - 108); + j++; + stackSize++; + } else if (value === 255) { // number (32 bit) + stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16) | + (data[j + 2] << 8) | data[j + 3]) / 65536; + j += 4; + stackSize++; + } else if (value === 19 || value === 20) { + state.hints += stackSize >> 1; + // skipping right amount of hints flag data + j += (state.hints + 7) >> 3; + stackSize %= 2; + validationCommand = CharstringValidationData[value]; + } else if (value === 10 || value === 29) { + var subrsIndex; + if (value === 10) { + subrsIndex = localSubrIndex; + } else { + subrsIndex = globalSubrIndex; + } + if (!subrsIndex) { + validationCommand = CharstringValidationData[value]; + warn('Missing subrsIndex for ' + validationCommand.id); + return false; + } + var bias = 32768; + if (subrsIndex.count < 1240) { + bias = 107; + } else if (subrsIndex.count < 33900) { + bias = 1131; + } + var subrNumber = stack[--stackSize] + bias; + if (subrNumber < 0 || subrNumber >= subrsIndex.count) { + validationCommand = CharstringValidationData[value]; + warn('Out of bounds subrIndex for ' + validationCommand.id); + return false; + } + state.stackSize = stackSize; + state.callDepth++; + var valid = this.parseCharString(state, subrsIndex.get(subrNumber), + localSubrIndex, globalSubrIndex); + if (!valid) { + return false; + } + state.callDepth--; + stackSize = state.stackSize; + continue; + } else if (value === 11) { + state.stackSize = stackSize; + return true; + } else { + validationCommand = CharstringValidationData[value]; + } + if (validationCommand) { + if (validationCommand.stem) { + state.hints += stackSize >> 1; + } + if ('min' in validationCommand) { + if (!state.undefStack && stackSize < validationCommand.min) { + warn('Not enough parameters for ' + validationCommand.id + + '; actual: ' + stackSize + + ', expected: ' + validationCommand.min); + return false; + } + } + if (state.firstStackClearing && validationCommand.stackClearing) { + state.firstStackClearing = false; + // the optional character width can be found before the first + // stack-clearing command arguments + stackSize -= validationCommand.min; + if (stackSize >= 2 && validationCommand.stem) { + // there are even amount of arguments for stem commands + stackSize %= 2; + } else if (stackSize > 1) { + warn('Found too many parameters for stack-clearing command'); + } + if (stackSize > 0 && stack[stackSize - 1] >= 0) { + state.width = stack[stackSize - 1]; + } + } + if ('stackDelta' in validationCommand) { + if ('stackFn' in validationCommand) { + validationCommand.stackFn(stack, stackSize); + } + stackSize += validationCommand.stackDelta; + } else if (validationCommand.stackClearing) { + stackSize = 0; + } else if (validationCommand.resetStack) { + stackSize = 0; + state.undefStack = false; + } else if (validationCommand.undefStack) { + stackSize = 0; + state.undefStack = true; + state.firstStackClearing = false; + } + } + } + state.stackSize = stackSize; + return true; + }, + parseCharStrings: function CFFParser_parseCharStrings(charStrings, + localSubrIndex, + globalSubrIndex, + fdSelect, + fdArray) { + var seacs = []; + var widths = []; + var count = charStrings.count; + for (var i = 0; i < count; i++) { + var charstring = charStrings.get(i); + var state = { + callDepth: 0, + stackSize: 0, + stack: [], + undefStack: true, + hints: 0, + firstStackClearing: true, + seac: null, + width: null + }; + var valid = true; + var localSubrToUse = null; + if (fdSelect && fdArray.length) { + var fdIndex = fdSelect.getFDIndex(i); + if (fdIndex === -1) { + warn('Glyph index is not in fd select.'); + valid = false; + } + if (fdIndex >= fdArray.length) { + warn('Invalid fd index for glyph index.'); + valid = false; + } + if (valid) { + localSubrToUse = fdArray[fdIndex].privateDict.subrsIndex; + } + } else if (localSubrIndex) { + localSubrToUse = localSubrIndex; + } + if (valid) { + valid = this.parseCharString(state, charstring, localSubrToUse, + globalSubrIndex); + } + if (state.width !== null) { + widths[i] = state.width; + } + if (state.seac !== null) { + seacs[i] = state.seac; + } + if (!valid) { + // resetting invalid charstring to single 'endchar' + charStrings.set(i, new Uint8Array([14])); + } + } + return { charStrings: charStrings, seacs: seacs, widths: widths }; + }, + emptyPrivateDictionary: + function CFFParser_emptyPrivateDictionary(parentDict) { + var privateDict = this.createDict(CFFPrivateDict, [], + parentDict.strings); + parentDict.setByKey(18, [0, 0]); + parentDict.privateDict = privateDict; + }, + parsePrivateDict: function CFFParser_parsePrivateDict(parentDict) { + // no private dict, do nothing + if (!parentDict.hasName('Private')) { + this.emptyPrivateDictionary(parentDict); + return; + } + var privateOffset = parentDict.getByName('Private'); + // make sure the params are formatted correctly + if (!isArray(privateOffset) || privateOffset.length !== 2) { + parentDict.removeByName('Private'); + return; + } + var size = privateOffset[0]; + var offset = privateOffset[1]; + // remove empty dicts or ones that refer to invalid location + if (size === 0 || offset >= this.bytes.length) { + this.emptyPrivateDictionary(parentDict); + return; + } - cff.names = [name]; + var privateDictEnd = offset + size; + var dictData = this.bytes.subarray(offset, privateDictEnd); + var dict = this.parseDict(dictData); + var privateDict = this.createDict(CFFPrivateDict, dict, + parentDict.strings); + parentDict.privateDict = privateDict; - var topDict = new CFFTopDict(); - // CFF strings IDs 0...390 are predefined names, so refering - // to entries in our own String INDEX starts at SID 391. - topDict.setByName('version', 391); - topDict.setByName('Notice', 392); - topDict.setByName('FullName', 393); - topDict.setByName('FamilyName', 394); - topDict.setByName('Weight', 395); - topDict.setByName('Encoding', null); // placeholder - topDict.setByName('FontMatrix', properties.fontMatrix); - topDict.setByName('FontBBox', properties.bbox); - topDict.setByName('charset', null); // placeholder - topDict.setByName('CharStrings', null); // placeholder - topDict.setByName('Private', null); // placeholder - cff.topDict = topDict; + // Parse the Subrs index also since it's relative to the private dict. + if (!privateDict.getByName('Subrs')) { + return; + } + var subrsOffset = privateDict.getByName('Subrs'); + var relativeOffset = offset + subrsOffset; + // Validate the offset. + if (subrsOffset === 0 || relativeOffset >= this.bytes.length) { + this.emptyPrivateDictionary(parentDict); + return; + } + var subrsIndex = this.parseIndex(relativeOffset); + privateDict.subrsIndex = subrsIndex.obj; + }, + parseCharsets: function CFFParser_parseCharsets(pos, length, strings, cid) { + if (pos === 0) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.ISO_ADOBE, + ISOAdobeCharset); + } else if (pos === 1) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT, + ExpertCharset); + } else if (pos === 2) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT_SUBSET, + ExpertSubsetCharset); + } - var strings = new CFFStrings(); - strings.add('Version 0.11'); // Version - strings.add('See original notice'); // Notice - strings.add(name); // FullName - strings.add(name); // FamilyName - strings.add('Medium'); // Weight - cff.strings = strings; + var bytes = this.bytes; + var start = pos; + var format = bytes[pos++]; + var charset = ['.notdef']; + var id, count, i; - cff.globalSubrIndex = new CFFIndex(); + // subtract 1 for the .notdef glyph + length -= 1; - var count = glyphs.length; - var charsetArray = [0]; - var i, ii; - for (i = 0; i < count; i++) { - var index = CFFStandardStrings.indexOf(charstrings[i].glyphName); - // TODO: Insert the string and correctly map it. Previously it was - // thought mapping names that aren't in the standard strings to .notdef - // was fine, however in issue818 when mapping them all to .notdef the - // adieresis glyph no longer worked. - if (index === -1) { - index = 0; + switch (format) { + case 0: + for (i = 0; i < length; i++) { + id = (bytes[pos++] << 8) | bytes[pos++]; + charset.push(cid ? id : strings.get(id)); + } + break; + case 1: + while (charset.length <= length) { + id = (bytes[pos++] << 8) | bytes[pos++]; + count = bytes[pos++]; + for (i = 0; i <= count; i++) { + charset.push(cid ? id++ : strings.get(id++)); + } + } + break; + case 2: + while (charset.length <= length) { + id = (bytes[pos++] << 8) | bytes[pos++]; + count = (bytes[pos++] << 8) | bytes[pos++]; + for (i = 0; i <= count; i++) { + charset.push(cid ? id++ : strings.get(id++)); + } + } + break; + default: + error('Unknown charset format'); } - charsetArray.push((index >> 8) & 0xff, index & 0xff); - } - cff.charset = new CFFCharset(false, 0, [], charsetArray); + // Raw won't be needed if we actually compile the charset. + var end = pos; + var raw = bytes.subarray(start, end); - var charStringsIndex = new CFFIndex(); - charStringsIndex.add([0x8B, 0x0E]); // .notdef - for (i = 0; i < count; i++) { - charStringsIndex.add(glyphs[i]); - } - cff.charStrings = charStringsIndex; + return new CFFCharset(false, format, charset, raw); + }, + parseEncoding: function CFFParser_parseEncoding(pos, + properties, + strings, + charset) { + var encoding = Object.create(null); + var bytes = this.bytes; + var predefined = false; + var hasSupplement = false; + var format, i, ii; + var raw = null; - var privateDict = new CFFPrivateDict(); - privateDict.setByName('Subrs', null); // placeholder - var fields = [ - 'BlueValues', - 'OtherBlues', - 'FamilyBlues', - 'FamilyOtherBlues', - 'StemSnapH', - 'StemSnapV', - 'BlueShift', - 'BlueFuzz', - 'BlueScale', - 'LanguageGroup', - 'ExpansionFactor', - 'ForceBold', - 'StdHW', - 'StdVW' - ]; - for (i = 0, ii = fields.length; i < ii; i++) { - var field = fields[i]; - if (!(field in properties.privateData)) { - continue; + function readSupplement() { + var supplementsCount = bytes[pos++]; + for (i = 0; i < supplementsCount; i++) { + var code = bytes[pos++]; + var sid = (bytes[pos++] << 8) + (bytes[pos++] & 0xff); + encoding[code] = charset.indexOf(strings.get(sid)); + } } - var value = properties.privateData[field]; - if (isArray(value)) { - // All of the private dictionary array data in CFF must be stored as - // "delta-encoded" numbers. - for (var j = value.length - 1; j > 0; j--) { - value[j] -= value[j - 1]; // ... difference from previous value + + if (pos === 0 || pos === 1) { + predefined = true; + format = pos; + var baseEncoding = pos ? ExpertEncoding : StandardEncoding; + for (i = 0, ii = charset.length; i < ii; i++) { + var index = baseEncoding.indexOf(charset[i]); + if (index !== -1) { + encoding[index] = i; + } + } + } else { + var dataStart = pos; + format = bytes[pos++]; + switch (format & 0x7f) { + case 0: + var glyphsCount = bytes[pos++]; + for (i = 1; i <= glyphsCount; i++) { + encoding[bytes[pos++]] = i; + } + break; + + case 1: + var rangesCount = bytes[pos++]; + var gid = 1; + for (i = 0; i < rangesCount; i++) { + var start = bytes[pos++]; + var left = bytes[pos++]; + for (var j = start; j <= start + left; j++) { + encoding[j] = gid++; + } + } + break; + + default: + error('Unknow encoding format: ' + format + ' in CFF'); + break; + } + var dataEnd = pos; + if (format & 0x80) { + // The font sanitizer does not support CFF encoding with a + // supplement, since the encoding is not really used to map + // between gid to glyph, let's overwrite what is declared in + // the top dictionary to let the sanitizer think the font use + // StandardEncoding, that's a lie but that's ok. + bytes[dataStart] &= 0x7f; + readSupplement(); + hasSupplement = true; } + raw = bytes.subarray(dataStart, dataEnd); } - privateDict.setByName(field, value); - } - cff.topDict.privateDict = privateDict; + format = format & 0x7f; + return new CFFEncoding(predefined, format, encoding, raw); + }, + parseFDSelect: function CFFParser_parseFDSelect(pos, length) { + var start = pos; + var bytes = this.bytes; + var format = bytes[pos++]; + var fdSelect = []; + var i; - var subrIndex = new CFFIndex(); - for (i = 0, ii = subrs.length; i < ii; i++) { - subrIndex.add(subrs[i]); + switch (format) { + case 0: + for (i = 0; i < length; ++i) { + var id = bytes[pos++]; + fdSelect.push(id); + } + break; + case 3: + var rangesCount = (bytes[pos++] << 8) | bytes[pos++]; + for (i = 0; i < rangesCount; ++i) { + var first = (bytes[pos++] << 8) | bytes[pos++]; + var fdIndex = bytes[pos++]; + var next = (bytes[pos] << 8) | bytes[pos + 1]; + for (var j = first; j < next; ++j) { + fdSelect.push(fdIndex); + } + } + // Advance past the sentinel(next). + pos += 2; + break; + default: + error('Unknown fdselect format ' + format); + break; + } + var end = pos; + return new CFFFDSelect(fdSelect, bytes.subarray(start, end)); } - privateDict.subrsIndex = subrIndex; + }; + return CFFParser; +})(); - var compiler = new CFFCompiler(cff); - return compiler.compile(); +// Compact Font Format +var CFF = (function CFFClosure() { + function CFF() { + this.header = null; + this.names = []; + this.topDict = null; + this.strings = new CFFStrings(); + this.globalSubrIndex = null; + + // The following could really be per font, but since we only have one font + // store them here. + this.encoding = null; + this.charset = null; + this.charStrings = null; + this.fdArray = []; + this.fdSelect = null; + + this.isCIDFont = false; } -}; + return CFF; +})(); -var CFFFont = (function CFFFontClosure() { - function CFFFont(file, properties) { - this.properties = properties; +var CFFHeader = (function CFFHeaderClosure() { + function CFFHeader(major, minor, hdrSize, offSize) { + this.major = major; + this.minor = minor; + this.hdrSize = hdrSize; + this.offSize = offSize; + } + return CFFHeader; +})(); - var parser = new CFFParser(file, properties); - this.cff = parser.parse(); - var compiler = new CFFCompiler(this.cff); - this.seacs = this.cff.seacs; - try { - this.data = compiler.compile(); - } catch (e) { - warn('Failed to compile font ' + properties.loadedName); - // There may have just been an issue with the compiler, set the data - // anyway and hope the font loaded. - this.data = file; - } +var CFFStrings = (function CFFStringsClosure() { + function CFFStrings() { + this.strings = []; } + CFFStrings.prototype = { + get: function CFFStrings_get(index) { + if (index >= 0 && index <= 390) { + return CFFStandardStrings[index]; + } + if (index - 391 <= this.strings.length) { + return this.strings[index - 391]; + } + return CFFStandardStrings[0]; + }, + add: function CFFStrings_add(value) { + this.strings.push(value); + }, + get count() { + return this.strings.length; + } + }; + return CFFStrings; +})(); - CFFFont.prototype = { - get numGlyphs() { - return this.cff.charStrings.count; +var CFFIndex = (function CFFIndexClosure() { + function CFFIndex() { + this.objects = []; + this.length = 0; + } + CFFIndex.prototype = { + add: function CFFIndex_add(data) { + this.length += data.length; + this.objects.push(data); }, - getCharset: function CFFFont_getCharset() { - return this.cff.charset.charset; + set: function CFFIndex_set(index, data) { + this.length += data.length - this.objects[index].length; + this.objects[index] = data; }, - getGlyphMapping: function CFFFont_getGlyphMapping() { - var cff = this.cff; - var properties = this.properties; - var charsets = cff.charset.charset; - var charCodeToGlyphId; - var glyphId; + get: function CFFIndex_get(index) { + return this.objects[index]; + }, + get count() { + return this.objects.length; + } + }; + return CFFIndex; +})(); - if (properties.composite) { - charCodeToGlyphId = Object.create(null); - if (cff.isCIDFont) { - // If the font is actually a CID font then we should use the charset - // to map CIDs to GIDs. - for (glyphId = 0; glyphId < charsets.length; glyphId++) { - var cid = charsets[glyphId]; - var charCode = properties.cMap.charCodeOf(cid); - charCodeToGlyphId[charCode] = glyphId; - } - } else { - // If it is NOT actually a CID font then CIDs should be mapped - // directly to GIDs. - for (glyphId = 0; glyphId < cff.charStrings.count; glyphId++) { - charCodeToGlyphId[glyphId] = glyphId; - } - } - return charCodeToGlyphId; +var CFFDict = (function CFFDictClosure() { + function CFFDict(tables, strings) { + this.keyToNameMap = tables.keyToNameMap; + this.nameToKeyMap = tables.nameToKeyMap; + this.defaults = tables.defaults; + this.types = tables.types; + this.opcodes = tables.opcodes; + this.order = tables.order; + this.strings = strings; + this.values = Object.create(null); + } + CFFDict.prototype = { + // value should always be an array + setByKey: function CFFDict_setByKey(key, value) { + if (!(key in this.keyToNameMap)) { + return false; } - - var encoding = cff.encoding ? cff.encoding.encoding : null; - charCodeToGlyphId = type1FontGlyphMapping(properties, encoding, charsets); - return charCodeToGlyphId; + // ignore empty values + if (value.length === 0) { + return true; + } + var type = this.types[key]; + // remove the array wrapping these types of values + if (type === 'num' || type === 'sid' || type === 'offset') { + value = value[0]; + } + this.values[key] = value; + return true; + }, + setByName: function CFFDict_setByName(name, value) { + if (!(name in this.nameToKeyMap)) { + error('Invalid dictionary name "' + name + '"'); + } + this.values[this.nameToKeyMap[name]] = value; + }, + hasName: function CFFDict_hasName(name) { + return this.nameToKeyMap[name] in this.values; + }, + getByName: function CFFDict_getByName(name) { + if (!(name in this.nameToKeyMap)) { + error('Invalid dictionary name "' + name + '"'); + } + var key = this.nameToKeyMap[name]; + if (!(key in this.values)) { + return this.defaults[key]; + } + return this.values[key]; + }, + removeByName: function CFFDict_removeByName(name) { + delete this.values[this.nameToKeyMap[name]]; + } + }; + CFFDict.createTables = function CFFDict_createTables(layout) { + var tables = { + keyToNameMap: {}, + nameToKeyMap: {}, + defaults: {}, + types: {}, + opcodes: {}, + order: [] + }; + for (var i = 0, ii = layout.length; i < ii; ++i) { + var entry = layout[i]; + var key = isArray(entry[0]) ? (entry[0][0] << 8) + entry[0][1] : entry[0]; + tables.keyToNameMap[key] = entry[1]; + tables.nameToKeyMap[entry[1]] = key; + tables.types[key] = entry[2]; + tables.defaults[key] = entry[3]; + tables.opcodes[key] = isArray(entry[0]) ? entry[0] : [entry[0]]; + tables.order.push(key); } + return tables; }; + return CFFDict; +})(); - return CFFFont; +var CFFTopDict = (function CFFTopDictClosure() { + var layout = [ + [[12, 30], 'ROS', ['sid', 'sid', 'num'], null], + [[12, 20], 'SyntheticBase', 'num', null], + [0, 'version', 'sid', null], + [1, 'Notice', 'sid', null], + [[12, 0], 'Copyright', 'sid', null], + [2, 'FullName', 'sid', null], + [3, 'FamilyName', 'sid', null], + [4, 'Weight', 'sid', null], + [[12, 1], 'isFixedPitch', 'num', 0], + [[12, 2], 'ItalicAngle', 'num', 0], + [[12, 3], 'UnderlinePosition', 'num', -100], + [[12, 4], 'UnderlineThickness', 'num', 50], + [[12, 5], 'PaintType', 'num', 0], + [[12, 6], 'CharstringType', 'num', 2], + [[12, 7], 'FontMatrix', ['num', 'num', 'num', 'num', 'num', 'num'], + [0.001, 0, 0, 0.001, 0, 0]], + [13, 'UniqueID', 'num', null], + [5, 'FontBBox', ['num', 'num', 'num', 'num'], [0, 0, 0, 0]], + [[12, 8], 'StrokeWidth', 'num', 0], + [14, 'XUID', 'array', null], + [15, 'charset', 'offset', 0], + [16, 'Encoding', 'offset', 0], + [17, 'CharStrings', 'offset', 0], + [18, 'Private', ['offset', 'offset'], null], + [[12, 21], 'PostScript', 'sid', null], + [[12, 22], 'BaseFontName', 'sid', null], + [[12, 23], 'BaseFontBlend', 'delta', null], + [[12, 31], 'CIDFontVersion', 'num', 0], + [[12, 32], 'CIDFontRevision', 'num', 0], + [[12, 33], 'CIDFontType', 'num', 0], + [[12, 34], 'CIDCount', 'num', 8720], + [[12, 35], 'UIDBase', 'num', null], + // XXX: CID Fonts on DirectWrite 6.1 only seem to work if FDSelect comes + // before FDArray. + [[12, 37], 'FDSelect', 'offset', null], + [[12, 36], 'FDArray', 'offset', null], + [[12, 38], 'FontName', 'sid', null] + ]; + var tables = null; + function CFFTopDict(strings) { + if (tables === null) { + tables = CFFDict.createTables(layout); + } + CFFDict.call(this, tables, strings); + this.privateDict = null; + } + CFFTopDict.prototype = Object.create(CFFDict.prototype); + return CFFTopDict; })(); -var CFFParser = (function CFFParserClosure() { - var CharstringValidationData = [ - null, - { id: 'hstem', min: 2, stackClearing: true, stem: true }, - null, - { id: 'vstem', min: 2, stackClearing: true, stem: true }, - { id: 'vmoveto', min: 1, stackClearing: true }, - { id: 'rlineto', min: 2, resetStack: true }, - { id: 'hlineto', min: 1, resetStack: true }, - { id: 'vlineto', min: 1, resetStack: true }, - { id: 'rrcurveto', min: 6, resetStack: true }, - null, - { id: 'callsubr', min: 1, undefStack: true }, - { id: 'return', min: 0, undefStack: true }, - null, // 12 - null, - { id: 'endchar', min: 0, stackClearing: true }, - null, - null, - null, - { id: 'hstemhm', min: 2, stackClearing: true, stem: true }, - { id: 'hintmask', min: 0, stackClearing: true }, - { id: 'cntrmask', min: 0, stackClearing: true }, - { id: 'rmoveto', min: 2, stackClearing: true }, - { id: 'hmoveto', min: 1, stackClearing: true }, - { id: 'vstemhm', min: 2, stackClearing: true, stem: true }, - { id: 'rcurveline', min: 8, resetStack: true }, - { id: 'rlinecurve', min: 8, resetStack: true }, - { id: 'vvcurveto', min: 4, resetStack: true }, - { id: 'hhcurveto', min: 4, resetStack: true }, - null, // shortint - { id: 'callgsubr', min: 1, undefStack: true }, - { id: 'vhcurveto', min: 4, resetStack: true }, - { id: 'hvcurveto', min: 4, resetStack: true } +var CFFPrivateDict = (function CFFPrivateDictClosure() { + var layout = [ + [6, 'BlueValues', 'delta', null], + [7, 'OtherBlues', 'delta', null], + [8, 'FamilyBlues', 'delta', null], + [9, 'FamilyOtherBlues', 'delta', null], + [[12, 9], 'BlueScale', 'num', 0.039625], + [[12, 10], 'BlueShift', 'num', 7], + [[12, 11], 'BlueFuzz', 'num', 1], + [10, 'StdHW', 'num', null], + [11, 'StdVW', 'num', null], + [[12, 12], 'StemSnapH', 'delta', null], + [[12, 13], 'StemSnapV', 'delta', null], + [[12, 14], 'ForceBold', 'num', 0], + [[12, 17], 'LanguageGroup', 'num', 0], + [[12, 18], 'ExpansionFactor', 'num', 0.06], + [[12, 19], 'initialRandomSeed', 'num', 0], + [20, 'defaultWidthX', 'num', 0], + [21, 'nominalWidthX', 'num', 0], + [19, 'Subrs', 'offset', null] ]; - var CharstringValidationData12 = [ - null, - null, - null, - { id: 'and', min: 2, stackDelta: -1 }, - { id: 'or', min: 2, stackDelta: -1 }, - { id: 'not', min: 1, stackDelta: 0 }, - null, - null, - null, - { id: 'abs', min: 1, stackDelta: 0 }, - { id: 'add', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] + stack[index - 1]; + var tables = null; + function CFFPrivateDict(strings) { + if (tables === null) { + tables = CFFDict.createTables(layout); + } + CFFDict.call(this, tables, strings); + this.subrsIndex = null; + } + CFFPrivateDict.prototype = Object.create(CFFDict.prototype); + return CFFPrivateDict; +})(); + +var CFFCharsetPredefinedTypes = { + ISO_ADOBE: 0, + EXPERT: 1, + EXPERT_SUBSET: 2 +}; +var CFFCharset = (function CFFCharsetClosure() { + function CFFCharset(predefined, format, charset, raw) { + this.predefined = predefined; + this.format = format; + this.charset = charset; + this.raw = raw; + } + return CFFCharset; +})(); + +var CFFEncoding = (function CFFEncodingClosure() { + function CFFEncoding(predefined, format, encoding, raw) { + this.predefined = predefined; + this.format = format; + this.encoding = encoding; + this.raw = raw; + } + return CFFEncoding; +})(); + +var CFFFDSelect = (function CFFFDSelectClosure() { + function CFFFDSelect(fdSelect, raw) { + this.fdSelect = fdSelect; + this.raw = raw; + } + CFFFDSelect.prototype = { + getFDIndex: function CFFFDSelect_get(glyphIndex) { + if (glyphIndex < 0 || glyphIndex >= this.fdSelect.length) { + return -1; } + return this.fdSelect[glyphIndex]; + } + }; + return CFFFDSelect; +})(); + +// Helper class to keep track of where an offset is within the data and helps +// filling in that offset once it's known. +var CFFOffsetTracker = (function CFFOffsetTrackerClosure() { + function CFFOffsetTracker() { + this.offsets = Object.create(null); + } + CFFOffsetTracker.prototype = { + isTracking: function CFFOffsetTracker_isTracking(key) { + return key in this.offsets; }, - { id: 'sub', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] - stack[index - 1]; + track: function CFFOffsetTracker_track(key, location) { + if (key in this.offsets) { + error('Already tracking location of ' + key); } + this.offsets[key] = location; }, - { id: 'div', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] / stack[index - 1]; + offset: function CFFOffsetTracker_offset(value) { + for (var key in this.offsets) { + this.offsets[key] += value; } }, - null, - { id: 'neg', min: 1, stackDelta: 0, - stackFn: function stack_div(stack, index) { - stack[index - 1] = -stack[index - 1]; + setEntryLocation: function CFFOffsetTracker_setEntryLocation(key, + values, + output) { + if (!(key in this.offsets)) { + error('Not tracking location of ' + key); } - }, - { id: 'eq', min: 2, stackDelta: -1 }, - null, - null, - { id: 'drop', min: 1, stackDelta: -1 }, - null, - { id: 'put', min: 2, stackDelta: -2 }, - { id: 'get', min: 1, stackDelta: 0 }, - { id: 'ifelse', min: 4, stackDelta: -3 }, - { id: 'random', min: 0, stackDelta: 1 }, - { id: 'mul', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] * stack[index - 1]; + var data = output.data; + var dataOffset = this.offsets[key]; + var size = 5; + for (var i = 0, ii = values.length; i < ii; ++i) { + var offset0 = i * size + dataOffset; + var offset1 = offset0 + 1; + var offset2 = offset0 + 2; + var offset3 = offset0 + 3; + var offset4 = offset0 + 4; + // It's easy to screw up offsets so perform this sanity check. + if (data[offset0] !== 0x1d || data[offset1] !== 0 || + data[offset2] !== 0 || data[offset3] !== 0 || data[offset4] !== 0) { + error('writing to an offset that is not empty'); + } + var value = values[i]; + data[offset0] = 0x1d; + data[offset1] = (value >> 24) & 0xFF; + data[offset2] = (value >> 16) & 0xFF; + data[offset3] = (value >> 8) & 0xFF; + data[offset4] = value & 0xFF; } - }, - null, - { id: 'sqrt', min: 1, stackDelta: 0 }, - { id: 'dup', min: 1, stackDelta: 1 }, - { id: 'exch', min: 2, stackDelta: 0 }, - { id: 'index', min: 2, stackDelta: 0 }, - { id: 'roll', min: 3, stackDelta: -2 }, - null, - null, - null, - { id: 'hflex', min: 7, resetStack: true }, - { id: 'flex', min: 13, resetStack: true }, - { id: 'hflex1', min: 9, resetStack: true }, - { id: 'flex1', min: 11, resetStack: true } - ]; + } + }; + return CFFOffsetTracker; +})(); - function CFFParser(file, properties) { - this.bytes = file.getBytes(); - this.properties = properties; +// Takes a CFF and converts it to the binary representation. +var CFFCompiler = (function CFFCompilerClosure() { + function CFFCompiler(cff) { + this.cff = cff; } - CFFParser.prototype = { - parse: function CFFParser_parse() { - var properties = this.properties; - var cff = new CFF(); - this.cff = cff; + CFFCompiler.prototype = { + compile: function CFFCompiler_compile() { + var cff = this.cff; + var output = { + data: [], + length: 0, + add: function CFFCompiler_add(data) { + this.data = this.data.concat(data); + this.length = this.data.length; + } + }; - // The first five sections must be in order, all the others are reached - // via offsets contained in one of the below. - var header = this.parseHeader(); - var nameIndex = this.parseIndex(header.endPos); - var topDictIndex = this.parseIndex(nameIndex.endPos); - var stringIndex = this.parseIndex(topDictIndex.endPos); - var globalSubrIndex = this.parseIndex(stringIndex.endPos); + // Compile the five entries that must be in order. + var header = this.compileHeader(cff.header); + output.add(header); - var topDictParsed = this.parseDict(topDictIndex.obj.get(0)); - var topDict = this.createDict(CFFTopDict, topDictParsed, cff.strings); + var nameIndex = this.compileNameIndex(cff.names); + output.add(nameIndex); - cff.header = header.obj; - cff.names = this.parseNameIndex(nameIndex.obj); - cff.strings = this.parseStringIndex(stringIndex.obj); - cff.topDict = topDict; - cff.globalSubrIndex = globalSubrIndex.obj; + if (cff.isCIDFont) { + // The spec is unclear on how font matrices should relate to each other + // when there is one in the main top dict and the sub top dicts. + // Windows handles this differently than linux and osx so we have to + // normalize to work on all. + // Rules based off of some mailing list discussions: + // - If main font has a matrix and subfont doesn't, use the main matrix. + // - If no main font matrix and there is a subfont matrix, use the + // subfont matrix. + // - If both have matrices, concat together. + // - If neither have matrices, use default. + // To make this work on all platforms we move the top matrix into each + // sub top dict and concat if necessary. + if (cff.topDict.hasName('FontMatrix')) { + var base = cff.topDict.getByName('FontMatrix'); + cff.topDict.removeByName('FontMatrix'); + for (var i = 0, ii = cff.fdArray.length; i < ii; i++) { + var subDict = cff.fdArray[i]; + var matrix = base.slice(0); + if (subDict.hasName('FontMatrix')) { + matrix = Util.transform(matrix, subDict.getByName('FontMatrix')); + } + subDict.setByName('FontMatrix', matrix); + } + } + } - this.parsePrivateDict(cff.topDict); + var compiled = this.compileTopDicts([cff.topDict], + output.length, + cff.isCIDFont); + output.add(compiled.output); + var topDictTracker = compiled.trackers[0]; - cff.isCIDFont = topDict.hasName('ROS'); + var stringIndex = this.compileStringIndex(cff.strings.strings); + output.add(stringIndex); - var charStringOffset = topDict.getByName('CharStrings'); - var charStringIndex = this.parseIndex(charStringOffset).obj; + var globalSubrIndex = this.compileIndex(cff.globalSubrIndex); + output.add(globalSubrIndex); - var fontMatrix = topDict.getByName('FontMatrix'); - if (fontMatrix) { - properties.fontMatrix = fontMatrix; + // Now start on the other entries that have no specfic order. + if (cff.encoding && cff.topDict.hasName('Encoding')) { + if (cff.encoding.predefined) { + topDictTracker.setEntryLocation('Encoding', [cff.encoding.format], + output); + } else { + var encoding = this.compileEncoding(cff.encoding); + topDictTracker.setEntryLocation('Encoding', [output.length], output); + output.add(encoding); + } } - var fontBBox = topDict.getByName('FontBBox'); - if (fontBBox) { - // adjusting ascent/descent - properties.ascent = fontBBox[3]; - properties.descent = fontBBox[1]; - properties.ascentScaled = true; + if (cff.charset && cff.topDict.hasName('charset')) { + if (cff.charset.predefined) { + topDictTracker.setEntryLocation('charset', [cff.charset.format], + output); + } else { + var charset = this.compileCharset(cff.charset); + topDictTracker.setEntryLocation('charset', [output.length], output); + output.add(charset); + } } - var charset, encoding; + var charStrings = this.compileCharStrings(cff.charStrings); + topDictTracker.setEntryLocation('CharStrings', [output.length], output); + output.add(charStrings); + if (cff.isCIDFont) { - var fdArrayIndex = this.parseIndex(topDict.getByName('FDArray')).obj; - for (var i = 0, ii = fdArrayIndex.count; i < ii; ++i) { - var dictRaw = fdArrayIndex.get(i); - var fontDict = this.createDict(CFFTopDict, this.parseDict(dictRaw), - cff.strings); - this.parsePrivateDict(fontDict); - cff.fdArray.push(fontDict); - } - // cid fonts don't have an encoding - encoding = null; - charset = this.parseCharsets(topDict.getByName('charset'), - charStringIndex.count, cff.strings, true); - cff.fdSelect = this.parseFDSelect(topDict.getByName('FDSelect'), - charStringIndex.count); - } else { - charset = this.parseCharsets(topDict.getByName('charset'), - charStringIndex.count, cff.strings, false); - encoding = this.parseEncoding(topDict.getByName('Encoding'), - properties, - cff.strings, charset.charset); + // For some reason FDSelect must be in front of FDArray on windows. OSX + // and linux don't seem to care. + topDictTracker.setEntryLocation('FDSelect', [output.length], output); + var fdSelect = this.compileFDSelect(cff.fdSelect.raw); + output.add(fdSelect); + // It is unclear if the sub font dictionary can have CID related + // dictionary keys, but the sanitizer doesn't like them so remove them. + compiled = this.compileTopDicts(cff.fdArray, output.length, true); + topDictTracker.setEntryLocation('FDArray', [output.length], output); + output.add(compiled.output); + var fontDictTrackers = compiled.trackers; + + this.compilePrivateDicts(cff.fdArray, fontDictTrackers, output); } - cff.charset = charset; - cff.encoding = encoding; + this.compilePrivateDicts([cff.topDict], [topDictTracker], output); - var charStringsAndSeacs = this.parseCharStrings( - charStringIndex, - topDict.privateDict.subrsIndex, - globalSubrIndex.obj, - cff.fdSelect, - cff.fdArray); - cff.charStrings = charStringsAndSeacs.charStrings; - cff.seacs = charStringsAndSeacs.seacs; - cff.widths = charStringsAndSeacs.widths; + // If the font data ends with INDEX whose object data is zero-length, + // the sanitizer will bail out. Add a dummy byte to avoid that. + output.add([0]); - return cff; + return output.data; }, - parseHeader: function CFFParser_parseHeader() { - var bytes = this.bytes; - var bytesLength = bytes.length; - var offset = 0; - - // Prevent an infinite loop, by checking that the offset is within the - // bounds of the bytes array. Necessary in empty, or invalid, font files. - while (offset < bytesLength && bytes[offset] !== 1) { - ++offset; - } - if (offset >= bytesLength) { - error('Invalid CFF header'); - } else if (offset !== 0) { - info('cff data is shifted'); - bytes = bytes.subarray(offset); - this.bytes = bytes; + encodeNumber: function CFFCompiler_encodeNumber(value) { + if (parseFloat(value) === parseInt(value, 10) && !isNaN(value)) { // isInt + return this.encodeInteger(value); + } else { + return this.encodeFloat(value); } - var major = bytes[0]; - var minor = bytes[1]; - var hdrSize = bytes[2]; - var offSize = bytes[3]; - var header = new CFFHeader(major, minor, hdrSize, offSize); - return { obj: header, endPos: hdrSize }; }, - parseDict: function CFFParser_parseDict(dict) { - var pos = 0; + encodeFloat: function CFFCompiler_encodeFloat(num) { + var value = num.toString(); - function parseOperand() { - var value = dict[pos++]; - if (value === 30) { - return parseFloatOperand(pos); - } else if (value === 28) { - value = dict[pos++]; - value = ((value << 24) | (dict[pos++] << 16)) >> 16; - return value; - } else if (value === 29) { - value = dict[pos++]; - value = (value << 8) | dict[pos++]; - value = (value << 8) | dict[pos++]; - value = (value << 8) | dict[pos++]; - return value; - } else if (value >= 32 && value <= 246) { - return value - 139; - } else if (value >= 247 && value <= 250) { - return ((value - 247) * 256) + dict[pos++] + 108; - } else if (value >= 251 && value <= 254) { - return -((value - 251) * 256) - dict[pos++] - 108; + // rounding inaccurate doubles + var m = /\.(\d*?)(?:9{5,20}|0{5,20})\d{0,2}(?:e(.+)|$)/.exec(value); + if (m) { + var epsilon = parseFloat('1e' + ((m[2] ? +m[2] : 0) + m[1].length)); + value = (Math.round(num * epsilon) / epsilon).toString(); + } + + var nibbles = ''; + var i, ii; + for (i = 0, ii = value.length; i < ii; ++i) { + var a = value[i]; + if (a === 'e') { + nibbles += value[++i] === '-' ? 'c' : 'b'; + } else if (a === '.') { + nibbles += 'a'; + } else if (a === '-') { + nibbles += 'e'; } else { - error('255 is not a valid DICT command'); + nibbles += a; } - return -1; } + nibbles += (nibbles.length & 1) ? 'f' : 'ff'; + var out = [30]; + for (i = 0, ii = nibbles.length; i < ii; i += 2) { + out.push(parseInt(nibbles.substr(i, 2), 16)); + } + return out; + }, + encodeInteger: function CFFCompiler_encodeInteger(value) { + var code; + if (value >= -107 && value <= 107) { + code = [value + 139]; + } else if (value >= 108 && value <= 1131) { + value = [value - 108]; + code = [(value >> 8) + 247, value & 0xFF]; + } else if (value >= -1131 && value <= -108) { + value = -value - 108; + code = [(value >> 8) + 251, value & 0xFF]; + } else if (value >= -32768 && value <= 32767) { + code = [0x1c, (value >> 8) & 0xFF, value & 0xFF]; + } else { + code = [0x1d, + (value >> 24) & 0xFF, + (value >> 16) & 0xFF, + (value >> 8) & 0xFF, + value & 0xFF]; + } + return code; + }, + compileHeader: function CFFCompiler_compileHeader(header) { + return [ + header.major, + header.minor, + header.hdrSize, + header.offSize + ]; + }, + compileNameIndex: function CFFCompiler_compileNameIndex(names) { + var nameIndex = new CFFIndex(); + for (var i = 0, ii = names.length; i < ii; ++i) { + nameIndex.add(stringToBytes(names[i])); + } + return this.compileIndex(nameIndex); + }, + compileTopDicts: function CFFCompiler_compileTopDicts(dicts, + length, + removeCidKeys) { + var fontDictTrackers = []; + var fdArrayIndex = new CFFIndex(); + for (var i = 0, ii = dicts.length; i < ii; ++i) { + var fontDict = dicts[i]; + if (removeCidKeys) { + fontDict.removeByName('CIDFontVersion'); + fontDict.removeByName('CIDFontRevision'); + fontDict.removeByName('CIDFontType'); + fontDict.removeByName('CIDCount'); + fontDict.removeByName('UIDBase'); + } + var fontDictTracker = new CFFOffsetTracker(); + var fontDictData = this.compileDict(fontDict, fontDictTracker); + fontDictTrackers.push(fontDictTracker); + fdArrayIndex.add(fontDictData); + fontDictTracker.offset(length); + } + fdArrayIndex = this.compileIndex(fdArrayIndex, fontDictTrackers); + return { + trackers: fontDictTrackers, + output: fdArrayIndex + }; + }, + compilePrivateDicts: function CFFCompiler_compilePrivateDicts(dicts, + trackers, + output) { + for (var i = 0, ii = dicts.length; i < ii; ++i) { + var fontDict = dicts[i]; + assert(fontDict.privateDict && fontDict.hasName('Private'), + 'There must be an private dictionary.'); + var privateDict = fontDict.privateDict; + var privateDictTracker = new CFFOffsetTracker(); + var privateDictData = this.compileDict(privateDict, privateDictTracker); - function parseFloatOperand() { - var str = ''; - var eof = 15; - var lookup = ['0', '1', '2', '3', '4', '5', '6', '7', '8', - '9', '.', 'E', 'E-', null, '-']; - var length = dict.length; - while (pos < length) { - var b = dict[pos++]; - var b1 = b >> 4; - var b2 = b & 15; + var outputLength = output.length; + privateDictTracker.offset(outputLength); + if (!privateDictData.length) { + // The private dictionary was empty, set the output length to zero to + // ensure the offset length isn't out of bounds in the eyes of the + // sanitizer. + outputLength = 0; + } - if (b1 === eof) { - break; - } - str += lookup[b1]; + trackers[i].setEntryLocation('Private', + [privateDictData.length, outputLength], + output); + output.add(privateDictData); - if (b2 === eof) { - break; - } - str += lookup[b2]; + if (privateDict.subrsIndex && privateDict.hasName('Subrs')) { + var subrs = this.compileIndex(privateDict.subrsIndex); + privateDictTracker.setEntryLocation('Subrs', [privateDictData.length], + output); + output.add(subrs); } - return parseFloat(str); } - - var operands = []; - var entries = []; - - pos = 0; - var end = dict.length; - while (pos < end) { - var b = dict[pos]; - if (b <= 21) { - if (b === 12) { - b = (b << 8) | dict[++pos]; - } - entries.push([b, operands]); - operands = []; - ++pos; - } else { - operands.push(parseOperand()); + }, + compileDict: function CFFCompiler_compileDict(dict, offsetTracker) { + var out = []; + // The dictionary keys must be in a certain order. + var order = dict.order; + for (var i = 0; i < order.length; ++i) { + var key = order[i]; + if (!(key in dict.values)) { + continue; + } + var values = dict.values[key]; + var types = dict.types[key]; + if (!isArray(types)) { + types = [types]; + } + if (!isArray(values)) { + values = [values]; } - } - return entries; - }, - parseIndex: function CFFParser_parseIndex(pos) { - var cffIndex = new CFFIndex(); - var bytes = this.bytes; - var count = (bytes[pos++] << 8) | bytes[pos++]; - var offsets = []; - var end = pos; - var i, ii; - if (count !== 0) { - var offsetSize = bytes[pos++]; - // add 1 for offset to determine size of last object - var startPos = pos + ((count + 1) * offsetSize) - 1; + // Remove any empty dict values. + if (values.length === 0) { + continue; + } - for (i = 0, ii = count + 1; i < ii; ++i) { - var offset = 0; - for (var j = 0; j < offsetSize; ++j) { - offset <<= 8; - offset += bytes[pos++]; + for (var j = 0, jj = types.length; j < jj; ++j) { + var type = types[j]; + var value = values[j]; + switch (type) { + case 'num': + case 'sid': + out = out.concat(this.encodeNumber(value)); + break; + case 'offset': + // For offsets we just insert a 32bit integer so we don't have to + // deal with figuring out the length of the offset when it gets + // replaced later on by the compiler. + var name = dict.keyToNameMap[key]; + // Some offsets have the offset and the length, so just record the + // position of the first one. + if (!offsetTracker.isTracking(name)) { + offsetTracker.track(name, out.length); + } + out = out.concat([0x1d, 0, 0, 0, 0]); + break; + case 'array': + case 'delta': + out = out.concat(this.encodeNumber(value)); + for (var k = 1, kk = values.length; k < kk; ++k) { + out = out.concat(this.encodeNumber(values[k])); + } + break; + default: + error('Unknown data type of ' + type); + break; } - offsets.push(startPos + offset); } - end = offsets[count]; - } - for (i = 0, ii = offsets.length - 1; i < ii; ++i) { - var offsetStart = offsets[i]; - var offsetEnd = offsets[i + 1]; - cffIndex.add(bytes.subarray(offsetStart, offsetEnd)); + out = out.concat(dict.opcodes[key]); } - return {obj: cffIndex, endPos: end}; + return out; }, - parseNameIndex: function CFFParser_parseNameIndex(index) { - var names = []; - for (var i = 0, ii = index.count; i < ii; ++i) { - var name = index.get(i); - // OTS doesn't allow names to be over 127 characters. - var length = Math.min(name.length, 127); - var data = []; - // OTS also only permits certain characters in the name. - for (var j = 0; j < length; ++j) { - var c = name[j]; - if (j === 0 && c === 0) { - data[j] = c; - continue; - } - if ((c < 33 || c > 126) || c === 91 /* [ */ || c === 93 /* ] */ || - c === 40 /* ( */ || c === 41 /* ) */ || c === 123 /* { */ || - c === 125 /* } */ || c === 60 /* < */ || c === 62 /* > */ || - c === 47 /* / */ || c === 37 /* % */ || c === 35 /* # */) { - data[j] = 95; - continue; - } - data[j] = c; - } - names.push(bytesToString(data)); + compileStringIndex: function CFFCompiler_compileStringIndex(strings) { + var stringIndex = new CFFIndex(); + for (var i = 0, ii = strings.length; i < ii; ++i) { + stringIndex.add(stringToBytes(strings[i])); } - return names; + return this.compileIndex(stringIndex); }, - parseStringIndex: function CFFParser_parseStringIndex(index) { - var strings = new CFFStrings(); - for (var i = 0, ii = index.count; i < ii; ++i) { - var data = index.get(i); - strings.add(bytesToString(data)); - } - return strings; + compileGlobalSubrIndex: function CFFCompiler_compileGlobalSubrIndex() { + var globalSubrIndex = this.cff.globalSubrIndex; + this.out.writeByteArray(this.compileIndex(globalSubrIndex)); }, - createDict: function CFFParser_createDict(Type, dict, strings) { - var cffDict = new Type(strings); - for (var i = 0, ii = dict.length; i < ii; ++i) { - var pair = dict[i]; - var key = pair[0]; - var value = pair[1]; - cffDict.setByKey(key, value); + compileCharStrings: function CFFCompiler_compileCharStrings(charStrings) { + return this.compileIndex(charStrings); + }, + compileCharset: function CFFCompiler_compileCharset(charset) { + return this.compileTypedArray(charset.raw); + }, + compileEncoding: function CFFCompiler_compileEncoding(encoding) { + return this.compileTypedArray(encoding.raw); + }, + compileFDSelect: function CFFCompiler_compileFDSelect(fdSelect) { + return this.compileTypedArray(fdSelect); + }, + compileTypedArray: function CFFCompiler_compileTypedArray(data) { + var out = []; + for (var i = 0, ii = data.length; i < ii; ++i) { + out[i] = data[i]; } - return cffDict; + return out; }, - parseCharString: function CFFParser_parseCharString(state, data, - localSubrIndex, - globalSubrIndex) { - if (state.callDepth > MAX_SUBR_NESTING) { - return false; + compileIndex: function CFFCompiler_compileIndex(index, trackers) { + trackers = trackers || []; + var objects = index.objects; + // First 2 bytes contains the number of objects contained into this index + var count = objects.length; + + // If there is no object, just create an index. This technically + // should just be [0, 0] but OTS has an issue with that. + if (count === 0) { + return [0, 0, 0]; } - var stackSize = state.stackSize; - var stack = state.stack; - var length = data.length; + var data = [(count >> 8) & 0xFF, count & 0xff]; - for (var j = 0; j < length;) { - var value = data[j++]; - var validationCommand = null; - if (value === 12) { - var q = data[j++]; - if (q === 0) { - // The CFF specification state that the 'dotsection' command - // (12, 0) is deprecated and treated as a no-op, but all Type2 - // charstrings processors should support them. Unfortunately - // the font sanitizer don't. As a workaround the sequence (12, 0) - // is replaced by a useless (0, hmoveto). - data[j - 2] = 139; - data[j - 1] = 22; - stackSize = 0; - } else { - validationCommand = CharstringValidationData12[q]; - } - } else if (value === 28) { // number (16 bit) - stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16)) >> 16; - j += 2; - stackSize++; - } else if (value === 14) { - if (stackSize >= 4) { - stackSize -= 4; - if (SEAC_ANALYSIS_ENABLED) { - state.seac = stack.slice(stackSize, stackSize + 4); - return false; - } - } - validationCommand = CharstringValidationData[value]; - } else if (value >= 32 && value <= 246) { // number - stack[stackSize] = value - 139; - stackSize++; - } else if (value >= 247 && value <= 254) { // number (+1 bytes) - stack[stackSize] = (value < 251 ? - ((value - 247) << 8) + data[j] + 108 : - -((value - 251) << 8) - data[j] - 108); - j++; - stackSize++; - } else if (value === 255) { // number (32 bit) - stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16) | - (data[j + 2] << 8) | data[j + 3]) / 65536; - j += 4; - stackSize++; - } else if (value === 19 || value === 20) { - state.hints += stackSize >> 1; - // skipping right amount of hints flag data - j += (state.hints + 7) >> 3; - stackSize %= 2; - validationCommand = CharstringValidationData[value]; - } else if (value === 10 || value === 29) { - var subrsIndex; - if (value === 10) { - subrsIndex = localSubrIndex; - } else { - subrsIndex = globalSubrIndex; - } - if (!subrsIndex) { - validationCommand = CharstringValidationData[value]; - warn('Missing subrsIndex for ' + validationCommand.id); - return false; - } - var bias = 32768; - if (subrsIndex.count < 1240) { - bias = 107; - } else if (subrsIndex.count < 33900) { - bias = 1131; - } - var subrNumber = stack[--stackSize] + bias; - if (subrNumber < 0 || subrNumber >= subrsIndex.count) { - validationCommand = CharstringValidationData[value]; - warn('Out of bounds subrIndex for ' + validationCommand.id); - return false; - } - state.stackSize = stackSize; - state.callDepth++; - var valid = this.parseCharString(state, subrsIndex.get(subrNumber), - localSubrIndex, globalSubrIndex); - if (!valid) { - return false; - } - state.callDepth--; - stackSize = state.stackSize; - continue; - } else if (value === 11) { - state.stackSize = stackSize; - return true; + var lastOffset = 1, i; + for (i = 0; i < count; ++i) { + lastOffset += objects[i].length; + } + + var offsetSize; + if (lastOffset < 0x100) { + offsetSize = 1; + } else if (lastOffset < 0x10000) { + offsetSize = 2; + } else if (lastOffset < 0x1000000) { + offsetSize = 3; + } else { + offsetSize = 4; + } + + // Next byte contains the offset size use to reference object in the file + data.push(offsetSize); + + // Add another offset after this one because we need a new offset + var relativeOffset = 1; + for (i = 0; i < count + 1; i++) { + if (offsetSize === 1) { + data.push(relativeOffset & 0xFF); + } else if (offsetSize === 2) { + data.push((relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); + } else if (offsetSize === 3) { + data.push((relativeOffset >> 16) & 0xFF, + (relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); } else { - validationCommand = CharstringValidationData[value]; + data.push((relativeOffset >>> 24) & 0xFF, + (relativeOffset >> 16) & 0xFF, + (relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); } - if (validationCommand) { - if (validationCommand.stem) { - state.hints += stackSize >> 1; - } - if ('min' in validationCommand) { - if (!state.undefStack && stackSize < validationCommand.min) { - warn('Not enough parameters for ' + validationCommand.id + - '; actual: ' + stackSize + - ', expected: ' + validationCommand.min); - return false; - } - } - if (state.firstStackClearing && validationCommand.stackClearing) { - state.firstStackClearing = false; - // the optional character width can be found before the first - // stack-clearing command arguments - stackSize -= validationCommand.min; - if (stackSize >= 2 && validationCommand.stem) { - // there are even amount of arguments for stem commands - stackSize %= 2; - } else if (stackSize > 1) { - warn('Found too many parameters for stack-clearing command'); - } - if (stackSize > 0 && stack[stackSize - 1] >= 0) { - state.width = stack[stackSize - 1]; - } - } - if ('stackDelta' in validationCommand) { - if ('stackFn' in validationCommand) { - validationCommand.stackFn(stack, stackSize); - } - stackSize += validationCommand.stackDelta; - } else if (validationCommand.stackClearing) { - stackSize = 0; - } else if (validationCommand.resetStack) { - stackSize = 0; - state.undefStack = false; - } else if (validationCommand.undefStack) { - stackSize = 0; - state.undefStack = true; - state.firstStackClearing = false; - } + + if (objects[i]) { + relativeOffset += objects[i].length; } } - state.stackSize = stackSize; - return true; - }, - parseCharStrings: function CFFParser_parseCharStrings(charStrings, - localSubrIndex, - globalSubrIndex, - fdSelect, - fdArray) { - var seacs = []; - var widths = []; - var count = charStrings.count; - for (var i = 0; i < count; i++) { - var charstring = charStrings.get(i); - var state = { - callDepth: 0, - stackSize: 0, - stack: [], - undefStack: true, - hints: 0, - firstStackClearing: true, - seac: null, - width: null - }; - var valid = true; - var localSubrToUse = null; - if (fdSelect && fdArray.length) { - var fdIndex = fdSelect.getFDIndex(i); - if (fdIndex === -1) { - warn('Glyph index is not in fd select.'); - valid = false; - } - if (fdIndex >= fdArray.length) { - warn('Invalid fd index for glyph index.'); - valid = false; - } - if (valid) { - localSubrToUse = fdArray[fdIndex].privateDict.subrsIndex; - } - } else if (localSubrIndex) { - localSubrToUse = localSubrIndex; - } - if (valid) { - valid = this.parseCharString(state, charstring, localSubrToUse, - globalSubrIndex); - } - if (state.width !== null) { - widths[i] = state.width; + + for (i = 0; i < count; i++) { + // Notify the tracker where the object will be offset in the data. + if (trackers[i]) { + trackers[i].offset(data.length); } - if (state.seac !== null) { - seacs[i] = state.seac; + for (var j = 0, jj = objects[i].length; j < jj; j++) { + data.push(objects[i][j]); } - if (!valid) { - // resetting invalid charstring to single 'endchar' - charStrings.set(i, new Uint8Array([14])); + } + return data; + } + }; + return CFFCompiler; +})(); + +function _enableSeacAnalysis(enabled) { + exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED = enabled; +} + +// Workaround for seac on Windows. +(function checkSeacSupport() { + if (typeof navigator !== 'undefined' && /Windows/.test(navigator.userAgent)) { + SEAC_ANALYSIS_ENABLED = true; + } +})(); + +// Workaround for Private Use Area characters in Chrome on Windows +// http://code.google.com/p/chromium/issues/detail?id=122465 +// https://github.com/mozilla/pdf.js/issues/1689 +(function checkChromeWindows() { + if (typeof navigator !== 'undefined' && + /Windows.*Chrome/.test(navigator.userAgent)) { + SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = true; + } +})(); + +exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED; +exports.CFFCompiler = CFFCompiler; +exports.CFFIndex = CFFIndex; +exports.CFFParser = CFFParser; +exports.CFFStrings = CFFStrings; +exports.ErrorFont = ErrorFont; +exports.FontFlags = FontFlags; +exports.Font = Font; +exports.IdentityToUnicodeMap = IdentityToUnicodeMap; +exports.ToUnicodeMap = ToUnicodeMap; +exports.Type1Parser = Type1Parser; +exports.getFontType = getFontType; +exports._enableSeacAnalysis = _enableSeacAnalysis; + +// TODO refactor to remove cyclic dependency on font_renderer.js +coreFontRenderer._setCoreFonts(exports); +})); + + +(function (root, factory) { + { + factory((root.pdfjsCoreFunction = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCorePsParser); + } +}(this, function (exports, sharedUtil, corePrimitives, corePsParser) { + +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isBool = sharedUtil.isBool; +var isDict = corePrimitives.isDict; +var isStream = corePrimitives.isStream; +var PostScriptLexer = corePsParser.PostScriptLexer; +var PostScriptParser = corePsParser.PostScriptParser; + +var PDFFunction = (function PDFFunctionClosure() { + var CONSTRUCT_SAMPLED = 0; + var CONSTRUCT_INTERPOLATED = 2; + var CONSTRUCT_STICHED = 3; + var CONSTRUCT_POSTSCRIPT = 4; + + return { + getSampleArray: function PDFFunction_getSampleArray(size, outputSize, bps, + str) { + var i, ii; + var length = 1; + for (i = 0, ii = size.length; i < ii; i++) { + length *= size[i]; + } + length *= outputSize; + + var array = new Array(length); + var codeSize = 0; + var codeBuf = 0; + // 32 is a valid bps so shifting won't work + var sampleMul = 1.0 / (Math.pow(2.0, bps) - 1); + + var strBytes = str.getBytes((length * bps + 7) / 8); + var strIdx = 0; + for (i = 0; i < length; i++) { + while (codeSize < bps) { + codeBuf <<= 8; + codeBuf |= strBytes[strIdx++]; + codeSize += 8; } + codeSize -= bps; + array[i] = (codeBuf >> codeSize) * sampleMul; + codeBuf &= (1 << codeSize) - 1; } - return { charStrings: charStrings, seacs: seacs, widths: widths }; + return array; }, - emptyPrivateDictionary: - function CFFParser_emptyPrivateDictionary(parentDict) { - var privateDict = this.createDict(CFFPrivateDict, [], - parentDict.strings); - parentDict.setByKey(18, [0, 0]); - parentDict.privateDict = privateDict; + + getIR: function PDFFunction_getIR(xref, fn) { + var dict = fn.dict; + if (!dict) { + dict = fn; + } + + var types = [this.constructSampled, + null, + this.constructInterpolated, + this.constructStiched, + this.constructPostScript]; + + var typeNum = dict.get('FunctionType'); + var typeFn = types[typeNum]; + if (!typeFn) { + error('Unknown type of function'); + } + + return typeFn.call(this, fn, dict, xref); + }, + + fromIR: function PDFFunction_fromIR(IR) { + var type = IR[0]; + switch (type) { + case CONSTRUCT_SAMPLED: + return this.constructSampledFromIR(IR); + case CONSTRUCT_INTERPOLATED: + return this.constructInterpolatedFromIR(IR); + case CONSTRUCT_STICHED: + return this.constructStichedFromIR(IR); + //case CONSTRUCT_POSTSCRIPT: + default: + return this.constructPostScriptFromIR(IR); + } }, - parsePrivateDict: function CFFParser_parsePrivateDict(parentDict) { - // no private dict, do nothing - if (!parentDict.hasName('Private')) { - this.emptyPrivateDictionary(parentDict); - return; + + parse: function PDFFunction_parse(xref, fn) { + var IR = this.getIR(xref, fn); + return this.fromIR(IR); + }, + + parseArray: function PDFFunction_parseArray(xref, fnObj) { + if (!isArray(fnObj)) { + // not an array -- parsing as regular function + return this.parse(xref, fnObj); } - var privateOffset = parentDict.getByName('Private'); - // make sure the params are formatted correctly - if (!isArray(privateOffset) || privateOffset.length !== 2) { - parentDict.removeByName('Private'); - return; + + var fnArray = []; + for (var j = 0, jj = fnObj.length; j < jj; j++) { + var obj = xref.fetchIfRef(fnObj[j]); + fnArray.push(PDFFunction.parse(xref, obj)); } - var size = privateOffset[0]; - var offset = privateOffset[1]; - // remove empty dicts or ones that refer to invalid location - if (size === 0 || offset >= this.bytes.length) { - this.emptyPrivateDictionary(parentDict); - return; + return function (src, srcOffset, dest, destOffset) { + for (var i = 0, ii = fnArray.length; i < ii; i++) { + fnArray[i](src, srcOffset, dest, destOffset + i); + } + }; + }, + + constructSampled: function PDFFunction_constructSampled(str, dict) { + function toMultiArray(arr) { + var inputLength = arr.length; + var out = []; + var index = 0; + for (var i = 0; i < inputLength; i += 2) { + out[index] = [arr[i], arr[i + 1]]; + ++index; + } + return out; } + var domain = dict.get('Domain'); + var range = dict.get('Range'); - var privateDictEnd = offset + size; - var dictData = this.bytes.subarray(offset, privateDictEnd); - var dict = this.parseDict(dictData); - var privateDict = this.createDict(CFFPrivateDict, dict, - parentDict.strings); - parentDict.privateDict = privateDict; + if (!domain || !range) { + error('No domain or range'); + } - // Parse the Subrs index also since it's relative to the private dict. - if (!privateDict.getByName('Subrs')) { - return; + var inputSize = domain.length / 2; + var outputSize = range.length / 2; + + domain = toMultiArray(domain); + range = toMultiArray(range); + + var size = dict.get('Size'); + var bps = dict.get('BitsPerSample'); + var order = dict.get('Order') || 1; + if (order !== 1) { + // No description how cubic spline interpolation works in PDF32000:2008 + // As in poppler, ignoring order, linear interpolation may work as good + info('No support for cubic spline interpolation: ' + order); } - var subrsOffset = privateDict.getByName('Subrs'); - var relativeOffset = offset + subrsOffset; - // Validate the offset. - if (subrsOffset === 0 || relativeOffset >= this.bytes.length) { - this.emptyPrivateDictionary(parentDict); - return; + + var encode = dict.get('Encode'); + if (!encode) { + encode = []; + for (var i = 0; i < inputSize; ++i) { + encode.push(0); + encode.push(size[i] - 1); + } } - var subrsIndex = this.parseIndex(relativeOffset); - privateDict.subrsIndex = subrsIndex.obj; + encode = toMultiArray(encode); + + var decode = dict.get('Decode'); + if (!decode) { + decode = range; + } else { + decode = toMultiArray(decode); + } + + var samples = this.getSampleArray(size, outputSize, bps, str); + + return [ + CONSTRUCT_SAMPLED, inputSize, domain, encode, decode, samples, size, + outputSize, Math.pow(2, bps) - 1, range + ]; }, - parseCharsets: function CFFParser_parseCharsets(pos, length, strings, cid) { - if (pos === 0) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.ISO_ADOBE, - ISOAdobeCharset); - } else if (pos === 1) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT, - ExpertCharset); - } else if (pos === 2) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT_SUBSET, - ExpertSubsetCharset); + + constructSampledFromIR: function PDFFunction_constructSampledFromIR(IR) { + // See chapter 3, page 109 of the PDF reference + function interpolate(x, xmin, xmax, ymin, ymax) { + return ymin + ((x - xmin) * ((ymax - ymin) / (xmax - xmin))); } - var bytes = this.bytes; - var start = pos; - var format = bytes[pos++]; - var charset = ['.notdef']; - var id, count, i; + return function constructSampledFromIRResult(src, srcOffset, + dest, destOffset) { + // See chapter 3, page 110 of the PDF reference. + var m = IR[1]; + var domain = IR[2]; + var encode = IR[3]; + var decode = IR[4]; + var samples = IR[5]; + var size = IR[6]; + var n = IR[7]; + //var mask = IR[8]; + var range = IR[9]; - // subtract 1 for the .notdef glyph - length -= 1; + // Building the cube vertices: its part and sample index + // http://rjwagner49.com/Mathematics/Interpolation.pdf + var cubeVertices = 1 << m; + var cubeN = new Float64Array(cubeVertices); + var cubeVertex = new Uint32Array(cubeVertices); + var i, j; + for (j = 0; j < cubeVertices; j++) { + cubeN[j] = 1; + } - switch (format) { - case 0: - for (i = 0; i < length; i++) { - id = (bytes[pos++] << 8) | bytes[pos++]; - charset.push(cid ? id : strings.get(id)); - } - break; - case 1: - while (charset.length <= length) { - id = (bytes[pos++] << 8) | bytes[pos++]; - count = bytes[pos++]; - for (i = 0; i <= count; i++) { - charset.push(cid ? id++ : strings.get(id++)); + var k = n, pos = 1; + // Map x_i to y_j for 0 <= i < m using the sampled function. + for (i = 0; i < m; ++i) { + // x_i' = min(max(x_i, Domain_2i), Domain_2i+1) + var domain_2i = domain[i][0]; + var domain_2i_1 = domain[i][1]; + var xi = Math.min(Math.max(src[srcOffset +i], domain_2i), + domain_2i_1); + + // e_i = Interpolate(x_i', Domain_2i, Domain_2i+1, + // Encode_2i, Encode_2i+1) + var e = interpolate(xi, domain_2i, domain_2i_1, + encode[i][0], encode[i][1]); + + // e_i' = min(max(e_i, 0), Size_i - 1) + var size_i = size[i]; + e = Math.min(Math.max(e, 0), size_i - 1); + + // Adjusting the cube: N and vertex sample index + var e0 = e < size_i - 1 ? Math.floor(e) : e - 1; // e1 = e0 + 1; + var n0 = e0 + 1 - e; // (e1 - e) / (e1 - e0); + var n1 = e - e0; // (e - e0) / (e1 - e0); + var offset0 = e0 * k; + var offset1 = offset0 + k; // e1 * k + for (j = 0; j < cubeVertices; j++) { + if (j & pos) { + cubeN[j] *= n1; + cubeVertex[j] += offset1; + } else { + cubeN[j] *= n0; + cubeVertex[j] += offset0; } } - break; - case 2: - while (charset.length <= length) { - id = (bytes[pos++] << 8) | bytes[pos++]; - count = (bytes[pos++] << 8) | bytes[pos++]; - for (i = 0; i <= count; i++) { - charset.push(cid ? id++ : strings.get(id++)); - } + + k *= size_i; + pos <<= 1; + } + + for (j = 0; j < n; ++j) { + // Sum all cube vertices' samples portions + var rj = 0; + for (i = 0; i < cubeVertices; i++) { + rj += samples[cubeVertex[i] + j] * cubeN[i]; } - break; - default: - error('Unknown charset format'); + + // r_j' = Interpolate(r_j, 0, 2^BitsPerSample - 1, + // Decode_2j, Decode_2j+1) + rj = interpolate(rj, 0, 1, decode[j][0], decode[j][1]); + + // y_j = min(max(r_j, range_2j), range_2j+1) + dest[destOffset + j] = Math.min(Math.max(rj, range[j][0]), + range[j][1]); + } + }; + }, + + constructInterpolated: function PDFFunction_constructInterpolated(str, + dict) { + var c0 = dict.get('C0') || [0]; + var c1 = dict.get('C1') || [1]; + var n = dict.get('N'); + + if (!isArray(c0) || !isArray(c1)) { + error('Illegal dictionary for interpolated function'); } - // Raw won't be needed if we actually compile the charset. - var end = pos; - var raw = bytes.subarray(start, end); - return new CFFCharset(false, format, charset, raw); + var length = c0.length; + var diff = []; + for (var i = 0; i < length; ++i) { + diff.push(c1[i] - c0[i]); + } + + return [CONSTRUCT_INTERPOLATED, c0, diff, n]; }, - parseEncoding: function CFFParser_parseEncoding(pos, - properties, - strings, - charset) { - var encoding = Object.create(null); - var bytes = this.bytes; - var predefined = false; - var hasSupplement = false; - var format, i, ii; - var raw = null; - function readSupplement() { - var supplementsCount = bytes[pos++]; - for (i = 0; i < supplementsCount; i++) { - var code = bytes[pos++]; - var sid = (bytes[pos++] << 8) + (bytes[pos++] & 0xff); - encoding[code] = charset.indexOf(strings.get(sid)); + constructInterpolatedFromIR: + function PDFFunction_constructInterpolatedFromIR(IR) { + var c0 = IR[1]; + var diff = IR[2]; + var n = IR[3]; + + var length = diff.length; + + return function constructInterpolatedFromIRResult(src, srcOffset, + dest, destOffset) { + var x = n === 1 ? src[srcOffset] : Math.pow(src[srcOffset], n); + + for (var j = 0; j < length; ++j) { + dest[destOffset + j] = c0[j] + (x * diff[j]); } + }; + }, + + constructStiched: function PDFFunction_constructStiched(fn, dict, xref) { + var domain = dict.get('Domain'); + + if (!domain) { + error('No domain'); } - if (pos === 0 || pos === 1) { - predefined = true; - format = pos; - var baseEncoding = pos ? ExpertEncoding : StandardEncoding; - for (i = 0, ii = charset.length; i < ii; i++) { - var index = baseEncoding.indexOf(charset[i]); - if (index !== -1) { - encoding[index] = i; + var inputSize = domain.length / 2; + if (inputSize !== 1) { + error('Bad domain for stiched function'); + } + + var fnRefs = dict.get('Functions'); + var fns = []; + for (var i = 0, ii = fnRefs.length; i < ii; ++i) { + fns.push(PDFFunction.getIR(xref, xref.fetchIfRef(fnRefs[i]))); + } + + var bounds = dict.get('Bounds'); + var encode = dict.get('Encode'); + + return [CONSTRUCT_STICHED, domain, bounds, encode, fns]; + }, + + constructStichedFromIR: function PDFFunction_constructStichedFromIR(IR) { + var domain = IR[1]; + var bounds = IR[2]; + var encode = IR[3]; + var fnsIR = IR[4]; + var fns = []; + var tmpBuf = new Float32Array(1); + + for (var i = 0, ii = fnsIR.length; i < ii; i++) { + fns.push(PDFFunction.fromIR(fnsIR[i])); + } + + return function constructStichedFromIRResult(src, srcOffset, + dest, destOffset) { + var clip = function constructStichedFromIRClip(v, min, max) { + if (v > max) { + v = max; + } else if (v < min) { + v = min; } - } - } else { - var dataStart = pos; - format = bytes[pos++]; - switch (format & 0x7f) { - case 0: - var glyphsCount = bytes[pos++]; - for (i = 1; i <= glyphsCount; i++) { - encoding[bytes[pos++]] = i; - } - break; + return v; + }; - case 1: - var rangesCount = bytes[pos++]; - var gid = 1; - for (i = 0; i < rangesCount; i++) { - var start = bytes[pos++]; - var left = bytes[pos++]; - for (var j = start; j <= start + left; j++) { - encoding[j] = gid++; - } - } + // clip to domain + var v = clip(src[srcOffset], domain[0], domain[1]); + // calulate which bound the value is in + for (var i = 0, ii = bounds.length; i < ii; ++i) { + if (v < bounds[i]) { break; + } + } - default: - error('Unknow encoding format: ' + format + ' in CFF'); - break; + // encode value into domain of function + var dmin = domain[0]; + if (i > 0) { + dmin = bounds[i - 1]; } - var dataEnd = pos; - if (format & 0x80) { - // The font sanitizer does not support CFF encoding with a - // supplement, since the encoding is not really used to map - // between gid to glyph, let's overwrite what is declared in - // the top dictionary to let the sanitizer think the font use - // StandardEncoding, that's a lie but that's ok. - bytes[dataStart] &= 0x7f; - readSupplement(); - hasSupplement = true; + var dmax = domain[1]; + if (i < bounds.length) { + dmax = bounds[i]; } - raw = bytes.subarray(dataStart, dataEnd); + + var rmin = encode[2 * i]; + var rmax = encode[2 * i + 1]; + + // Prevent the value from becoming NaN as a result + // of division by zero (fixes issue6113.pdf). + tmpBuf[0] = dmin === dmax ? rmin : + rmin + (v - dmin) * (rmax - rmin) / (dmax - dmin); + + // call the appropriate function + fns[i](tmpBuf, 0, dest, destOffset); + }; + }, + + constructPostScript: function PDFFunction_constructPostScript(fn, dict, + xref) { + var domain = dict.get('Domain'); + var range = dict.get('Range'); + + if (!domain) { + error('No domain.'); } - format = format & 0x7f; - return new CFFEncoding(predefined, format, encoding, raw); + + if (!range) { + error('No range.'); + } + + var lexer = new PostScriptLexer(fn); + var parser = new PostScriptParser(lexer); + var code = parser.parse(); + + return [CONSTRUCT_POSTSCRIPT, domain, range, code]; }, - parseFDSelect: function CFFParser_parseFDSelect(pos, length) { - var start = pos; - var bytes = this.bytes; - var format = bytes[pos++]; - var fdSelect = []; - var i; - switch (format) { - case 0: - for (i = 0; i < length; ++i) { - var id = bytes[pos++]; - fdSelect.push(id); - } - break; - case 3: - var rangesCount = (bytes[pos++] << 8) | bytes[pos++]; - for (i = 0; i < rangesCount; ++i) { - var first = (bytes[pos++] << 8) | bytes[pos++]; - var fdIndex = bytes[pos++]; - var next = (bytes[pos] << 8) | bytes[pos + 1]; - for (var j = first; j < next; ++j) { - fdSelect.push(fdIndex); - } - } - // Advance past the sentinel(next). - pos += 2; - break; - default: - error('Unknown fdselect format ' + format); - break; + constructPostScriptFromIR: function PDFFunction_constructPostScriptFromIR( + IR) { + var domain = IR[1]; + var range = IR[2]; + var code = IR[3]; + + var compiled = (new PostScriptCompiler()).compile(code, domain, range); + if (compiled) { + // Compiled function consists of simple expressions such as addition, + // subtraction, Math.max, and also contains 'var' and 'return' + // statements. See the generation in the PostScriptCompiler below. + /*jshint -W054 */ + return new Function('src', 'srcOffset', 'dest', 'destOffset', compiled); } - var end = pos; - return new CFFFDSelect(fdSelect, bytes.subarray(start, end)); - } - }; - return CFFParser; -})(); -// Compact Font Format -var CFF = (function CFFClosure() { - function CFF() { - this.header = null; - this.names = []; - this.topDict = null; - this.strings = new CFFStrings(); - this.globalSubrIndex = null; + info('Unable to compile PS function'); - // The following could really be per font, but since we only have one font - // store them here. - this.encoding = null; - this.charset = null; - this.charStrings = null; - this.fdArray = []; - this.fdSelect = null; + var numOutputs = range.length >> 1; + var numInputs = domain.length >> 1; + var evaluator = new PostScriptEvaluator(code); + // Cache the values for a big speed up, the cache size is limited though + // since the number of possible values can be huge from a PS function. + var cache = Object.create(null); + // The MAX_CACHE_SIZE is set to ~4x the maximum number of distinct values + // seen in our tests. + var MAX_CACHE_SIZE = 2048 * 4; + var cache_available = MAX_CACHE_SIZE; + var tmpBuf = new Float32Array(numInputs); - this.isCIDFont = false; - } - return CFF; -})(); + return function constructPostScriptFromIRResult(src, srcOffset, + dest, destOffset) { + var i, value; + var key = ''; + var input = tmpBuf; + for (i = 0; i < numInputs; i++) { + value = src[srcOffset + i]; + input[i] = value; + key += value + '_'; + } -var CFFHeader = (function CFFHeaderClosure() { - function CFFHeader(major, minor, hdrSize, offSize) { - this.major = major; - this.minor = minor; - this.hdrSize = hdrSize; - this.offSize = offSize; - } - return CFFHeader; -})(); + var cachedValue = cache[key]; + if (cachedValue !== undefined) { + dest.set(cachedValue, destOffset); + return; + } -var CFFStrings = (function CFFStringsClosure() { - function CFFStrings() { - this.strings = []; - } - CFFStrings.prototype = { - get: function CFFStrings_get(index) { - if (index >= 0 && index <= 390) { - return CFFStandardStrings[index]; - } - if (index - 391 <= this.strings.length) { - return this.strings[index - 391]; - } - return CFFStandardStrings[0]; - }, - add: function CFFStrings_add(value) { - this.strings.push(value); - }, - get count() { - return this.strings.length; + var output = new Float32Array(numOutputs); + var stack = evaluator.execute(input); + var stackIndex = stack.length - numOutputs; + for (i = 0; i < numOutputs; i++) { + value = stack[stackIndex + i]; + var bound = range[i * 2]; + if (value < bound) { + value = bound; + } else { + bound = range[i * 2 +1]; + if (value > bound) { + value = bound; + } + } + output[i] = value; + } + if (cache_available > 0) { + cache_available--; + cache[key] = output; + } + dest.set(output, destOffset); + }; } }; - return CFFStrings; })(); -var CFFIndex = (function CFFIndexClosure() { - function CFFIndex() { - this.objects = []; - this.length = 0; +function isPDFFunction(v) { + var fnDict; + if (typeof v !== 'object') { + return false; + } else if (isDict(v)) { + fnDict = v; + } else if (isStream(v)) { + fnDict = v.dict; + } else { + return false; } - CFFIndex.prototype = { - add: function CFFIndex_add(data) { - this.length += data.length; - this.objects.push(data); - }, - set: function CFFIndex_set(index, data) { - this.length += data.length - this.objects[index].length; - this.objects[index] = data; - }, - get: function CFFIndex_get(index) { - return this.objects[index]; - }, - get count() { - return this.objects.length; - } - }; - return CFFIndex; -})(); + return fnDict.has('FunctionType'); +} -var CFFDict = (function CFFDictClosure() { - function CFFDict(tables, strings) { - this.keyToNameMap = tables.keyToNameMap; - this.nameToKeyMap = tables.nameToKeyMap; - this.defaults = tables.defaults; - this.types = tables.types; - this.opcodes = tables.opcodes; - this.order = tables.order; - this.strings = strings; - this.values = Object.create(null); +var PostScriptStack = (function PostScriptStackClosure() { + var MAX_STACK_SIZE = 100; + function PostScriptStack(initialStack) { + this.stack = !initialStack ? [] : + Array.prototype.slice.call(initialStack, 0); } - CFFDict.prototype = { - // value should always be an array - setByKey: function CFFDict_setByKey(key, value) { - if (!(key in this.keyToNameMap)) { - return false; - } - // ignore empty values - if (value.length === 0) { - return true; + + PostScriptStack.prototype = { + push: function PostScriptStack_push(value) { + if (this.stack.length >= MAX_STACK_SIZE) { + error('PostScript function stack overflow.'); } - var type = this.types[key]; - // remove the array wrapping these types of values - if (type === 'num' || type === 'sid' || type === 'offset') { - value = value[0]; + this.stack.push(value); + }, + pop: function PostScriptStack_pop() { + if (this.stack.length <= 0) { + error('PostScript function stack underflow.'); } - this.values[key] = value; - return true; + return this.stack.pop(); }, - setByName: function CFFDict_setByName(name, value) { - if (!(name in this.nameToKeyMap)) { - error('Invalid dictionary name "' + name + '"'); + copy: function PostScriptStack_copy(n) { + if (this.stack.length + n >= MAX_STACK_SIZE) { + error('PostScript function stack overflow.'); + } + var stack = this.stack; + for (var i = stack.length - n, j = n - 1; j >= 0; j--, i++) { + stack.push(stack[i]); } - this.values[this.nameToKeyMap[name]] = value; }, - hasName: function CFFDict_hasName(name) { - return this.nameToKeyMap[name] in this.values; + index: function PostScriptStack_index(n) { + this.push(this.stack[this.stack.length - n - 1]); }, - getByName: function CFFDict_getByName(name) { - if (!(name in this.nameToKeyMap)) { - error('Invalid dictionary name "' + name + '"'); + // rotate the last n stack elements p times + roll: function PostScriptStack_roll(n, p) { + var stack = this.stack; + var l = stack.length - n; + var r = stack.length - 1, c = l + (p - Math.floor(p / n) * n), i, j, t; + for (i = l, j = r; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; } - var key = this.nameToKeyMap[name]; - if (!(key in this.values)) { - return this.defaults[key]; + for (i = l, j = c - 1; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; + } + for (i = c, j = r; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; } - return this.values[key]; - }, - removeByName: function CFFDict_removeByName(name) { - delete this.values[this.nameToKeyMap[name]]; - } - }; - CFFDict.createTables = function CFFDict_createTables(layout) { - var tables = { - keyToNameMap: {}, - nameToKeyMap: {}, - defaults: {}, - types: {}, - opcodes: {}, - order: [] - }; - for (var i = 0, ii = layout.length; i < ii; ++i) { - var entry = layout[i]; - var key = isArray(entry[0]) ? (entry[0][0] << 8) + entry[0][1] : entry[0]; - tables.keyToNameMap[key] = entry[1]; - tables.nameToKeyMap[entry[1]] = key; - tables.types[key] = entry[2]; - tables.defaults[key] = entry[3]; - tables.opcodes[key] = isArray(entry[0]) ? entry[0] : [entry[0]]; - tables.order.push(key); } - return tables; }; - return CFFDict; -})(); - -var CFFTopDict = (function CFFTopDictClosure() { - var layout = [ - [[12, 30], 'ROS', ['sid', 'sid', 'num'], null], - [[12, 20], 'SyntheticBase', 'num', null], - [0, 'version', 'sid', null], - [1, 'Notice', 'sid', null], - [[12, 0], 'Copyright', 'sid', null], - [2, 'FullName', 'sid', null], - [3, 'FamilyName', 'sid', null], - [4, 'Weight', 'sid', null], - [[12, 1], 'isFixedPitch', 'num', 0], - [[12, 2], 'ItalicAngle', 'num', 0], - [[12, 3], 'UnderlinePosition', 'num', -100], - [[12, 4], 'UnderlineThickness', 'num', 50], - [[12, 5], 'PaintType', 'num', 0], - [[12, 6], 'CharstringType', 'num', 2], - [[12, 7], 'FontMatrix', ['num', 'num', 'num', 'num', 'num', 'num'], - [0.001, 0, 0, 0.001, 0, 0]], - [13, 'UniqueID', 'num', null], - [5, 'FontBBox', ['num', 'num', 'num', 'num'], [0, 0, 0, 0]], - [[12, 8], 'StrokeWidth', 'num', 0], - [14, 'XUID', 'array', null], - [15, 'charset', 'offset', 0], - [16, 'Encoding', 'offset', 0], - [17, 'CharStrings', 'offset', 0], - [18, 'Private', ['offset', 'offset'], null], - [[12, 21], 'PostScript', 'sid', null], - [[12, 22], 'BaseFontName', 'sid', null], - [[12, 23], 'BaseFontBlend', 'delta', null], - [[12, 31], 'CIDFontVersion', 'num', 0], - [[12, 32], 'CIDFontRevision', 'num', 0], - [[12, 33], 'CIDFontType', 'num', 0], - [[12, 34], 'CIDCount', 'num', 8720], - [[12, 35], 'UIDBase', 'num', null], - // XXX: CID Fonts on DirectWrite 6.1 only seem to work if FDSelect comes - // before FDArray. - [[12, 37], 'FDSelect', 'offset', null], - [[12, 36], 'FDArray', 'offset', null], - [[12, 38], 'FontName', 'sid', null] - ]; - var tables = null; - function CFFTopDict(strings) { - if (tables === null) { - tables = CFFDict.createTables(layout); - } - CFFDict.call(this, tables, strings); - this.privateDict = null; - } - CFFTopDict.prototype = Object.create(CFFDict.prototype); - return CFFTopDict; -})(); - -var CFFPrivateDict = (function CFFPrivateDictClosure() { - var layout = [ - [6, 'BlueValues', 'delta', null], - [7, 'OtherBlues', 'delta', null], - [8, 'FamilyBlues', 'delta', null], - [9, 'FamilyOtherBlues', 'delta', null], - [[12, 9], 'BlueScale', 'num', 0.039625], - [[12, 10], 'BlueShift', 'num', 7], - [[12, 11], 'BlueFuzz', 'num', 1], - [10, 'StdHW', 'num', null], - [11, 'StdVW', 'num', null], - [[12, 12], 'StemSnapH', 'delta', null], - [[12, 13], 'StemSnapV', 'delta', null], - [[12, 14], 'ForceBold', 'num', 0], - [[12, 17], 'LanguageGroup', 'num', 0], - [[12, 18], 'ExpansionFactor', 'num', 0.06], - [[12, 19], 'initialRandomSeed', 'num', 0], - [20, 'defaultWidthX', 'num', 0], - [21, 'nominalWidthX', 'num', 0], - [19, 'Subrs', 'offset', null] - ]; - var tables = null; - function CFFPrivateDict(strings) { - if (tables === null) { - tables = CFFDict.createTables(layout); - } - CFFDict.call(this, tables, strings); - this.subrsIndex = null; - } - CFFPrivateDict.prototype = Object.create(CFFDict.prototype); - return CFFPrivateDict; -})(); - -var CFFCharsetPredefinedTypes = { - ISO_ADOBE: 0, - EXPERT: 1, - EXPERT_SUBSET: 2 -}; -var CFFCharset = (function CFFCharsetClosure() { - function CFFCharset(predefined, format, charset, raw) { - this.predefined = predefined; - this.format = format; - this.charset = charset; - this.raw = raw; - } - return CFFCharset; + return PostScriptStack; })(); - -var CFFEncoding = (function CFFEncodingClosure() { - function CFFEncoding(predefined, format, encoding, raw) { - this.predefined = predefined; - this.format = format; - this.encoding = encoding; - this.raw = raw; +var PostScriptEvaluator = (function PostScriptEvaluatorClosure() { + function PostScriptEvaluator(operators) { + this.operators = operators; } - return CFFEncoding; -})(); + PostScriptEvaluator.prototype = { + execute: function PostScriptEvaluator_execute(initialStack) { + var stack = new PostScriptStack(initialStack); + var counter = 0; + var operators = this.operators; + var length = operators.length; + var operator, a, b; + while (counter < length) { + operator = operators[counter++]; + if (typeof operator === 'number') { + // Operator is really an operand and should be pushed to the stack. + stack.push(operator); + continue; + } + switch (operator) { + // non standard ps operators + case 'jz': // jump if false + b = stack.pop(); + a = stack.pop(); + if (!a) { + counter = b; + } + break; + case 'j': // jump + a = stack.pop(); + counter = a; + break; -var CFFFDSelect = (function CFFFDSelectClosure() { - function CFFFDSelect(fdSelect, raw) { - this.fdSelect = fdSelect; - this.raw = raw; - } - CFFFDSelect.prototype = { - getFDIndex: function CFFFDSelect_get(glyphIndex) { - if (glyphIndex < 0 || glyphIndex >= this.fdSelect.length) { - return -1; + // all ps operators in alphabetical order (excluding if/ifelse) + case 'abs': + a = stack.pop(); + stack.push(Math.abs(a)); + break; + case 'add': + b = stack.pop(); + a = stack.pop(); + stack.push(a + b); + break; + case 'and': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a && b); + } else { + stack.push(a & b); + } + break; + case 'atan': + a = stack.pop(); + stack.push(Math.atan(a)); + break; + case 'bitshift': + b = stack.pop(); + a = stack.pop(); + if (a > 0) { + stack.push(a << b); + } else { + stack.push(a >> b); + } + break; + case 'ceiling': + a = stack.pop(); + stack.push(Math.ceil(a)); + break; + case 'copy': + a = stack.pop(); + stack.copy(a); + break; + case 'cos': + a = stack.pop(); + stack.push(Math.cos(a)); + break; + case 'cvi': + a = stack.pop() | 0; + stack.push(a); + break; + case 'cvr': + // noop + break; + case 'div': + b = stack.pop(); + a = stack.pop(); + stack.push(a / b); + break; + case 'dup': + stack.copy(1); + break; + case 'eq': + b = stack.pop(); + a = stack.pop(); + stack.push(a === b); + break; + case 'exch': + stack.roll(2, 1); + break; + case 'exp': + b = stack.pop(); + a = stack.pop(); + stack.push(Math.pow(a, b)); + break; + case 'false': + stack.push(false); + break; + case 'floor': + a = stack.pop(); + stack.push(Math.floor(a)); + break; + case 'ge': + b = stack.pop(); + a = stack.pop(); + stack.push(a >= b); + break; + case 'gt': + b = stack.pop(); + a = stack.pop(); + stack.push(a > b); + break; + case 'idiv': + b = stack.pop(); + a = stack.pop(); + stack.push((a / b) | 0); + break; + case 'index': + a = stack.pop(); + stack.index(a); + break; + case 'le': + b = stack.pop(); + a = stack.pop(); + stack.push(a <= b); + break; + case 'ln': + a = stack.pop(); + stack.push(Math.log(a)); + break; + case 'log': + a = stack.pop(); + stack.push(Math.log(a) / Math.LN10); + break; + case 'lt': + b = stack.pop(); + a = stack.pop(); + stack.push(a < b); + break; + case 'mod': + b = stack.pop(); + a = stack.pop(); + stack.push(a % b); + break; + case 'mul': + b = stack.pop(); + a = stack.pop(); + stack.push(a * b); + break; + case 'ne': + b = stack.pop(); + a = stack.pop(); + stack.push(a !== b); + break; + case 'neg': + a = stack.pop(); + stack.push(-a); + break; + case 'not': + a = stack.pop(); + if (isBool(a)) { + stack.push(!a); + } else { + stack.push(~a); + } + break; + case 'or': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a || b); + } else { + stack.push(a | b); + } + break; + case 'pop': + stack.pop(); + break; + case 'roll': + b = stack.pop(); + a = stack.pop(); + stack.roll(a, b); + break; + case 'round': + a = stack.pop(); + stack.push(Math.round(a)); + break; + case 'sin': + a = stack.pop(); + stack.push(Math.sin(a)); + break; + case 'sqrt': + a = stack.pop(); + stack.push(Math.sqrt(a)); + break; + case 'sub': + b = stack.pop(); + a = stack.pop(); + stack.push(a - b); + break; + case 'true': + stack.push(true); + break; + case 'truncate': + a = stack.pop(); + a = a < 0 ? Math.ceil(a) : Math.floor(a); + stack.push(a); + break; + case 'xor': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a !== b); + } else { + stack.push(a ^ b); + } + break; + default: + error('Unknown operator ' + operator); + break; + } } - return this.fdSelect[glyphIndex]; + return stack.stack; } }; - return CFFFDSelect; + return PostScriptEvaluator; })(); -// Helper class to keep track of where an offset is within the data and helps -// filling in that offset once it's known. -var CFFOffsetTracker = (function CFFOffsetTrackerClosure() { - function CFFOffsetTracker() { - this.offsets = Object.create(null); +// Most of the PDFs functions consist of simple operations such as: +// roll, exch, sub, cvr, pop, index, dup, mul, if, gt, add. +// +// We can compile most of such programs, and at the same moment, we can +// optimize some expressions using basic math properties. Keeping track of +// min/max values will allow us to avoid extra Math.min/Math.max calls. +var PostScriptCompiler = (function PostScriptCompilerClosure() { + function AstNode(type) { + this.type = type; } - CFFOffsetTracker.prototype = { - isTracking: function CFFOffsetTracker_isTracking(key) { - return key in this.offsets; - }, - track: function CFFOffsetTracker_track(key, location) { - if (key in this.offsets) { - error('Already tracking location of ' + key); - } - this.offsets[key] = location; - }, - offset: function CFFOffsetTracker_offset(value) { - for (var key in this.offsets) { - this.offsets[key] += value; - } - }, - setEntryLocation: function CFFOffsetTracker_setEntryLocation(key, - values, - output) { - if (!(key in this.offsets)) { - error('Not tracking location of ' + key); - } - var data = output.data; - var dataOffset = this.offsets[key]; - var size = 5; - for (var i = 0, ii = values.length; i < ii; ++i) { - var offset0 = i * size + dataOffset; - var offset1 = offset0 + 1; - var offset2 = offset0 + 2; - var offset3 = offset0 + 3; - var offset4 = offset0 + 4; - // It's easy to screw up offsets so perform this sanity check. - if (data[offset0] !== 0x1d || data[offset1] !== 0 || - data[offset2] !== 0 || data[offset3] !== 0 || data[offset4] !== 0) { - error('writing to an offset that is not empty'); - } - var value = values[i]; - data[offset0] = 0x1d; - data[offset1] = (value >> 24) & 0xFF; - data[offset2] = (value >> 16) & 0xFF; - data[offset3] = (value >> 8) & 0xFF; - data[offset4] = value & 0xFF; - } - } + AstNode.prototype.visit = function (visitor) { + throw new Error('abstract method'); }; - return CFFOffsetTracker; -})(); -// Takes a CFF and converts it to the binary representation. -var CFFCompiler = (function CFFCompilerClosure() { - function CFFCompiler(cff) { - this.cff = cff; + function AstArgument(index, min, max) { + AstNode.call(this, 'args'); + this.index = index; + this.min = min; + this.max = max; } - CFFCompiler.prototype = { - compile: function CFFCompiler_compile() { - var cff = this.cff; - var output = { - data: [], - length: 0, - add: function CFFCompiler_add(data) { - this.data = this.data.concat(data); - this.length = this.data.length; - } - }; - - // Compile the five entries that must be in order. - var header = this.compileHeader(cff.header); - output.add(header); - - var nameIndex = this.compileNameIndex(cff.names); - output.add(nameIndex); - - if (cff.isCIDFont) { - // The spec is unclear on how font matrices should relate to each other - // when there is one in the main top dict and the sub top dicts. - // Windows handles this differently than linux and osx so we have to - // normalize to work on all. - // Rules based off of some mailing list discussions: - // - If main font has a matrix and subfont doesn't, use the main matrix. - // - If no main font matrix and there is a subfont matrix, use the - // subfont matrix. - // - If both have matrices, concat together. - // - If neither have matrices, use default. - // To make this work on all platforms we move the top matrix into each - // sub top dict and concat if necessary. - if (cff.topDict.hasName('FontMatrix')) { - var base = cff.topDict.getByName('FontMatrix'); - cff.topDict.removeByName('FontMatrix'); - for (var i = 0, ii = cff.fdArray.length; i < ii; i++) { - var subDict = cff.fdArray[i]; - var matrix = base.slice(0); - if (subDict.hasName('FontMatrix')) { - matrix = Util.transform(matrix, subDict.getByName('FontMatrix')); - } - subDict.setByName('FontMatrix', matrix); - } - } - } - - var compiled = this.compileTopDicts([cff.topDict], - output.length, - cff.isCIDFont); - output.add(compiled.output); - var topDictTracker = compiled.trackers[0]; - - var stringIndex = this.compileStringIndex(cff.strings.strings); - output.add(stringIndex); - - var globalSubrIndex = this.compileIndex(cff.globalSubrIndex); - output.add(globalSubrIndex); - - // Now start on the other entries that have no specfic order. - if (cff.encoding && cff.topDict.hasName('Encoding')) { - if (cff.encoding.predefined) { - topDictTracker.setEntryLocation('Encoding', [cff.encoding.format], - output); - } else { - var encoding = this.compileEncoding(cff.encoding); - topDictTracker.setEntryLocation('Encoding', [output.length], output); - output.add(encoding); - } - } - - if (cff.charset && cff.topDict.hasName('charset')) { - if (cff.charset.predefined) { - topDictTracker.setEntryLocation('charset', [cff.charset.format], - output); - } else { - var charset = this.compileCharset(cff.charset); - topDictTracker.setEntryLocation('charset', [output.length], output); - output.add(charset); - } - } - - var charStrings = this.compileCharStrings(cff.charStrings); - topDictTracker.setEntryLocation('CharStrings', [output.length], output); - output.add(charStrings); - - if (cff.isCIDFont) { - // For some reason FDSelect must be in front of FDArray on windows. OSX - // and linux don't seem to care. - topDictTracker.setEntryLocation('FDSelect', [output.length], output); - var fdSelect = this.compileFDSelect(cff.fdSelect.raw); - output.add(fdSelect); - // It is unclear if the sub font dictionary can have CID related - // dictionary keys, but the sanitizer doesn't like them so remove them. - compiled = this.compileTopDicts(cff.fdArray, output.length, true); - topDictTracker.setEntryLocation('FDArray', [output.length], output); - output.add(compiled.output); - var fontDictTrackers = compiled.trackers; + AstArgument.prototype = Object.create(AstNode.prototype); + AstArgument.prototype.visit = function (visitor) { + visitor.visitArgument(this); + }; - this.compilePrivateDicts(cff.fdArray, fontDictTrackers, output); - } + function AstLiteral(number) { + AstNode.call(this, 'literal'); + this.number = number; + this.min = number; + this.max = number; + } + AstLiteral.prototype = Object.create(AstNode.prototype); + AstLiteral.prototype.visit = function (visitor) { + visitor.visitLiteral(this); + }; - this.compilePrivateDicts([cff.topDict], [topDictTracker], output); + function AstBinaryOperation(op, arg1, arg2, min, max) { + AstNode.call(this, 'binary'); + this.op = op; + this.arg1 = arg1; + this.arg2 = arg2; + this.min = min; + this.max = max; + } + AstBinaryOperation.prototype = Object.create(AstNode.prototype); + AstBinaryOperation.prototype.visit = function (visitor) { + visitor.visitBinaryOperation(this); + }; - // If the font data ends with INDEX whose object data is zero-length, - // the sanitizer will bail out. Add a dummy byte to avoid that. - output.add([0]); + function AstMin(arg, max) { + AstNode.call(this, 'max'); + this.arg = arg; + this.min = arg.min; + this.max = max; + } + AstMin.prototype = Object.create(AstNode.prototype); + AstMin.prototype.visit = function (visitor) { + visitor.visitMin(this); + }; - return output.data; - }, - encodeNumber: function CFFCompiler_encodeNumber(value) { - if (parseFloat(value) === parseInt(value, 10) && !isNaN(value)) { // isInt - return this.encodeInteger(value); - } else { - return this.encodeFloat(value); - } - }, - encodeFloat: function CFFCompiler_encodeFloat(num) { - var value = num.toString(); + function AstVariable(index, min, max) { + AstNode.call(this, 'var'); + this.index = index; + this.min = min; + this.max = max; + } + AstVariable.prototype = Object.create(AstNode.prototype); + AstVariable.prototype.visit = function (visitor) { + visitor.visitVariable(this); + }; - // rounding inaccurate doubles - var m = /\.(\d*?)(?:9{5,20}|0{5,20})\d{0,2}(?:e(.+)|$)/.exec(value); - if (m) { - var epsilon = parseFloat('1e' + ((m[2] ? +m[2] : 0) + m[1].length)); - value = (Math.round(num * epsilon) / epsilon).toString(); - } + function AstVariableDefinition(variable, arg) { + AstNode.call(this, 'definition'); + this.variable = variable; + this.arg = arg; + } + AstVariableDefinition.prototype = Object.create(AstNode.prototype); + AstVariableDefinition.prototype.visit = function (visitor) { + visitor.visitVariableDefinition(this); + }; - var nibbles = ''; - var i, ii; - for (i = 0, ii = value.length; i < ii; ++i) { - var a = value[i]; - if (a === 'e') { - nibbles += value[++i] === '-' ? 'c' : 'b'; - } else if (a === '.') { - nibbles += 'a'; - } else if (a === '-') { - nibbles += 'e'; - } else { - nibbles += a; - } - } - nibbles += (nibbles.length & 1) ? 'f' : 'ff'; - var out = [30]; - for (i = 0, ii = nibbles.length; i < ii; i += 2) { - out.push(parseInt(nibbles.substr(i, 2), 16)); - } - return out; + function ExpressionBuilderVisitor() { + this.parts = []; + } + ExpressionBuilderVisitor.prototype = { + visitArgument: function (arg) { + this.parts.push('Math.max(', arg.min, ', Math.min(', + arg.max, ', src[srcOffset + ', arg.index, ']))'); }, - encodeInteger: function CFFCompiler_encodeInteger(value) { - var code; - if (value >= -107 && value <= 107) { - code = [value + 139]; - } else if (value >= 108 && value <= 1131) { - value = [value - 108]; - code = [(value >> 8) + 247, value & 0xFF]; - } else if (value >= -1131 && value <= -108) { - value = -value - 108; - code = [(value >> 8) + 251, value & 0xFF]; - } else if (value >= -32768 && value <= 32767) { - code = [0x1c, (value >> 8) & 0xFF, value & 0xFF]; - } else { - code = [0x1d, - (value >> 24) & 0xFF, - (value >> 16) & 0xFF, - (value >> 8) & 0xFF, - value & 0xFF]; - } - return code; + visitVariable: function (variable) { + this.parts.push('v', variable.index); }, - compileHeader: function CFFCompiler_compileHeader(header) { - return [ - header.major, - header.minor, - header.hdrSize, - header.offSize - ]; + visitLiteral: function (literal) { + this.parts.push(literal.number); }, - compileNameIndex: function CFFCompiler_compileNameIndex(names) { - var nameIndex = new CFFIndex(); - for (var i = 0, ii = names.length; i < ii; ++i) { - nameIndex.add(stringToBytes(names[i])); - } - return this.compileIndex(nameIndex); + visitBinaryOperation: function (operation) { + this.parts.push('('); + operation.arg1.visit(this); + this.parts.push(' ', operation.op, ' '); + operation.arg2.visit(this); + this.parts.push(')'); }, - compileTopDicts: function CFFCompiler_compileTopDicts(dicts, - length, - removeCidKeys) { - var fontDictTrackers = []; - var fdArrayIndex = new CFFIndex(); - for (var i = 0, ii = dicts.length; i < ii; ++i) { - var fontDict = dicts[i]; - if (removeCidKeys) { - fontDict.removeByName('CIDFontVersion'); - fontDict.removeByName('CIDFontRevision'); - fontDict.removeByName('CIDFontType'); - fontDict.removeByName('CIDCount'); - fontDict.removeByName('UIDBase'); - } - var fontDictTracker = new CFFOffsetTracker(); - var fontDictData = this.compileDict(fontDict, fontDictTracker); - fontDictTrackers.push(fontDictTracker); - fdArrayIndex.add(fontDictData); - fontDictTracker.offset(length); - } - fdArrayIndex = this.compileIndex(fdArrayIndex, fontDictTrackers); - return { - trackers: fontDictTrackers, - output: fdArrayIndex - }; + visitVariableDefinition: function (definition) { + this.parts.push('var '); + definition.variable.visit(this); + this.parts.push(' = '); + definition.arg.visit(this); + this.parts.push(';'); }, - compilePrivateDicts: function CFFCompiler_compilePrivateDicts(dicts, - trackers, - output) { - for (var i = 0, ii = dicts.length; i < ii; ++i) { - var fontDict = dicts[i]; - assert(fontDict.privateDict && fontDict.hasName('Private'), - 'There must be an private dictionary.'); - var privateDict = fontDict.privateDict; - var privateDictTracker = new CFFOffsetTracker(); - var privateDictData = this.compileDict(privateDict, privateDictTracker); - - var outputLength = output.length; - privateDictTracker.offset(outputLength); - if (!privateDictData.length) { - // The private dictionary was empty, set the output length to zero to - // ensure the offset length isn't out of bounds in the eyes of the - // sanitizer. - outputLength = 0; - } - - trackers[i].setEntryLocation('Private', - [privateDictData.length, outputLength], - output); - output.add(privateDictData); - - if (privateDict.subrsIndex && privateDict.hasName('Subrs')) { - var subrs = this.compileIndex(privateDict.subrsIndex); - privateDictTracker.setEntryLocation('Subrs', [privateDictData.length], - output); - output.add(subrs); - } - } + visitMin: function (max) { + this.parts.push('Math.min('); + max.arg.visit(this); + this.parts.push(', ', max.max, ')'); }, - compileDict: function CFFCompiler_compileDict(dict, offsetTracker) { - var out = []; - // The dictionary keys must be in a certain order. - var order = dict.order; - for (var i = 0; i < order.length; ++i) { - var key = order[i]; - if (!(key in dict.values)) { - continue; - } - var values = dict.values[key]; - var types = dict.types[key]; - if (!isArray(types)) { - types = [types]; - } - if (!isArray(values)) { - values = [values]; - } - - // Remove any empty dict values. - if (values.length === 0) { - continue; - } + toString: function () { + return this.parts.join(''); + } + }; - for (var j = 0, jj = types.length; j < jj; ++j) { - var type = types[j]; - var value = values[j]; - switch (type) { - case 'num': - case 'sid': - out = out.concat(this.encodeNumber(value)); - break; - case 'offset': - // For offsets we just insert a 32bit integer so we don't have to - // deal with figuring out the length of the offset when it gets - // replaced later on by the compiler. - var name = dict.keyToNameMap[key]; - // Some offsets have the offset and the length, so just record the - // position of the first one. - if (!offsetTracker.isTracking(name)) { - offsetTracker.track(name, out.length); - } - out = out.concat([0x1d, 0, 0, 0, 0]); - break; - case 'array': - case 'delta': - out = out.concat(this.encodeNumber(value)); - for (var k = 1, kk = values.length; k < kk; ++k) { - out = out.concat(this.encodeNumber(values[k])); - } - break; - default: - error('Unknown data type of ' + type); - break; - } - } - out = out.concat(dict.opcodes[key]); - } - return out; - }, - compileStringIndex: function CFFCompiler_compileStringIndex(strings) { - var stringIndex = new CFFIndex(); - for (var i = 0, ii = strings.length; i < ii; ++i) { - stringIndex.add(stringToBytes(strings[i])); - } - return this.compileIndex(stringIndex); - }, - compileGlobalSubrIndex: function CFFCompiler_compileGlobalSubrIndex() { - var globalSubrIndex = this.cff.globalSubrIndex; - this.out.writeByteArray(this.compileIndex(globalSubrIndex)); - }, - compileCharStrings: function CFFCompiler_compileCharStrings(charStrings) { - return this.compileIndex(charStrings); - }, - compileCharset: function CFFCompiler_compileCharset(charset) { - return this.compileTypedArray(charset.raw); - }, - compileEncoding: function CFFCompiler_compileEncoding(encoding) { - return this.compileTypedArray(encoding.raw); - }, - compileFDSelect: function CFFCompiler_compileFDSelect(fdSelect) { - return this.compileTypedArray(fdSelect); - }, - compileTypedArray: function CFFCompiler_compileTypedArray(data) { - var out = []; - for (var i = 0, ii = data.length; i < ii; ++i) { - out[i] = data[i]; - } - return out; - }, - compileIndex: function CFFCompiler_compileIndex(index, trackers) { - trackers = trackers || []; - var objects = index.objects; - // First 2 bytes contains the number of objects contained into this index - var count = objects.length; + function buildAddOperation(num1, num2) { + if (num2.type === 'literal' && num2.number === 0) { + // optimization: second operand is 0 + return num1; + } + if (num1.type === 'literal' && num1.number === 0) { + // optimization: first operand is 0 + return num2; + } + if (num2.type === 'literal' && num1.type === 'literal') { + // optimization: operands operand are literals + return new AstLiteral(num1.number + num2.number); + } + return new AstBinaryOperation('+', num1, num2, + num1.min + num2.min, num1.max + num2.max); + } - // If there is no object, just create an index. This technically - // should just be [0, 0] but OTS has an issue with that. - if (count === 0) { - return [0, 0, 0]; + function buildMulOperation(num1, num2) { + if (num2.type === 'literal') { + // optimization: second operands is a literal... + if (num2.number === 0) { + return new AstLiteral(0); // and it's 0 + } else if (num2.number === 1) { + return num1; // and it's 1 + } else if (num1.type === 'literal') { + // ... and first operands is a literal too + return new AstLiteral(num1.number * num2.number); } - - var data = [(count >> 8) & 0xFF, count & 0xff]; - - var lastOffset = 1, i; - for (i = 0; i < count; ++i) { - lastOffset += objects[i].length; + } + if (num1.type === 'literal') { + // optimization: first operands is a literal... + if (num1.number === 0) { + return new AstLiteral(0); // and it's 0 + } else if (num1.number === 1) { + return num2; // and it's 1 } + } + var min = Math.min(num1.min * num2.min, num1.min * num2.max, + num1.max * num2.min, num1.max * num2.max); + var max = Math.max(num1.min * num2.min, num1.min * num2.max, + num1.max * num2.min, num1.max * num2.max); + return new AstBinaryOperation('*', num1, num2, min, max); + } - var offsetSize; - if (lastOffset < 0x100) { - offsetSize = 1; - } else if (lastOffset < 0x10000) { - offsetSize = 2; - } else if (lastOffset < 0x1000000) { - offsetSize = 3; - } else { - offsetSize = 4; + function buildSubOperation(num1, num2) { + if (num2.type === 'literal') { + // optimization: second operands is a literal... + if (num2.number === 0) { + return num1; // ... and it's 0 + } else if (num1.type === 'literal') { + // ... and first operands is a literal too + return new AstLiteral(num1.number - num2.number); } + } + if (num2.type === 'binary' && num2.op === '-' && + num1.type === 'literal' && num1.number === 1 && + num2.arg1.type === 'literal' && num2.arg1.number === 1) { + // optimization for case: 1 - (1 - x) + return num2.arg2; + } + return new AstBinaryOperation('-', num1, num2, + num1.min - num2.max, num1.max - num2.min); + } - // Next byte contains the offset size use to reference object in the file - data.push(offsetSize); + function buildMinOperation(num1, max) { + if (num1.min >= max) { + // optimization: num1 min value is not less than required max + return new AstLiteral(max); // just returning max + } else if (num1.max <= max) { + // optimization: num1 max value is not greater than required max + return num1; // just returning an argument + } + return new AstMin(num1, max); + } - // Add another offset after this one because we need a new offset - var relativeOffset = 1; - for (i = 0; i < count + 1; i++) { - if (offsetSize === 1) { - data.push(relativeOffset & 0xFF); - } else if (offsetSize === 2) { - data.push((relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); - } else if (offsetSize === 3) { - data.push((relativeOffset >> 16) & 0xFF, - (relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); - } else { - data.push((relativeOffset >>> 24) & 0xFF, - (relativeOffset >> 16) & 0xFF, - (relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); + function PostScriptCompiler() {} + PostScriptCompiler.prototype = { + compile: function PostScriptCompiler_compile(code, domain, range) { + var stack = []; + var i, ii; + var instructions = []; + var inputSize = domain.length >> 1, outputSize = range.length >> 1; + var lastRegister = 0; + var n, j; + var num1, num2, ast1, ast2, tmpVar, item; + for (i = 0; i < inputSize; i++) { + stack.push(new AstArgument(i, domain[i * 2], domain[i * 2 + 1])); + } + + for (i = 0, ii = code.length; i < ii; i++) { + item = code[i]; + if (typeof item === 'number') { + stack.push(new AstLiteral(item)); + continue; } - if (objects[i]) { - relativeOffset += objects[i].length; + switch (item) { + case 'add': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildAddOperation(num1, num2)); + break; + case 'cvr': + if (stack.length < 1) { + return null; + } + break; + case 'mul': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildMulOperation(num1, num2)); + break; + case 'sub': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildSubOperation(num1, num2)); + break; + case 'exch': + if (stack.length < 2) { + return null; + } + ast1 = stack.pop(); ast2 = stack.pop(); + stack.push(ast1, ast2); + break; + case 'pop': + if (stack.length < 1) { + return null; + } + stack.pop(); + break; + case 'index': + if (stack.length < 1) { + return null; + } + num1 = stack.pop(); + if (num1.type !== 'literal') { + return null; + } + n = num1.number; + if (n < 0 || (n|0) !== n || stack.length < n) { + return null; + } + ast1 = stack[stack.length - n - 1]; + if (ast1.type === 'literal' || ast1.type === 'var') { + stack.push(ast1); + break; + } + tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); + stack[stack.length - n - 1] = tmpVar; + stack.push(tmpVar); + instructions.push(new AstVariableDefinition(tmpVar, ast1)); + break; + case 'dup': + if (stack.length < 1) { + return null; + } + if (typeof code[i + 1] === 'number' && code[i + 2] === 'gt' && + code[i + 3] === i + 7 && code[i + 4] === 'jz' && + code[i + 5] === 'pop' && code[i + 6] === code[i + 1]) { + // special case of the commands sequence for the min operation + num1 = stack.pop(); + stack.push(buildMinOperation(num1, code[i + 1])); + i += 6; + break; + } + ast1 = stack[stack.length - 1]; + if (ast1.type === 'literal' || ast1.type === 'var') { + // we don't have to save into intermediate variable a literal or + // variable. + stack.push(ast1); + break; + } + tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); + stack[stack.length - 1] = tmpVar; + stack.push(tmpVar); + instructions.push(new AstVariableDefinition(tmpVar, ast1)); + break; + case 'roll': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + if (num2.type !== 'literal' || num1.type !== 'literal') { + // both roll operands must be numbers + return null; + } + j = num2.number; + n = num1.number; + if (n <= 0 || (n|0) !== n || (j|0) !== j || stack.length < n) { + // ... and integers + return null; + } + j = ((j % n) + n) % n; + if (j === 0) { + break; // just skipping -- there are nothing to rotate + } + Array.prototype.push.apply(stack, + stack.splice(stack.length - n, n - j)); + break; + default: + return null; // unsupported operator } } - for (i = 0; i < count; i++) { - // Notify the tracker where the object will be offset in the data. - if (trackers[i]) { - trackers[i].offset(data.length); + if (stack.length !== outputSize) { + return null; + } + + var result = []; + instructions.forEach(function (instruction) { + var statementBuilder = new ExpressionBuilderVisitor(); + instruction.visit(statementBuilder); + result.push(statementBuilder.toString()); + }); + stack.forEach(function (expr, i) { + var statementBuilder = new ExpressionBuilderVisitor(); + expr.visit(statementBuilder); + var min = range[i * 2], max = range[i * 2 + 1]; + var out = [statementBuilder.toString()]; + if (min > expr.min) { + out.unshift('Math.max(', min, ', '); + out.push(')'); } - for (var j = 0, jj = objects[i].length; j < jj; j++) { - data.push(objects[i][j]); + if (max < expr.max) { + out.unshift('Math.min(', max, ', '); + out.push(')'); } - } - return data; + out.unshift('dest[destOffset + ', i, '] = '); + out.push(';'); + result.push(out.join('')); + }); + return result.join('\n'); } }; - return CFFCompiler; -})(); - -function _enableSeacAnalysis(enabled) { - exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED = enabled; -} - -// Workaround for seac on Windows. -(function checkSeacSupport() { - if (typeof navigator !== 'undefined' && /Windows/.test(navigator.userAgent)) { - SEAC_ANALYSIS_ENABLED = true; - } -})(); -// Workaround for Private Use Area characters in Chrome on Windows -// http://code.google.com/p/chromium/issues/detail?id=122465 -// https://github.com/mozilla/pdf.js/issues/1689 -(function checkChromeWindows() { - if (typeof navigator !== 'undefined' && - /Windows.*Chrome/.test(navigator.userAgent)) { - SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = true; - } + return PostScriptCompiler; })(); -exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED; -exports.CFFCompiler = CFFCompiler; -exports.CFFIndex = CFFIndex; -exports.CFFParser = CFFParser; -exports.CFFStrings = CFFStrings; -exports.ErrorFont = ErrorFont; -exports.FontFlags = FontFlags; -exports.Font = Font; -exports.IdentityToUnicodeMap = IdentityToUnicodeMap; -exports.ToUnicodeMap = ToUnicodeMap; -exports.Type1Parser = Type1Parser; -exports.getFontType = getFontType; -exports._enableSeacAnalysis = _enableSeacAnalysis; - -// TODO refactor to remove cyclic dependency on font_renderer.js -coreFontRenderer._setCoreFonts(exports); +exports.isPDFFunction = isPDFFunction; +exports.PDFFunction = PDFFunction; +exports.PostScriptEvaluator = PostScriptEvaluator; +exports.PostScriptCompiler = PostScriptCompiler; })); (function (root, factory) { { - factory((root.pdfjsCoreFunction = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCorePsParser); + factory((root.pdfjsCoreColorSpace = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreFunction, root.pdfjsCoreStream); } -}(this, function (exports, sharedUtil, corePrimitives, corePsParser) { +}(this, function (exports, sharedUtil, corePrimitives, coreFunction, + coreStream) { var error = sharedUtil.error; var info = sharedUtil.info; var isArray = sharedUtil.isArray; -var isBool = sharedUtil.isBool; +var isString = sharedUtil.isString; +var shadow = sharedUtil.shadow; +var warn = sharedUtil.warn; var isDict = corePrimitives.isDict; +var isName = corePrimitives.isName; var isStream = corePrimitives.isStream; -var PostScriptLexer = corePsParser.PostScriptLexer; -var PostScriptParser = corePsParser.PostScriptParser; - -var PDFFunction = (function PDFFunctionClosure() { - var CONSTRUCT_SAMPLED = 0; - var CONSTRUCT_INTERPOLATED = 2; - var CONSTRUCT_STICHED = 3; - var CONSTRUCT_POSTSCRIPT = 4; - - return { - getSampleArray: function PDFFunction_getSampleArray(size, outputSize, bps, - str) { - var i, ii; - var length = 1; - for (i = 0, ii = size.length; i < ii; i++) { - length *= size[i]; - } - length *= outputSize; - - var array = new Array(length); - var codeSize = 0; - var codeBuf = 0; - // 32 is a valid bps so shifting won't work - var sampleMul = 1.0 / (Math.pow(2.0, bps) - 1); - - var strBytes = str.getBytes((length * bps + 7) / 8); - var strIdx = 0; - for (i = 0; i < length; i++) { - while (codeSize < bps) { - codeBuf <<= 8; - codeBuf |= strBytes[strIdx++]; - codeSize += 8; - } - codeSize -= bps; - array[i] = (codeBuf >> codeSize) * sampleMul; - codeBuf &= (1 << codeSize) - 1; - } - return array; - }, - - getIR: function PDFFunction_getIR(xref, fn) { - var dict = fn.dict; - if (!dict) { - dict = fn; - } - - var types = [this.constructSampled, - null, - this.constructInterpolated, - this.constructStiched, - this.constructPostScript]; - - var typeNum = dict.get('FunctionType'); - var typeFn = types[typeNum]; - if (!typeFn) { - error('Unknown type of function'); - } - - return typeFn.call(this, fn, dict, xref); - }, - - fromIR: function PDFFunction_fromIR(IR) { - var type = IR[0]; - switch (type) { - case CONSTRUCT_SAMPLED: - return this.constructSampledFromIR(IR); - case CONSTRUCT_INTERPOLATED: - return this.constructInterpolatedFromIR(IR); - case CONSTRUCT_STICHED: - return this.constructStichedFromIR(IR); - //case CONSTRUCT_POSTSCRIPT: - default: - return this.constructPostScriptFromIR(IR); - } - }, - - parse: function PDFFunction_parse(xref, fn) { - var IR = this.getIR(xref, fn); - return this.fromIR(IR); - }, - - parseArray: function PDFFunction_parseArray(xref, fnObj) { - if (!isArray(fnObj)) { - // not an array -- parsing as regular function - return this.parse(xref, fnObj); - } - - var fnArray = []; - for (var j = 0, jj = fnObj.length; j < jj; j++) { - var obj = xref.fetchIfRef(fnObj[j]); - fnArray.push(PDFFunction.parse(xref, obj)); - } - return function (src, srcOffset, dest, destOffset) { - for (var i = 0, ii = fnArray.length; i < ii; i++) { - fnArray[i](src, srcOffset, dest, destOffset + i); - } - }; - }, - - constructSampled: function PDFFunction_constructSampled(str, dict) { - function toMultiArray(arr) { - var inputLength = arr.length; - var out = []; - var index = 0; - for (var i = 0; i < inputLength; i += 2) { - out[index] = [arr[i], arr[i + 1]]; - ++index; - } - return out; - } - var domain = dict.get('Domain'); - var range = dict.get('Range'); - - if (!domain || !range) { - error('No domain or range'); - } - - var inputSize = domain.length / 2; - var outputSize = range.length / 2; - - domain = toMultiArray(domain); - range = toMultiArray(range); - - var size = dict.get('Size'); - var bps = dict.get('BitsPerSample'); - var order = dict.get('Order') || 1; - if (order !== 1) { - // No description how cubic spline interpolation works in PDF32000:2008 - // As in poppler, ignoring order, linear interpolation may work as good - info('No support for cubic spline interpolation: ' + order); - } - - var encode = dict.get('Encode'); - if (!encode) { - encode = []; - for (var i = 0; i < inputSize; ++i) { - encode.push(0); - encode.push(size[i] - 1); - } - } - encode = toMultiArray(encode); - - var decode = dict.get('Decode'); - if (!decode) { - decode = range; - } else { - decode = toMultiArray(decode); - } - - var samples = this.getSampleArray(size, outputSize, bps, str); - - return [ - CONSTRUCT_SAMPLED, inputSize, domain, encode, decode, samples, size, - outputSize, Math.pow(2, bps) - 1, range - ]; - }, - - constructSampledFromIR: function PDFFunction_constructSampledFromIR(IR) { - // See chapter 3, page 109 of the PDF reference - function interpolate(x, xmin, xmax, ymin, ymax) { - return ymin + ((x - xmin) * ((ymax - ymin) / (xmax - xmin))); - } - - return function constructSampledFromIRResult(src, srcOffset, - dest, destOffset) { - // See chapter 3, page 110 of the PDF reference. - var m = IR[1]; - var domain = IR[2]; - var encode = IR[3]; - var decode = IR[4]; - var samples = IR[5]; - var size = IR[6]; - var n = IR[7]; - //var mask = IR[8]; - var range = IR[9]; - - // Building the cube vertices: its part and sample index - // http://rjwagner49.com/Mathematics/Interpolation.pdf - var cubeVertices = 1 << m; - var cubeN = new Float64Array(cubeVertices); - var cubeVertex = new Uint32Array(cubeVertices); - var i, j; - for (j = 0; j < cubeVertices; j++) { - cubeN[j] = 1; - } - - var k = n, pos = 1; - // Map x_i to y_j for 0 <= i < m using the sampled function. - for (i = 0; i < m; ++i) { - // x_i' = min(max(x_i, Domain_2i), Domain_2i+1) - var domain_2i = domain[i][0]; - var domain_2i_1 = domain[i][1]; - var xi = Math.min(Math.max(src[srcOffset +i], domain_2i), - domain_2i_1); +var PDFFunction = coreFunction.PDFFunction; - // e_i = Interpolate(x_i', Domain_2i, Domain_2i+1, - // Encode_2i, Encode_2i+1) - var e = interpolate(xi, domain_2i, domain_2i_1, - encode[i][0], encode[i][1]); +var coreImage; // see _setCoreImage below +var PDFImage; // = coreImage.PDFImage; - // e_i' = min(max(e_i, 0), Size_i - 1) - var size_i = size[i]; - e = Math.min(Math.max(e, 0), size_i - 1); +var ColorSpace = (function ColorSpaceClosure() { + // Constructor should define this.numComps, this.defaultColor, this.name + function ColorSpace() { + error('should not call ColorSpace constructor'); + } - // Adjusting the cube: N and vertex sample index - var e0 = e < size_i - 1 ? Math.floor(e) : e - 1; // e1 = e0 + 1; - var n0 = e0 + 1 - e; // (e1 - e) / (e1 - e0); - var n1 = e - e0; // (e - e0) / (e1 - e0); - var offset0 = e0 * k; - var offset1 = offset0 + k; // e1 * k - for (j = 0; j < cubeVertices; j++) { - if (j & pos) { - cubeN[j] *= n1; - cubeVertex[j] += offset1; - } else { - cubeN[j] *= n0; - cubeVertex[j] += offset0; - } - } + ColorSpace.prototype = { + /** + * Converts the color value to the RGB color. The color components are + * located in the src array starting from the srcOffset. Returns the array + * of the rgb components, each value ranging from [0,255]. + */ + getRgb: function ColorSpace_getRgb(src, srcOffset) { + var rgb = new Uint8Array(3); + this.getRgbItem(src, srcOffset, rgb, 0); + return rgb; + }, + /** + * Converts the color value to the RGB color, similar to the getRgb method. + * The result placed into the dest array starting from the destOffset. + */ + getRgbItem: function ColorSpace_getRgbItem(src, srcOffset, + dest, destOffset) { + error('Should not call ColorSpace.getRgbItem'); + }, + /** + * Converts the specified number of the color values to the RGB colors. + * The colors are located in the src array starting from the srcOffset. + * The result is placed into the dest array starting from the destOffset. + * The src array items shall be in [0,2^bits) range, the dest array items + * will be in [0,255] range. alpha01 indicates how many alpha components + * there are in the dest array; it will be either 0 (RGB array) or 1 (RGBA + * array). + */ + getRgbBuffer: function ColorSpace_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + error('Should not call ColorSpace.getRgbBuffer'); + }, + /** + * Determines the number of bytes required to store the result of the + * conversion done by the getRgbBuffer method. As in getRgbBuffer, + * |alpha01| is either 0 (RGB output) or 1 (RGBA output). + */ + getOutputLength: function ColorSpace_getOutputLength(inputLength, + alpha01) { + error('Should not call ColorSpace.getOutputLength'); + }, + /** + * Returns true if source data will be equal the result/output data. + */ + isPassthrough: function ColorSpace_isPassthrough(bits) { + return false; + }, + /** + * Fills in the RGB colors in the destination buffer. alpha01 indicates + * how many alpha components there are in the dest array; it will be either + * 0 (RGB array) or 1 (RGBA array). + */ + fillRgb: function ColorSpace_fillRgb(dest, originalWidth, + originalHeight, width, height, + actualHeight, bpc, comps, alpha01) { + var count = originalWidth * originalHeight; + var rgbBuf = null; + var numComponentColors = 1 << bpc; + var needsResizing = originalHeight !== height || originalWidth !== width; + var i, ii; - k *= size_i; - pos <<= 1; + if (this.isPassthrough(bpc)) { + rgbBuf = comps; + } else if (this.numComps === 1 && count > numComponentColors && + this.name !== 'DeviceGray' && this.name !== 'DeviceRGB') { + // Optimization: create a color map when there is just one component and + // we are converting more colors than the size of the color map. We + // don't build the map if the colorspace is gray or rgb since those + // methods are faster than building a map. This mainly offers big speed + // ups for indexed and alternate colorspaces. + // + // TODO it may be worth while to cache the color map. While running + // testing I never hit a cache so I will leave that out for now (perhaps + // we are reparsing colorspaces too much?). + var allColors = bpc <= 8 ? new Uint8Array(numComponentColors) : + new Uint16Array(numComponentColors); + var key; + for (i = 0; i < numComponentColors; i++) { + allColors[i] = i; } + var colorMap = new Uint8Array(numComponentColors * 3); + this.getRgbBuffer(allColors, 0, numComponentColors, colorMap, 0, bpc, + /* alpha01 = */ 0); - for (j = 0; j < n; ++j) { - // Sum all cube vertices' samples portions - var rj = 0; - for (i = 0; i < cubeVertices; i++) { - rj += samples[cubeVertex[i] + j] * cubeN[i]; + var destPos, rgbPos; + if (!needsResizing) { + // Fill in the RGB values directly into |dest|. + destPos = 0; + for (i = 0; i < count; ++i) { + key = comps[i] * 3; + dest[destPos++] = colorMap[key]; + dest[destPos++] = colorMap[key + 1]; + dest[destPos++] = colorMap[key + 2]; + destPos += alpha01; + } + } else { + rgbBuf = new Uint8Array(count * 3); + rgbPos = 0; + for (i = 0; i < count; ++i) { + key = comps[i] * 3; + rgbBuf[rgbPos++] = colorMap[key]; + rgbBuf[rgbPos++] = colorMap[key + 1]; + rgbBuf[rgbPos++] = colorMap[key + 2]; } - - // r_j' = Interpolate(r_j, 0, 2^BitsPerSample - 1, - // Decode_2j, Decode_2j+1) - rj = interpolate(rj, 0, 1, decode[j][0], decode[j][1]); - - // y_j = min(max(r_j, range_2j), range_2j+1) - dest[destOffset + j] = Math.min(Math.max(rj, range[j][0]), - range[j][1]); } - }; - }, - - constructInterpolated: function PDFFunction_constructInterpolated(str, - dict) { - var c0 = dict.get('C0') || [0]; - var c1 = dict.get('C1') || [1]; - var n = dict.get('N'); - - if (!isArray(c0) || !isArray(c1)) { - error('Illegal dictionary for interpolated function'); + } else { + if (!needsResizing) { + // Fill in the RGB values directly into |dest|. + this.getRgbBuffer(comps, 0, width * actualHeight, dest, 0, bpc, + alpha01); + } else { + rgbBuf = new Uint8Array(count * 3); + this.getRgbBuffer(comps, 0, count, rgbBuf, 0, bpc, + /* alpha01 = */ 0); + } } - var length = c0.length; - var diff = []; - for (var i = 0; i < length; ++i) { - diff.push(c1[i] - c0[i]); + if (rgbBuf) { + if (needsResizing) { + PDFImage.resize(rgbBuf, bpc, 3, originalWidth, originalHeight, width, + height, dest, alpha01); + } else { + rgbPos = 0; + destPos = 0; + for (i = 0, ii = width * actualHeight; i < ii; i++) { + dest[destPos++] = rgbBuf[rgbPos++]; + dest[destPos++] = rgbBuf[rgbPos++]; + dest[destPos++] = rgbBuf[rgbPos++]; + destPos += alpha01; + } + } } - - return [CONSTRUCT_INTERPOLATED, c0, diff, n]; }, + /** + * True if the colorspace has components in the default range of [0, 1]. + * This should be true for all colorspaces except for lab color spaces + * which are [0,100], [-128, 127], [-128, 127]. + */ + usesZeroToOneRange: true + }; - constructInterpolatedFromIR: - function PDFFunction_constructInterpolatedFromIR(IR) { - var c0 = IR[1]; - var diff = IR[2]; - var n = IR[3]; - - var length = diff.length; + ColorSpace.parse = function ColorSpace_parse(cs, xref, res) { + var IR = ColorSpace.parseToIR(cs, xref, res); + if (IR instanceof AlternateCS) { + return IR; + } + return ColorSpace.fromIR(IR); + }; - return function constructInterpolatedFromIRResult(src, srcOffset, - dest, destOffset) { - var x = n === 1 ? src[srcOffset] : Math.pow(src[srcOffset], n); + ColorSpace.fromIR = function ColorSpace_fromIR(IR) { + var name = isArray(IR) ? IR[0] : IR; + var whitePoint, blackPoint, gamma; - for (var j = 0; j < length; ++j) { - dest[destOffset + j] = c0[j] + (x * diff[j]); + switch (name) { + case 'DeviceGrayCS': + return this.singletons.gray; + case 'DeviceRgbCS': + return this.singletons.rgb; + case 'DeviceCmykCS': + return this.singletons.cmyk; + case 'CalGrayCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + gamma = IR[3]; + return new CalGrayCS(whitePoint, blackPoint, gamma); + case 'CalRGBCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + gamma = IR[3]; + var matrix = IR[4]; + return new CalRGBCS(whitePoint, blackPoint, gamma, matrix); + case 'PatternCS': + var basePatternCS = IR[1]; + if (basePatternCS) { + basePatternCS = ColorSpace.fromIR(basePatternCS); } - }; - }, + return new PatternCS(basePatternCS); + case 'IndexedCS': + var baseIndexedCS = IR[1]; + var hiVal = IR[2]; + var lookup = IR[3]; + return new IndexedCS(ColorSpace.fromIR(baseIndexedCS), hiVal, lookup); + case 'AlternateCS': + var numComps = IR[1]; + var alt = IR[2]; + var tintFnIR = IR[3]; - constructStiched: function PDFFunction_constructStiched(fn, dict, xref) { - var domain = dict.get('Domain'); + return new AlternateCS(numComps, ColorSpace.fromIR(alt), + PDFFunction.fromIR(tintFnIR)); + case 'LabCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + var range = IR[3]; + return new LabCS(whitePoint, blackPoint, range); + default: + error('Unknown name ' + name); + } + return null; + }; - if (!domain) { - error('No domain'); + ColorSpace.parseToIR = function ColorSpace_parseToIR(cs, xref, res) { + if (isName(cs)) { + var colorSpaces = res.get('ColorSpace'); + if (isDict(colorSpaces)) { + var refcs = colorSpaces.get(cs.name); + if (refcs) { + cs = refcs; + } } + } - var inputSize = domain.length / 2; - if (inputSize !== 1) { - error('Bad domain for stiched function'); + cs = xref.fetchIfRef(cs); + var mode; + + if (isName(cs)) { + mode = cs.name; + this.mode = mode; + + switch (mode) { + case 'DeviceGray': + case 'G': + return 'DeviceGrayCS'; + case 'DeviceRGB': + case 'RGB': + return 'DeviceRgbCS'; + case 'DeviceCMYK': + case 'CMYK': + return 'DeviceCmykCS'; + case 'Pattern': + return ['PatternCS', null]; + default: + error('unrecognized colorspace ' + mode); } + } else if (isArray(cs)) { + mode = xref.fetchIfRef(cs[0]).name; + this.mode = mode; + var numComps, params, alt, whitePoint, blackPoint, gamma; - var fnRefs = dict.get('Functions'); - var fns = []; - for (var i = 0, ii = fnRefs.length; i < ii; ++i) { - fns.push(PDFFunction.getIR(xref, xref.fetchIfRef(fnRefs[i]))); + switch (mode) { + case 'DeviceGray': + case 'G': + return 'DeviceGrayCS'; + case 'DeviceRGB': + case 'RGB': + return 'DeviceRgbCS'; + case 'DeviceCMYK': + case 'CMYK': + return 'DeviceCmykCS'; + case 'CalGray': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + gamma = params.get('Gamma'); + return ['CalGrayCS', whitePoint, blackPoint, gamma]; + case 'CalRGB': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + gamma = params.get('Gamma'); + var matrix = params.get('Matrix'); + return ['CalRGBCS', whitePoint, blackPoint, gamma, matrix]; + case 'ICCBased': + var stream = xref.fetchIfRef(cs[1]); + var dict = stream.dict; + numComps = dict.get('N'); + alt = dict.get('Alternate'); + if (alt) { + var altIR = ColorSpace.parseToIR(alt, xref, res); + // Parse the /Alternate CS to ensure that the number of components + // are correct, and also (indirectly) that it is not a PatternCS. + var altCS = ColorSpace.fromIR(altIR); + if (altCS.numComps === numComps) { + return altIR; + } + warn('ICCBased color space: Ignoring incorrect /Alternate entry.'); + } + if (numComps === 1) { + return 'DeviceGrayCS'; + } else if (numComps === 3) { + return 'DeviceRgbCS'; + } else if (numComps === 4) { + return 'DeviceCmykCS'; + } + break; + case 'Pattern': + var basePatternCS = cs[1] || null; + if (basePatternCS) { + basePatternCS = ColorSpace.parseToIR(basePatternCS, xref, res); + } + return ['PatternCS', basePatternCS]; + case 'Indexed': + case 'I': + var baseIndexedCS = ColorSpace.parseToIR(cs[1], xref, res); + var hiVal = xref.fetchIfRef(cs[2]) + 1; + var lookup = xref.fetchIfRef(cs[3]); + if (isStream(lookup)) { + lookup = lookup.getBytes(); + } + return ['IndexedCS', baseIndexedCS, hiVal, lookup]; + case 'Separation': + case 'DeviceN': + var name = xref.fetchIfRef(cs[1]); + numComps = 1; + if (isName(name)) { + numComps = 1; + } else if (isArray(name)) { + numComps = name.length; + } + alt = ColorSpace.parseToIR(cs[2], xref, res); + var tintFnIR = PDFFunction.getIR(xref, xref.fetchIfRef(cs[3])); + return ['AlternateCS', numComps, alt, tintFnIR]; + case 'Lab': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + var range = params.get('Range'); + return ['LabCS', whitePoint, blackPoint, range]; + default: + error('unimplemented color space object "' + mode + '"'); } + } else { + error('unrecognized color space object: "' + cs + '"'); + } + return null; + }; + /** + * Checks if a decode map matches the default decode map for a color space. + * This handles the general decode maps where there are two values per + * component. e.g. [0, 1, 0, 1, 0, 1] for a RGB color. + * This does not handle Lab, Indexed, or Pattern decode maps since they are + * slightly different. + * @param {Array} decode Decode map (usually from an image). + * @param {Number} n Number of components the color space has. + */ + ColorSpace.isDefaultDecode = function ColorSpace_isDefaultDecode(decode, n) { + if (!isArray(decode)) { + return true; + } - var bounds = dict.get('Bounds'); - var encode = dict.get('Encode'); + if (n * 2 !== decode.length) { + warn('The decode map is not the correct length'); + return true; + } + for (var i = 0, ii = decode.length; i < ii; i += 2) { + if (decode[i] !== 0 || decode[i + 1] !== 1) { + return false; + } + } + return true; + }; - return [CONSTRUCT_STICHED, domain, bounds, encode, fns]; + ColorSpace.singletons = { + get gray() { + return shadow(this, 'gray', new DeviceGrayCS()); + }, + get rgb() { + return shadow(this, 'rgb', new DeviceRgbCS()); }, + get cmyk() { + return shadow(this, 'cmyk', new DeviceCmykCS()); + } + }; - constructStichedFromIR: function PDFFunction_constructStichedFromIR(IR) { - var domain = IR[1]; - var bounds = IR[2]; - var encode = IR[3]; - var fnsIR = IR[4]; - var fns = []; - var tmpBuf = new Float32Array(1); + return ColorSpace; +})(); - for (var i = 0, ii = fnsIR.length; i < ii; i++) { - fns.push(PDFFunction.fromIR(fnsIR[i])); - } +/** + * Alternate color space handles both Separation and DeviceN color spaces. A + * Separation color space is actually just a DeviceN with one color component. + * Both color spaces use a tinting function to convert colors to a base color + * space. + */ +var AlternateCS = (function AlternateCSClosure() { + function AlternateCS(numComps, base, tintFn) { + this.name = 'Alternate'; + this.numComps = numComps; + this.defaultColor = new Float32Array(numComps); + for (var i = 0; i < numComps; ++i) { + this.defaultColor[i] = 1; + } + this.base = base; + this.tintFn = tintFn; + this.tmpBuf = new Float32Array(base.numComps); + } - return function constructStichedFromIRResult(src, srcOffset, - dest, destOffset) { - var clip = function constructStichedFromIRClip(v, min, max) { - if (v > max) { - v = max; - } else if (v < min) { - v = min; - } - return v; - }; + AlternateCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function AlternateCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var tmpBuf = this.tmpBuf; + this.tintFn(src, srcOffset, tmpBuf, 0); + this.base.getRgbItem(tmpBuf, 0, dest, destOffset); + }, + getRgbBuffer: function AlternateCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var tintFn = this.tintFn; + var base = this.base; + var scale = 1 / ((1 << bits) - 1); + var baseNumComps = base.numComps; + var usesZeroToOneRange = base.usesZeroToOneRange; + var isPassthrough = (base.isPassthrough(8) || !usesZeroToOneRange) && + alpha01 === 0; + var pos = isPassthrough ? destOffset : 0; + var baseBuf = isPassthrough ? dest : new Uint8Array(baseNumComps * count); + var numComps = this.numComps; - // clip to domain - var v = clip(src[srcOffset], domain[0], domain[1]); - // calulate which bound the value is in - for (var i = 0, ii = bounds.length; i < ii; ++i) { - if (v < bounds[i]) { - break; + var scaled = new Float32Array(numComps); + var tinted = new Float32Array(baseNumComps); + var i, j; + if (usesZeroToOneRange) { + for (i = 0; i < count; i++) { + for (j = 0; j < numComps; j++) { + scaled[j] = src[srcOffset++] * scale; + } + tintFn(scaled, 0, tinted, 0); + for (j = 0; j < baseNumComps; j++) { + baseBuf[pos++] = tinted[j] * 255; } } - - // encode value into domain of function - var dmin = domain[0]; - if (i > 0) { - dmin = bounds[i - 1]; - } - var dmax = domain[1]; - if (i < bounds.length) { - dmax = bounds[i]; + } else { + for (i = 0; i < count; i++) { + for (j = 0; j < numComps; j++) { + scaled[j] = src[srcOffset++] * scale; + } + tintFn(scaled, 0, tinted, 0); + base.getRgbItem(tinted, 0, baseBuf, pos); + pos += baseNumComps; } + } + if (!isPassthrough) { + base.getRgbBuffer(baseBuf, 0, count, dest, destOffset, 8, alpha01); + } + }, + getOutputLength: function AlternateCS_getOutputLength(inputLength, + alpha01) { + return this.base.getOutputLength(inputLength * + this.base.numComps / this.numComps, + alpha01); + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function AlternateCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true + }; - var rmin = encode[2 * i]; - var rmax = encode[2 * i + 1]; + return AlternateCS; +})(); - // Prevent the value from becoming NaN as a result - // of division by zero (fixes issue6113.pdf). - tmpBuf[0] = dmin === dmax ? rmin : - rmin + (v - dmin) * (rmax - rmin) / (dmax - dmin); +var PatternCS = (function PatternCSClosure() { + function PatternCS(baseCS) { + this.name = 'Pattern'; + this.base = baseCS; + } + PatternCS.prototype = {}; - // call the appropriate function - fns[i](tmpBuf, 0, dest, destOffset); - }; - }, + return PatternCS; +})(); - constructPostScript: function PDFFunction_constructPostScript(fn, dict, - xref) { - var domain = dict.get('Domain'); - var range = dict.get('Range'); +var IndexedCS = (function IndexedCSClosure() { + function IndexedCS(base, highVal, lookup) { + this.name = 'Indexed'; + this.numComps = 1; + this.defaultColor = new Uint8Array([0]); + this.base = base; + this.highVal = highVal; - if (!domain) { - error('No domain.'); - } + var baseNumComps = base.numComps; + var length = baseNumComps * highVal; + var lookupArray; - if (!range) { - error('No range.'); + if (isStream(lookup)) { + lookupArray = new Uint8Array(length); + var bytes = lookup.getBytes(length); + lookupArray.set(bytes); + } else if (isString(lookup)) { + lookupArray = new Uint8Array(length); + for (var i = 0; i < length; ++i) { + lookupArray[i] = lookup.charCodeAt(i); } + } else if (lookup instanceof Uint8Array || lookup instanceof Array) { + lookupArray = lookup; + } else { + error('Unrecognized lookup table: ' + lookup); + } + this.lookup = lookupArray; + } - var lexer = new PostScriptLexer(fn); - var parser = new PostScriptParser(lexer); - var code = parser.parse(); - - return [CONSTRUCT_POSTSCRIPT, domain, range, code]; + IndexedCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function IndexedCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var numComps = this.base.numComps; + var start = src[srcOffset] * numComps; + this.base.getRgbItem(this.lookup, start, dest, destOffset); }, + getRgbBuffer: function IndexedCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var base = this.base; + var numComps = base.numComps; + var outputDelta = base.getOutputLength(numComps, alpha01); + var lookup = this.lookup; - constructPostScriptFromIR: function PDFFunction_constructPostScriptFromIR( - IR) { - var domain = IR[1]; - var range = IR[2]; - var code = IR[3]; - - var compiled = (new PostScriptCompiler()).compile(code, domain, range); - if (compiled) { - // Compiled function consists of simple expressions such as addition, - // subtraction, Math.max, and also contains 'var' and 'return' - // statements. See the generation in the PostScriptCompiler below. - /*jshint -W054 */ - return new Function('src', 'srcOffset', 'dest', 'destOffset', compiled); + for (var i = 0; i < count; ++i) { + var lookupPos = src[srcOffset++] * numComps; + base.getRgbBuffer(lookup, lookupPos, 1, dest, destOffset, 8, alpha01); + destOffset += outputDelta; } - - info('Unable to compile PS function'); - - var numOutputs = range.length >> 1; - var numInputs = domain.length >> 1; - var evaluator = new PostScriptEvaluator(code); - // Cache the values for a big speed up, the cache size is limited though - // since the number of possible values can be huge from a PS function. - var cache = Object.create(null); - // The MAX_CACHE_SIZE is set to ~4x the maximum number of distinct values - // seen in our tests. - var MAX_CACHE_SIZE = 2048 * 4; - var cache_available = MAX_CACHE_SIZE; - var tmpBuf = new Float32Array(numInputs); - - return function constructPostScriptFromIRResult(src, srcOffset, - dest, destOffset) { - var i, value; - var key = ''; - var input = tmpBuf; - for (i = 0; i < numInputs; i++) { - value = src[srcOffset + i]; - input[i] = value; - key += value + '_'; - } - - var cachedValue = cache[key]; - if (cachedValue !== undefined) { - dest.set(cachedValue, destOffset); - return; - } - - var output = new Float32Array(numOutputs); - var stack = evaluator.execute(input); - var stackIndex = stack.length - numOutputs; - for (i = 0; i < numOutputs; i++) { - value = stack[stackIndex + i]; - var bound = range[i * 2]; - if (value < bound) { - value = bound; - } else { - bound = range[i * 2 +1]; - if (value > bound) { - value = bound; - } - } - output[i] = value; - } - if (cache_available > 0) { - cache_available--; - cache[key] = output; - } - dest.set(output, destOffset); - }; - } + }, + getOutputLength: function IndexedCS_getOutputLength(inputLength, alpha01) { + return this.base.getOutputLength(inputLength * this.base.numComps, + alpha01); + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function IndexedCS_isDefaultDecode(decodeMap) { + // indexed color maps shouldn't be changed + return true; + }, + usesZeroToOneRange: true }; + return IndexedCS; })(); -function isPDFFunction(v) { - var fnDict; - if (typeof v !== 'object') { - return false; - } else if (isDict(v)) { - fnDict = v; - } else if (isStream(v)) { - fnDict = v.dict; - } else { - return false; - } - return fnDict.has('FunctionType'); -} - -var PostScriptStack = (function PostScriptStackClosure() { - var MAX_STACK_SIZE = 100; - function PostScriptStack(initialStack) { - this.stack = !initialStack ? [] : - Array.prototype.slice.call(initialStack, 0); +var DeviceGrayCS = (function DeviceGrayCSClosure() { + function DeviceGrayCS() { + this.name = 'DeviceGray'; + this.numComps = 1; + this.defaultColor = new Float32Array([0]); } - PostScriptStack.prototype = { - push: function PostScriptStack_push(value) { - if (this.stack.length >= MAX_STACK_SIZE) { - error('PostScript function stack overflow.'); - } - this.stack.push(value); + DeviceGrayCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceGrayCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var c = (src[srcOffset] * 255) | 0; + c = c < 0 ? 0 : c > 255 ? 255 : c; + dest[destOffset] = dest[destOffset + 1] = dest[destOffset + 2] = c; }, - pop: function PostScriptStack_pop() { - if (this.stack.length <= 0) { - error('PostScript function stack underflow.'); + getRgbBuffer: function DeviceGrayCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 255 / ((1 << bits) - 1); + var j = srcOffset, q = destOffset; + for (var i = 0; i < count; ++i) { + var c = (scale * src[j++]) | 0; + dest[q++] = c; + dest[q++] = c; + dest[q++] = c; + q += alpha01; } - return this.stack.pop(); }, - copy: function PostScriptStack_copy(n) { - if (this.stack.length + n >= MAX_STACK_SIZE) { - error('PostScript function stack overflow.'); - } - var stack = this.stack; - for (var i = stack.length - n, j = n - 1; j >= 0; j--, i++) { - stack.push(stack[i]); - } + getOutputLength: function DeviceGrayCS_getOutputLength(inputLength, + alpha01) { + return inputLength * (3 + alpha01); }, - index: function PostScriptStack_index(n) { - this.push(this.stack[this.stack.length - n - 1]); + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceGrayCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - // rotate the last n stack elements p times - roll: function PostScriptStack_roll(n, p) { - var stack = this.stack; - var l = stack.length - n; - var r = stack.length - 1, c = l + (p - Math.floor(p / n) * n), i, j, t; - for (i = l, j = r; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - for (i = l, j = c - 1; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - for (i = c, j = r; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - } + usesZeroToOneRange: true }; - return PostScriptStack; + return DeviceGrayCS; })(); -var PostScriptEvaluator = (function PostScriptEvaluatorClosure() { - function PostScriptEvaluator(operators) { - this.operators = operators; - } - PostScriptEvaluator.prototype = { - execute: function PostScriptEvaluator_execute(initialStack) { - var stack = new PostScriptStack(initialStack); - var counter = 0; - var operators = this.operators; - var length = operators.length; - var operator, a, b; - while (counter < length) { - operator = operators[counter++]; - if (typeof operator === 'number') { - // Operator is really an operand and should be pushed to the stack. - stack.push(operator); - continue; - } - switch (operator) { - // non standard ps operators - case 'jz': // jump if false - b = stack.pop(); - a = stack.pop(); - if (!a) { - counter = b; - } - break; - case 'j': // jump - a = stack.pop(); - counter = a; - break; - // all ps operators in alphabetical order (excluding if/ifelse) - case 'abs': - a = stack.pop(); - stack.push(Math.abs(a)); - break; - case 'add': - b = stack.pop(); - a = stack.pop(); - stack.push(a + b); - break; - case 'and': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a && b); - } else { - stack.push(a & b); - } - break; - case 'atan': - a = stack.pop(); - stack.push(Math.atan(a)); - break; - case 'bitshift': - b = stack.pop(); - a = stack.pop(); - if (a > 0) { - stack.push(a << b); - } else { - stack.push(a >> b); - } - break; - case 'ceiling': - a = stack.pop(); - stack.push(Math.ceil(a)); - break; - case 'copy': - a = stack.pop(); - stack.copy(a); - break; - case 'cos': - a = stack.pop(); - stack.push(Math.cos(a)); - break; - case 'cvi': - a = stack.pop() | 0; - stack.push(a); - break; - case 'cvr': - // noop - break; - case 'div': - b = stack.pop(); - a = stack.pop(); - stack.push(a / b); - break; - case 'dup': - stack.copy(1); - break; - case 'eq': - b = stack.pop(); - a = stack.pop(); - stack.push(a === b); - break; - case 'exch': - stack.roll(2, 1); - break; - case 'exp': - b = stack.pop(); - a = stack.pop(); - stack.push(Math.pow(a, b)); - break; - case 'false': - stack.push(false); - break; - case 'floor': - a = stack.pop(); - stack.push(Math.floor(a)); - break; - case 'ge': - b = stack.pop(); - a = stack.pop(); - stack.push(a >= b); - break; - case 'gt': - b = stack.pop(); - a = stack.pop(); - stack.push(a > b); - break; - case 'idiv': - b = stack.pop(); - a = stack.pop(); - stack.push((a / b) | 0); - break; - case 'index': - a = stack.pop(); - stack.index(a); - break; - case 'le': - b = stack.pop(); - a = stack.pop(); - stack.push(a <= b); - break; - case 'ln': - a = stack.pop(); - stack.push(Math.log(a)); - break; - case 'log': - a = stack.pop(); - stack.push(Math.log(a) / Math.LN10); - break; - case 'lt': - b = stack.pop(); - a = stack.pop(); - stack.push(a < b); - break; - case 'mod': - b = stack.pop(); - a = stack.pop(); - stack.push(a % b); - break; - case 'mul': - b = stack.pop(); - a = stack.pop(); - stack.push(a * b); - break; - case 'ne': - b = stack.pop(); - a = stack.pop(); - stack.push(a !== b); - break; - case 'neg': - a = stack.pop(); - stack.push(-a); - break; - case 'not': - a = stack.pop(); - if (isBool(a)) { - stack.push(!a); - } else { - stack.push(~a); - } - break; - case 'or': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a || b); - } else { - stack.push(a | b); - } - break; - case 'pop': - stack.pop(); - break; - case 'roll': - b = stack.pop(); - a = stack.pop(); - stack.roll(a, b); - break; - case 'round': - a = stack.pop(); - stack.push(Math.round(a)); - break; - case 'sin': - a = stack.pop(); - stack.push(Math.sin(a)); - break; - case 'sqrt': - a = stack.pop(); - stack.push(Math.sqrt(a)); - break; - case 'sub': - b = stack.pop(); - a = stack.pop(); - stack.push(a - b); - break; - case 'true': - stack.push(true); - break; - case 'truncate': - a = stack.pop(); - a = a < 0 ? Math.ceil(a) : Math.floor(a); - stack.push(a); - break; - case 'xor': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a !== b); - } else { - stack.push(a ^ b); - } - break; - default: - error('Unknown operator ' + operator); - break; - } +var DeviceRgbCS = (function DeviceRgbCSClosure() { + function DeviceRgbCS() { + this.name = 'DeviceRGB'; + this.numComps = 3; + this.defaultColor = new Float32Array([0, 0, 0]); + } + DeviceRgbCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceRgbCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var r = (src[srcOffset] * 255) | 0; + var g = (src[srcOffset + 1] * 255) | 0; + var b = (src[srcOffset + 2] * 255) | 0; + dest[destOffset] = r < 0 ? 0 : r > 255 ? 255 : r; + dest[destOffset + 1] = g < 0 ? 0 : g > 255 ? 255 : g; + dest[destOffset + 2] = b < 0 ? 0 : b > 255 ? 255 : b; + }, + getRgbBuffer: function DeviceRgbCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + if (bits === 8 && alpha01 === 0) { + dest.set(src.subarray(srcOffset, srcOffset + count * 3), destOffset); + return; } - return stack.stack; - } + var scale = 255 / ((1 << bits) - 1); + var j = srcOffset, q = destOffset; + for (var i = 0; i < count; ++i) { + dest[q++] = (scale * src[j++]) | 0; + dest[q++] = (scale * src[j++]) | 0; + dest[q++] = (scale * src[j++]) | 0; + q += alpha01; + } + }, + getOutputLength: function DeviceRgbCS_getOutputLength(inputLength, + alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; + }, + isPassthrough: function DeviceRgbCS_isPassthrough(bits) { + return bits === 8; + }, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceRgbCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true }; - return PostScriptEvaluator; + return DeviceRgbCS; })(); -// Most of the PDFs functions consist of simple operations such as: -// roll, exch, sub, cvr, pop, index, dup, mul, if, gt, add. -// -// We can compile most of such programs, and at the same moment, we can -// optimize some expressions using basic math properties. Keeping track of -// min/max values will allow us to avoid extra Math.min/Math.max calls. -var PostScriptCompiler = (function PostScriptCompilerClosure() { - function AstNode(type) { - this.type = type; - } - AstNode.prototype.visit = function (visitor) { - throw new Error('abstract method'); - }; +var DeviceCmykCS = (function DeviceCmykCSClosure() { + // The coefficients below was found using numerical analysis: the method of + // steepest descent for the sum((f_i - color_value_i)^2) for r/g/b colors, + // where color_value is the tabular value from the table of sampled RGB colors + // from CMYK US Web Coated (SWOP) colorspace, and f_i is the corresponding + // CMYK color conversion using the estimation below: + // f(A, B,.. N) = Acc+Bcm+Ccy+Dck+c+Fmm+Gmy+Hmk+Im+Jyy+Kyk+Ly+Mkk+Nk+255 + function convertToRgb(src, srcOffset, srcScale, dest, destOffset) { + var c = src[srcOffset + 0] * srcScale; + var m = src[srcOffset + 1] * srcScale; + var y = src[srcOffset + 2] * srcScale; + var k = src[srcOffset + 3] * srcScale; - function AstArgument(index, min, max) { - AstNode.call(this, 'args'); - this.index = index; - this.min = min; - this.max = max; - } - AstArgument.prototype = Object.create(AstNode.prototype); - AstArgument.prototype.visit = function (visitor) { - visitor.visitArgument(this); - }; + var r = + (c * (-4.387332384609988 * c + 54.48615194189176 * m + + 18.82290502165302 * y + 212.25662451639585 * k + + -285.2331026137004) + + m * (1.7149763477362134 * m - 5.6096736904047315 * y + + -17.873870861415444 * k - 5.497006427196366) + + y * (-2.5217340131683033 * y - 21.248923337353073 * k + + 17.5119270841813) + + k * (-21.86122147463605 * k - 189.48180835922747) + 255) | 0; + var g = + (c * (8.841041422036149 * c + 60.118027045597366 * m + + 6.871425592049007 * y + 31.159100130055922 * k + + -79.2970844816548) + + m * (-15.310361306967817 * m + 17.575251261109482 * y + + 131.35250912493976 * k - 190.9453302588951) + + y * (4.444339102852739 * y + 9.8632861493405 * k - 24.86741582555878) + + k * (-20.737325471181034 * k - 187.80453709719578) + 255) | 0; + var b = + (c * (0.8842522430003296 * c + 8.078677503112928 * m + + 30.89978309703729 * y - 0.23883238689178934 * k + + -14.183576799673286) + + m * (10.49593273432072 * m + 63.02378494754052 * y + + 50.606957656360734 * k - 112.23884253719248) + + y * (0.03296041114873217 * y + 115.60384449646641 * k + + -193.58209356861505) + + k * (-22.33816807309886 * k - 180.12613974708367) + 255) | 0; - function AstLiteral(number) { - AstNode.call(this, 'literal'); - this.number = number; - this.min = number; - this.max = number; + dest[destOffset] = r > 255 ? 255 : r < 0 ? 0 : r; + dest[destOffset + 1] = g > 255 ? 255 : g < 0 ? 0 : g; + dest[destOffset + 2] = b > 255 ? 255 : b < 0 ? 0 : b; } - AstLiteral.prototype = Object.create(AstNode.prototype); - AstLiteral.prototype.visit = function (visitor) { - visitor.visitLiteral(this); - }; - function AstBinaryOperation(op, arg1, arg2, min, max) { - AstNode.call(this, 'binary'); - this.op = op; - this.arg1 = arg1; - this.arg2 = arg2; - this.min = min; - this.max = max; + function DeviceCmykCS() { + this.name = 'DeviceCMYK'; + this.numComps = 4; + this.defaultColor = new Float32Array([0, 0, 0, 1]); } - AstBinaryOperation.prototype = Object.create(AstNode.prototype); - AstBinaryOperation.prototype.visit = function (visitor) { - visitor.visitBinaryOperation(this); + DeviceCmykCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceCmykCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(src, srcOffset, 1, dest, destOffset); + }, + getRgbBuffer: function DeviceCmykCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + for (var i = 0; i < count; i++) { + convertToRgb(src, srcOffset, scale, dest, destOffset); + srcOffset += 4; + destOffset += 3 + alpha01; + } + }, + getOutputLength: function DeviceCmykCS_getOutputLength(inputLength, + alpha01) { + return (inputLength / 4 * (3 + alpha01)) | 0; + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceCmykCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true }; - function AstMin(arg, max) { - AstNode.call(this, 'max'); - this.arg = arg; - this.min = arg.min; - this.max = max; - } - AstMin.prototype = Object.create(AstNode.prototype); - AstMin.prototype.visit = function (visitor) { - visitor.visitMin(this); - }; + return DeviceCmykCS; +})(); - function AstVariable(index, min, max) { - AstNode.call(this, 'var'); - this.index = index; - this.min = min; - this.max = max; - } - AstVariable.prototype = Object.create(AstNode.prototype); - AstVariable.prototype.visit = function (visitor) { - visitor.visitVariable(this); - }; +// +// CalGrayCS: Based on "PDF Reference, Sixth Ed", p.245 +// +var CalGrayCS = (function CalGrayCSClosure() { + function CalGrayCS(whitePoint, blackPoint, gamma) { + this.name = 'CalGray'; + this.numComps = 1; + this.defaultColor = new Float32Array([0]); - function AstVariableDefinition(variable, arg) { - AstNode.call(this, 'definition'); - this.variable = variable; - this.arg = arg; + if (!whitePoint) { + error('WhitePoint missing - required for color space CalGray'); + } + blackPoint = blackPoint || [0, 0, 0]; + gamma = gamma || 1; + + // Translate arguments to spec variables. + this.XW = whitePoint[0]; + this.YW = whitePoint[1]; + this.ZW = whitePoint[2]; + + this.XB = blackPoint[0]; + this.YB = blackPoint[1]; + this.ZB = blackPoint[2]; + + this.G = gamma; + + // Validate variables as per spec. + if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { + error('Invalid WhitePoint components for ' + this.name + + ', no fallback available'); + } + + if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { + info('Invalid BlackPoint for ' + this.name + ', falling back to default'); + this.XB = this.YB = this.ZB = 0; + } + + if (this.XB !== 0 || this.YB !== 0 || this.ZB !== 0) { + warn(this.name + ', BlackPoint: XB: ' + this.XB + ', YB: ' + this.YB + + ', ZB: ' + this.ZB + ', only default values are supported.'); + } + + if (this.G < 1) { + info('Invalid Gamma: ' + this.G + ' for ' + this.name + + ', falling back to default'); + this.G = 1; + } } - AstVariableDefinition.prototype = Object.create(AstNode.prototype); - AstVariableDefinition.prototype.visit = function (visitor) { - visitor.visitVariableDefinition(this); - }; - function ExpressionBuilderVisitor() { - this.parts = []; + function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { + // A represents a gray component of a calibrated gray space. + // A <---> AG in the spec + var A = src[srcOffset] * scale; + var AG = Math.pow(A, cs.G); + + // Computes L as per spec. ( = cs.YW * AG ) + // Except if other than default BlackPoint values are used. + var L = cs.YW * AG; + // http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html, Ch 4. + // Convert values to rgb range [0, 255]. + var val = Math.max(295.8 * Math.pow(L, 0.333333333333333333) - 40.8, 0) | 0; + dest[destOffset] = val; + dest[destOffset + 1] = val; + dest[destOffset + 2] = val; } - ExpressionBuilderVisitor.prototype = { - visitArgument: function (arg) { - this.parts.push('Math.max(', arg.min, ', Math.min(', - arg.max, ', src[srcOffset + ', arg.index, ']))'); - }, - visitVariable: function (variable) { - this.parts.push('v', variable.index); - }, - visitLiteral: function (literal) { - this.parts.push(literal.number); + + CalGrayCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function CalGrayCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(this, src, srcOffset, dest, destOffset, 1); }, - visitBinaryOperation: function (operation) { - this.parts.push('('); - operation.arg1.visit(this); - this.parts.push(' ', operation.op, ' '); - operation.arg2.visit(this); - this.parts.push(')'); + getRgbBuffer: function CalGrayCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + + for (var i = 0; i < count; ++i) { + convertToRgb(this, src, srcOffset, dest, destOffset, scale); + srcOffset += 1; + destOffset += 3 + alpha01; + } }, - visitVariableDefinition: function (definition) { - this.parts.push('var '); - definition.variable.visit(this); - this.parts.push(' = '); - definition.arg.visit(this); - this.parts.push(';'); + getOutputLength: function CalGrayCS_getOutputLength(inputLength, alpha01) { + return inputLength * (3 + alpha01); }, - visitMin: function (max) { - this.parts.push('Math.min('); - max.arg.visit(this); - this.parts.push(', ', max.max, ')'); + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function CalGrayCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - toString: function () { - return this.parts.join(''); - } + usesZeroToOneRange: true }; + return CalGrayCS; +})(); - function buildAddOperation(num1, num2) { - if (num2.type === 'literal' && num2.number === 0) { - // optimization: second operand is 0 - return num1; +// +// CalRGBCS: Based on "PDF Reference, Sixth Ed", p.247 +// +var CalRGBCS = (function CalRGBCSClosure() { + + // See http://www.brucelindbloom.com/index.html?Eqn_ChromAdapt.html for these + // matrices. + var BRADFORD_SCALE_MATRIX = new Float32Array([ + 0.8951, 0.2664, -0.1614, + -0.7502, 1.7135, 0.0367, + 0.0389, -0.0685, 1.0296]); + + var BRADFORD_SCALE_INVERSE_MATRIX = new Float32Array([ + 0.9869929, -0.1470543, 0.1599627, + 0.4323053, 0.5183603, 0.0492912, + -0.0085287, 0.0400428, 0.9684867]); + + // See http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html. + var SRGB_D65_XYZ_TO_RGB_MATRIX = new Float32Array([ + 3.2404542, -1.5371385, -0.4985314, + -0.9692660, 1.8760108, 0.0415560, + 0.0556434, -0.2040259, 1.0572252]); + + var FLAT_WHITEPOINT_MATRIX = new Float32Array([1, 1, 1]); + + var tempNormalizeMatrix = new Float32Array(3); + var tempConvertMatrix1 = new Float32Array(3); + var tempConvertMatrix2 = new Float32Array(3); + + var DECODE_L_CONSTANT = Math.pow(((8 + 16) / 116), 3) / 8.0; + + function CalRGBCS(whitePoint, blackPoint, gamma, matrix) { + this.name = 'CalRGB'; + this.numComps = 3; + this.defaultColor = new Float32Array(3); + + if (!whitePoint) { + error('WhitePoint missing - required for color space CalRGB'); } - if (num1.type === 'literal' && num1.number === 0) { - // optimization: first operand is 0 - return num2; + blackPoint = blackPoint || new Float32Array(3); + gamma = gamma || new Float32Array([1, 1, 1]); + matrix = matrix || new Float32Array([1, 0, 0, 0, 1, 0, 0, 0, 1]); + + // Translate arguments to spec variables. + var XW = whitePoint[0]; + var YW = whitePoint[1]; + var ZW = whitePoint[2]; + this.whitePoint = whitePoint; + + var XB = blackPoint[0]; + var YB = blackPoint[1]; + var ZB = blackPoint[2]; + this.blackPoint = blackPoint; + + this.GR = gamma[0]; + this.GG = gamma[1]; + this.GB = gamma[2]; + + this.MXA = matrix[0]; + this.MYA = matrix[1]; + this.MZA = matrix[2]; + this.MXB = matrix[3]; + this.MYB = matrix[4]; + this.MZB = matrix[5]; + this.MXC = matrix[6]; + this.MYC = matrix[7]; + this.MZC = matrix[8]; + + // Validate variables as per spec. + if (XW < 0 || ZW < 0 || YW !== 1) { + error('Invalid WhitePoint components for ' + this.name + + ', no fallback available'); } - if (num2.type === 'literal' && num1.type === 'literal') { - // optimization: operands operand are literals - return new AstLiteral(num1.number + num2.number); + + if (XB < 0 || YB < 0 || ZB < 0) { + info('Invalid BlackPoint for ' + this.name + ' [' + XB + ', ' + YB + + ', ' + ZB + '], falling back to default'); + this.blackPoint = new Float32Array(3); } - return new AstBinaryOperation('+', num1, num2, - num1.min + num2.min, num1.max + num2.max); - } - function buildMulOperation(num1, num2) { - if (num2.type === 'literal') { - // optimization: second operands is a literal... - if (num2.number === 0) { - return new AstLiteral(0); // and it's 0 - } else if (num2.number === 1) { - return num1; // and it's 1 - } else if (num1.type === 'literal') { - // ... and first operands is a literal too - return new AstLiteral(num1.number * num2.number); - } + if (this.GR < 0 || this.GG < 0 || this.GB < 0) { + info('Invalid Gamma [' + this.GR + ', ' + this.GG + ', ' + this.GB + + '] for ' + this.name + ', falling back to default'); + this.GR = this.GG = this.GB = 1; } - if (num1.type === 'literal') { - // optimization: first operands is a literal... - if (num1.number === 0) { - return new AstLiteral(0); // and it's 0 - } else if (num1.number === 1) { - return num2; // and it's 1 - } + + if (this.MXA < 0 || this.MYA < 0 || this.MZA < 0 || + this.MXB < 0 || this.MYB < 0 || this.MZB < 0 || + this.MXC < 0 || this.MYC < 0 || this.MZC < 0) { + info('Invalid Matrix for ' + this.name + ' [' + + this.MXA + ', ' + this.MYA + ', ' + this.MZA + + this.MXB + ', ' + this.MYB + ', ' + this.MZB + + this.MXC + ', ' + this.MYC + ', ' + this.MZC + + '], falling back to default'); + this.MXA = this.MYB = this.MZC = 1; + this.MXB = this.MYA = this.MZA = this.MXC = this.MYC = this.MZB = 0; } - var min = Math.min(num1.min * num2.min, num1.min * num2.max, - num1.max * num2.min, num1.max * num2.max); - var max = Math.max(num1.min * num2.min, num1.min * num2.max, - num1.max * num2.min, num1.max * num2.max); - return new AstBinaryOperation('*', num1, num2, min, max); } - function buildSubOperation(num1, num2) { - if (num2.type === 'literal') { - // optimization: second operands is a literal... - if (num2.number === 0) { - return num1; // ... and it's 0 - } else if (num1.type === 'literal') { - // ... and first operands is a literal too - return new AstLiteral(num1.number - num2.number); - } + function matrixProduct(a, b, result) { + result[0] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; + result[1] = a[3] * b[0] + a[4] * b[1] + a[5] * b[2]; + result[2] = a[6] * b[0] + a[7] * b[1] + a[8] * b[2]; + } + + function convertToFlat(sourceWhitePoint, LMS, result) { + result[0] = LMS[0] * 1 / sourceWhitePoint[0]; + result[1] = LMS[1] * 1 / sourceWhitePoint[1]; + result[2] = LMS[2] * 1 / sourceWhitePoint[2]; + } + + function convertToD65(sourceWhitePoint, LMS, result) { + var D65X = 0.95047; + var D65Y = 1; + var D65Z = 1.08883; + + result[0] = LMS[0] * D65X / sourceWhitePoint[0]; + result[1] = LMS[1] * D65Y / sourceWhitePoint[1]; + result[2] = LMS[2] * D65Z / sourceWhitePoint[2]; + } + + function sRGBTransferFunction(color) { + // See http://en.wikipedia.org/wiki/SRGB. + if (color <= 0.0031308){ + return adjustToRange(0, 1, 12.92 * color); } - if (num2.type === 'binary' && num2.op === '-' && - num1.type === 'literal' && num1.number === 1 && - num2.arg1.type === 'literal' && num2.arg1.number === 1) { - // optimization for case: 1 - (1 - x) - return num2.arg2; + + return adjustToRange(0, 1, (1 + 0.055) * Math.pow(color, 1 / 2.4) - 0.055); + } + + function adjustToRange(min, max, value) { + return Math.max(min, Math.min(max, value)); + } + + function decodeL(L) { + if (L < 0) { + return -decodeL(-L); } - return new AstBinaryOperation('-', num1, num2, - num1.min - num2.max, num1.max - num2.min); + + if (L > 8.0) { + return Math.pow(((L + 16) / 116), 3); + } + + return L * DECODE_L_CONSTANT; + } + + function compensateBlackPoint(sourceBlackPoint, XYZ_Flat, result) { + + // In case the blackPoint is already the default blackPoint then there is + // no need to do compensation. + if (sourceBlackPoint[0] === 0 && + sourceBlackPoint[1] === 0 && + sourceBlackPoint[2] === 0) { + result[0] = XYZ_Flat[0]; + result[1] = XYZ_Flat[1]; + result[2] = XYZ_Flat[2]; + return; + } + + // For the blackPoint calculation details, please see + // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ + // AdobeBPC.pdf. + // The destination blackPoint is the default blackPoint [0, 0, 0]. + var zeroDecodeL = decodeL(0); + + var X_DST = zeroDecodeL; + var X_SRC = decodeL(sourceBlackPoint[0]); + + var Y_DST = zeroDecodeL; + var Y_SRC = decodeL(sourceBlackPoint[1]); + + var Z_DST = zeroDecodeL; + var Z_SRC = decodeL(sourceBlackPoint[2]); + + var X_Scale = (1 - X_DST) / (1 - X_SRC); + var X_Offset = 1 - X_Scale; + + var Y_Scale = (1 - Y_DST) / (1 - Y_SRC); + var Y_Offset = 1 - Y_Scale; + + var Z_Scale = (1 - Z_DST) / (1 - Z_SRC); + var Z_Offset = 1 - Z_Scale; + + result[0] = XYZ_Flat[0] * X_Scale + X_Offset; + result[1] = XYZ_Flat[1] * Y_Scale + Y_Offset; + result[2] = XYZ_Flat[2] * Z_Scale + Z_Offset; } - function buildMinOperation(num1, max) { - if (num1.min >= max) { - // optimization: num1 min value is not less than required max - return new AstLiteral(max); // just returning max - } else if (num1.max <= max) { - // optimization: num1 max value is not greater than required max - return num1; // just returning an argument + function normalizeWhitePointToFlat(sourceWhitePoint, XYZ_In, result) { + + // In case the whitePoint is already flat then there is no need to do + // normalization. + if (sourceWhitePoint[0] === 1 && sourceWhitePoint[2] === 1) { + result[0] = XYZ_In[0]; + result[1] = XYZ_In[1]; + result[2] = XYZ_In[2]; + return; } - return new AstMin(num1, max); + + var LMS = result; + matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + + var LMS_Flat = tempNormalizeMatrix; + convertToFlat(sourceWhitePoint, LMS, LMS_Flat); + + matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_Flat, result); } - function PostScriptCompiler() {} - PostScriptCompiler.prototype = { - compile: function PostScriptCompiler_compile(code, domain, range) { - var stack = []; - var i, ii; - var instructions = []; - var inputSize = domain.length >> 1, outputSize = range.length >> 1; - var lastRegister = 0; - var n, j; - var num1, num2, ast1, ast2, tmpVar, item; - for (i = 0; i < inputSize; i++) { - stack.push(new AstArgument(i, domain[i * 2], domain[i * 2 + 1])); - } + function normalizeWhitePointToD65(sourceWhitePoint, XYZ_In, result) { - for (i = 0, ii = code.length; i < ii; i++) { - item = code[i]; - if (typeof item === 'number') { - stack.push(new AstLiteral(item)); - continue; - } + var LMS = result; + matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); - switch (item) { - case 'add': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildAddOperation(num1, num2)); - break; - case 'cvr': - if (stack.length < 1) { - return null; - } - break; - case 'mul': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildMulOperation(num1, num2)); - break; - case 'sub': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildSubOperation(num1, num2)); - break; - case 'exch': - if (stack.length < 2) { - return null; - } - ast1 = stack.pop(); ast2 = stack.pop(); - stack.push(ast1, ast2); - break; - case 'pop': - if (stack.length < 1) { - return null; - } - stack.pop(); - break; - case 'index': - if (stack.length < 1) { - return null; - } - num1 = stack.pop(); - if (num1.type !== 'literal') { - return null; - } - n = num1.number; - if (n < 0 || (n|0) !== n || stack.length < n) { - return null; - } - ast1 = stack[stack.length - n - 1]; - if (ast1.type === 'literal' || ast1.type === 'var') { - stack.push(ast1); - break; - } - tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); - stack[stack.length - n - 1] = tmpVar; - stack.push(tmpVar); - instructions.push(new AstVariableDefinition(tmpVar, ast1)); - break; - case 'dup': - if (stack.length < 1) { - return null; - } - if (typeof code[i + 1] === 'number' && code[i + 2] === 'gt' && - code[i + 3] === i + 7 && code[i + 4] === 'jz' && - code[i + 5] === 'pop' && code[i + 6] === code[i + 1]) { - // special case of the commands sequence for the min operation - num1 = stack.pop(); - stack.push(buildMinOperation(num1, code[i + 1])); - i += 6; - break; - } - ast1 = stack[stack.length - 1]; - if (ast1.type === 'literal' || ast1.type === 'var') { - // we don't have to save into intermediate variable a literal or - // variable. - stack.push(ast1); - break; - } - tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); - stack[stack.length - 1] = tmpVar; - stack.push(tmpVar); - instructions.push(new AstVariableDefinition(tmpVar, ast1)); - break; - case 'roll': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - if (num2.type !== 'literal' || num1.type !== 'literal') { - // both roll operands must be numbers - return null; - } - j = num2.number; - n = num1.number; - if (n <= 0 || (n|0) !== n || (j|0) !== j || stack.length < n) { - // ... and integers - return null; - } - j = ((j % n) + n) % n; - if (j === 0) { - break; // just skipping -- there are nothing to rotate - } - Array.prototype.push.apply(stack, - stack.splice(stack.length - n, n - j)); - break; - default: - return null; // unsupported operator - } - } + var LMS_D65 = tempNormalizeMatrix; + convertToD65(sourceWhitePoint, LMS, LMS_D65); - if (stack.length !== outputSize) { - return null; - } + matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_D65, result); + } - var result = []; - instructions.forEach(function (instruction) { - var statementBuilder = new ExpressionBuilderVisitor(); - instruction.visit(statementBuilder); - result.push(statementBuilder.toString()); - }); - stack.forEach(function (expr, i) { - var statementBuilder = new ExpressionBuilderVisitor(); - expr.visit(statementBuilder); - var min = range[i * 2], max = range[i * 2 + 1]; - var out = [statementBuilder.toString()]; - if (min > expr.min) { - out.unshift('Math.max(', min, ', '); - out.push(')'); - } - if (max < expr.max) { - out.unshift('Math.min(', max, ', '); - out.push(')'); - } - out.unshift('dest[destOffset + ', i, '] = '); - out.push(';'); - result.push(out.join('')); - }); - return result.join('\n'); - } - }; + function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { + // A, B and C represent a red, green and blue components of a calibrated + // rgb space. + var A = adjustToRange(0, 1, src[srcOffset] * scale); + var B = adjustToRange(0, 1, src[srcOffset + 1] * scale); + var C = adjustToRange(0, 1, src[srcOffset + 2] * scale); - return PostScriptCompiler; -})(); + // A <---> AGR in the spec + // B <---> BGG in the spec + // C <---> CGB in the spec + var AGR = Math.pow(A, cs.GR); + var BGG = Math.pow(B, cs.GG); + var CGB = Math.pow(C, cs.GB); -exports.isPDFFunction = isPDFFunction; -exports.PDFFunction = PDFFunction; -exports.PostScriptEvaluator = PostScriptEvaluator; -exports.PostScriptCompiler = PostScriptCompiler; -})); + // Computes intermediate variables L, M, N as per spec. + // To decode X, Y, Z values map L, M, N directly to them. + var X = cs.MXA * AGR + cs.MXB * BGG + cs.MXC * CGB; + var Y = cs.MYA * AGR + cs.MYB * BGG + cs.MYC * CGB; + var Z = cs.MZA * AGR + cs.MZB * BGG + cs.MZC * CGB; + // The following calculations are based on this document: + // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ + // AdobeBPC.pdf. + var XYZ = tempConvertMatrix1; + XYZ[0] = X; + XYZ[1] = Y; + XYZ[2] = Z; + var XYZ_Flat = tempConvertMatrix2; -(function (root, factory) { - { - factory((root.pdfjsCoreColorSpace = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreFunction, root.pdfjsCoreStream); - } -}(this, function (exports, sharedUtil, corePrimitives, coreFunction, - coreStream) { + normalizeWhitePointToFlat(cs.whitePoint, XYZ, XYZ_Flat); -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isString = sharedUtil.isString; -var shadow = sharedUtil.shadow; -var warn = sharedUtil.warn; -var isDict = corePrimitives.isDict; -var isName = corePrimitives.isName; -var isStream = corePrimitives.isStream; -var PDFFunction = coreFunction.PDFFunction; + var XYZ_Black = tempConvertMatrix1; + compensateBlackPoint(cs.blackPoint, XYZ_Flat, XYZ_Black); -var coreImage; // see _setCoreImage below -var PDFImage; // = coreImage.PDFImage; + var XYZ_D65 = tempConvertMatrix2; + normalizeWhitePointToD65(FLAT_WHITEPOINT_MATRIX, XYZ_Black, XYZ_D65); -var ColorSpace = (function ColorSpaceClosure() { - // Constructor should define this.numComps, this.defaultColor, this.name - function ColorSpace() { - error('should not call ColorSpace constructor'); + var SRGB = tempConvertMatrix1; + matrixProduct(SRGB_D65_XYZ_TO_RGB_MATRIX, XYZ_D65, SRGB); + + var sR = sRGBTransferFunction(SRGB[0]); + var sG = sRGBTransferFunction(SRGB[1]); + var sB = sRGBTransferFunction(SRGB[2]); + + // Convert the values to rgb range [0, 255]. + dest[destOffset] = Math.round(sR * 255); + dest[destOffset + 1] = Math.round(sG * 255); + dest[destOffset + 2] = Math.round(sB * 255); } - ColorSpace.prototype = { - /** - * Converts the color value to the RGB color. The color components are - * located in the src array starting from the srcOffset. Returns the array - * of the rgb components, each value ranging from [0,255]. - */ - getRgb: function ColorSpace_getRgb(src, srcOffset) { + CalRGBCS.prototype = { + getRgb: function CalRGBCS_getRgb(src, srcOffset) { var rgb = new Uint8Array(3); this.getRgbItem(src, srcOffset, rgb, 0); return rgb; }, - /** - * Converts the color value to the RGB color, similar to the getRgb method. - * The result placed into the dest array starting from the destOffset. - */ - getRgbItem: function ColorSpace_getRgbItem(src, srcOffset, - dest, destOffset) { - error('Should not call ColorSpace.getRgbItem'); + getRgbItem: function CalRGBCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(this, src, srcOffset, dest, destOffset, 1); }, - /** - * Converts the specified number of the color values to the RGB colors. - * The colors are located in the src array starting from the srcOffset. - * The result is placed into the dest array starting from the destOffset. - * The src array items shall be in [0,2^bits) range, the dest array items - * will be in [0,255] range. alpha01 indicates how many alpha components - * there are in the dest array; it will be either 0 (RGB array) or 1 (RGBA - * array). - */ - getRgbBuffer: function ColorSpace_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - error('Should not call ColorSpace.getRgbBuffer'); + getRgbBuffer: function CalRGBCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + + for (var i = 0; i < count; ++i) { + convertToRgb(this, src, srcOffset, dest, destOffset, scale); + srcOffset += 3; + destOffset += 3 + alpha01; + } }, - /** - * Determines the number of bytes required to store the result of the - * conversion done by the getRgbBuffer method. As in getRgbBuffer, - * |alpha01| is either 0 (RGB output) or 1 (RGBA output). - */ - getOutputLength: function ColorSpace_getOutputLength(inputLength, - alpha01) { - error('Should not call ColorSpace.getOutputLength'); + getOutputLength: function CalRGBCS_getOutputLength(inputLength, alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; }, - /** - * Returns true if source data will be equal the result/output data. - */ - isPassthrough: function ColorSpace_isPassthrough(bits) { - return false; + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function CalRGBCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - /** - * Fills in the RGB colors in the destination buffer. alpha01 indicates - * how many alpha components there are in the dest array; it will be either - * 0 (RGB array) or 1 (RGBA array). - */ - fillRgb: function ColorSpace_fillRgb(dest, originalWidth, - originalHeight, width, height, - actualHeight, bpc, comps, alpha01) { - var count = originalWidth * originalHeight; - var rgbBuf = null; - var numComponentColors = 1 << bpc; - var needsResizing = originalHeight !== height || originalWidth !== width; - var i, ii; + usesZeroToOneRange: true + }; + return CalRGBCS; +})(); - if (this.isPassthrough(bpc)) { - rgbBuf = comps; - } else if (this.numComps === 1 && count > numComponentColors && - this.name !== 'DeviceGray' && this.name !== 'DeviceRGB') { - // Optimization: create a color map when there is just one component and - // we are converting more colors than the size of the color map. We - // don't build the map if the colorspace is gray or rgb since those - // methods are faster than building a map. This mainly offers big speed - // ups for indexed and alternate colorspaces. - // - // TODO it may be worth while to cache the color map. While running - // testing I never hit a cache so I will leave that out for now (perhaps - // we are reparsing colorspaces too much?). - var allColors = bpc <= 8 ? new Uint8Array(numComponentColors) : - new Uint16Array(numComponentColors); - var key; - for (i = 0; i < numComponentColors; i++) { - allColors[i] = i; - } - var colorMap = new Uint8Array(numComponentColors * 3); - this.getRgbBuffer(allColors, 0, numComponentColors, colorMap, 0, bpc, - /* alpha01 = */ 0); +// +// LabCS: Based on "PDF Reference, Sixth Ed", p.250 +// +var LabCS = (function LabCSClosure() { + function LabCS(whitePoint, blackPoint, range) { + this.name = 'Lab'; + this.numComps = 3; + this.defaultColor = new Float32Array([0, 0, 0]); - var destPos, rgbPos; - if (!needsResizing) { - // Fill in the RGB values directly into |dest|. - destPos = 0; - for (i = 0; i < count; ++i) { - key = comps[i] * 3; - dest[destPos++] = colorMap[key]; - dest[destPos++] = colorMap[key + 1]; - dest[destPos++] = colorMap[key + 2]; - destPos += alpha01; - } - } else { - rgbBuf = new Uint8Array(count * 3); - rgbPos = 0; - for (i = 0; i < count; ++i) { - key = comps[i] * 3; - rgbBuf[rgbPos++] = colorMap[key]; - rgbBuf[rgbPos++] = colorMap[key + 1]; - rgbBuf[rgbPos++] = colorMap[key + 2]; - } - } - } else { - if (!needsResizing) { - // Fill in the RGB values directly into |dest|. - this.getRgbBuffer(comps, 0, width * actualHeight, dest, 0, bpc, - alpha01); - } else { - rgbBuf = new Uint8Array(count * 3); - this.getRgbBuffer(comps, 0, count, rgbBuf, 0, bpc, - /* alpha01 = */ 0); - } - } + if (!whitePoint) { + error('WhitePoint missing - required for color space Lab'); + } + blackPoint = blackPoint || [0, 0, 0]; + range = range || [-100, 100, -100, 100]; + + // Translate args to spec variables + this.XW = whitePoint[0]; + this.YW = whitePoint[1]; + this.ZW = whitePoint[2]; + this.amin = range[0]; + this.amax = range[1]; + this.bmin = range[2]; + this.bmax = range[3]; + + // These are here just for completeness - the spec doesn't offer any + // formulas that use BlackPoint in Lab + this.XB = blackPoint[0]; + this.YB = blackPoint[1]; + this.ZB = blackPoint[2]; + + // Validate vars as per spec + if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { + error('Invalid WhitePoint components, no fallback available'); + } + + if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { + info('Invalid BlackPoint, falling back to default'); + this.XB = this.YB = this.ZB = 0; + } + + if (this.amin > this.amax || this.bmin > this.bmax) { + info('Invalid Range, falling back to defaults'); + this.amin = -100; + this.amax = 100; + this.bmin = -100; + this.bmax = 100; + } + } + + // Function g(x) from spec + function fn_g(x) { + if (x >= 6 / 29) { + return x * x * x; + } else { + return (108 / 841) * (x - 4 / 29); + } + } + + function decode(value, high1, low2, high2) { + return low2 + (value) * (high2 - low2) / (high1); + } + + // If decoding is needed maxVal should be 2^bits per component - 1. + function convertToRgb(cs, src, srcOffset, maxVal, dest, destOffset) { + // XXX: Lab input is in the range of [0, 100], [amin, amax], [bmin, bmax] + // not the usual [0, 1]. If a command like setFillColor is used the src + // values will already be within the correct range. However, if we are + // converting an image we have to map the values to the correct range given + // above. + // Ls,as,bs <---> L*,a*,b* in the spec + var Ls = src[srcOffset]; + var as = src[srcOffset + 1]; + var bs = src[srcOffset + 2]; + if (maxVal !== false) { + Ls = decode(Ls, maxVal, 0, 100); + as = decode(as, maxVal, cs.amin, cs.amax); + bs = decode(bs, maxVal, cs.bmin, cs.bmax); + } + + // Adjust limits of 'as' and 'bs' + as = as > cs.amax ? cs.amax : as < cs.amin ? cs.amin : as; + bs = bs > cs.bmax ? cs.bmax : bs < cs.bmin ? cs.bmin : bs; + + // Computes intermediate variables X,Y,Z as per spec + var M = (Ls + 16) / 116; + var L = M + (as / 500); + var N = M - (bs / 200); + + var X = cs.XW * fn_g(L); + var Y = cs.YW * fn_g(M); + var Z = cs.ZW * fn_g(N); + + var r, g, b; + // Using different conversions for D50 and D65 white points, + // per http://www.color.org/srgb.pdf + if (cs.ZW < 1) { + // Assuming D50 (X=0.9642, Y=1.00, Z=0.8249) + r = X * 3.1339 + Y * -1.6170 + Z * -0.4906; + g = X * -0.9785 + Y * 1.9160 + Z * 0.0333; + b = X * 0.0720 + Y * -0.2290 + Z * 1.4057; + } else { + // Assuming D65 (X=0.9505, Y=1.00, Z=1.0888) + r = X * 3.2406 + Y * -1.5372 + Z * -0.4986; + g = X * -0.9689 + Y * 1.8758 + Z * 0.0415; + b = X * 0.0557 + Y * -0.2040 + Z * 1.0570; + } + // clamp color values to [0,1] range then convert to [0,255] range. + dest[destOffset] = r <= 0 ? 0 : r >= 1 ? 255 : Math.sqrt(r) * 255 | 0; + dest[destOffset + 1] = g <= 0 ? 0 : g >= 1 ? 255 : Math.sqrt(g) * 255 | 0; + dest[destOffset + 2] = b <= 0 ? 0 : b >= 1 ? 255 : Math.sqrt(b) * 255 | 0; + } - if (rgbBuf) { - if (needsResizing) { - PDFImage.resize(rgbBuf, bpc, 3, originalWidth, originalHeight, width, - height, dest, alpha01); - } else { - rgbPos = 0; - destPos = 0; - for (i = 0, ii = width * actualHeight; i < ii; i++) { - dest[destPos++] = rgbBuf[rgbPos++]; - dest[destPos++] = rgbBuf[rgbPos++]; - dest[destPos++] = rgbBuf[rgbPos++]; - destPos += alpha01; - } - } + LabCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function LabCS_getRgbItem(src, srcOffset, dest, destOffset) { + convertToRgb(this, src, srcOffset, false, dest, destOffset); + }, + getRgbBuffer: function LabCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var maxVal = (1 << bits) - 1; + for (var i = 0; i < count; i++) { + convertToRgb(this, src, srcOffset, maxVal, dest, destOffset); + srcOffset += 3; + destOffset += 3 + alpha01; } }, - /** - * True if the colorspace has components in the default range of [0, 1]. - * This should be true for all colorspaces except for lab color spaces - * which are [0,100], [-128, 127], [-128, 127]. - */ - usesZeroToOneRange: true + getOutputLength: function LabCS_getOutputLength(inputLength, alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function LabCS_isDefaultDecode(decodeMap) { + // XXX: Decoding is handled with the lab conversion because of the strange + // ranges that are used. + return true; + }, + usesZeroToOneRange: false }; + return LabCS; +})(); - ColorSpace.parse = function ColorSpace_parse(cs, xref, res) { - var IR = ColorSpace.parseToIR(cs, xref, res); - if (IR instanceof AlternateCS) { - return IR; - } - return ColorSpace.fromIR(IR); - }; +// TODO refactor to remove dependency on image.js +function _setCoreImage(coreImage_) { + coreImage = coreImage_; + PDFImage = coreImage_.PDFImage; +} +exports._setCoreImage = _setCoreImage; - ColorSpace.fromIR = function ColorSpace_fromIR(IR) { - var name = isArray(IR) ? IR[0] : IR; - var whitePoint, blackPoint, gamma; +exports.ColorSpace = ColorSpace; - switch (name) { - case 'DeviceGrayCS': - return this.singletons.gray; - case 'DeviceRgbCS': - return this.singletons.rgb; - case 'DeviceCmykCS': - return this.singletons.cmyk; - case 'CalGrayCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - gamma = IR[3]; - return new CalGrayCS(whitePoint, blackPoint, gamma); - case 'CalRGBCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - gamma = IR[3]; - var matrix = IR[4]; - return new CalRGBCS(whitePoint, blackPoint, gamma, matrix); - case 'PatternCS': - var basePatternCS = IR[1]; - if (basePatternCS) { - basePatternCS = ColorSpace.fromIR(basePatternCS); - } - return new PatternCS(basePatternCS); - case 'IndexedCS': - var baseIndexedCS = IR[1]; - var hiVal = IR[2]; - var lookup = IR[3]; - return new IndexedCS(ColorSpace.fromIR(baseIndexedCS), hiVal, lookup); - case 'AlternateCS': - var numComps = IR[1]; - var alt = IR[2]; - var tintFnIR = IR[3]; +// TODO refactor to remove dependency on colorspace.js +coreStream._setCoreColorSpace(exports); +})); - return new AlternateCS(numComps, ColorSpace.fromIR(alt), - PDFFunction.fromIR(tintFnIR)); - case 'LabCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - var range = IR[3]; - return new LabCS(whitePoint, blackPoint, range); - default: - error('Unknown name ' + name); + +(function (root, factory) { + { + factory((root.pdfjsCoreImage = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreColorSpace, root.pdfjsCoreStream, + root.pdfjsCoreJpx); + } +}(this, function (exports, sharedUtil, corePrimitives, coreColorSpace, + coreStream, coreJpx) { + +var ImageKind = sharedUtil.ImageKind; +var assert = sharedUtil.assert; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var warn = sharedUtil.warn; +var Name = corePrimitives.Name; +var isStream = corePrimitives.isStream; +var ColorSpace = coreColorSpace.ColorSpace; +var DecodeStream = coreStream.DecodeStream; +var Stream = coreStream.Stream; +var JpegStream = coreStream.JpegStream; +var JpxImage = coreJpx.JpxImage; + +var PDFImage = (function PDFImageClosure() { + /** + * Decode the image in the main thread if it supported. Resovles the promise + * when the image data is ready. + */ + function handleImageData(handler, xref, res, image) { + if (image instanceof JpegStream && image.isNativelyDecodable(xref, res)) { + // For natively supported jpegs send them to the main thread for decoding. + var dict = image.dict; + var colorSpace = dict.get('ColorSpace', 'CS'); + colorSpace = ColorSpace.parse(colorSpace, xref, res); + var numComps = colorSpace.numComps; + var decodePromise = handler.sendWithPromise('JpegDecode', + [image.getIR(), numComps]); + return decodePromise.then(function (message) { + var data = message.data; + return new Stream(data, 0, data.length, image.dict); + }); + } else { + return Promise.resolve(image); } - return null; - }; + } - ColorSpace.parseToIR = function ColorSpace_parseToIR(cs, xref, res) { - if (isName(cs)) { - var colorSpaces = res.get('ColorSpace'); - if (isDict(colorSpaces)) { - var refcs = colorSpaces.get(cs.name); - if (refcs) { - cs = refcs; - } + /** + * Decode and clamp a value. The formula is different from the spec because we + * don't decode to float range [0,1], we decode it in the [0,max] range. + */ + function decodeAndClamp(value, addend, coefficient, max) { + value = addend + value * coefficient; + // Clamp the value to the range + return (value < 0 ? 0 : (value > max ? max : value)); + } + + function PDFImage(xref, res, image, inline, smask, mask, isMask) { + this.image = image; + var dict = image.dict; + if (dict.has('Filter')) { + var filter = dict.get('Filter').name; + if (filter === 'JPXDecode') { + var jpxImage = new JpxImage(); + jpxImage.parseImageProperties(image.stream); + image.stream.reset(); + image.bitsPerComponent = jpxImage.bitsPerComponent; + image.numComps = jpxImage.componentsCount; + } else if (filter === 'JBIG2Decode') { + image.bitsPerComponent = 1; + image.numComps = 1; } } + // TODO cache rendered images? - cs = xref.fetchIfRef(cs); - var mode; + this.width = dict.get('Width', 'W'); + this.height = dict.get('Height', 'H'); - if (isName(cs)) { - mode = cs.name; - this.mode = mode; + if (this.width < 1 || this.height < 1) { + error('Invalid image width: ' + this.width + ' or height: ' + + this.height); + } - switch (mode) { - case 'DeviceGray': - case 'G': - return 'DeviceGrayCS'; - case 'DeviceRGB': - case 'RGB': - return 'DeviceRgbCS'; - case 'DeviceCMYK': - case 'CMYK': - return 'DeviceCmykCS'; - case 'Pattern': - return ['PatternCS', null]; - default: - error('unrecognized colorspace ' + mode); - } - } else if (isArray(cs)) { - mode = xref.fetchIfRef(cs[0]).name; - this.mode = mode; - var numComps, params, alt, whitePoint, blackPoint, gamma; + this.interpolate = dict.get('Interpolate', 'I') || false; + this.imageMask = dict.get('ImageMask', 'IM') || false; + this.matte = dict.get('Matte') || false; - switch (mode) { - case 'DeviceGray': - case 'G': - return 'DeviceGrayCS'; - case 'DeviceRGB': - case 'RGB': - return 'DeviceRgbCS'; - case 'DeviceCMYK': - case 'CMYK': - return 'DeviceCmykCS'; - case 'CalGray': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - gamma = params.get('Gamma'); - return ['CalGrayCS', whitePoint, blackPoint, gamma]; - case 'CalRGB': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - gamma = params.get('Gamma'); - var matrix = params.get('Matrix'); - return ['CalRGBCS', whitePoint, blackPoint, gamma, matrix]; - case 'ICCBased': - var stream = xref.fetchIfRef(cs[1]); - var dict = stream.dict; - numComps = dict.get('N'); - alt = dict.get('Alternate'); - if (alt) { - var altIR = ColorSpace.parseToIR(alt, xref, res); - // Parse the /Alternate CS to ensure that the number of components - // are correct, and also (indirectly) that it is not a PatternCS. - var altCS = ColorSpace.fromIR(altIR); - if (altCS.numComps === numComps) { - return altIR; - } - warn('ICCBased color space: Ignoring incorrect /Alternate entry.'); - } - if (numComps === 1) { - return 'DeviceGrayCS'; - } else if (numComps === 3) { - return 'DeviceRgbCS'; - } else if (numComps === 4) { - return 'DeviceCmykCS'; - } - break; - case 'Pattern': - var basePatternCS = cs[1] || null; - if (basePatternCS) { - basePatternCS = ColorSpace.parseToIR(basePatternCS, xref, res); - } - return ['PatternCS', basePatternCS]; - case 'Indexed': - case 'I': - var baseIndexedCS = ColorSpace.parseToIR(cs[1], xref, res); - var hiVal = xref.fetchIfRef(cs[2]) + 1; - var lookup = xref.fetchIfRef(cs[3]); - if (isStream(lookup)) { - lookup = lookup.getBytes(); - } - return ['IndexedCS', baseIndexedCS, hiVal, lookup]; - case 'Separation': - case 'DeviceN': - var name = xref.fetchIfRef(cs[1]); - numComps = 1; - if (isName(name)) { - numComps = 1; - } else if (isArray(name)) { - numComps = name.length; - } - alt = ColorSpace.parseToIR(cs[2], xref, res); - var tintFnIR = PDFFunction.getIR(xref, xref.fetchIfRef(cs[3])); - return ['AlternateCS', numComps, alt, tintFnIR]; - case 'Lab': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - var range = params.get('Range'); - return ['LabCS', whitePoint, blackPoint, range]; - default: - error('unimplemented color space object "' + mode + '"'); + var bitsPerComponent = image.bitsPerComponent; + if (!bitsPerComponent) { + bitsPerComponent = dict.get('BitsPerComponent', 'BPC'); + if (!bitsPerComponent) { + if (this.imageMask) { + bitsPerComponent = 1; + } else { + error('Bits per component missing in image: ' + this.imageMask); + } } - } else { - error('unrecognized color space object: "' + cs + '"'); } - return null; - }; - /** - * Checks if a decode map matches the default decode map for a color space. - * This handles the general decode maps where there are two values per - * component. e.g. [0, 1, 0, 1, 0, 1] for a RGB color. - * This does not handle Lab, Indexed, or Pattern decode maps since they are - * slightly different. - * @param {Array} decode Decode map (usually from an image). - * @param {Number} n Number of components the color space has. - */ - ColorSpace.isDefaultDecode = function ColorSpace_isDefaultDecode(decode, n) { - if (!isArray(decode)) { - return true; + this.bpc = bitsPerComponent; + + if (!this.imageMask) { + var colorSpace = dict.get('ColorSpace', 'CS'); + if (!colorSpace) { + info('JPX images (which do not require color spaces)'); + switch (image.numComps) { + case 1: + colorSpace = Name.get('DeviceGray'); + break; + case 3: + colorSpace = Name.get('DeviceRGB'); + break; + case 4: + colorSpace = Name.get('DeviceCMYK'); + break; + default: + error('JPX images with ' + this.numComps + + ' color components not supported.'); + } + } + this.colorSpace = ColorSpace.parse(colorSpace, xref, res); + this.numComps = this.colorSpace.numComps; } - if (n * 2 !== decode.length) { - warn('The decode map is not the correct length'); - return true; + this.decode = dict.get('Decode', 'D'); + this.needsDecode = false; + if (this.decode && + ((this.colorSpace && !this.colorSpace.isDefaultDecode(this.decode)) || + (isMask && !ColorSpace.isDefaultDecode(this.decode, 1)))) { + this.needsDecode = true; + // Do some preprocessing to avoid more math. + var max = (1 << bitsPerComponent) - 1; + this.decodeCoefficients = []; + this.decodeAddends = []; + for (var i = 0, j = 0; i < this.decode.length; i += 2, ++j) { + var dmin = this.decode[i]; + var dmax = this.decode[i + 1]; + this.decodeCoefficients[j] = dmax - dmin; + this.decodeAddends[j] = max * dmin; + } } - for (var i = 0, ii = decode.length; i < ii; i += 2) { - if (decode[i] !== 0 || decode[i + 1] !== 1) { - return false; + + if (smask) { + this.smask = new PDFImage(xref, res, smask, false); + } else if (mask) { + if (isStream(mask)) { + var maskDict = mask.dict, imageMask = maskDict.get('ImageMask', 'IM'); + if (!imageMask) { + warn('Ignoring /Mask in image without /ImageMask.'); + } else { + this.mask = new PDFImage(xref, res, mask, false, null, null, true); + } + } else { + // Color key mask (just an array). + this.mask = mask; } } - return true; - }; + } + /** + * Handles processing of image data and returns the Promise that is resolved + * with a PDFImage when the image is ready to be used. + */ + PDFImage.buildImage = function PDFImage_buildImage(handler, xref, + res, image, inline) { + var imagePromise = handleImageData(handler, xref, res, image); + var smaskPromise; + var maskPromise; - ColorSpace.singletons = { - get gray() { - return shadow(this, 'gray', new DeviceGrayCS()); - }, - get rgb() { - return shadow(this, 'rgb', new DeviceRgbCS()); - }, - get cmyk() { - return shadow(this, 'cmyk', new DeviceCmykCS()); + var smask = image.dict.get('SMask'); + var mask = image.dict.get('Mask'); + + if (smask) { + smaskPromise = handleImageData(handler, xref, res, smask); + maskPromise = Promise.resolve(null); + } else { + smaskPromise = Promise.resolve(null); + if (mask) { + if (isStream(mask)) { + maskPromise = handleImageData(handler, xref, res, mask); + } else if (isArray(mask)) { + maskPromise = Promise.resolve(mask); + } else { + warn('Unsupported mask format.'); + maskPromise = Promise.resolve(null); + } + } else { + maskPromise = Promise.resolve(null); + } } + return Promise.all([imagePromise, smaskPromise, maskPromise]).then( + function(results) { + var imageData = results[0]; + var smaskData = results[1]; + var maskData = results[2]; + return new PDFImage(xref, res, imageData, inline, smaskData, maskData); + }); }; - return ColorSpace; -})(); + /** + * Resize an image using the nearest neighbor algorithm. Currently only + * supports one and three component images. + * @param {TypedArray} pixels The original image with one component. + * @param {Number} bpc Number of bits per component. + * @param {Number} components Number of color components, 1 or 3 is supported. + * @param {Number} w1 Original width. + * @param {Number} h1 Original height. + * @param {Number} w2 New width. + * @param {Number} h2 New height. + * @param {TypedArray} dest (Optional) The destination buffer. + * @param {Number} alpha01 (Optional) Size reserved for the alpha channel. + * @return {TypedArray} Resized image data. + */ + PDFImage.resize = function PDFImage_resize(pixels, bpc, components, + w1, h1, w2, h2, dest, alpha01) { -/** - * Alternate color space handles both Separation and DeviceN color spaces. A - * Separation color space is actually just a DeviceN with one color component. - * Both color spaces use a tinting function to convert colors to a base color - * space. - */ -var AlternateCS = (function AlternateCSClosure() { - function AlternateCS(numComps, base, tintFn) { - this.name = 'Alternate'; - this.numComps = numComps; - this.defaultColor = new Float32Array(numComps); - for (var i = 0; i < numComps; ++i) { - this.defaultColor[i] = 1; + if (components !== 1 && components !== 3) { + error('Unsupported component count for resizing.'); } - this.base = base; - this.tintFn = tintFn; - this.tmpBuf = new Float32Array(base.numComps); - } - AlternateCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function AlternateCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var tmpBuf = this.tmpBuf; - this.tintFn(src, srcOffset, tmpBuf, 0); - this.base.getRgbItem(tmpBuf, 0, dest, destOffset); - }, - getRgbBuffer: function AlternateCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var tintFn = this.tintFn; - var base = this.base; - var scale = 1 / ((1 << bits) - 1); - var baseNumComps = base.numComps; - var usesZeroToOneRange = base.usesZeroToOneRange; - var isPassthrough = (base.isPassthrough(8) || !usesZeroToOneRange) && - alpha01 === 0; - var pos = isPassthrough ? destOffset : 0; - var baseBuf = isPassthrough ? dest : new Uint8Array(baseNumComps * count); - var numComps = this.numComps; + var length = w2 * h2 * components; + var temp = dest ? dest : (bpc <= 8 ? new Uint8Array(length) : + (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); + var xRatio = w1 / w2; + var yRatio = h1 / h2; + var i, j, py, newIndex = 0, oldIndex; + var xScaled = new Uint16Array(w2); + var w1Scanline = w1 * components; + if (alpha01 !== 1) { + alpha01 = 0; + } - var scaled = new Float32Array(numComps); - var tinted = new Float32Array(baseNumComps); - var i, j; - if (usesZeroToOneRange) { - for (i = 0; i < count; i++) { - for (j = 0; j < numComps; j++) { - scaled[j] = src[srcOffset++] * scale; - } - tintFn(scaled, 0, tinted, 0); - for (j = 0; j < baseNumComps; j++) { - baseBuf[pos++] = tinted[j] * 255; - } - } - } else { - for (i = 0; i < count; i++) { - for (j = 0; j < numComps; j++) { - scaled[j] = src[srcOffset++] * scale; - } - tintFn(scaled, 0, tinted, 0); - base.getRgbItem(tinted, 0, baseBuf, pos); - pos += baseNumComps; + for (j = 0; j < w2; j++) { + xScaled[j] = Math.floor(j * xRatio) * components; + } + + if (components === 1) { + for (i = 0; i < h2; i++) { + py = Math.floor(i * yRatio) * w1Scanline; + for (j = 0; j < w2; j++) { + oldIndex = py + xScaled[j]; + temp[newIndex++] = pixels[oldIndex]; } } - if (!isPassthrough) { - base.getRgbBuffer(baseBuf, 0, count, dest, destOffset, 8, alpha01); + } else if (components === 3) { + for (i = 0; i < h2; i++) { + py = Math.floor(i * yRatio) * w1Scanline; + for (j = 0; j < w2; j++) { + oldIndex = py + xScaled[j]; + temp[newIndex++] = pixels[oldIndex++]; + temp[newIndex++] = pixels[oldIndex++]; + temp[newIndex++] = pixels[oldIndex++]; + newIndex += alpha01; + } } - }, - getOutputLength: function AlternateCS_getOutputLength(inputLength, - alpha01) { - return this.base.getOutputLength(inputLength * - this.base.numComps / this.numComps, - alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function AlternateCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true + } + return temp; }; - return AlternateCS; -})(); - -var PatternCS = (function PatternCSClosure() { - function PatternCS(baseCS) { - this.name = 'Pattern'; - this.base = baseCS; - } - PatternCS.prototype = {}; + PDFImage.createMask = + function PDFImage_createMask(imgArray, width, height, + imageIsFromDecodeStream, inverseDecode) { - return PatternCS; -})(); + // |imgArray| might not contain full data for every pixel of the mask, so + // we need to distinguish between |computedLength| and |actualLength|. + // In particular, if inverseDecode is true, then the array we return must + // have a length of |computedLength|. -var IndexedCS = (function IndexedCSClosure() { - function IndexedCS(base, highVal, lookup) { - this.name = 'Indexed'; - this.numComps = 1; - this.defaultColor = new Uint8Array([0]); - this.base = base; - this.highVal = highVal; + var computedLength = ((width + 7) >> 3) * height; + var actualLength = imgArray.byteLength; + var haveFullData = computedLength === actualLength; + var data, i; - var baseNumComps = base.numComps; - var length = baseNumComps * highVal; - var lookupArray; + if (imageIsFromDecodeStream && (!inverseDecode || haveFullData)) { + // imgArray came from a DecodeStream and its data is in an appropriate + // form, so we can just transfer it. + data = imgArray; + } else if (!inverseDecode) { + data = new Uint8Array(actualLength); + data.set(imgArray); + } else { + data = new Uint8Array(computedLength); + data.set(imgArray); + for (i = actualLength; i < computedLength; i++) { + data[i] = 0xff; + } + } - if (isStream(lookup)) { - lookupArray = new Uint8Array(length); - var bytes = lookup.getBytes(length); - lookupArray.set(bytes); - } else if (isString(lookup)) { - lookupArray = new Uint8Array(length); - for (var i = 0; i < length; ++i) { - lookupArray[i] = lookup.charCodeAt(i); + // If necessary, invert the original mask data (but not any extra we might + // have added above). It's safe to modify the array -- whether it's the + // original or a copy, we're about to transfer it anyway, so nothing else + // in this thread can be relying on its contents. + if (inverseDecode) { + for (i = 0; i < actualLength; i++) { + data[i] = ~data[i]; } - } else if (lookup instanceof Uint8Array || lookup instanceof Array) { - lookupArray = lookup; - } else { - error('Unrecognized lookup table: ' + lookup); } - this.lookup = lookupArray; - } - IndexedCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function IndexedCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var numComps = this.base.numComps; - var start = src[srcOffset] * numComps; - this.base.getRgbItem(this.lookup, start, dest, destOffset); - }, - getRgbBuffer: function IndexedCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var base = this.base; - var numComps = base.numComps; - var outputDelta = base.getOutputLength(numComps, alpha01); - var lookup = this.lookup; + return {data: data, width: width, height: height}; + }; - for (var i = 0; i < count; ++i) { - var lookupPos = src[srcOffset++] * numComps; - base.getRgbBuffer(lookup, lookupPos, 1, dest, destOffset, 8, alpha01); - destOffset += outputDelta; - } - }, - getOutputLength: function IndexedCS_getOutputLength(inputLength, alpha01) { - return this.base.getOutputLength(inputLength * this.base.numComps, - alpha01); + PDFImage.prototype = { + get drawWidth() { + return Math.max(this.width, + this.smask && this.smask.width || 0, + this.mask && this.mask.width || 0); }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function IndexedCS_isDefaultDecode(decodeMap) { - // indexed color maps shouldn't be changed - return true; + + get drawHeight() { + return Math.max(this.height, + this.smask && this.smask.height || 0, + this.mask && this.mask.height || 0); }, - usesZeroToOneRange: true - }; - return IndexedCS; -})(); -var DeviceGrayCS = (function DeviceGrayCSClosure() { - function DeviceGrayCS() { - this.name = 'DeviceGray'; - this.numComps = 1; - this.defaultColor = new Float32Array([0]); - } + decodeBuffer: function PDFImage_decodeBuffer(buffer) { + var bpc = this.bpc; + var numComps = this.numComps; - DeviceGrayCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceGrayCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var c = (src[srcOffset] * 255) | 0; - c = c < 0 ? 0 : c > 255 ? 255 : c; - dest[destOffset] = dest[destOffset + 1] = dest[destOffset + 2] = c; - }, - getRgbBuffer: function DeviceGrayCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 255 / ((1 << bits) - 1); - var j = srcOffset, q = destOffset; - for (var i = 0; i < count; ++i) { - var c = (scale * src[j++]) | 0; - dest[q++] = c; - dest[q++] = c; - dest[q++] = c; - q += alpha01; - } - }, - getOutputLength: function DeviceGrayCS_getOutputLength(inputLength, - alpha01) { - return inputLength * (3 + alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceGrayCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceGrayCS; -})(); + var decodeAddends = this.decodeAddends; + var decodeCoefficients = this.decodeCoefficients; + var max = (1 << bpc) - 1; + var i, ii; -var DeviceRgbCS = (function DeviceRgbCSClosure() { - function DeviceRgbCS() { - this.name = 'DeviceRGB'; - this.numComps = 3; - this.defaultColor = new Float32Array([0, 0, 0]); - } - DeviceRgbCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceRgbCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var r = (src[srcOffset] * 255) | 0; - var g = (src[srcOffset + 1] * 255) | 0; - var b = (src[srcOffset + 2] * 255) | 0; - dest[destOffset] = r < 0 ? 0 : r > 255 ? 255 : r; - dest[destOffset + 1] = g < 0 ? 0 : g > 255 ? 255 : g; - dest[destOffset + 2] = b < 0 ? 0 : b > 255 ? 255 : b; - }, - getRgbBuffer: function DeviceRgbCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - if (bits === 8 && alpha01 === 0) { - dest.set(src.subarray(srcOffset, srcOffset + count * 3), destOffset); + if (bpc === 1) { + // If the buffer needed decode that means it just needs to be inverted. + for (i = 0, ii = buffer.length; i < ii; i++) { + buffer[i] = +!(buffer[i]); + } return; } - var scale = 255 / ((1 << bits) - 1); - var j = srcOffset, q = destOffset; - for (var i = 0; i < count; ++i) { - dest[q++] = (scale * src[j++]) | 0; - dest[q++] = (scale * src[j++]) | 0; - dest[q++] = (scale * src[j++]) | 0; - q += alpha01; + var index = 0; + for (i = 0, ii = this.width * this.height; i < ii; i++) { + for (var j = 0; j < numComps; j++) { + buffer[index] = decodeAndClamp(buffer[index], decodeAddends[j], + decodeCoefficients[j], max); + index++; + } } }, - getOutputLength: function DeviceRgbCS_getOutputLength(inputLength, - alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: function DeviceRgbCS_isPassthrough(bits) { - return bits === 8; - }, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceRgbCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceRgbCS; -})(); - -var DeviceCmykCS = (function DeviceCmykCSClosure() { - // The coefficients below was found using numerical analysis: the method of - // steepest descent for the sum((f_i - color_value_i)^2) for r/g/b colors, - // where color_value is the tabular value from the table of sampled RGB colors - // from CMYK US Web Coated (SWOP) colorspace, and f_i is the corresponding - // CMYK color conversion using the estimation below: - // f(A, B,.. N) = Acc+Bcm+Ccy+Dck+c+Fmm+Gmy+Hmk+Im+Jyy+Kyk+Ly+Mkk+Nk+255 - function convertToRgb(src, srcOffset, srcScale, dest, destOffset) { - var c = src[srcOffset + 0] * srcScale; - var m = src[srcOffset + 1] * srcScale; - var y = src[srcOffset + 2] * srcScale; - var k = src[srcOffset + 3] * srcScale; - - var r = - (c * (-4.387332384609988 * c + 54.48615194189176 * m + - 18.82290502165302 * y + 212.25662451639585 * k + - -285.2331026137004) + - m * (1.7149763477362134 * m - 5.6096736904047315 * y + - -17.873870861415444 * k - 5.497006427196366) + - y * (-2.5217340131683033 * y - 21.248923337353073 * k + - 17.5119270841813) + - k * (-21.86122147463605 * k - 189.48180835922747) + 255) | 0; - var g = - (c * (8.841041422036149 * c + 60.118027045597366 * m + - 6.871425592049007 * y + 31.159100130055922 * k + - -79.2970844816548) + - m * (-15.310361306967817 * m + 17.575251261109482 * y + - 131.35250912493976 * k - 190.9453302588951) + - y * (4.444339102852739 * y + 9.8632861493405 * k - 24.86741582555878) + - k * (-20.737325471181034 * k - 187.80453709719578) + 255) | 0; - var b = - (c * (0.8842522430003296 * c + 8.078677503112928 * m + - 30.89978309703729 * y - 0.23883238689178934 * k + - -14.183576799673286) + - m * (10.49593273432072 * m + 63.02378494754052 * y + - 50.606957656360734 * k - 112.23884253719248) + - y * (0.03296041114873217 * y + 115.60384449646641 * k + - -193.58209356861505) + - k * (-22.33816807309886 * k - 180.12613974708367) + 255) | 0; - dest[destOffset] = r > 255 ? 255 : r < 0 ? 0 : r; - dest[destOffset + 1] = g > 255 ? 255 : g < 0 ? 0 : g; - dest[destOffset + 2] = b > 255 ? 255 : b < 0 ? 0 : b; - } + getComponents: function PDFImage_getComponents(buffer) { + var bpc = this.bpc; - function DeviceCmykCS() { - this.name = 'DeviceCMYK'; - this.numComps = 4; - this.defaultColor = new Float32Array([0, 0, 0, 1]); - } - DeviceCmykCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceCmykCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(src, srcOffset, 1, dest, destOffset); - }, - getRgbBuffer: function DeviceCmykCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; i++) { - convertToRgb(src, srcOffset, scale, dest, destOffset); - srcOffset += 4; - destOffset += 3 + alpha01; + // This image doesn't require any extra work. + if (bpc === 8) { + return buffer; } - }, - getOutputLength: function DeviceCmykCS_getOutputLength(inputLength, - alpha01) { - return (inputLength / 4 * (3 + alpha01)) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceCmykCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceCmykCS; -})(); + var width = this.width; + var height = this.height; + var numComps = this.numComps; -// -// CalGrayCS: Based on "PDF Reference, Sixth Ed", p.245 -// -var CalGrayCS = (function CalGrayCSClosure() { - function CalGrayCS(whitePoint, blackPoint, gamma) { - this.name = 'CalGray'; - this.numComps = 1; - this.defaultColor = new Float32Array([0]); + var length = width * height * numComps; + var bufferPos = 0; + var output = (bpc <= 8 ? new Uint8Array(length) : + (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); + var rowComps = width * numComps; - if (!whitePoint) { - error('WhitePoint missing - required for color space CalGray'); - } - blackPoint = blackPoint || [0, 0, 0]; - gamma = gamma || 1; + var max = (1 << bpc) - 1; + var i = 0, ii, buf; - // Translate arguments to spec variables. - this.XW = whitePoint[0]; - this.YW = whitePoint[1]; - this.ZW = whitePoint[2]; + if (bpc === 1) { + // Optimization for reading 1 bpc images. + var mask, loop1End, loop2End; + for (var j = 0; j < height; j++) { + loop1End = i + (rowComps & ~7); + loop2End = i + rowComps; - this.XB = blackPoint[0]; - this.YB = blackPoint[1]; - this.ZB = blackPoint[2]; + // unroll loop for all full bytes + while (i < loop1End) { + buf = buffer[bufferPos++]; + output[i] = (buf >> 7) & 1; + output[i + 1] = (buf >> 6) & 1; + output[i + 2] = (buf >> 5) & 1; + output[i + 3] = (buf >> 4) & 1; + output[i + 4] = (buf >> 3) & 1; + output[i + 5] = (buf >> 2) & 1; + output[i + 6] = (buf >> 1) & 1; + output[i + 7] = buf & 1; + i += 8; + } - this.G = gamma; + // handle remaing bits + if (i < loop2End) { + buf = buffer[bufferPos++]; + mask = 128; + while (i < loop2End) { + output[i++] = +!!(buf & mask); + mask >>= 1; + } + } + } + } else { + // The general case that handles all other bpc values. + var bits = 0; + buf = 0; + for (i = 0, ii = length; i < ii; ++i) { + if (i % rowComps === 0) { + buf = 0; + bits = 0; + } - // Validate variables as per spec. - if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { - error('Invalid WhitePoint components for ' + this.name + - ', no fallback available'); - } + while (bits < bpc) { + buf = (buf << 8) | buffer[bufferPos++]; + bits += 8; + } - if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { - info('Invalid BlackPoint for ' + this.name + ', falling back to default'); - this.XB = this.YB = this.ZB = 0; - } + var remainingBits = bits - bpc; + var value = buf >> remainingBits; + output[i] = (value < 0 ? 0 : (value > max ? max : value)); + buf = buf & ((1 << remainingBits) - 1); + bits = remainingBits; + } + } + return output; + }, - if (this.XB !== 0 || this.YB !== 0 || this.ZB !== 0) { - warn(this.name + ', BlackPoint: XB: ' + this.XB + ', YB: ' + this.YB + - ', ZB: ' + this.ZB + ', only default values are supported.'); - } + fillOpacity: function PDFImage_fillOpacity(rgbaBuf, width, height, + actualHeight, image) { + var smask = this.smask; + var mask = this.mask; + var alphaBuf, sw, sh, i, ii, j; - if (this.G < 1) { - info('Invalid Gamma: ' + this.G + ' for ' + this.name + - ', falling back to default'); - this.G = 1; - } - } + if (smask) { + sw = smask.width; + sh = smask.height; + alphaBuf = new Uint8Array(sw * sh); + smask.fillGrayBuffer(alphaBuf); + if (sw !== width || sh !== height) { + alphaBuf = PDFImage.resize(alphaBuf, smask.bpc, 1, sw, sh, width, + height); + } + } else if (mask) { + if (mask instanceof PDFImage) { + sw = mask.width; + sh = mask.height; + alphaBuf = new Uint8Array(sw * sh); + mask.numComps = 1; + mask.fillGrayBuffer(alphaBuf); - function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { - // A represents a gray component of a calibrated gray space. - // A <---> AG in the spec - var A = src[srcOffset] * scale; - var AG = Math.pow(A, cs.G); + // Need to invert values in rgbaBuf + for (i = 0, ii = sw * sh; i < ii; ++i) { + alphaBuf[i] = 255 - alphaBuf[i]; + } - // Computes L as per spec. ( = cs.YW * AG ) - // Except if other than default BlackPoint values are used. - var L = cs.YW * AG; - // http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html, Ch 4. - // Convert values to rgb range [0, 255]. - var val = Math.max(295.8 * Math.pow(L, 0.333333333333333333) - 40.8, 0) | 0; - dest[destOffset] = val; - dest[destOffset + 1] = val; - dest[destOffset + 2] = val; - } + if (sw !== width || sh !== height) { + alphaBuf = PDFImage.resize(alphaBuf, mask.bpc, 1, sw, sh, width, + height); + } + } else if (isArray(mask)) { + // Color key mask: if any of the compontents are outside the range + // then they should be painted. + alphaBuf = new Uint8Array(width * height); + var numComps = this.numComps; + for (i = 0, ii = width * height; i < ii; ++i) { + var opacity = 0; + var imageOffset = i * numComps; + for (j = 0; j < numComps; ++j) { + var color = image[imageOffset + j]; + var maskOffset = j * 2; + if (color < mask[maskOffset] || color > mask[maskOffset + 1]) { + opacity = 255; + break; + } + } + alphaBuf[i] = opacity; + } + } else { + error('Unknown mask format.'); + } + } - CalGrayCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function CalGrayCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(this, src, srcOffset, dest, destOffset, 1); + if (alphaBuf) { + for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { + rgbaBuf[j] = alphaBuf[i]; + } + } else { + // No mask. + for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { + rgbaBuf[j] = 255; + } + } }, - getRgbBuffer: function CalGrayCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; ++i) { - convertToRgb(this, src, srcOffset, dest, destOffset, scale); - srcOffset += 1; - destOffset += 3 + alpha01; + undoPreblend: function PDFImage_undoPreblend(buffer, width, height) { + var matte = this.smask && this.smask.matte; + if (!matte) { + return; + } + var matteRgb = this.colorSpace.getRgb(matte, 0); + var matteR = matteRgb[0]; + var matteG = matteRgb[1]; + var matteB = matteRgb[2]; + var length = width * height * 4; + var r, g, b; + for (var i = 0; i < length; i += 4) { + var alpha = buffer[i + 3]; + if (alpha === 0) { + // according formula we have to get Infinity in all components + // making it white (typical paper color) should be okay + buffer[i] = 255; + buffer[i + 1] = 255; + buffer[i + 2] = 255; + continue; + } + var k = 255 / alpha; + r = (buffer[i] - matteR) * k + matteR; + g = (buffer[i + 1] - matteG) * k + matteG; + b = (buffer[i + 2] - matteB) * k + matteB; + buffer[i] = r <= 0 ? 0 : r >= 255 ? 255 : r | 0; + buffer[i + 1] = g <= 0 ? 0 : g >= 255 ? 255 : g | 0; + buffer[i + 2] = b <= 0 ? 0 : b >= 255 ? 255 : b | 0; } }, - getOutputLength: function CalGrayCS_getOutputLength(inputLength, alpha01) { - return inputLength * (3 + alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function CalGrayCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return CalGrayCS; -})(); - -// -// CalRGBCS: Based on "PDF Reference, Sixth Ed", p.247 -// -var CalRGBCS = (function CalRGBCSClosure() { - - // See http://www.brucelindbloom.com/index.html?Eqn_ChromAdapt.html for these - // matrices. - var BRADFORD_SCALE_MATRIX = new Float32Array([ - 0.8951, 0.2664, -0.1614, - -0.7502, 1.7135, 0.0367, - 0.0389, -0.0685, 1.0296]); - - var BRADFORD_SCALE_INVERSE_MATRIX = new Float32Array([ - 0.9869929, -0.1470543, 0.1599627, - 0.4323053, 0.5183603, 0.0492912, - -0.0085287, 0.0400428, 0.9684867]); - - // See http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html. - var SRGB_D65_XYZ_TO_RGB_MATRIX = new Float32Array([ - 3.2404542, -1.5371385, -0.4985314, - -0.9692660, 1.8760108, 0.0415560, - 0.0556434, -0.2040259, 1.0572252]); - var FLAT_WHITEPOINT_MATRIX = new Float32Array([1, 1, 1]); + createImageData: function PDFImage_createImageData(forceRGBA) { + var drawWidth = this.drawWidth; + var drawHeight = this.drawHeight; + var imgData = { // other fields are filled in below + width: drawWidth, + height: drawHeight + }; - var tempNormalizeMatrix = new Float32Array(3); - var tempConvertMatrix1 = new Float32Array(3); - var tempConvertMatrix2 = new Float32Array(3); + var numComps = this.numComps; + var originalWidth = this.width; + var originalHeight = this.height; + var bpc = this.bpc; - var DECODE_L_CONSTANT = Math.pow(((8 + 16) / 116), 3) / 8.0; + // Rows start at byte boundary. + var rowBytes = (originalWidth * numComps * bpc + 7) >> 3; + var imgArray; - function CalRGBCS(whitePoint, blackPoint, gamma, matrix) { - this.name = 'CalRGB'; - this.numComps = 3; - this.defaultColor = new Float32Array(3); + if (!forceRGBA) { + // If it is a 1-bit-per-pixel grayscale (i.e. black-and-white) image + // without any complications, we pass a same-sized copy to the main + // thread rather than expanding by 32x to RGBA form. This saves *lots* + // of memory for many scanned documents. It's also much faster. + // + // Similarly, if it is a 24-bit-per pixel RGB image without any + // complications, we avoid expanding by 1.333x to RGBA form. + var kind; + if (this.colorSpace.name === 'DeviceGray' && bpc === 1) { + kind = ImageKind.GRAYSCALE_1BPP; + } else if (this.colorSpace.name === 'DeviceRGB' && bpc === 8 && + !this.needsDecode) { + kind = ImageKind.RGB_24BPP; + } + if (kind && !this.smask && !this.mask && + drawWidth === originalWidth && drawHeight === originalHeight) { + imgData.kind = kind; - if (!whitePoint) { - error('WhitePoint missing - required for color space CalRGB'); - } - blackPoint = blackPoint || new Float32Array(3); - gamma = gamma || new Float32Array([1, 1, 1]); - matrix = matrix || new Float32Array([1, 0, 0, 0, 1, 0, 0, 0, 1]); + imgArray = this.getImageBytes(originalHeight * rowBytes); + // If imgArray came from a DecodeStream, we're safe to transfer it + // (and thus detach its underlying buffer) because it will constitute + // the entire DecodeStream's data. But if it came from a Stream, we + // need to copy it because it'll only be a portion of the Stream's + // data, and the rest will be read later on. + if (this.image instanceof DecodeStream) { + imgData.data = imgArray; + } else { + var newArray = new Uint8Array(imgArray.length); + newArray.set(imgArray); + imgData.data = newArray; + } + if (this.needsDecode) { + // Invert the buffer (which must be grayscale if we reached here). + assert(kind === ImageKind.GRAYSCALE_1BPP); + var buffer = imgData.data; + for (var i = 0, ii = buffer.length; i < ii; i++) { + buffer[i] ^= 0xff; + } + } + return imgData; + } + if (this.image instanceof JpegStream && !this.smask && !this.mask && + (this.colorSpace.name === 'DeviceGray' || + this.colorSpace.name === 'DeviceRGB' || + this.colorSpace.name === 'DeviceCMYK')) { + imgData.kind = ImageKind.RGB_24BPP; + imgData.data = this.getImageBytes(originalHeight * rowBytes, + drawWidth, drawHeight, true); + return imgData; + } + } - // Translate arguments to spec variables. - var XW = whitePoint[0]; - var YW = whitePoint[1]; - var ZW = whitePoint[2]; - this.whitePoint = whitePoint; + imgArray = this.getImageBytes(originalHeight * rowBytes); + // imgArray can be incomplete (e.g. after CCITT fax encoding). + var actualHeight = 0 | (imgArray.length / rowBytes * + drawHeight / originalHeight); - var XB = blackPoint[0]; - var YB = blackPoint[1]; - var ZB = blackPoint[2]; - this.blackPoint = blackPoint; + var comps = this.getComponents(imgArray); - this.GR = gamma[0]; - this.GG = gamma[1]; - this.GB = gamma[2]; + // If opacity data is present, use RGBA_32BPP form. Otherwise, use the + // more compact RGB_24BPP form if allowable. + var alpha01, maybeUndoPreblend; + if (!forceRGBA && !this.smask && !this.mask) { + imgData.kind = ImageKind.RGB_24BPP; + imgData.data = new Uint8Array(drawWidth * drawHeight * 3); + alpha01 = 0; + maybeUndoPreblend = false; + } else { + imgData.kind = ImageKind.RGBA_32BPP; + imgData.data = new Uint8Array(drawWidth * drawHeight * 4); + alpha01 = 1; + maybeUndoPreblend = true; - this.MXA = matrix[0]; - this.MYA = matrix[1]; - this.MZA = matrix[2]; - this.MXB = matrix[3]; - this.MYB = matrix[4]; - this.MZB = matrix[5]; - this.MXC = matrix[6]; - this.MYC = matrix[7]; - this.MZC = matrix[8]; + // Color key masking (opacity) must be performed before decoding. + this.fillOpacity(imgData.data, drawWidth, drawHeight, actualHeight, + comps); + } - // Validate variables as per spec. - if (XW < 0 || ZW < 0 || YW !== 1) { - error('Invalid WhitePoint components for ' + this.name + - ', no fallback available'); - } + if (this.needsDecode) { + this.decodeBuffer(comps); + } + this.colorSpace.fillRgb(imgData.data, originalWidth, originalHeight, + drawWidth, drawHeight, actualHeight, bpc, comps, + alpha01); + if (maybeUndoPreblend) { + this.undoPreblend(imgData.data, drawWidth, actualHeight); + } - if (XB < 0 || YB < 0 || ZB < 0) { - info('Invalid BlackPoint for ' + this.name + ' [' + XB + ', ' + YB + - ', ' + ZB + '], falling back to default'); - this.blackPoint = new Float32Array(3); - } + return imgData; + }, - if (this.GR < 0 || this.GG < 0 || this.GB < 0) { - info('Invalid Gamma [' + this.GR + ', ' + this.GG + ', ' + this.GB + - '] for ' + this.name + ', falling back to default'); - this.GR = this.GG = this.GB = 1; - } + fillGrayBuffer: function PDFImage_fillGrayBuffer(buffer) { + var numComps = this.numComps; + if (numComps !== 1) { + error('Reading gray scale from a color image: ' + numComps); + } - if (this.MXA < 0 || this.MYA < 0 || this.MZA < 0 || - this.MXB < 0 || this.MYB < 0 || this.MZB < 0 || - this.MXC < 0 || this.MYC < 0 || this.MZC < 0) { - info('Invalid Matrix for ' + this.name + ' [' + - this.MXA + ', ' + this.MYA + ', ' + this.MZA + - this.MXB + ', ' + this.MYB + ', ' + this.MZB + - this.MXC + ', ' + this.MYC + ', ' + this.MZC + - '], falling back to default'); - this.MXA = this.MYB = this.MZC = 1; - this.MXB = this.MYA = this.MZA = this.MXC = this.MYC = this.MZB = 0; - } - } + var width = this.width; + var height = this.height; + var bpc = this.bpc; - function matrixProduct(a, b, result) { - result[0] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; - result[1] = a[3] * b[0] + a[4] * b[1] + a[5] * b[2]; - result[2] = a[6] * b[0] + a[7] * b[1] + a[8] * b[2]; - } + // rows start at byte boundary + var rowBytes = (width * numComps * bpc + 7) >> 3; + var imgArray = this.getImageBytes(height * rowBytes); - function convertToFlat(sourceWhitePoint, LMS, result) { - result[0] = LMS[0] * 1 / sourceWhitePoint[0]; - result[1] = LMS[1] * 1 / sourceWhitePoint[1]; - result[2] = LMS[2] * 1 / sourceWhitePoint[2]; - } + var comps = this.getComponents(imgArray); + var i, length; - function convertToD65(sourceWhitePoint, LMS, result) { - var D65X = 0.95047; - var D65Y = 1; - var D65Z = 1.08883; + if (bpc === 1) { + // inline decoding (= inversion) for 1 bpc images + length = width * height; + if (this.needsDecode) { + // invert and scale to {0, 255} + for (i = 0; i < length; ++i) { + buffer[i] = (comps[i] - 1) & 255; + } + } else { + // scale to {0, 255} + for (i = 0; i < length; ++i) { + buffer[i] = (-comps[i]) & 255; + } + } + return; + } - result[0] = LMS[0] * D65X / sourceWhitePoint[0]; - result[1] = LMS[1] * D65Y / sourceWhitePoint[1]; - result[2] = LMS[2] * D65Z / sourceWhitePoint[2]; - } + if (this.needsDecode) { + this.decodeBuffer(comps); + } + length = width * height; + // we aren't using a colorspace so we need to scale the value + var scale = 255 / ((1 << bpc) - 1); + for (i = 0; i < length; ++i) { + buffer[i] = (scale * comps[i]) | 0; + } + }, - function sRGBTransferFunction(color) { - // See http://en.wikipedia.org/wiki/SRGB. - if (color <= 0.0031308){ - return adjustToRange(0, 1, 12.92 * color); + getImageBytes: function PDFImage_getImageBytes(length, + drawWidth, drawHeight, + forceRGB) { + this.image.reset(); + this.image.drawWidth = drawWidth || this.width; + this.image.drawHeight = drawHeight || this.height; + this.image.forceRGB = !!forceRGB; + return this.image.getBytes(length); } + }; + return PDFImage; +})(); - return adjustToRange(0, 1, (1 + 0.055) * Math.pow(color, 1 / 2.4) - 0.055); - } - - function adjustToRange(min, max, value) { - return Math.max(min, Math.min(max, value)); - } +exports.PDFImage = PDFImage; - function decodeL(L) { - if (L < 0) { - return -decodeL(-L); - } +// TODO refactor to remove dependency on colorspace.js +coreColorSpace._setCoreImage(exports); +})); - if (L > 8.0) { - return Math.pow(((L + 16) / 116), 3); - } - return L * DECODE_L_CONSTANT; +(function (root, factory) { + { + factory((root.pdfjsCoreObj = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreCrypto, root.pdfjsCoreParser, + root.pdfjsCoreChunkedStream, root.pdfjsCoreColorSpace); } +}(this, function (exports, sharedUtil, corePrimitives, coreCrypto, coreParser, + coreChunkedStream, coreColorSpace) { - function compensateBlackPoint(sourceBlackPoint, XYZ_Flat, result) { - - // In case the blackPoint is already the default blackPoint then there is - // no need to do compensation. - if (sourceBlackPoint[0] === 0 && - sourceBlackPoint[1] === 0 && - sourceBlackPoint[2] === 0) { - result[0] = XYZ_Flat[0]; - result[1] = XYZ_Flat[1]; - result[2] = XYZ_Flat[2]; - return; - } +var InvalidPDFException = sharedUtil.InvalidPDFException; +var MissingDataException = sharedUtil.MissingDataException; +var XRefParseException = sharedUtil.XRefParseException; +var assert = sharedUtil.assert; +var bytesToString = sharedUtil.bytesToString; +var createPromiseCapability = sharedUtil.createPromiseCapability; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isInt = sharedUtil.isInt; +var isString = sharedUtil.isString; +var shadow = sharedUtil.shadow; +var stringToPDFString = sharedUtil.stringToPDFString; +var stringToUTF8String = sharedUtil.stringToUTF8String; +var warn = sharedUtil.warn; +var isValidUrl = sharedUtil.isValidUrl; +var Util = sharedUtil.Util; +var Ref = corePrimitives.Ref; +var RefSet = corePrimitives.RefSet; +var RefSetCache = corePrimitives.RefSetCache; +var isName = corePrimitives.isName; +var isCmd = corePrimitives.isCmd; +var isDict = corePrimitives.isDict; +var isRef = corePrimitives.isRef; +var isStream = corePrimitives.isStream; +var CipherTransformFactory = coreCrypto.CipherTransformFactory; +var Lexer = coreParser.Lexer; +var Parser = coreParser.Parser; +var ChunkedStream = coreChunkedStream.ChunkedStream; +var ColorSpace = coreColorSpace.ColorSpace; - // For the blackPoint calculation details, please see - // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ - // AdobeBPC.pdf. - // The destination blackPoint is the default blackPoint [0, 0, 0]. - var zeroDecodeL = decodeL(0); +var Catalog = (function CatalogClosure() { + function Catalog(pdfManager, xref, pageFactory) { + this.pdfManager = pdfManager; + this.xref = xref; + this.catDict = xref.getCatalogObj(); + this.fontCache = new RefSetCache(); + assert(isDict(this.catDict), + 'catalog object is not a dictionary'); - var X_DST = zeroDecodeL; - var X_SRC = decodeL(sourceBlackPoint[0]); + // TODO refactor to move getPage() to the PDFDocument. + this.pageFactory = pageFactory; + this.pagePromises = []; + } - var Y_DST = zeroDecodeL; - var Y_SRC = decodeL(sourceBlackPoint[1]); + Catalog.prototype = { + get metadata() { + var streamRef = this.catDict.getRaw('Metadata'); + if (!isRef(streamRef)) { + return shadow(this, 'metadata', null); + } - var Z_DST = zeroDecodeL; - var Z_SRC = decodeL(sourceBlackPoint[2]); + var encryptMetadata = (!this.xref.encrypt ? false : + this.xref.encrypt.encryptMetadata); - var X_Scale = (1 - X_DST) / (1 - X_SRC); - var X_Offset = 1 - X_Scale; + var stream = this.xref.fetch(streamRef, !encryptMetadata); + var metadata; + if (stream && isDict(stream.dict)) { + var type = stream.dict.get('Type'); + var subtype = stream.dict.get('Subtype'); - var Y_Scale = (1 - Y_DST) / (1 - Y_SRC); - var Y_Offset = 1 - Y_Scale; + if (isName(type) && isName(subtype) && + type.name === 'Metadata' && subtype.name === 'XML') { + // XXX: This should examine the charset the XML document defines, + // however since there are currently no real means to decode + // arbitrary charsets, let's just hope that the author of the PDF + // was reasonable enough to stick with the XML default charset, + // which is UTF-8. + try { + metadata = stringToUTF8String(bytesToString(stream.getBytes())); + } catch (e) { + info('Skipping invalid metadata.'); + } + } + } - var Z_Scale = (1 - Z_DST) / (1 - Z_SRC); - var Z_Offset = 1 - Z_Scale; + return shadow(this, 'metadata', metadata); + }, + get toplevelPagesDict() { + var pagesObj = this.catDict.get('Pages'); + assert(isDict(pagesObj), 'invalid top-level pages dictionary'); + // shadow the prototype getter + return shadow(this, 'toplevelPagesDict', pagesObj); + }, + get documentOutline() { + var obj = null; + try { + obj = this.readDocumentOutline(); + } catch (ex) { + if (ex instanceof MissingDataException) { + throw ex; + } + warn('Unable to read document outline'); + } + return shadow(this, 'documentOutline', obj); + }, + readDocumentOutline: function Catalog_readDocumentOutline() { + var obj = this.catDict.get('Outlines'); + if (!isDict(obj)) { + return null; + } + obj = obj.getRaw('First'); + if (!isRef(obj)) { + return null; + } + var root = { items: [] }; + var queue = [{obj: obj, parent: root}]; + // To avoid recursion, keep track of the already processed items. + var processed = new RefSet(); + processed.put(obj); + var xref = this.xref, blackColor = new Uint8Array(3); - result[0] = XYZ_Flat[0] * X_Scale + X_Offset; - result[1] = XYZ_Flat[1] * Y_Scale + Y_Offset; - result[2] = XYZ_Flat[2] * Z_Scale + Z_Offset; - } + while (queue.length > 0) { + var i = queue.shift(); + var outlineDict = xref.fetchIfRef(i.obj); + if (outlineDict === null) { + continue; + } + assert(outlineDict.has('Title'), 'Invalid outline item'); - function normalizeWhitePointToFlat(sourceWhitePoint, XYZ_In, result) { + var actionDict = outlineDict.get('A'), dest = null, url = null; + if (actionDict) { + var destEntry = actionDict.get('D'); + if (destEntry) { + dest = destEntry; + } else { + var uriEntry = actionDict.get('URI'); + if (isString(uriEntry) && isValidUrl(uriEntry, false)) { + url = uriEntry; + } + } + } else if (outlineDict.has('Dest')) { + dest = outlineDict.getRaw('Dest'); + if (isName(dest)) { + dest = dest.name; + } + } + var title = outlineDict.get('Title'); + var flags = outlineDict.get('F') || 0; + + var color = outlineDict.get('C'), rgbColor = blackColor; + // We only need to parse the color when it's valid, and non-default. + if (isArray(color) && color.length === 3 && + (color[0] !== 0 || color[1] !== 0 || color[2] !== 0)) { + rgbColor = ColorSpace.singletons.rgb.getRgb(color, 0); + } + var outlineItem = { + dest: dest, + url: url, + title: stringToPDFString(title), + color: rgbColor, + count: outlineDict.get('Count'), + bold: !!(flags & 2), + italic: !!(flags & 1), + items: [] + }; + i.parent.items.push(outlineItem); + obj = outlineDict.getRaw('First'); + if (isRef(obj) && !processed.has(obj)) { + queue.push({obj: obj, parent: outlineItem}); + processed.put(obj); + } + obj = outlineDict.getRaw('Next'); + if (isRef(obj) && !processed.has(obj)) { + queue.push({obj: obj, parent: i.parent}); + processed.put(obj); + } + } + return (root.items.length > 0 ? root.items : null); + }, + get numPages() { + var obj = this.toplevelPagesDict.get('Count'); + assert( + isInt(obj), + 'page count in top level pages object is not an integer' + ); + // shadow the prototype getter + return shadow(this, 'num', obj); + }, + get destinations() { + function fetchDestination(dest) { + return isDict(dest) ? dest.get('D') : dest; + } - // In case the whitePoint is already flat then there is no need to do - // normalization. - if (sourceWhitePoint[0] === 1 && sourceWhitePoint[2] === 1) { - result[0] = XYZ_In[0]; - result[1] = XYZ_In[1]; - result[2] = XYZ_In[2]; - return; - } + var xref = this.xref; + var dests = {}, nameTreeRef, nameDictionaryRef; + var obj = this.catDict.get('Names'); + if (obj && obj.has('Dests')) { + nameTreeRef = obj.getRaw('Dests'); + } else if (this.catDict.has('Dests')) { + nameDictionaryRef = this.catDict.get('Dests'); + } - var LMS = result; - matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + if (nameDictionaryRef) { + // reading simple destination dictionary + obj = nameDictionaryRef; + obj.forEach(function catalogForEach(key, value) { + if (!value) { + return; + } + dests[key] = fetchDestination(value); + }); + } + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + var names = nameTree.getAll(); + for (var name in names) { + dests[name] = fetchDestination(names[name]); + } + } + return shadow(this, 'destinations', dests); + }, + getDestination: function Catalog_getDestination(destinationId) { + function fetchDestination(dest) { + return isDict(dest) ? dest.get('D') : dest; + } - var LMS_Flat = tempNormalizeMatrix; - convertToFlat(sourceWhitePoint, LMS, LMS_Flat); + var xref = this.xref; + var dest = null, nameTreeRef, nameDictionaryRef; + var obj = this.catDict.get('Names'); + if (obj && obj.has('Dests')) { + nameTreeRef = obj.getRaw('Dests'); + } else if (this.catDict.has('Dests')) { + nameDictionaryRef = this.catDict.get('Dests'); + } - matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_Flat, result); - } + if (nameDictionaryRef) { // Simple destination dictionary. + var value = nameDictionaryRef.get(destinationId); + if (value) { + dest = fetchDestination(value); + } + } + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + dest = fetchDestination(nameTree.get(destinationId)); + } + return dest; + }, - function normalizeWhitePointToD65(sourceWhitePoint, XYZ_In, result) { + get pageLabels() { + var obj = null; + try { + obj = this.readPageLabels(); + } catch (ex) { + if (ex instanceof MissingDataException) { + throw ex; + } + warn('Unable to read page labels.'); + } + return shadow(this, 'pageLabels', obj); + }, + readPageLabels: function Catalog_readPageLabels() { + var obj = this.catDict.getRaw('PageLabels'); + if (!obj) { + return null; + } + var pageLabels = new Array(this.numPages); + var style = null; + var prefix = ''; + var start = 1; - var LMS = result; - matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + var numberTree = new NumberTree(obj, this.xref); + var nums = numberTree.getAll(); + var currentLabel = '', currentIndex = 1; - var LMS_D65 = tempNormalizeMatrix; - convertToD65(sourceWhitePoint, LMS, LMS_D65); + for (var i = 0, ii = this.numPages; i < ii; i++) { + if (i in nums) { + var labelDict = nums[i]; + assert(isDict(labelDict), 'The PageLabel is not a dictionary.'); - matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_D65, result); - } + var type = labelDict.get('Type'); + assert(!type || (isName(type) && type.name === 'PageLabel'), + 'Invalid type in PageLabel dictionary.'); - function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { - // A, B and C represent a red, green and blue components of a calibrated - // rgb space. - var A = adjustToRange(0, 1, src[srcOffset] * scale); - var B = adjustToRange(0, 1, src[srcOffset + 1] * scale); - var C = adjustToRange(0, 1, src[srcOffset + 2] * scale); + var s = labelDict.get('S'); + assert(!s || isName(s), 'Invalid style in PageLabel dictionary.'); + style = (s ? s.name : null); - // A <---> AGR in the spec - // B <---> BGG in the spec - // C <---> CGB in the spec - var AGR = Math.pow(A, cs.GR); - var BGG = Math.pow(B, cs.GG); - var CGB = Math.pow(C, cs.GB); + prefix = labelDict.get('P') || ''; + assert(isString(prefix), 'Invalid prefix in PageLabel dictionary.'); - // Computes intermediate variables L, M, N as per spec. - // To decode X, Y, Z values map L, M, N directly to them. - var X = cs.MXA * AGR + cs.MXB * BGG + cs.MXC * CGB; - var Y = cs.MYA * AGR + cs.MYB * BGG + cs.MYC * CGB; - var Z = cs.MZA * AGR + cs.MZB * BGG + cs.MZC * CGB; + start = labelDict.get('St') || 1; + assert(isInt(start), 'Invalid start in PageLabel dictionary.'); + currentIndex = start; + } - // The following calculations are based on this document: - // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ - // AdobeBPC.pdf. - var XYZ = tempConvertMatrix1; - XYZ[0] = X; - XYZ[1] = Y; - XYZ[2] = Z; - var XYZ_Flat = tempConvertMatrix2; + switch (style) { + case 'D': + currentLabel = currentIndex; + break; + case 'R': + case 'r': + currentLabel = Util.toRoman(currentIndex, style === 'r'); + break; + case 'A': + case 'a': + var LIMIT = 26; // Use only the characters A--Z, or a--z. + var A_UPPER_CASE = 0x41, A_LOWER_CASE = 0x61; - normalizeWhitePointToFlat(cs.whitePoint, XYZ, XYZ_Flat); + var baseCharCode = (style === 'a' ? A_LOWER_CASE : A_UPPER_CASE); + var letterIndex = currentIndex - 1; + var character = String.fromCharCode(baseCharCode + + (letterIndex % LIMIT)); + var charBuf = []; + for (var j = 0, jj = (letterIndex / LIMIT) | 0; j <= jj; j++) { + charBuf.push(character); + } + currentLabel = charBuf.join(''); + break; + default: + assert(!style, + 'Invalid style "' + style + '" in PageLabel dictionary.'); + } + pageLabels[i] = prefix + currentLabel; - var XYZ_Black = tempConvertMatrix1; - compensateBlackPoint(cs.blackPoint, XYZ_Flat, XYZ_Black); + currentLabel = ''; + currentIndex++; + } + return pageLabels; + }, - var XYZ_D65 = tempConvertMatrix2; - normalizeWhitePointToD65(FLAT_WHITEPOINT_MATRIX, XYZ_Black, XYZ_D65); + get attachments() { + var xref = this.xref; + var attachments = null, nameTreeRef; + var obj = this.catDict.get('Names'); + if (obj) { + nameTreeRef = obj.getRaw('EmbeddedFiles'); + } - var SRGB = tempConvertMatrix1; - matrixProduct(SRGB_D65_XYZ_TO_RGB_MATRIX, XYZ_D65, SRGB); + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + var names = nameTree.getAll(); + for (var name in names) { + var fs = new FileSpec(names[name], xref); + if (!attachments) { + attachments = Object.create(null); + } + attachments[stringToPDFString(name)] = fs.serializable; + } + } + return shadow(this, 'attachments', attachments); + }, + get javaScript() { + var xref = this.xref; + var obj = this.catDict.get('Names'); - var sR = sRGBTransferFunction(SRGB[0]); - var sG = sRGBTransferFunction(SRGB[1]); - var sB = sRGBTransferFunction(SRGB[2]); + var javaScript = []; + function appendIfJavaScriptDict(jsDict) { + var type = jsDict.get('S'); + if (!isName(type) || type.name !== 'JavaScript') { + return; + } + var js = jsDict.get('JS'); + if (isStream(js)) { + js = bytesToString(js.getBytes()); + } else if (!isString(js)) { + return; + } + javaScript.push(stringToPDFString(js)); + } + if (obj && obj.has('JavaScript')) { + var nameTree = new NameTree(obj.getRaw('JavaScript'), xref); + var names = nameTree.getAll(); + for (var name in names) { + // We don't really use the JavaScript right now. This code is + // defensive so we don't cause errors on document load. + var jsDict = names[name]; + if (isDict(jsDict)) { + appendIfJavaScriptDict(jsDict); + } + } + } - // Convert the values to rgb range [0, 255]. - dest[destOffset] = Math.round(sR * 255); - dest[destOffset + 1] = Math.round(sG * 255); - dest[destOffset + 2] = Math.round(sB * 255); - } + // Append OpenAction actions to javaScript array + var openactionDict = this.catDict.get('OpenAction'); + if (isDict(openactionDict, 'Action')) { + var actionType = openactionDict.get('S'); + if (isName(actionType) && actionType.name === 'Named') { + // The named Print action is not a part of the PDF 1.7 specification, + // but is supported by many PDF readers/writers (including Adobe's). + var action = openactionDict.get('N'); + if (isName(action) && action.name === 'Print') { + javaScript.push('print({});'); + } + } else { + appendIfJavaScriptDict(openactionDict); + } + } - CalRGBCS.prototype = { - getRgb: function CalRGBCS_getRgb(src, srcOffset) { - var rgb = new Uint8Array(3); - this.getRgbItem(src, srcOffset, rgb, 0); - return rgb; + return shadow(this, 'javaScript', javaScript); }, - getRgbItem: function CalRGBCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(this, src, srcOffset, dest, destOffset, 1); + + cleanup: function Catalog_cleanup() { + var promises = []; + this.fontCache.forEach(function (promise) { + promises.push(promise); + }); + return Promise.all(promises).then(function (translatedFonts) { + for (var i = 0, ii = translatedFonts.length; i < ii; i++) { + var font = translatedFonts[i].dict; + delete font.translated; + } + this.fontCache.clear(); + }.bind(this)); }, - getRgbBuffer: function CalRGBCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; ++i) { - convertToRgb(this, src, srcOffset, dest, destOffset, scale); - srcOffset += 3; - destOffset += 3 + alpha01; + getPage: function Catalog_getPage(pageIndex) { + if (!(pageIndex in this.pagePromises)) { + this.pagePromises[pageIndex] = this.getPageDict(pageIndex).then( + function (a) { + var dict = a[0]; + var ref = a[1]; + return this.pageFactory.createPage(pageIndex, dict, ref, + this.fontCache); + }.bind(this) + ); } + return this.pagePromises[pageIndex]; }, - getOutputLength: function CalRGBCS_getOutputLength(inputLength, alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function CalRGBCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return CalRGBCS; -})(); - -// -// LabCS: Based on "PDF Reference, Sixth Ed", p.250 -// -var LabCS = (function LabCSClosure() { - function LabCS(whitePoint, blackPoint, range) { - this.name = 'Lab'; - this.numComps = 3; - this.defaultColor = new Float32Array([0, 0, 0]); - - if (!whitePoint) { - error('WhitePoint missing - required for color space Lab'); - } - blackPoint = blackPoint || [0, 0, 0]; - range = range || [-100, 100, -100, 100]; - - // Translate args to spec variables - this.XW = whitePoint[0]; - this.YW = whitePoint[1]; - this.ZW = whitePoint[2]; - this.amin = range[0]; - this.amax = range[1]; - this.bmin = range[2]; - this.bmax = range[3]; - - // These are here just for completeness - the spec doesn't offer any - // formulas that use BlackPoint in Lab - this.XB = blackPoint[0]; - this.YB = blackPoint[1]; - this.ZB = blackPoint[2]; - // Validate vars as per spec - if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { - error('Invalid WhitePoint components, no fallback available'); - } - - if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { - info('Invalid BlackPoint, falling back to default'); - this.XB = this.YB = this.ZB = 0; - } - - if (this.amin > this.amax || this.bmin > this.bmax) { - info('Invalid Range, falling back to defaults'); - this.amin = -100; - this.amax = 100; - this.bmin = -100; - this.bmax = 100; - } - } + getPageDict: function Catalog_getPageDict(pageIndex) { + var capability = createPromiseCapability(); + var nodesToVisit = [this.catDict.getRaw('Pages')]; + var currentPageIndex = 0; + var xref = this.xref; + var checkAllKids = false; - // Function g(x) from spec - function fn_g(x) { - if (x >= 6 / 29) { - return x * x * x; - } else { - return (108 / 841) * (x - 4 / 29); - } - } + function next() { + while (nodesToVisit.length) { + var currentNode = nodesToVisit.pop(); - function decode(value, high1, low2, high2) { - return low2 + (value) * (high2 - low2) / (high1); - } + if (isRef(currentNode)) { + xref.fetchAsync(currentNode).then(function (obj) { + if (isDict(obj, 'Page') || (isDict(obj) && !obj.has('Kids'))) { + if (pageIndex === currentPageIndex) { + capability.resolve([obj, currentNode]); + } else { + currentPageIndex++; + next(); + } + return; + } + nodesToVisit.push(obj); + next(); + }, capability.reject); + return; + } - // If decoding is needed maxVal should be 2^bits per component - 1. - function convertToRgb(cs, src, srcOffset, maxVal, dest, destOffset) { - // XXX: Lab input is in the range of [0, 100], [amin, amax], [bmin, bmax] - // not the usual [0, 1]. If a command like setFillColor is used the src - // values will already be within the correct range. However, if we are - // converting an image we have to map the values to the correct range given - // above. - // Ls,as,bs <---> L*,a*,b* in the spec - var Ls = src[srcOffset]; - var as = src[srcOffset + 1]; - var bs = src[srcOffset + 2]; - if (maxVal !== false) { - Ls = decode(Ls, maxVal, 0, 100); - as = decode(as, maxVal, cs.amin, cs.amax); - bs = decode(bs, maxVal, cs.bmin, cs.bmax); - } + // Must be a child page dictionary. + assert( + isDict(currentNode), + 'page dictionary kid reference points to wrong type of object' + ); + var count = currentNode.get('Count'); + // If the current node doesn't have any children, avoid getting stuck + // in an empty node further down in the tree (see issue5644.pdf). + if (count === 0) { + checkAllKids = true; + } + // Skip nodes where the page can't be. + if (currentPageIndex + count <= pageIndex) { + currentPageIndex += count; + continue; + } - // Adjust limits of 'as' and 'bs' - as = as > cs.amax ? cs.amax : as < cs.amin ? cs.amin : as; - bs = bs > cs.bmax ? cs.bmax : bs < cs.bmin ? cs.bmin : bs; + var kids = currentNode.get('Kids'); + assert(isArray(kids), 'page dictionary kids object is not an array'); + if (!checkAllKids && count === kids.length) { + // Nodes that don't have the page have been skipped and this is the + // bottom of the tree which means the page requested must be a + // descendant of this pages node. Ideally we would just resolve the + // promise with the page ref here, but there is the case where more + // pages nodes could link to single a page (see issue 3666 pdf). To + // handle this push it back on the queue so if it is a pages node it + // will be descended into. + nodesToVisit = [kids[pageIndex - currentPageIndex]]; + currentPageIndex = pageIndex; + continue; + } else { + for (var last = kids.length - 1; last >= 0; last--) { + nodesToVisit.push(kids[last]); + } + } + } + capability.reject('Page index ' + pageIndex + ' not found.'); + } + next(); + return capability.promise; + }, - // Computes intermediate variables X,Y,Z as per spec - var M = (Ls + 16) / 116; - var L = M + (as / 500); - var N = M - (bs / 200); + getPageIndex: function Catalog_getPageIndex(ref) { + // The page tree nodes have the count of all the leaves below them. To get + // how many pages are before we just have to walk up the tree and keep + // adding the count of siblings to the left of the node. + var xref = this.xref; + function pagesBeforeRef(kidRef) { + var total = 0; + var parentRef; + return xref.fetchAsync(kidRef).then(function (node) { + if (!node) { + return null; + } + parentRef = node.getRaw('Parent'); + return node.getAsync('Parent'); + }).then(function (parent) { + if (!parent) { + return null; + } + return parent.getAsync('Kids'); + }).then(function (kids) { + if (!kids) { + return null; + } + var kidPromises = []; + var found = false; + for (var i = 0; i < kids.length; i++) { + var kid = kids[i]; + assert(isRef(kid), 'kids must be a ref'); + if (kid.num === kidRef.num) { + found = true; + break; + } + kidPromises.push(xref.fetchAsync(kid).then(function (kid) { + if (kid.has('Count')) { + var count = kid.get('Count'); + total += count; + } else { // page leaf node + total++; + } + })); + } + if (!found) { + error('kid ref not found in parents kids'); + } + return Promise.all(kidPromises).then(function () { + return [total, parentRef]; + }); + }); + } - var X = cs.XW * fn_g(L); - var Y = cs.YW * fn_g(M); - var Z = cs.ZW * fn_g(N); + var total = 0; + function next(ref) { + return pagesBeforeRef(ref).then(function (args) { + if (!args) { + return total; + } + var count = args[0]; + var parentRef = args[1]; + total += count; + return next(parentRef); + }); + } - var r, g, b; - // Using different conversions for D50 and D65 white points, - // per http://www.color.org/srgb.pdf - if (cs.ZW < 1) { - // Assuming D50 (X=0.9642, Y=1.00, Z=0.8249) - r = X * 3.1339 + Y * -1.6170 + Z * -0.4906; - g = X * -0.9785 + Y * 1.9160 + Z * 0.0333; - b = X * 0.0720 + Y * -0.2290 + Z * 1.4057; - } else { - // Assuming D65 (X=0.9505, Y=1.00, Z=1.0888) - r = X * 3.2406 + Y * -1.5372 + Z * -0.4986; - g = X * -0.9689 + Y * 1.8758 + Z * 0.0415; - b = X * 0.0557 + Y * -0.2040 + Z * 1.0570; + return next(ref); } - // clamp color values to [0,1] range then convert to [0,255] range. - dest[destOffset] = r <= 0 ? 0 : r >= 1 ? 255 : Math.sqrt(r) * 255 | 0; - dest[destOffset + 1] = g <= 0 ? 0 : g >= 1 ? 255 : Math.sqrt(g) * 255 | 0; - dest[destOffset + 2] = b <= 0 ? 0 : b >= 1 ? 255 : Math.sqrt(b) * 255 | 0; + }; + + return Catalog; +})(); + +var XRef = (function XRefClosure() { + function XRef(stream, password) { + this.stream = stream; + this.entries = []; + this.xrefstms = Object.create(null); + // prepare the XRef cache + this.cache = []; + this.password = password; + this.stats = { + streamTypes: [], + fontTypes: [] + }; } - LabCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function LabCS_getRgbItem(src, srcOffset, dest, destOffset) { - convertToRgb(this, src, srcOffset, false, dest, destOffset); + XRef.prototype = { + setStartXRef: function XRef_setStartXRef(startXRef) { + // Store the starting positions of xref tables as we process them + // so we can recover from missing data errors + this.startXRefQueue = [startXRef]; }, - getRgbBuffer: function LabCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var maxVal = (1 << bits) - 1; - for (var i = 0; i < count; i++) { - convertToRgb(this, src, srcOffset, maxVal, dest, destOffset); - srcOffset += 3; - destOffset += 3 + alpha01; + + parse: function XRef_parse(recoveryMode) { + var trailerDict; + if (!recoveryMode) { + trailerDict = this.readXRef(); + } else { + warn('Indexing all PDF objects'); + trailerDict = this.indexObjects(); + } + trailerDict.assignXref(this); + this.trailer = trailerDict; + var encrypt = trailerDict.get('Encrypt'); + if (encrypt) { + var ids = trailerDict.get('ID'); + var fileId = (ids && ids.length) ? ids[0] : ''; + this.encrypt = new CipherTransformFactory(encrypt, fileId, + this.password); + } + + // get the root dictionary (catalog) object + if (!(this.root = trailerDict.get('Root'))) { + error('Invalid root reference'); } }, - getOutputLength: function LabCS_getOutputLength(inputLength, alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function LabCS_isDefaultDecode(decodeMap) { - // XXX: Decoding is handled with the lab conversion because of the strange - // ranges that are used. - return true; - }, - usesZeroToOneRange: false - }; - return LabCS; -})(); -// TODO refactor to remove dependency on image.js -function _setCoreImage(coreImage_) { - coreImage = coreImage_; - PDFImage = coreImage_.PDFImage; -} -exports._setCoreImage = _setCoreImage; + processXRefTable: function XRef_processXRefTable(parser) { + if (!('tableState' in this)) { + // Stores state of the table as we process it so we can resume + // from middle of table in case of missing data error + this.tableState = { + entryNum: 0, + streamPos: parser.lexer.stream.pos, + parserBuf1: parser.buf1, + parserBuf2: parser.buf2 + }; + } -exports.ColorSpace = ColorSpace; + var obj = this.readXRefTable(parser); -// TODO refactor to remove dependency on colorspace.js -coreStream._setCoreColorSpace(exports); -})); + // Sanity check + if (!isCmd(obj, 'trailer')) { + error('Invalid XRef table: could not find trailer dictionary'); + } + // Read trailer dictionary, e.g. + // trailer + // << /Size 22 + // /Root 20R + // /Info 10R + // /ID [ <81b14aafa313db63dbd6f981e49f94f4> ] + // >> + // The parser goes through the entire stream << ... >> and provides + // a getter interface for the key-value table + var dict = parser.getObj(); + // The pdflib PDF generator can generate a nested trailer dictionary + if (!isDict(dict) && dict.dict) { + dict = dict.dict; + } + if (!isDict(dict)) { + error('Invalid XRef table: could not parse trailer dictionary'); + } + delete this.tableState; -(function (root, factory) { - { - factory((root.pdfjsCoreImage = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreColorSpace, root.pdfjsCoreStream, - root.pdfjsCoreJpx); - } -}(this, function (exports, sharedUtil, corePrimitives, coreColorSpace, - coreStream, coreJpx) { + return dict; + }, -var ImageKind = sharedUtil.ImageKind; -var assert = sharedUtil.assert; -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var warn = sharedUtil.warn; -var Name = corePrimitives.Name; -var isStream = corePrimitives.isStream; -var ColorSpace = coreColorSpace.ColorSpace; -var DecodeStream = coreStream.DecodeStream; -var Stream = coreStream.Stream; -var JpegStream = coreStream.JpegStream; -var JpxImage = coreJpx.JpxImage; + readXRefTable: function XRef_readXRefTable(parser) { + // Example of cross-reference table: + // xref + // 0 1 <-- subsection header (first obj #, obj count) + // 0000000000 65535 f <-- actual object (offset, generation #, f/n) + // 23 2 <-- subsection header ... and so on ... + // 0000025518 00002 n + // 0000025635 00000 n + // trailer + // ... -var PDFImage = (function PDFImageClosure() { - /** - * Decode the image in the main thread if it supported. Resovles the promise - * when the image data is ready. - */ - function handleImageData(handler, xref, res, image) { - if (image instanceof JpegStream && image.isNativelyDecodable(xref, res)) { - // For natively supported jpegs send them to the main thread for decoding. - var dict = image.dict; - var colorSpace = dict.get('ColorSpace', 'CS'); - colorSpace = ColorSpace.parse(colorSpace, xref, res); - var numComps = colorSpace.numComps; - var decodePromise = handler.sendWithPromise('JpegDecode', - [image.getIR(), numComps]); - return decodePromise.then(function (message) { - var data = message.data; - return new Stream(data, 0, data.length, image.dict); - }); - } else { - return Promise.resolve(image); - } - } + var stream = parser.lexer.stream; + var tableState = this.tableState; + stream.pos = tableState.streamPos; + parser.buf1 = tableState.parserBuf1; + parser.buf2 = tableState.parserBuf2; - /** - * Decode and clamp a value. The formula is different from the spec because we - * don't decode to float range [0,1], we decode it in the [0,max] range. - */ - function decodeAndClamp(value, addend, coefficient, max) { - value = addend + value * coefficient; - // Clamp the value to the range - return (value < 0 ? 0 : (value > max ? max : value)); - } + // Outer loop is over subsection headers + var obj; - function PDFImage(xref, res, image, inline, smask, mask, isMask) { - this.image = image; - var dict = image.dict; - if (dict.has('Filter')) { - var filter = dict.get('Filter').name; - if (filter === 'JPXDecode') { - var jpxImage = new JpxImage(); - jpxImage.parseImageProperties(image.stream); - image.stream.reset(); - image.bitsPerComponent = jpxImage.bitsPerComponent; - image.numComps = jpxImage.componentsCount; - } else if (filter === 'JBIG2Decode') { - image.bitsPerComponent = 1; - image.numComps = 1; - } - } - // TODO cache rendered images? + while (true) { + if (!('firstEntryNum' in tableState) || !('entryCount' in tableState)) { + if (isCmd(obj = parser.getObj(), 'trailer')) { + break; + } + tableState.firstEntryNum = obj; + tableState.entryCount = parser.getObj(); + } - this.width = dict.get('Width', 'W'); - this.height = dict.get('Height', 'H'); + var first = tableState.firstEntryNum; + var count = tableState.entryCount; + if (!isInt(first) || !isInt(count)) { + error('Invalid XRef table: wrong types in subsection header'); + } + // Inner loop is over objects themselves + for (var i = tableState.entryNum; i < count; i++) { + tableState.streamPos = stream.pos; + tableState.entryNum = i; + tableState.parserBuf1 = parser.buf1; + tableState.parserBuf2 = parser.buf2; - if (this.width < 1 || this.height < 1) { - error('Invalid image width: ' + this.width + ' or height: ' + - this.height); - } + var entry = {}; + entry.offset = parser.getObj(); + entry.gen = parser.getObj(); + var type = parser.getObj(); - this.interpolate = dict.get('Interpolate', 'I') || false; - this.imageMask = dict.get('ImageMask', 'IM') || false; - this.matte = dict.get('Matte') || false; + if (isCmd(type, 'f')) { + entry.free = true; + } else if (isCmd(type, 'n')) { + entry.uncompressed = true; + } - var bitsPerComponent = image.bitsPerComponent; - if (!bitsPerComponent) { - bitsPerComponent = dict.get('BitsPerComponent', 'BPC'); - if (!bitsPerComponent) { - if (this.imageMask) { - bitsPerComponent = 1; - } else { - error('Bits per component missing in image: ' + this.imageMask); + // Validate entry obj + if (!isInt(entry.offset) || !isInt(entry.gen) || + !(entry.free || entry.uncompressed)) { + error('Invalid entry in XRef subsection: ' + first + ', ' + count); + } + + if (!this.entries[i + first]) { + this.entries[i + first] = entry; + } } + + tableState.entryNum = 0; + tableState.streamPos = stream.pos; + tableState.parserBuf1 = parser.buf1; + tableState.parserBuf2 = parser.buf2; + delete tableState.firstEntryNum; + delete tableState.entryCount; } - } - this.bpc = bitsPerComponent; - if (!this.imageMask) { - var colorSpace = dict.get('ColorSpace', 'CS'); - if (!colorSpace) { - info('JPX images (which do not require color spaces)'); - switch (image.numComps) { - case 1: - colorSpace = Name.get('DeviceGray'); - break; - case 3: - colorSpace = Name.get('DeviceRGB'); - break; - case 4: - colorSpace = Name.get('DeviceCMYK'); - break; - default: - error('JPX images with ' + this.numComps + - ' color components not supported.'); - } + // Per issue 3248: hp scanners generate bad XRef + if (first === 1 && this.entries[1] && this.entries[1].free) { + // shifting the entries + this.entries.shift(); } - this.colorSpace = ColorSpace.parse(colorSpace, xref, res); - this.numComps = this.colorSpace.numComps; - } - this.decode = dict.get('Decode', 'D'); - this.needsDecode = false; - if (this.decode && - ((this.colorSpace && !this.colorSpace.isDefaultDecode(this.decode)) || - (isMask && !ColorSpace.isDefaultDecode(this.decode, 1)))) { - this.needsDecode = true; - // Do some preprocessing to avoid more math. - var max = (1 << bitsPerComponent) - 1; - this.decodeCoefficients = []; - this.decodeAddends = []; - for (var i = 0, j = 0; i < this.decode.length; i += 2, ++j) { - var dmin = this.decode[i]; - var dmax = this.decode[i + 1]; - this.decodeCoefficients[j] = dmax - dmin; - this.decodeAddends[j] = max * dmin; + // Sanity check: as per spec, first object must be free + if (this.entries[0] && !this.entries[0].free) { + error('Invalid XRef table: unexpected first object'); } - } + return obj; + }, - if (smask) { - this.smask = new PDFImage(xref, res, smask, false); - } else if (mask) { - if (isStream(mask)) { - var maskDict = mask.dict, imageMask = maskDict.get('ImageMask', 'IM'); - if (!imageMask) { - warn('Ignoring /Mask in image without /ImageMask.'); - } else { - this.mask = new PDFImage(xref, res, mask, false, null, null, true); + processXRefStream: function XRef_processXRefStream(stream) { + if (!('streamState' in this)) { + // Stores state of the stream as we process it so we can resume + // from middle of stream in case of missing data error + var streamParameters = stream.dict; + var byteWidths = streamParameters.get('W'); + var range = streamParameters.get('Index'); + if (!range) { + range = [0, streamParameters.get('Size')]; } - } else { - // Color key mask (just an array). - this.mask = mask; + + this.streamState = { + entryRanges: range, + byteWidths: byteWidths, + entryNum: 0, + streamPos: stream.pos + }; } - } - } - /** - * Handles processing of image data and returns the Promise that is resolved - * with a PDFImage when the image is ready to be used. - */ - PDFImage.buildImage = function PDFImage_buildImage(handler, xref, - res, image, inline) { - var imagePromise = handleImageData(handler, xref, res, image); - var smaskPromise; - var maskPromise; + this.readXRefStream(stream); + delete this.streamState; - var smask = image.dict.get('SMask'); - var mask = image.dict.get('Mask'); + return stream.dict; + }, - if (smask) { - smaskPromise = handleImageData(handler, xref, res, smask); - maskPromise = Promise.resolve(null); - } else { - smaskPromise = Promise.resolve(null); - if (mask) { - if (isStream(mask)) { - maskPromise = handleImageData(handler, xref, res, mask); - } else if (isArray(mask)) { - maskPromise = Promise.resolve(mask); - } else { - warn('Unsupported mask format.'); - maskPromise = Promise.resolve(null); + readXRefStream: function XRef_readXRefStream(stream) { + var i, j; + var streamState = this.streamState; + stream.pos = streamState.streamPos; + + var byteWidths = streamState.byteWidths; + var typeFieldWidth = byteWidths[0]; + var offsetFieldWidth = byteWidths[1]; + var generationFieldWidth = byteWidths[2]; + + var entryRanges = streamState.entryRanges; + while (entryRanges.length > 0) { + var first = entryRanges[0]; + var n = entryRanges[1]; + + if (!isInt(first) || !isInt(n)) { + error('Invalid XRef range fields: ' + first + ', ' + n); } - } else { - maskPromise = Promise.resolve(null); + if (!isInt(typeFieldWidth) || !isInt(offsetFieldWidth) || + !isInt(generationFieldWidth)) { + error('Invalid XRef entry fields length: ' + first + ', ' + n); + } + for (i = streamState.entryNum; i < n; ++i) { + streamState.entryNum = i; + streamState.streamPos = stream.pos; + + var type = 0, offset = 0, generation = 0; + for (j = 0; j < typeFieldWidth; ++j) { + type = (type << 8) | stream.getByte(); + } + // if type field is absent, its default value is 1 + if (typeFieldWidth === 0) { + type = 1; + } + for (j = 0; j < offsetFieldWidth; ++j) { + offset = (offset << 8) | stream.getByte(); + } + for (j = 0; j < generationFieldWidth; ++j) { + generation = (generation << 8) | stream.getByte(); + } + var entry = {}; + entry.offset = offset; + entry.gen = generation; + switch (type) { + case 0: + entry.free = true; + break; + case 1: + entry.uncompressed = true; + break; + case 2: + break; + default: + error('Invalid XRef entry type: ' + type); + } + if (!this.entries[first + i]) { + this.entries[first + i] = entry; + } + } + + streamState.entryNum = 0; + streamState.streamPos = stream.pos; + entryRanges.splice(0, 2); } - } - return Promise.all([imagePromise, smaskPromise, maskPromise]).then( - function(results) { - var imageData = results[0]; - var smaskData = results[1]; - var maskData = results[2]; - return new PDFImage(xref, res, imageData, inline, smaskData, maskData); - }); - }; + }, - /** - * Resize an image using the nearest neighbor algorithm. Currently only - * supports one and three component images. - * @param {TypedArray} pixels The original image with one component. - * @param {Number} bpc Number of bits per component. - * @param {Number} components Number of color components, 1 or 3 is supported. - * @param {Number} w1 Original width. - * @param {Number} h1 Original height. - * @param {Number} w2 New width. - * @param {Number} h2 New height. - * @param {TypedArray} dest (Optional) The destination buffer. - * @param {Number} alpha01 (Optional) Size reserved for the alpha channel. - * @return {TypedArray} Resized image data. - */ - PDFImage.resize = function PDFImage_resize(pixels, bpc, components, - w1, h1, w2, h2, dest, alpha01) { + indexObjects: function XRef_indexObjects() { + // Simple scan through the PDF content to find objects, + // trailers and XRef streams. + var TAB = 0x9, LF = 0xA, CR = 0xD, SPACE = 0x20; + var PERCENT = 0x25, LT = 0x3C; + + function readToken(data, offset) { + var token = '', ch = data[offset]; + while (ch !== LF && ch !== CR && ch !== LT) { + if (++offset >= data.length) { + break; + } + token += String.fromCharCode(ch); + ch = data[offset]; + } + return token; + } + function skipUntil(data, offset, what) { + var length = what.length, dataLength = data.length; + var skipped = 0; + // finding byte sequence + while (offset < dataLength) { + var i = 0; + while (i < length && data[offset + i] === what[i]) { + ++i; + } + if (i >= length) { + break; // sequence found + } + offset++; + skipped++; + } + return skipped; + } + var objRegExp = /^(\d+)\s+(\d+)\s+obj\b/; + var trailerBytes = new Uint8Array([116, 114, 97, 105, 108, 101, 114]); + var startxrefBytes = new Uint8Array([115, 116, 97, 114, 116, 120, 114, + 101, 102]); + var endobjBytes = new Uint8Array([101, 110, 100, 111, 98, 106]); + var xrefBytes = new Uint8Array([47, 88, 82, 101, 102]); - if (components !== 1 && components !== 3) { - error('Unsupported component count for resizing.'); - } + // Clear out any existing entries, since they may be bogus. + this.entries.length = 0; - var length = w2 * h2 * components; - var temp = dest ? dest : (bpc <= 8 ? new Uint8Array(length) : - (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); - var xRatio = w1 / w2; - var yRatio = h1 / h2; - var i, j, py, newIndex = 0, oldIndex; - var xScaled = new Uint16Array(w2); - var w1Scanline = w1 * components; - if (alpha01 !== 1) { - alpha01 = 0; - } + var stream = this.stream; + stream.pos = 0; + var buffer = stream.getBytes(); + var position = stream.start, length = buffer.length; + var trailers = [], xrefStms = []; + while (position < length) { + var ch = buffer[position]; + if (ch === TAB || ch === LF || ch === CR || ch === SPACE) { + ++position; + continue; + } + if (ch === PERCENT) { // %-comment + do { + ++position; + if (position >= length) { + break; + } + ch = buffer[position]; + } while (ch !== LF && ch !== CR); + continue; + } + var token = readToken(buffer, position); + var m; + if (token.indexOf('xref') === 0 && + (token.length === 4 || /\s/.test(token[4]))) { + position += skipUntil(buffer, position, trailerBytes); + trailers.push(position); + position += skipUntil(buffer, position, startxrefBytes); + } else if ((m = objRegExp.exec(token))) { + if (typeof this.entries[m[1]] === 'undefined') { + this.entries[m[1]] = { + offset: position - stream.start, + gen: m[2] | 0, + uncompressed: true + }; + } + var contentLength = skipUntil(buffer, position, endobjBytes) + 7; + var content = buffer.subarray(position, position + contentLength); - for (j = 0; j < w2; j++) { - xScaled[j] = Math.floor(j * xRatio) * components; - } + // checking XRef stream suspect + // (it shall have '/XRef' and next char is not a letter) + var xrefTagOffset = skipUntil(content, 0, xrefBytes); + if (xrefTagOffset < contentLength && + content[xrefTagOffset + 5] < 64) { + xrefStms.push(position - stream.start); + this.xrefstms[position - stream.start] = 1; // Avoid recursion + } - if (components === 1) { - for (i = 0; i < h2; i++) { - py = Math.floor(i * yRatio) * w1Scanline; - for (j = 0; j < w2; j++) { - oldIndex = py + xScaled[j]; - temp[newIndex++] = pixels[oldIndex]; + position += contentLength; + } else if (token.indexOf('trailer') === 0 && + (token.length === 7 || /\s/.test(token[7]))) { + trailers.push(position); + position += skipUntil(buffer, position, startxrefBytes); + } else { + position += token.length + 1; } } - } else if (components === 3) { - for (i = 0; i < h2; i++) { - py = Math.floor(i * yRatio) * w1Scanline; - for (j = 0; j < w2; j++) { - oldIndex = py + xScaled[j]; - temp[newIndex++] = pixels[oldIndex++]; - temp[newIndex++] = pixels[oldIndex++]; - temp[newIndex++] = pixels[oldIndex++]; - newIndex += alpha01; + // reading XRef streams + var i, ii; + for (i = 0, ii = xrefStms.length; i < ii; ++i) { + this.startXRefQueue.push(xrefStms[i]); + this.readXRef(/* recoveryMode */ true); + } + // finding main trailer + var dict; + for (i = 0, ii = trailers.length; i < ii; ++i) { + stream.pos = trailers[i]; + var parser = new Parser(new Lexer(stream), true, this); + var obj = parser.getObj(); + if (!isCmd(obj, 'trailer')) { + continue; + } + // read the trailer dictionary + if (!isDict(dict = parser.getObj())) { + continue; + } + // taking the first one with 'ID' + if (dict.has('ID')) { + return dict; } } - } - return temp; - }; - - PDFImage.createMask = - function PDFImage_createMask(imgArray, width, height, - imageIsFromDecodeStream, inverseDecode) { - - // |imgArray| might not contain full data for every pixel of the mask, so - // we need to distinguish between |computedLength| and |actualLength|. - // In particular, if inverseDecode is true, then the array we return must - // have a length of |computedLength|. + // no tailer with 'ID', taking last one (if exists) + if (dict) { + return dict; + } + // nothing helps + // calling error() would reject worker with an UnknownErrorException. + throw new InvalidPDFException('Invalid PDF structure'); + }, - var computedLength = ((width + 7) >> 3) * height; - var actualLength = imgArray.byteLength; - var haveFullData = computedLength === actualLength; - var data, i; + readXRef: function XRef_readXRef(recoveryMode) { + var stream = this.stream; - if (imageIsFromDecodeStream && (!inverseDecode || haveFullData)) { - // imgArray came from a DecodeStream and its data is in an appropriate - // form, so we can just transfer it. - data = imgArray; - } else if (!inverseDecode) { - data = new Uint8Array(actualLength); - data.set(imgArray); - } else { - data = new Uint8Array(computedLength); - data.set(imgArray); - for (i = actualLength; i < computedLength; i++) { - data[i] = 0xff; - } - } + try { + while (this.startXRefQueue.length) { + var startXRef = this.startXRefQueue[0]; - // If necessary, invert the original mask data (but not any extra we might - // have added above). It's safe to modify the array -- whether it's the - // original or a copy, we're about to transfer it anyway, so nothing else - // in this thread can be relying on its contents. - if (inverseDecode) { - for (i = 0; i < actualLength; i++) { - data[i] = ~data[i]; - } - } + stream.pos = startXRef + stream.start; - return {data: data, width: width, height: height}; - }; + var parser = new Parser(new Lexer(stream), true, this); + var obj = parser.getObj(); + var dict; - PDFImage.prototype = { - get drawWidth() { - return Math.max(this.width, - this.smask && this.smask.width || 0, - this.mask && this.mask.width || 0); - }, + // Get dictionary + if (isCmd(obj, 'xref')) { + // Parse end-of-file XRef + dict = this.processXRefTable(parser); + if (!this.topDict) { + this.topDict = dict; + } - get drawHeight() { - return Math.max(this.height, - this.smask && this.smask.height || 0, - this.mask && this.mask.height || 0); - }, + // Recursively get other XRefs 'XRefStm', if any + obj = dict.get('XRefStm'); + if (isInt(obj)) { + var pos = obj; + // ignore previously loaded xref streams + // (possible infinite recursion) + if (!(pos in this.xrefstms)) { + this.xrefstms[pos] = 1; + this.startXRefQueue.push(pos); + } + } + } else if (isInt(obj)) { + // Parse in-stream XRef + if (!isInt(parser.getObj()) || + !isCmd(parser.getObj(), 'obj') || + !isStream(obj = parser.getObj())) { + error('Invalid XRef stream'); + } + dict = this.processXRefStream(obj); + if (!this.topDict) { + this.topDict = dict; + } + if (!dict) { + error('Failed to read XRef stream'); + } + } else { + error('Invalid XRef stream header'); + } - decodeBuffer: function PDFImage_decodeBuffer(buffer) { - var bpc = this.bpc; - var numComps = this.numComps; + // Recursively get previous dictionary, if any + obj = dict.get('Prev'); + if (isInt(obj)) { + this.startXRefQueue.push(obj); + } else if (isRef(obj)) { + // The spec says Prev must not be a reference, i.e. "/Prev NNN" + // This is a fallback for non-compliant PDFs, i.e. "/Prev NNN 0 R" + this.startXRefQueue.push(obj.num); + } - var decodeAddends = this.decodeAddends; - var decodeCoefficients = this.decodeCoefficients; - var max = (1 << bpc) - 1; - var i, ii; + this.startXRefQueue.shift(); + } - if (bpc === 1) { - // If the buffer needed decode that means it just needs to be inverted. - for (i = 0, ii = buffer.length; i < ii; i++) { - buffer[i] = +!(buffer[i]); + return this.topDict; + } catch (e) { + if (e instanceof MissingDataException) { + throw e; } - return; + info('(while reading XRef): ' + e); } - var index = 0; - for (i = 0, ii = this.width * this.height; i < ii; i++) { - for (var j = 0; j < numComps; j++) { - buffer[index] = decodeAndClamp(buffer[index], decodeAddends[j], - decodeCoefficients[j], max); - index++; - } + + if (recoveryMode) { + return; } + throw new XRefParseException(); }, - getComponents: function PDFImage_getComponents(buffer) { - var bpc = this.bpc; - - // This image doesn't require any extra work. - if (bpc === 8) { - return buffer; + getEntry: function XRef_getEntry(i) { + var xrefEntry = this.entries[i]; + if (xrefEntry && !xrefEntry.free && xrefEntry.offset) { + return xrefEntry; } + return null; + }, - var width = this.width; - var height = this.height; - var numComps = this.numComps; - - var length = width * height * numComps; - var bufferPos = 0; - var output = (bpc <= 8 ? new Uint8Array(length) : - (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); - var rowComps = width * numComps; + fetchIfRef: function XRef_fetchIfRef(obj) { + if (!isRef(obj)) { + return obj; + } + return this.fetch(obj); + }, - var max = (1 << bpc) - 1; - var i = 0, ii, buf; + fetch: function XRef_fetch(ref, suppressEncryption) { + assert(isRef(ref), 'ref object is not a reference'); + var num = ref.num; + if (num in this.cache) { + var cacheEntry = this.cache[num]; + return cacheEntry; + } - if (bpc === 1) { - // Optimization for reading 1 bpc images. - var mask, loop1End, loop2End; - for (var j = 0; j < height; j++) { - loop1End = i + (rowComps & ~7); - loop2End = i + rowComps; + var xrefEntry = this.getEntry(num); - // unroll loop for all full bytes - while (i < loop1End) { - buf = buffer[bufferPos++]; - output[i] = (buf >> 7) & 1; - output[i + 1] = (buf >> 6) & 1; - output[i + 2] = (buf >> 5) & 1; - output[i + 3] = (buf >> 4) & 1; - output[i + 4] = (buf >> 3) & 1; - output[i + 5] = (buf >> 2) & 1; - output[i + 6] = (buf >> 1) & 1; - output[i + 7] = buf & 1; - i += 8; - } + // the referenced entry can be free + if (xrefEntry === null) { + return (this.cache[num] = null); + } - // handle remaing bits - if (i < loop2End) { - buf = buffer[bufferPos++]; - mask = 128; - while (i < loop2End) { - output[i++] = +!!(buf & mask); - mask >>= 1; - } - } - } + if (xrefEntry.uncompressed) { + xrefEntry = this.fetchUncompressed(ref, xrefEntry, suppressEncryption); } else { - // The general case that handles all other bpc values. - var bits = 0; - buf = 0; - for (i = 0, ii = length; i < ii; ++i) { - if (i % rowComps === 0) { - buf = 0; - bits = 0; - } + xrefEntry = this.fetchCompressed(xrefEntry, suppressEncryption); + } + if (isDict(xrefEntry)){ + xrefEntry.objId = ref.toString(); + } else if (isStream(xrefEntry)) { + xrefEntry.dict.objId = ref.toString(); + } + return xrefEntry; + }, - while (bits < bpc) { - buf = (buf << 8) | buffer[bufferPos++]; - bits += 8; + fetchUncompressed: function XRef_fetchUncompressed(ref, xrefEntry, + suppressEncryption) { + var gen = ref.gen; + var num = ref.num; + if (xrefEntry.gen !== gen) { + error('inconsistent generation in XRef'); + } + var stream = this.stream.makeSubStream(xrefEntry.offset + + this.stream.start); + var parser = new Parser(new Lexer(stream), true, this); + var obj1 = parser.getObj(); + var obj2 = parser.getObj(); + var obj3 = parser.getObj(); + if (!isInt(obj1) || parseInt(obj1, 10) !== num || + !isInt(obj2) || parseInt(obj2, 10) !== gen || + !isCmd(obj3)) { + error('bad XRef entry'); + } + if (!isCmd(obj3, 'obj')) { + // some bad PDFs use "obj1234" and really mean 1234 + if (obj3.cmd.indexOf('obj') === 0) { + num = parseInt(obj3.cmd.substring(3), 10); + if (!isNaN(num)) { + return num; } - - var remainingBits = bits - bpc; - var value = buf >> remainingBits; - output[i] = (value < 0 ? 0 : (value > max ? max : value)); - buf = buf & ((1 << remainingBits) - 1); - bits = remainingBits; } + error('bad XRef entry'); } - return output; + if (this.encrypt && !suppressEncryption) { + xrefEntry = parser.getObj(this.encrypt.createCipherTransform(num, gen)); + } else { + xrefEntry = parser.getObj(); + } + if (!isStream(xrefEntry)) { + this.cache[num] = xrefEntry; + } + return xrefEntry; }, - fillOpacity: function PDFImage_fillOpacity(rgbaBuf, width, height, - actualHeight, image) { - var smask = this.smask; - var mask = this.mask; - var alphaBuf, sw, sh, i, ii, j; - - if (smask) { - sw = smask.width; - sh = smask.height; - alphaBuf = new Uint8Array(sw * sh); - smask.fillGrayBuffer(alphaBuf); - if (sw !== width || sh !== height) { - alphaBuf = PDFImage.resize(alphaBuf, smask.bpc, 1, sw, sh, width, - height); + fetchCompressed: function XRef_fetchCompressed(xrefEntry, + suppressEncryption) { + var tableOffset = xrefEntry.offset; + var stream = this.fetch(new Ref(tableOffset, 0)); + if (!isStream(stream)) { + error('bad ObjStm stream'); + } + var first = stream.dict.get('First'); + var n = stream.dict.get('N'); + if (!isInt(first) || !isInt(n)) { + error('invalid first and n parameters for ObjStm stream'); + } + var parser = new Parser(new Lexer(stream), false, this); + parser.allowStreams = true; + var i, entries = [], num, nums = []; + // read the object numbers to populate cache + for (i = 0; i < n; ++i) { + num = parser.getObj(); + if (!isInt(num)) { + error('invalid object number in the ObjStm stream: ' + num); } - } else if (mask) { - if (mask instanceof PDFImage) { - sw = mask.width; - sh = mask.height; - alphaBuf = new Uint8Array(sw * sh); - mask.numComps = 1; - mask.fillGrayBuffer(alphaBuf); - - // Need to invert values in rgbaBuf - for (i = 0, ii = sw * sh; i < ii; ++i) { - alphaBuf[i] = 255 - alphaBuf[i]; - } - - if (sw !== width || sh !== height) { - alphaBuf = PDFImage.resize(alphaBuf, mask.bpc, 1, sw, sh, width, - height); - } - } else if (isArray(mask)) { - // Color key mask: if any of the compontents are outside the range - // then they should be painted. - alphaBuf = new Uint8Array(width * height); - var numComps = this.numComps; - for (i = 0, ii = width * height; i < ii; ++i) { - var opacity = 0; - var imageOffset = i * numComps; - for (j = 0; j < numComps; ++j) { - var color = image[imageOffset + j]; - var maskOffset = j * 2; - if (color < mask[maskOffset] || color > mask[maskOffset + 1]) { - opacity = 255; - break; - } - } - alphaBuf[i] = opacity; - } - } else { - error('Unknown mask format.'); + nums.push(num); + var offset = parser.getObj(); + if (!isInt(offset)) { + error('invalid object offset in the ObjStm stream: ' + offset); } } - - if (alphaBuf) { - for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { - rgbaBuf[j] = alphaBuf[i]; + // read stream objects for cache + for (i = 0; i < n; ++i) { + entries.push(parser.getObj()); + num = nums[i]; + var entry = this.entries[num]; + if (entry && entry.offset === tableOffset && entry.gen === i) { + this.cache[num] = entries[i]; } - } else { - // No mask. - for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { - rgbaBuf[j] = 255; + } + xrefEntry = entries[xrefEntry.gen]; + if (xrefEntry === undefined) { + error('bad XRef entry for compressed object'); + } + return xrefEntry; + }, + + fetchIfRefAsync: function XRef_fetchIfRefAsync(obj) { + if (!isRef(obj)) { + return Promise.resolve(obj); + } + return this.fetchAsync(obj); + }, + + fetchAsync: function XRef_fetchAsync(ref, suppressEncryption) { + var streamManager = this.stream.manager; + var xref = this; + return new Promise(function tryFetch(resolve, reject) { + try { + resolve(xref.fetch(ref, suppressEncryption)); + } catch (e) { + if (e instanceof MissingDataException) { + streamManager.requestRange(e.begin, e.end).then(function () { + tryFetch(resolve, reject); + }, reject); + return; + } + reject(e); } - } + }); }, - undoPreblend: function PDFImage_undoPreblend(buffer, width, height) { - var matte = this.smask && this.smask.matte; - if (!matte) { - return; + getCatalogObj: function XRef_getCatalogObj() { + return this.root; + } + }; + + return XRef; +})(); + +/** + * A NameTree/NumberTree is like a Dict but has some advantageous properties, + * see the specification (7.9.6 and 7.9.7) for additional details. + * TODO: implement all the Dict functions and make this more efficient. + */ +var NameOrNumberTree = (function NameOrNumberTreeClosure() { + function NameOrNumberTree(root, xref) { + throw new Error('Cannot initialize NameOrNumberTree.'); + } + + NameOrNumberTree.prototype = { + getAll: function NameOrNumberTree_getAll() { + var dict = Object.create(null); + if (!this.root) { + return dict; } - var matteRgb = this.colorSpace.getRgb(matte, 0); - var matteR = matteRgb[0]; - var matteG = matteRgb[1]; - var matteB = matteRgb[2]; - var length = width * height * 4; - var r, g, b; - for (var i = 0; i < length; i += 4) { - var alpha = buffer[i + 3]; - if (alpha === 0) { - // according formula we have to get Infinity in all components - // making it white (typical paper color) should be okay - buffer[i] = 255; - buffer[i + 1] = 255; - buffer[i + 2] = 255; + var xref = this.xref; + // Reading Name/Number tree. + var processed = new RefSet(); + processed.put(this.root); + var queue = [this.root]; + while (queue.length > 0) { + var i, n; + var obj = xref.fetchIfRef(queue.shift()); + if (!isDict(obj)) { continue; } - var k = 255 / alpha; - r = (buffer[i] - matteR) * k + matteR; - g = (buffer[i + 1] - matteG) * k + matteG; - b = (buffer[i + 2] - matteB) * k + matteB; - buffer[i] = r <= 0 ? 0 : r >= 255 ? 255 : r | 0; - buffer[i + 1] = g <= 0 ? 0 : g >= 255 ? 255 : g | 0; - buffer[i + 2] = b <= 0 ? 0 : b >= 255 ? 255 : b | 0; + if (obj.has('Kids')) { + var kids = obj.get('Kids'); + for (i = 0, n = kids.length; i < n; i++) { + var kid = kids[i]; + assert(!processed.has(kid), + 'Duplicate entry in "' + this._type + '" tree.'); + queue.push(kid); + processed.put(kid); + } + continue; + } + var entries = obj.get(this._type); + if (isArray(entries)) { + for (i = 0, n = entries.length; i < n; i += 2) { + dict[xref.fetchIfRef(entries[i])] = xref.fetchIfRef(entries[i + 1]); + } + } } + return dict; }, - createImageData: function PDFImage_createImageData(forceRGBA) { - var drawWidth = this.drawWidth; - var drawHeight = this.drawHeight; - var imgData = { // other fields are filled in below - width: drawWidth, - height: drawHeight - }; + get: function NameOrNumberTree_get(key) { + if (!this.root) { + return null; + } - var numComps = this.numComps; - var originalWidth = this.width; - var originalHeight = this.height; - var bpc = this.bpc; + var xref = this.xref; + var kidsOrEntries = xref.fetchIfRef(this.root); + var loopCount = 0; + var MAX_LEVELS = 10; + var l, r, m; - // Rows start at byte boundary. - var rowBytes = (originalWidth * numComps * bpc + 7) >> 3; - var imgArray; + // Perform a binary search to quickly find the entry that + // contains the key we are looking for. + while (kidsOrEntries.has('Kids')) { + if (++loopCount > MAX_LEVELS) { + warn('Search depth limit reached for "' + this._type + '" tree.'); + return null; + } - if (!forceRGBA) { - // If it is a 1-bit-per-pixel grayscale (i.e. black-and-white) image - // without any complications, we pass a same-sized copy to the main - // thread rather than expanding by 32x to RGBA form. This saves *lots* - // of memory for many scanned documents. It's also much faster. - // - // Similarly, if it is a 24-bit-per pixel RGB image without any - // complications, we avoid expanding by 1.333x to RGBA form. - var kind; - if (this.colorSpace.name === 'DeviceGray' && bpc === 1) { - kind = ImageKind.GRAYSCALE_1BPP; - } else if (this.colorSpace.name === 'DeviceRGB' && bpc === 8 && - !this.needsDecode) { - kind = ImageKind.RGB_24BPP; + var kids = kidsOrEntries.get('Kids'); + if (!isArray(kids)) { + return null; } - if (kind && !this.smask && !this.mask && - drawWidth === originalWidth && drawHeight === originalHeight) { - imgData.kind = kind; - imgArray = this.getImageBytes(originalHeight * rowBytes); - // If imgArray came from a DecodeStream, we're safe to transfer it - // (and thus detach its underlying buffer) because it will constitute - // the entire DecodeStream's data. But if it came from a Stream, we - // need to copy it because it'll only be a portion of the Stream's - // data, and the rest will be read later on. - if (this.image instanceof DecodeStream) { - imgData.data = imgArray; + l = 0; + r = kids.length - 1; + while (l <= r) { + m = (l + r) >> 1; + var kid = xref.fetchIfRef(kids[m]); + var limits = kid.get('Limits'); + + if (key < xref.fetchIfRef(limits[0])) { + r = m - 1; + } else if (key > xref.fetchIfRef(limits[1])) { + l = m + 1; } else { - var newArray = new Uint8Array(imgArray.length); - newArray.set(imgArray); - imgData.data = newArray; - } - if (this.needsDecode) { - // Invert the buffer (which must be grayscale if we reached here). - assert(kind === ImageKind.GRAYSCALE_1BPP); - var buffer = imgData.data; - for (var i = 0, ii = buffer.length; i < ii; i++) { - buffer[i] ^= 0xff; - } + kidsOrEntries = xref.fetchIfRef(kids[m]); + break; } - return imgData; } - if (this.image instanceof JpegStream && !this.smask && !this.mask && - (this.colorSpace.name === 'DeviceGray' || - this.colorSpace.name === 'DeviceRGB' || - this.colorSpace.name === 'DeviceCMYK')) { - imgData.kind = ImageKind.RGB_24BPP; - imgData.data = this.getImageBytes(originalHeight * rowBytes, - drawWidth, drawHeight, true); - return imgData; + if (l > r) { + return null; } } - imgArray = this.getImageBytes(originalHeight * rowBytes); - // imgArray can be incomplete (e.g. after CCITT fax encoding). - var actualHeight = 0 | (imgArray.length / rowBytes * - drawHeight / originalHeight); + // If we get here, then we have found the right entry. Now go through the + // entries in the dictionary until we find the key we're looking for. + var entries = kidsOrEntries.get(this._type); + if (isArray(entries)) { + // Perform a binary search to reduce the lookup time. + l = 0; + r = entries.length - 2; + while (l <= r) { + // Check only even indices (0, 2, 4, ...) because the + // odd indices contain the actual data. + m = (l + r) & ~1; + var currentKey = xref.fetchIfRef(entries[m]); + if (key < currentKey) { + r = m - 2; + } else if (key > currentKey) { + l = m + 2; + } else { + return xref.fetchIfRef(entries[m + 1]); + } + } + } + return null; + } + }; + return NameOrNumberTree; +})(); - var comps = this.getComponents(imgArray); +var NameTree = (function NameTreeClosure() { + function NameTree(root, xref) { + this.root = root; + this.xref = xref; + this._type = 'Names'; + } - // If opacity data is present, use RGBA_32BPP form. Otherwise, use the - // more compact RGB_24BPP form if allowable. - var alpha01, maybeUndoPreblend; - if (!forceRGBA && !this.smask && !this.mask) { - imgData.kind = ImageKind.RGB_24BPP; - imgData.data = new Uint8Array(drawWidth * drawHeight * 3); - alpha01 = 0; - maybeUndoPreblend = false; - } else { - imgData.kind = ImageKind.RGBA_32BPP; - imgData.data = new Uint8Array(drawWidth * drawHeight * 4); - alpha01 = 1; - maybeUndoPreblend = true; + Util.inherit(NameTree, NameOrNumberTree, {}); - // Color key masking (opacity) must be performed before decoding. - this.fillOpacity(imgData.data, drawWidth, drawHeight, actualHeight, - comps); + return NameTree; +})(); + +var NumberTree = (function NumberTreeClosure() { + function NumberTree(root, xref) { + this.root = root; + this.xref = xref; + this._type = 'Nums'; + } + + Util.inherit(NumberTree, NameOrNumberTree, {}); + + return NumberTree; +})(); + +/** + * "A PDF file can refer to the contents of another file by using a File + * Specification (PDF 1.1)", see the spec (7.11) for more details. + * NOTE: Only embedded files are supported (as part of the attachments support) + * TODO: support the 'URL' file system (with caching if !/V), portable + * collections attributes and related files (/RF) + */ +var FileSpec = (function FileSpecClosure() { + function FileSpec(root, xref) { + if (!root || !isDict(root)) { + return; + } + this.xref = xref; + this.root = root; + if (root.has('FS')) { + this.fs = root.get('FS'); + } + this.description = root.has('Desc') ? + stringToPDFString(root.get('Desc')) : + ''; + if (root.has('RF')) { + warn('Related file specifications are not supported'); + } + this.contentAvailable = true; + if (!root.has('EF')) { + this.contentAvailable = false; + warn('Non-embedded file specifications are not supported'); + } + } + + function pickPlatformItem(dict) { + // Look for the filename in this order: + // UF, F, Unix, Mac, DOS + if (dict.has('UF')) { + return dict.get('UF'); + } else if (dict.has('F')) { + return dict.get('F'); + } else if (dict.has('Unix')) { + return dict.get('Unix'); + } else if (dict.has('Mac')) { + return dict.get('Mac'); + } else if (dict.has('DOS')) { + return dict.get('DOS'); + } else { + return null; + } + } + + FileSpec.prototype = { + get filename() { + if (!this._filename && this.root) { + var filename = pickPlatformItem(this.root) || 'unnamed'; + this._filename = stringToPDFString(filename). + replace(/\\\\/g, '\\'). + replace(/\\\//g, '/'). + replace(/\\/g, '/'); + } + return this._filename; + }, + get content() { + if (!this.contentAvailable) { + return null; + } + if (!this.contentRef && this.root) { + this.contentRef = pickPlatformItem(this.root.get('EF')); + } + var content = null; + if (this.contentRef) { + var xref = this.xref; + var fileObj = xref.fetchIfRef(this.contentRef); + if (fileObj && isStream(fileObj)) { + content = fileObj.getBytes(); + } else { + warn('Embedded file specification points to non-existing/invalid ' + + 'content'); + } + } else { + warn('Embedded file specification does not have a content'); } + return content; + }, + get serializable() { + return { + filename: this.filename, + content: this.content + }; + } + }; + return FileSpec; +})(); - if (this.needsDecode) { - this.decodeBuffer(comps); +/** + * A helper for loading missing data in object graphs. It traverses the graph + * depth first and queues up any objects that have missing data. Once it has + * has traversed as many objects that are available it attempts to bundle the + * missing data requests and then resume from the nodes that weren't ready. + * + * NOTE: It provides protection from circular references by keeping track of + * of loaded references. However, you must be careful not to load any graphs + * that have references to the catalog or other pages since that will cause the + * entire PDF document object graph to be traversed. + */ +var ObjectLoader = (function() { + function mayHaveChildren(value) { + return isRef(value) || isDict(value) || isArray(value) || isStream(value); + } + + function addChildren(node, nodesToVisit) { + var value; + if (isDict(node) || isStream(node)) { + var map; + if (isDict(node)) { + map = node.map; + } else { + map = node.dict.map; } - this.colorSpace.fillRgb(imgData.data, originalWidth, originalHeight, - drawWidth, drawHeight, actualHeight, bpc, comps, - alpha01); - if (maybeUndoPreblend) { - this.undoPreblend(imgData.data, drawWidth, actualHeight); + for (var key in map) { + value = map[key]; + if (mayHaveChildren(value)) { + nodesToVisit.push(value); + } + } + } else if (isArray(node)) { + for (var i = 0, ii = node.length; i < ii; i++) { + value = node[i]; + if (mayHaveChildren(value)) { + nodesToVisit.push(value); + } } + } + } - return imgData; - }, + function ObjectLoader(obj, keys, xref) { + this.obj = obj; + this.keys = keys; + this.xref = xref; + this.refSet = null; + this.capability = null; + } - fillGrayBuffer: function PDFImage_fillGrayBuffer(buffer) { - var numComps = this.numComps; - if (numComps !== 1) { - error('Reading gray scale from a color image: ' + numComps); + ObjectLoader.prototype = { + load: function ObjectLoader_load() { + var keys = this.keys; + this.capability = createPromiseCapability(); + // Don't walk the graph if all the data is already loaded. + if (!(this.xref.stream instanceof ChunkedStream) || + this.xref.stream.getMissingChunks().length === 0) { + this.capability.resolve(); + return this.capability.promise; } - var width = this.width; - var height = this.height; - var bpc = this.bpc; + this.refSet = new RefSet(); + // Setup the initial nodes to visit. + var nodesToVisit = []; + for (var i = 0; i < keys.length; i++) { + nodesToVisit.push(this.obj[keys[i]]); + } - // rows start at byte boundary - var rowBytes = (width * numComps * bpc + 7) >> 3; - var imgArray = this.getImageBytes(height * rowBytes); + this._walk(nodesToVisit); + return this.capability.promise; + }, - var comps = this.getComponents(imgArray); - var i, length; + _walk: function ObjectLoader_walk(nodesToVisit) { + var nodesToRevisit = []; + var pendingRequests = []; + // DFS walk of the object graph. + while (nodesToVisit.length) { + var currentNode = nodesToVisit.pop(); - if (bpc === 1) { - // inline decoding (= inversion) for 1 bpc images - length = width * height; - if (this.needsDecode) { - // invert and scale to {0, 255} - for (i = 0; i < length; ++i) { - buffer[i] = (comps[i] - 1) & 255; + // Only references or chunked streams can cause missing data exceptions. + if (isRef(currentNode)) { + // Skip nodes that have already been visited. + if (this.refSet.has(currentNode)) { + continue; } - } else { - // scale to {0, 255} - for (i = 0; i < length; ++i) { - buffer[i] = (-comps[i]) & 255; + try { + var ref = currentNode; + this.refSet.put(ref); + currentNode = this.xref.fetch(currentNode); + } catch (e) { + if (!(e instanceof MissingDataException)) { + throw e; + } + nodesToRevisit.push(currentNode); + pendingRequests.push({ begin: e.begin, end: e.end }); + } + } + if (currentNode && currentNode.getBaseStreams) { + var baseStreams = currentNode.getBaseStreams(); + var foundMissingData = false; + for (var i = 0; i < baseStreams.length; i++) { + var stream = baseStreams[i]; + if (stream.getMissingChunks && stream.getMissingChunks().length) { + foundMissingData = true; + pendingRequests.push({ + begin: stream.start, + end: stream.end + }); + } + } + if (foundMissingData) { + nodesToRevisit.push(currentNode); } } - return; - } - if (this.needsDecode) { - this.decodeBuffer(comps); - } - length = width * height; - // we aren't using a colorspace so we need to scale the value - var scale = 255 / ((1 << bpc) - 1); - for (i = 0; i < length; ++i) { - buffer[i] = (scale * comps[i]) | 0; + addChildren(currentNode, nodesToVisit); } - }, - getImageBytes: function PDFImage_getImageBytes(length, - drawWidth, drawHeight, - forceRGB) { - this.image.reset(); - this.image.drawWidth = drawWidth || this.width; - this.image.drawHeight = drawHeight || this.height; - this.image.forceRGB = !!forceRGB; - return this.image.getBytes(length); + if (pendingRequests.length) { + this.xref.stream.manager.requestRanges(pendingRequests).then( + function pendingRequestCallback() { + nodesToVisit = nodesToRevisit; + for (var i = 0; i < nodesToRevisit.length; i++) { + var node = nodesToRevisit[i]; + // Remove any reference nodes from the currrent refset so they + // aren't skipped when we revist them. + if (isRef(node)) { + this.refSet.remove(node); + } + } + this._walk(nodesToVisit); + }.bind(this), this.capability.reject); + return; + } + // Everything is loaded. + this.refSet = null; + this.capability.resolve(); } }; - return PDFImage; -})(); -exports.PDFImage = PDFImage; + return ObjectLoader; +})(); -// TODO refactor to remove dependency on colorspace.js -coreColorSpace._setCoreImage(exports); +exports.Catalog = Catalog; +exports.ObjectLoader = ObjectLoader; +exports.XRef = XRef; })); diff --git a/build/pdf.js b/build/pdf.js index ec741f26a..39a6a1421 100644 --- a/build/pdf.js +++ b/build/pdf.js @@ -28,8 +28,8 @@ factory((root.pdfjsDistBuildPdf = {})); // Use strict in our context only - users might not want it 'use strict'; -var pdfjsVersion = '1.4.76'; -var pdfjsBuild = '45fea88'; +var pdfjsVersion = '1.4.79'; +var pdfjsBuild = '9eedfc1'; var pdfjsFilePath = typeof document !== 'undefined' && document.currentScript ? @@ -8978,7 +8978,7 @@ var PDFDocumentProxy = (function PDFDocumentProxyClosure() { * title: string, * bold: boolean, * italic: boolean, - * color: rgb array, + * color: rgb Uint8Array, * dest: dest obj, * url: string, * items: array of more items like this diff --git a/build/pdf.worker.js b/build/pdf.worker.js index 9e63ae4ef..256ffe8b5 100644 --- a/build/pdf.worker.js +++ b/build/pdf.worker.js @@ -28,8 +28,8 @@ factory((root.pdfjsDistBuildPdfWorker = {})); // Use strict in our context only - users might not want it 'use strict'; -var pdfjsVersion = '1.4.76'; -var pdfjsBuild = '45fea88'; +var pdfjsVersion = '1.4.79'; +var pdfjsBuild = '9eedfc1'; var pdfjsFilePath = typeof document !== 'undefined' && document.currentScript ? @@ -25740,4803 +25740,3232 @@ exports.IdentityCMap = IdentityCMap; (function (root, factory) { { - factory((root.pdfjsCoreObj = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreCrypto, root.pdfjsCoreParser, - root.pdfjsCoreChunkedStream); + factory((root.pdfjsCorePsParser = {}), root.pdfjsSharedUtil, + root.pdfjsCoreParser); } -}(this, function (exports, sharedUtil, corePrimitives, coreCrypto, coreParser, - coreChunkedStream) { +}(this, function (exports, sharedUtil, coreParser) { -var InvalidPDFException = sharedUtil.InvalidPDFException; -var MissingDataException = sharedUtil.MissingDataException; -var XRefParseException = sharedUtil.XRefParseException; -var assert = sharedUtil.assert; -var bytesToString = sharedUtil.bytesToString; -var createPromiseCapability = sharedUtil.createPromiseCapability; var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isInt = sharedUtil.isInt; -var isString = sharedUtil.isString; -var shadow = sharedUtil.shadow; -var stringToPDFString = sharedUtil.stringToPDFString; -var stringToUTF8String = sharedUtil.stringToUTF8String; -var warn = sharedUtil.warn; -var isValidUrl = sharedUtil.isValidUrl; -var Util = sharedUtil.Util; -var Ref = corePrimitives.Ref; -var RefSet = corePrimitives.RefSet; -var RefSetCache = corePrimitives.RefSetCache; -var isName = corePrimitives.isName; -var isCmd = corePrimitives.isCmd; -var isDict = corePrimitives.isDict; -var isRef = corePrimitives.isRef; -var isStream = corePrimitives.isStream; -var CipherTransformFactory = coreCrypto.CipherTransformFactory; +var EOF = coreParser.EOF; var Lexer = coreParser.Lexer; -var Parser = coreParser.Parser; -var ChunkedStream = coreChunkedStream.ChunkedStream; - -var Catalog = (function CatalogClosure() { - function Catalog(pdfManager, xref, pageFactory) { - this.pdfManager = pdfManager; - this.xref = xref; - this.catDict = xref.getCatalogObj(); - this.fontCache = new RefSetCache(); - assert(isDict(this.catDict), - 'catalog object is not a dictionary'); - // TODO refactor to move getPage() to the PDFDocument. - this.pageFactory = pageFactory; - this.pagePromises = []; +var PostScriptParser = (function PostScriptParserClosure() { + function PostScriptParser(lexer) { + this.lexer = lexer; + this.operators = []; + this.token = null; + this.prev = null; } - - Catalog.prototype = { - get metadata() { - var streamRef = this.catDict.getRaw('Metadata'); - if (!isRef(streamRef)) { - return shadow(this, 'metadata', null); - } - - var encryptMetadata = (!this.xref.encrypt ? false : - this.xref.encrypt.encryptMetadata); - - var stream = this.xref.fetch(streamRef, !encryptMetadata); - var metadata; - if (stream && isDict(stream.dict)) { - var type = stream.dict.get('Type'); - var subtype = stream.dict.get('Subtype'); - - if (isName(type) && isName(subtype) && - type.name === 'Metadata' && subtype.name === 'XML') { - // XXX: This should examine the charset the XML document defines, - // however since there are currently no real means to decode - // arbitrary charsets, let's just hope that the author of the PDF - // was reasonable enough to stick with the XML default charset, - // which is UTF-8. - try { - metadata = stringToUTF8String(bytesToString(stream.getBytes())); - } catch (e) { - info('Skipping invalid metadata.'); - } - } - } - - return shadow(this, 'metadata', metadata); - }, - get toplevelPagesDict() { - var pagesObj = this.catDict.get('Pages'); - assert(isDict(pagesObj), 'invalid top-level pages dictionary'); - // shadow the prototype getter - return shadow(this, 'toplevelPagesDict', pagesObj); + PostScriptParser.prototype = { + nextToken: function PostScriptParser_nextToken() { + this.prev = this.token; + this.token = this.lexer.getToken(); }, - get documentOutline() { - var obj = null; - try { - obj = this.readDocumentOutline(); - } catch (ex) { - if (ex instanceof MissingDataException) { - throw ex; - } - warn('Unable to read document outline'); + accept: function PostScriptParser_accept(type) { + if (this.token.type === type) { + this.nextToken(); + return true; } - return shadow(this, 'documentOutline', obj); + return false; }, - readDocumentOutline: function Catalog_readDocumentOutline() { - var xref = this.xref; - var obj = this.catDict.get('Outlines'); - var root = { items: [] }; - if (isDict(obj)) { - obj = obj.getRaw('First'); - var processed = new RefSet(); - if (isRef(obj)) { - var queue = [{obj: obj, parent: root}]; - // to avoid recursion keeping track of the items - // in the processed dictionary - processed.put(obj); - while (queue.length > 0) { - var i = queue.shift(); - var outlineDict = xref.fetchIfRef(i.obj); - if (outlineDict === null) { - continue; - } - if (!outlineDict.has('Title')) { - error('Invalid outline item'); - } - var actionDict = outlineDict.get('A'), dest = null, url = null; - if (actionDict) { - var destEntry = actionDict.get('D'); - if (destEntry) { - dest = destEntry; - } else { - var uriEntry = actionDict.get('URI'); - if (isString(uriEntry) && isValidUrl(uriEntry, false)) { - url = uriEntry; - } - } - } else if (outlineDict.has('Dest')) { - dest = outlineDict.getRaw('Dest'); - if (isName(dest)) { - dest = dest.name; - } - } - var title = outlineDict.get('Title'); - var outlineItem = { - dest: dest, - url: url, - title: stringToPDFString(title), - color: outlineDict.get('C') || [0, 0, 0], - count: outlineDict.get('Count'), - bold: !!(outlineDict.get('F') & 2), - italic: !!(outlineDict.get('F') & 1), - items: [] - }; - i.parent.items.push(outlineItem); - obj = outlineDict.getRaw('First'); - if (isRef(obj) && !processed.has(obj)) { - queue.push({obj: obj, parent: outlineItem}); - processed.put(obj); - } - obj = outlineDict.getRaw('Next'); - if (isRef(obj) && !processed.has(obj)) { - queue.push({obj: obj, parent: i.parent}); - processed.put(obj); - } - } - } + expect: function PostScriptParser_expect(type) { + if (this.accept(type)) { + return true; } - return (root.items.length > 0 ? root.items : null); + error('Unexpected symbol: found ' + this.token.type + ' expected ' + + type + '.'); }, - get numPages() { - var obj = this.toplevelPagesDict.get('Count'); - assert( - isInt(obj), - 'page count in top level pages object is not an integer' - ); - // shadow the prototype getter - return shadow(this, 'num', obj); + parse: function PostScriptParser_parse() { + this.nextToken(); + this.expect(PostScriptTokenTypes.LBRACE); + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + return this.operators; }, - get destinations() { - function fetchDestination(dest) { - return isDict(dest) ? dest.get('D') : dest; - } - - var xref = this.xref; - var dests = {}, nameTreeRef, nameDictionaryRef; - var obj = this.catDict.get('Names'); - if (obj && obj.has('Dests')) { - nameTreeRef = obj.getRaw('Dests'); - } else if (this.catDict.has('Dests')) { - nameDictionaryRef = this.catDict.get('Dests'); - } - - if (nameDictionaryRef) { - // reading simple destination dictionary - obj = nameDictionaryRef; - obj.forEach(function catalogForEach(key, value) { - if (!value) { - return; - } - dests[key] = fetchDestination(value); - }); - } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - var names = nameTree.getAll(); - for (var name in names) { - dests[name] = fetchDestination(names[name]); + parseBlock: function PostScriptParser_parseBlock() { + while (true) { + if (this.accept(PostScriptTokenTypes.NUMBER)) { + this.operators.push(this.prev.value); + } else if (this.accept(PostScriptTokenTypes.OPERATOR)) { + this.operators.push(this.prev.value); + } else if (this.accept(PostScriptTokenTypes.LBRACE)) { + this.parseCondition(); + } else { + return; } } - return shadow(this, 'destinations', dests); }, - getDestination: function Catalog_getDestination(destinationId) { - function fetchDestination(dest) { - return isDict(dest) ? dest.get('D') : dest; - } - - var xref = this.xref; - var dest = null, nameTreeRef, nameDictionaryRef; - var obj = this.catDict.get('Names'); - if (obj && obj.has('Dests')) { - nameTreeRef = obj.getRaw('Dests'); - } else if (this.catDict.has('Dests')) { - nameDictionaryRef = this.catDict.get('Dests'); - } + parseCondition: function PostScriptParser_parseCondition() { + // Add two place holders that will be updated later + var conditionLocation = this.operators.length; + this.operators.push(null, null); - if (nameDictionaryRef) { // Simple destination dictionary. - var value = nameDictionaryRef.get(destinationId); - if (value) { - dest = fetchDestination(value); - } - } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - dest = fetchDestination(nameTree.get(destinationId)); - } - return dest; - }, + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + if (this.accept(PostScriptTokenTypes.IF)) { + // The true block is right after the 'if' so it just falls through on + // true else it jumps and skips the true block. + this.operators[conditionLocation] = this.operators.length; + this.operators[conditionLocation + 1] = 'jz'; + } else if (this.accept(PostScriptTokenTypes.LBRACE)) { + var jumpLocation = this.operators.length; + this.operators.push(null, null); + var endOfTrue = this.operators.length; + this.parseBlock(); + this.expect(PostScriptTokenTypes.RBRACE); + this.expect(PostScriptTokenTypes.IFELSE); + // The jump is added at the end of the true block to skip the false + // block. + this.operators[jumpLocation] = this.operators.length; + this.operators[jumpLocation + 1] = 'j'; - get pageLabels() { - var obj = null; - try { - obj = this.readPageLabels(); - } catch (ex) { - if (ex instanceof MissingDataException) { - throw ex; - } - warn('Unable to read page labels.'); - } - return shadow(this, 'pageLabels', obj); - }, - readPageLabels: function Catalog_readPageLabels() { - var obj = this.catDict.getRaw('PageLabels'); - if (!obj) { - return null; + this.operators[conditionLocation] = endOfTrue; + this.operators[conditionLocation + 1] = 'jz'; + } else { + error('PS Function: error parsing conditional.'); } - var pageLabels = new Array(this.numPages); - var style = null; - var prefix = ''; - var start = 1; - - var numberTree = new NumberTree(obj, this.xref); - var nums = numberTree.getAll(); - var currentLabel = '', currentIndex = 1; - - for (var i = 0, ii = this.numPages; i < ii; i++) { - if (i in nums) { - var labelDict = nums[i]; - assert(isDict(labelDict), 'The PageLabel is not a dictionary.'); + } + }; + return PostScriptParser; +})(); - var type = labelDict.get('Type'); - assert(!type || (isName(type) && type.name === 'PageLabel'), - 'Invalid type in PageLabel dictionary.'); +var PostScriptTokenTypes = { + LBRACE: 0, + RBRACE: 1, + NUMBER: 2, + OPERATOR: 3, + IF: 4, + IFELSE: 5 +}; - var s = labelDict.get('S'); - assert(!s || isName(s), 'Invalid style in PageLabel dictionary.'); - style = (s ? s.name : null); +var PostScriptToken = (function PostScriptTokenClosure() { + function PostScriptToken(type, value) { + this.type = type; + this.value = value; + } - prefix = labelDict.get('P') || ''; - assert(isString(prefix), 'Invalid prefix in PageLabel dictionary.'); + var opCache = Object.create(null); - start = labelDict.get('St') || 1; - assert(isInt(start), 'Invalid start in PageLabel dictionary.'); - currentIndex = start; - } + PostScriptToken.getOperator = function PostScriptToken_getOperator(op) { + var opValue = opCache[op]; + if (opValue) { + return opValue; + } + return opCache[op] = new PostScriptToken(PostScriptTokenTypes.OPERATOR, op); + }; - switch (style) { - case 'D': - currentLabel = currentIndex; - break; - case 'R': - case 'r': - currentLabel = Util.toRoman(currentIndex, style === 'r'); - break; - case 'A': - case 'a': - var LIMIT = 26; // Use only the characters A--Z, or a--z. - var A_UPPER_CASE = 0x41, A_LOWER_CASE = 0x61; + PostScriptToken.LBRACE = new PostScriptToken(PostScriptTokenTypes.LBRACE, + '{'); + PostScriptToken.RBRACE = new PostScriptToken(PostScriptTokenTypes.RBRACE, + '}'); + PostScriptToken.IF = new PostScriptToken(PostScriptTokenTypes.IF, 'IF'); + PostScriptToken.IFELSE = new PostScriptToken(PostScriptTokenTypes.IFELSE, + 'IFELSE'); + return PostScriptToken; +})(); - var baseCharCode = (style === 'a' ? A_LOWER_CASE : A_UPPER_CASE); - var letterIndex = currentIndex - 1; - var character = String.fromCharCode(baseCharCode + - (letterIndex % LIMIT)); - var charBuf = []; - for (var j = 0, jj = (letterIndex / LIMIT) | 0; j <= jj; j++) { - charBuf.push(character); - } - currentLabel = charBuf.join(''); - break; - default: - assert(!style, - 'Invalid style "' + style + '" in PageLabel dictionary.'); - } - pageLabels[i] = prefix + currentLabel; +var PostScriptLexer = (function PostScriptLexerClosure() { + function PostScriptLexer(stream) { + this.stream = stream; + this.nextChar(); - currentLabel = ''; - currentIndex++; - } - return pageLabels; + this.strBuf = []; + } + PostScriptLexer.prototype = { + nextChar: function PostScriptLexer_nextChar() { + return (this.currentChar = this.stream.getByte()); }, + getToken: function PostScriptLexer_getToken() { + var comment = false; + var ch = this.currentChar; - get attachments() { - var xref = this.xref; - var attachments = null, nameTreeRef; - var obj = this.catDict.get('Names'); - if (obj) { - nameTreeRef = obj.getRaw('EmbeddedFiles'); - } + // skip comments + while (true) { + if (ch < 0) { + return EOF; + } - if (nameTreeRef) { - var nameTree = new NameTree(nameTreeRef, xref); - var names = nameTree.getAll(); - for (var name in names) { - var fs = new FileSpec(names[name], xref); - if (!attachments) { - attachments = Object.create(null); + if (comment) { + if (ch === 0x0A || ch === 0x0D) { + comment = false; } - attachments[stringToPDFString(name)] = fs.serializable; + } else if (ch === 0x25) { // '%' + comment = true; + } else if (!Lexer.isSpace(ch)) { + break; } + ch = this.nextChar(); } - return shadow(this, 'attachments', attachments); - }, - get javaScript() { - var xref = this.xref; - var obj = this.catDict.get('Names'); + switch (ch | 0) { + case 0x30: case 0x31: case 0x32: case 0x33: case 0x34: // '0'-'4' + case 0x35: case 0x36: case 0x37: case 0x38: case 0x39: // '5'-'9' + case 0x2B: case 0x2D: case 0x2E: // '+', '-', '.' + return new PostScriptToken(PostScriptTokenTypes.NUMBER, + this.getNumber()); + case 0x7B: // '{' + this.nextChar(); + return PostScriptToken.LBRACE; + case 0x7D: // '}' + this.nextChar(); + return PostScriptToken.RBRACE; + } + // operator + var strBuf = this.strBuf; + strBuf.length = 0; + strBuf[0] = String.fromCharCode(ch); - var javaScript = []; - function appendIfJavaScriptDict(jsDict) { - var type = jsDict.get('S'); - if (!isName(type) || type.name !== 'JavaScript') { - return; - } - var js = jsDict.get('JS'); - if (isStream(js)) { - js = bytesToString(js.getBytes()); - } else if (!isString(js)) { - return; - } - javaScript.push(stringToPDFString(js)); + while ((ch = this.nextChar()) >= 0 && // and 'A'-'Z', 'a'-'z' + ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { + strBuf.push(String.fromCharCode(ch)); } - if (obj && obj.has('JavaScript')) { - var nameTree = new NameTree(obj.getRaw('JavaScript'), xref); - var names = nameTree.getAll(); - for (var name in names) { - // We don't really use the JavaScript right now. This code is - // defensive so we don't cause errors on document load. - var jsDict = names[name]; - if (isDict(jsDict)) { - appendIfJavaScriptDict(jsDict); - } - } + var str = strBuf.join(''); + switch (str.toLowerCase()) { + case 'if': + return PostScriptToken.IF; + case 'ifelse': + return PostScriptToken.IFELSE; + default: + return PostScriptToken.getOperator(str); } + }, + getNumber: function PostScriptLexer_getNumber() { + var ch = this.currentChar; + var strBuf = this.strBuf; + strBuf.length = 0; + strBuf[0] = String.fromCharCode(ch); - // Append OpenAction actions to javaScript array - var openactionDict = this.catDict.get('OpenAction'); - if (isDict(openactionDict, 'Action')) { - var actionType = openactionDict.get('S'); - if (isName(actionType) && actionType.name === 'Named') { - // The named Print action is not a part of the PDF 1.7 specification, - // but is supported by many PDF readers/writers (including Adobe's). - var action = openactionDict.get('N'); - if (isName(action) && action.name === 'Print') { - javaScript.push('print({});'); - } + while ((ch = this.nextChar()) >= 0) { + if ((ch >= 0x30 && ch <= 0x39) || // '0'-'9' + ch === 0x2D || ch === 0x2E) { // '-', '.' + strBuf.push(String.fromCharCode(ch)); } else { - appendIfJavaScriptDict(openactionDict); + break; } } - - return shadow(this, 'javaScript', javaScript); - }, - - cleanup: function Catalog_cleanup() { - var promises = []; - this.fontCache.forEach(function (promise) { - promises.push(promise); - }); - return Promise.all(promises).then(function (translatedFonts) { - for (var i = 0, ii = translatedFonts.length; i < ii; i++) { - var font = translatedFonts[i].dict; - delete font.translated; - } - this.fontCache.clear(); - }.bind(this)); - }, - - getPage: function Catalog_getPage(pageIndex) { - if (!(pageIndex in this.pagePromises)) { - this.pagePromises[pageIndex] = this.getPageDict(pageIndex).then( - function (a) { - var dict = a[0]; - var ref = a[1]; - return this.pageFactory.createPage(pageIndex, dict, ref, - this.fontCache); - }.bind(this) - ); + var value = parseFloat(strBuf.join('')); + if (isNaN(value)) { + error('Invalid floating point number: ' + value); } - return this.pagePromises[pageIndex]; - }, + return value; + } + }; + return PostScriptLexer; +})(); - getPageDict: function Catalog_getPageDict(pageIndex) { - var capability = createPromiseCapability(); - var nodesToVisit = [this.catDict.getRaw('Pages')]; - var currentPageIndex = 0; - var xref = this.xref; - var checkAllKids = false; +exports.PostScriptLexer = PostScriptLexer; +exports.PostScriptParser = PostScriptParser; +})); - function next() { - while (nodesToVisit.length) { - var currentNode = nodesToVisit.pop(); - if (isRef(currentNode)) { - xref.fetchAsync(currentNode).then(function (obj) { - if (isDict(obj, 'Page') || (isDict(obj) && !obj.has('Kids'))) { - if (pageIndex === currentPageIndex) { - capability.resolve([obj, currentNode]); - } else { - currentPageIndex++; - next(); - } - return; - } - nodesToVisit.push(obj); - next(); - }, capability.reject); - return; - } +(function (root, factory) { + { + factory((root.pdfjsCoreFonts = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreStream, root.pdfjsCoreParser, + root.pdfjsCoreCMap, root.pdfjsCoreGlyphList, root.pdfjsCoreCharsets, + root.pdfjsCoreFontRenderer, root.pdfjsCoreEncodings, + root.pdfjsCoreStandardFonts, root.pdfjsCoreUnicode); + } +}(this, function (exports, sharedUtil, corePrimitives, coreStream, coreParser, + coreCMap, coreGlyphList, coreCharsets, coreFontRenderer, + coreEncodings, coreStandardFonts, coreUnicode) { - // Must be a child page dictionary. - assert( - isDict(currentNode), - 'page dictionary kid reference points to wrong type of object' - ); - var count = currentNode.get('Count'); - // If the current node doesn't have any children, avoid getting stuck - // in an empty node further down in the tree (see issue5644.pdf). - if (count === 0) { - checkAllKids = true; - } - // Skip nodes where the page can't be. - if (currentPageIndex + count <= pageIndex) { - currentPageIndex += count; - continue; - } +var FONT_IDENTITY_MATRIX = sharedUtil.FONT_IDENTITY_MATRIX; +var FontType = sharedUtil.FontType; +var Util = sharedUtil.Util; +var assert = sharedUtil.assert; +var bytesToString = sharedUtil.bytesToString; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isInt = sharedUtil.isInt; +var isNum = sharedUtil.isNum; +var readUint32 = sharedUtil.readUint32; +var shadow = sharedUtil.shadow; +var stringToBytes = sharedUtil.stringToBytes; +var string32 = sharedUtil.string32; +var warn = sharedUtil.warn; +var Name = corePrimitives.Name; +var Stream = coreStream.Stream; +var Lexer = coreParser.Lexer; +var CMapFactory = coreCMap.CMapFactory; +var IdentityCMap = coreCMap.IdentityCMap; +var getGlyphsUnicode = coreGlyphList.getGlyphsUnicode; +var getDingbatsGlyphsUnicode = coreGlyphList.getDingbatsGlyphsUnicode; +var ISOAdobeCharset = coreCharsets.ISOAdobeCharset; +var ExpertCharset = coreCharsets.ExpertCharset; +var ExpertSubsetCharset = coreCharsets.ExpertSubsetCharset; +var FontRendererFactory = coreFontRenderer.FontRendererFactory; +var WinAnsiEncoding = coreEncodings.WinAnsiEncoding; +var StandardEncoding = coreEncodings.StandardEncoding; +var MacRomanEncoding = coreEncodings.MacRomanEncoding; +var SymbolSetEncoding = coreEncodings.SymbolSetEncoding; +var ZapfDingbatsEncoding = coreEncodings.ZapfDingbatsEncoding; +var ExpertEncoding = coreEncodings.ExpertEncoding; +var getEncoding = coreEncodings.getEncoding; +var getStdFontMap = coreStandardFonts.getStdFontMap; +var getNonStdFontMap = coreStandardFonts.getNonStdFontMap; +var getGlyphMapForStandardFonts = coreStandardFonts.getGlyphMapForStandardFonts; +var getSupplementalGlyphMapForArialBlack = + coreStandardFonts.getSupplementalGlyphMapForArialBlack; +var getUnicodeRangeFor = coreUnicode.getUnicodeRangeFor; +var mapSpecialUnicodeValues = coreUnicode.mapSpecialUnicodeValues; - var kids = currentNode.get('Kids'); - assert(isArray(kids), 'page dictionary kids object is not an array'); - if (!checkAllKids && count === kids.length) { - // Nodes that don't have the page have been skipped and this is the - // bottom of the tree which means the page requested must be a - // descendant of this pages node. Ideally we would just resolve the - // promise with the page ref here, but there is the case where more - // pages nodes could link to single a page (see issue 3666 pdf). To - // handle this push it back on the queue so if it is a pages node it - // will be descended into. - nodesToVisit = [kids[pageIndex - currentPageIndex]]; - currentPageIndex = pageIndex; - continue; - } else { - for (var last = kids.length - 1; last >= 0; last--) { - nodesToVisit.push(kids[last]); - } - } - } - capability.reject('Page index ' + pageIndex + ' not found.'); - } - next(); - return capability.promise; - }, +// Unicode Private Use Area +var PRIVATE_USE_OFFSET_START = 0xE000; +var PRIVATE_USE_OFFSET_END = 0xF8FF; +var SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = false; - getPageIndex: function Catalog_getPageIndex(ref) { - // The page tree nodes have the count of all the leaves below them. To get - // how many pages are before we just have to walk up the tree and keep - // adding the count of siblings to the left of the node. - var xref = this.xref; - function pagesBeforeRef(kidRef) { - var total = 0; - var parentRef; - return xref.fetchAsync(kidRef).then(function (node) { - if (!node) { - return null; - } - parentRef = node.getRaw('Parent'); - return node.getAsync('Parent'); - }).then(function (parent) { - if (!parent) { - return null; - } - return parent.getAsync('Kids'); - }).then(function (kids) { - if (!kids) { - return null; - } - var kidPromises = []; - var found = false; - for (var i = 0; i < kids.length; i++) { - var kid = kids[i]; - assert(isRef(kid), 'kids must be a ref'); - if (kid.num === kidRef.num) { - found = true; - break; - } - kidPromises.push(xref.fetchAsync(kid).then(function (kid) { - if (kid.has('Count')) { - var count = kid.get('Count'); - total += count; - } else { // page leaf node - total++; - } - })); - } - if (!found) { - error('kid ref not found in parents kids'); - } - return Promise.all(kidPromises).then(function () { - return [total, parentRef]; - }); - }); - } +// PDF Glyph Space Units are one Thousandth of a TextSpace Unit +// except for Type 3 fonts +var PDF_GLYPH_SPACE_UNITS = 1000; - var total = 0; - function next(ref) { - return pagesBeforeRef(ref).then(function (args) { - if (!args) { - return total; - } - var count = args[0]; - var parentRef = args[1]; - total += count; - return next(parentRef); - }); - } +// Hinting is currently disabled due to unknown problems on windows +// in tracemonkey and various other pdfs with type1 fonts. +var HINTING_ENABLED = false; - return next(ref); - } +// Accented charactars are not displayed properly on windows, using this flag +// to control analysis of seac charstrings. +var SEAC_ANALYSIS_ENABLED = false; + +// Maximum subroutine call depth of type 2 chartrings. Matches OTS. +var MAX_SUBR_NESTING = 10; + +var FontFlags = { + FixedPitch: 1, + Serif: 2, + Symbolic: 4, + Script: 8, + Nonsymbolic: 32, + Italic: 64, + AllCap: 65536, + SmallCap: 131072, + ForceBold: 262144 +}; + +var MacStandardGlyphOrdering = [ + '.notdef', '.null', 'nonmarkingreturn', 'space', 'exclam', 'quotedbl', + 'numbersign', 'dollar', 'percent', 'ampersand', 'quotesingle', 'parenleft', + 'parenright', 'asterisk', 'plus', 'comma', 'hyphen', 'period', 'slash', + 'zero', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', + 'nine', 'colon', 'semicolon', 'less', 'equal', 'greater', 'question', 'at', + 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', + 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'bracketleft', + 'backslash', 'bracketright', 'asciicircum', 'underscore', 'grave', 'a', 'b', + 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', + 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 'braceleft', 'bar', 'braceright', + 'asciitilde', 'Adieresis', 'Aring', 'Ccedilla', 'Eacute', 'Ntilde', + 'Odieresis', 'Udieresis', 'aacute', 'agrave', 'acircumflex', 'adieresis', + 'atilde', 'aring', 'ccedilla', 'eacute', 'egrave', 'ecircumflex', 'edieresis', + 'iacute', 'igrave', 'icircumflex', 'idieresis', 'ntilde', 'oacute', 'ograve', + 'ocircumflex', 'odieresis', 'otilde', 'uacute', 'ugrave', 'ucircumflex', + 'udieresis', 'dagger', 'degree', 'cent', 'sterling', 'section', 'bullet', + 'paragraph', 'germandbls', 'registered', 'copyright', 'trademark', 'acute', + 'dieresis', 'notequal', 'AE', 'Oslash', 'infinity', 'plusminus', 'lessequal', + 'greaterequal', 'yen', 'mu', 'partialdiff', 'summation', 'product', 'pi', + 'integral', 'ordfeminine', 'ordmasculine', 'Omega', 'ae', 'oslash', + 'questiondown', 'exclamdown', 'logicalnot', 'radical', 'florin', + 'approxequal', 'Delta', 'guillemotleft', 'guillemotright', 'ellipsis', + 'nonbreakingspace', 'Agrave', 'Atilde', 'Otilde', 'OE', 'oe', 'endash', + 'emdash', 'quotedblleft', 'quotedblright', 'quoteleft', 'quoteright', + 'divide', 'lozenge', 'ydieresis', 'Ydieresis', 'fraction', 'currency', + 'guilsinglleft', 'guilsinglright', 'fi', 'fl', 'daggerdbl', 'periodcentered', + 'quotesinglbase', 'quotedblbase', 'perthousand', 'Acircumflex', + 'Ecircumflex', 'Aacute', 'Edieresis', 'Egrave', 'Iacute', 'Icircumflex', + 'Idieresis', 'Igrave', 'Oacute', 'Ocircumflex', 'apple', 'Ograve', 'Uacute', + 'Ucircumflex', 'Ugrave', 'dotlessi', 'circumflex', 'tilde', 'macron', + 'breve', 'dotaccent', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', + 'Lslash', 'lslash', 'Scaron', 'scaron', 'Zcaron', 'zcaron', 'brokenbar', + 'Eth', 'eth', 'Yacute', 'yacute', 'Thorn', 'thorn', 'minus', 'multiply', + 'onesuperior', 'twosuperior', 'threesuperior', 'onehalf', 'onequarter', + 'threequarters', 'franc', 'Gbreve', 'gbreve', 'Idotaccent', 'Scedilla', + 'scedilla', 'Cacute', 'cacute', 'Ccaron', 'ccaron', 'dcroat']; + +function adjustWidths(properties) { + if (!properties.fontMatrix) { + return; + } + if (properties.fontMatrix[0] === FONT_IDENTITY_MATRIX[0]) { + return; + } + // adjusting width to fontMatrix scale + var scale = 0.001 / properties.fontMatrix[0]; + var glyphsWidths = properties.widths; + for (var glyph in glyphsWidths) { + glyphsWidths[glyph] *= scale; + } + properties.defaultWidth *= scale; +} + +function getFontType(type, subtype) { + switch (type) { + case 'Type1': + return subtype === 'Type1C' ? FontType.TYPE1C : FontType.TYPE1; + case 'CIDFontType0': + return subtype === 'CIDFontType0C' ? FontType.CIDFONTTYPE0C : + FontType.CIDFONTTYPE0; + case 'OpenType': + return FontType.OPENTYPE; + case 'TrueType': + return FontType.TRUETYPE; + case 'CIDFontType2': + return FontType.CIDFONTTYPE2; + case 'MMType1': + return FontType.MMTYPE1; + case 'Type0': + return FontType.TYPE0; + default: + return FontType.UNKNOWN; + } +} + +var Glyph = (function GlyphClosure() { + function Glyph(fontChar, unicode, accent, width, vmetric, operatorListId, + isSpace) { + this.fontChar = fontChar; + this.unicode = unicode; + this.accent = accent; + this.width = width; + this.vmetric = vmetric; + this.operatorListId = operatorListId; + this.isSpace = isSpace; + } + + Glyph.prototype.matchesForCache = function(fontChar, unicode, accent, width, + vmetric, operatorListId, isSpace) { + return this.fontChar === fontChar && + this.unicode === unicode && + this.accent === accent && + this.width === width && + this.vmetric === vmetric && + this.operatorListId === operatorListId && + this.isSpace === isSpace; }; - return Catalog; + return Glyph; })(); -var XRef = (function XRefClosure() { - function XRef(stream, password) { - this.stream = stream; - this.entries = []; - this.xrefstms = Object.create(null); - // prepare the XRef cache - this.cache = []; - this.password = password; - this.stats = { - streamTypes: [], - fontTypes: [] - }; +var ToUnicodeMap = (function ToUnicodeMapClosure() { + function ToUnicodeMap(cmap) { + // The elements of this._map can be integers or strings, depending on how + // |cmap| was created. + this._map = cmap; } - XRef.prototype = { - setStartXRef: function XRef_setStartXRef(startXRef) { - // Store the starting positions of xref tables as we process them - // so we can recover from missing data errors - this.startXRefQueue = [startXRef]; + ToUnicodeMap.prototype = { + get length() { + return this._map.length; }, - parse: function XRef_parse(recoveryMode) { - var trailerDict; - if (!recoveryMode) { - trailerDict = this.readXRef(); - } else { - warn('Indexing all PDF objects'); - trailerDict = this.indexObjects(); - } - trailerDict.assignXref(this); - this.trailer = trailerDict; - var encrypt = trailerDict.get('Encrypt'); - if (encrypt) { - var ids = trailerDict.get('ID'); - var fileId = (ids && ids.length) ? ids[0] : ''; - this.encrypt = new CipherTransformFactory(encrypt, fileId, - this.password); + forEach: function(callback) { + for (var charCode in this._map) { + callback(charCode, this._map[charCode].charCodeAt(0)); } + }, - // get the root dictionary (catalog) object - if (!(this.root = trailerDict.get('Root'))) { - error('Invalid root reference'); - } + has: function(i) { + return this._map[i] !== undefined; }, - processXRefTable: function XRef_processXRefTable(parser) { - if (!('tableState' in this)) { - // Stores state of the table as we process it so we can resume - // from middle of table in case of missing data error - this.tableState = { - entryNum: 0, - streamPos: parser.lexer.stream.pos, - parserBuf1: parser.buf1, - parserBuf2: parser.buf2 - }; - } + get: function(i) { + return this._map[i]; + }, - var obj = this.readXRefTable(parser); + charCodeOf: function(v) { + return this._map.indexOf(v); + } + }; - // Sanity check - if (!isCmd(obj, 'trailer')) { - error('Invalid XRef table: could not find trailer dictionary'); - } - // Read trailer dictionary, e.g. - // trailer - // << /Size 22 - // /Root 20R - // /Info 10R - // /ID [ <81b14aafa313db63dbd6f981e49f94f4> ] - // >> - // The parser goes through the entire stream << ... >> and provides - // a getter interface for the key-value table - var dict = parser.getObj(); + return ToUnicodeMap; +})(); - // The pdflib PDF generator can generate a nested trailer dictionary - if (!isDict(dict) && dict.dict) { - dict = dict.dict; - } - if (!isDict(dict)) { - error('Invalid XRef table: could not parse trailer dictionary'); - } - delete this.tableState; +var IdentityToUnicodeMap = (function IdentityToUnicodeMapClosure() { + function IdentityToUnicodeMap(firstChar, lastChar) { + this.firstChar = firstChar; + this.lastChar = lastChar; + } - return dict; + IdentityToUnicodeMap.prototype = { + get length() { + return (this.lastChar + 1) - this.firstChar; }, - readXRefTable: function XRef_readXRefTable(parser) { - // Example of cross-reference table: - // xref - // 0 1 <-- subsection header (first obj #, obj count) - // 0000000000 65535 f <-- actual object (offset, generation #, f/n) - // 23 2 <-- subsection header ... and so on ... - // 0000025518 00002 n - // 0000025635 00000 n - // trailer - // ... - - var stream = parser.lexer.stream; - var tableState = this.tableState; - stream.pos = tableState.streamPos; - parser.buf1 = tableState.parserBuf1; - parser.buf2 = tableState.parserBuf2; - - // Outer loop is over subsection headers - var obj; + forEach: function (callback) { + for (var i = this.firstChar, ii = this.lastChar; i <= ii; i++) { + callback(i, i); + } + }, - while (true) { - if (!('firstEntryNum' in tableState) || !('entryCount' in tableState)) { - if (isCmd(obj = parser.getObj(), 'trailer')) { - break; - } - tableState.firstEntryNum = obj; - tableState.entryCount = parser.getObj(); - } + has: function (i) { + return this.firstChar <= i && i <= this.lastChar; + }, - var first = tableState.firstEntryNum; - var count = tableState.entryCount; - if (!isInt(first) || !isInt(count)) { - error('Invalid XRef table: wrong types in subsection header'); - } - // Inner loop is over objects themselves - for (var i = tableState.entryNum; i < count; i++) { - tableState.streamPos = stream.pos; - tableState.entryNum = i; - tableState.parserBuf1 = parser.buf1; - tableState.parserBuf2 = parser.buf2; + get: function (i) { + if (this.firstChar <= i && i <= this.lastChar) { + return String.fromCharCode(i); + } + return undefined; + }, - var entry = {}; - entry.offset = parser.getObj(); - entry.gen = parser.getObj(); - var type = parser.getObj(); + charCodeOf: function (v) { + return (isInt(v) && v >= this.firstChar && v <= this.lastChar) ? v : -1; + } + }; - if (isCmd(type, 'f')) { - entry.free = true; - } else if (isCmd(type, 'n')) { - entry.uncompressed = true; - } + return IdentityToUnicodeMap; +})(); - // Validate entry obj - if (!isInt(entry.offset) || !isInt(entry.gen) || - !(entry.free || entry.uncompressed)) { - error('Invalid entry in XRef subsection: ' + first + ', ' + count); - } +var OpenTypeFileBuilder = (function OpenTypeFileBuilderClosure() { + function writeInt16(dest, offset, num) { + dest[offset] = (num >> 8) & 0xFF; + dest[offset + 1] = num & 0xFF; + } - if (!this.entries[i + first]) { - this.entries[i + first] = entry; - } - } + function writeInt32(dest, offset, num) { + dest[offset] = (num >> 24) & 0xFF; + dest[offset + 1] = (num >> 16) & 0xFF; + dest[offset + 2] = (num >> 8) & 0xFF; + dest[offset + 3] = num & 0xFF; + } - tableState.entryNum = 0; - tableState.streamPos = stream.pos; - tableState.parserBuf1 = parser.buf1; - tableState.parserBuf2 = parser.buf2; - delete tableState.firstEntryNum; - delete tableState.entryCount; + function writeData(dest, offset, data) { + var i, ii; + if (data instanceof Uint8Array) { + dest.set(data, offset); + } else if (typeof data === 'string') { + for (i = 0, ii = data.length; i < ii; i++) { + dest[offset++] = data.charCodeAt(i) & 0xFF; } + } else { + // treating everything else as array + for (i = 0, ii = data.length; i < ii; i++) { + dest[offset++] = data[i] & 0xFF; + } + } + } - // Per issue 3248: hp scanners generate bad XRef - if (first === 1 && this.entries[1] && this.entries[1].free) { - // shifting the entries - this.entries.shift(); - } - - // Sanity check: as per spec, first object must be free - if (this.entries[0] && !this.entries[0].free) { - error('Invalid XRef table: unexpected first object'); - } - return obj; - }, - - processXRefStream: function XRef_processXRefStream(stream) { - if (!('streamState' in this)) { - // Stores state of the stream as we process it so we can resume - // from middle of stream in case of missing data error - var streamParameters = stream.dict; - var byteWidths = streamParameters.get('W'); - var range = streamParameters.get('Index'); - if (!range) { - range = [0, streamParameters.get('Size')]; - } - - this.streamState = { - entryRanges: range, - byteWidths: byteWidths, - entryNum: 0, - streamPos: stream.pos - }; - } - this.readXRefStream(stream); - delete this.streamState; + function OpenTypeFileBuilder(sfnt) { + this.sfnt = sfnt; + this.tables = Object.create(null); + } - return stream.dict; - }, + OpenTypeFileBuilder.getSearchParams = + function OpenTypeFileBuilder_getSearchParams(entriesCount, entrySize) { + var maxPower2 = 1, log2 = 0; + while ((maxPower2 ^ entriesCount) > maxPower2) { + maxPower2 <<= 1; + log2++; + } + var searchRange = maxPower2 * entrySize; + return { + range: searchRange, + entry: log2, + rangeShift: entrySize * entriesCount - searchRange + }; + }; - readXRefStream: function XRef_readXRefStream(stream) { - var i, j; - var streamState = this.streamState; - stream.pos = streamState.streamPos; + var OTF_HEADER_SIZE = 12; + var OTF_TABLE_ENTRY_SIZE = 16; - var byteWidths = streamState.byteWidths; - var typeFieldWidth = byteWidths[0]; - var offsetFieldWidth = byteWidths[1]; - var generationFieldWidth = byteWidths[2]; + OpenTypeFileBuilder.prototype = { + toArray: function OpenTypeFileBuilder_toArray() { + var sfnt = this.sfnt; - var entryRanges = streamState.entryRanges; - while (entryRanges.length > 0) { - var first = entryRanges[0]; - var n = entryRanges[1]; + // Tables needs to be written by ascendant alphabetic order + var tables = this.tables; + var tablesNames = Object.keys(tables); + tablesNames.sort(); + var numTables = tablesNames.length; - if (!isInt(first) || !isInt(n)) { - error('Invalid XRef range fields: ' + first + ', ' + n); - } - if (!isInt(typeFieldWidth) || !isInt(offsetFieldWidth) || - !isInt(generationFieldWidth)) { - error('Invalid XRef entry fields length: ' + first + ', ' + n); - } - for (i = streamState.entryNum; i < n; ++i) { - streamState.entryNum = i; - streamState.streamPos = stream.pos; + var i, j, jj, table, tableName; + // layout the tables data + var offset = OTF_HEADER_SIZE + numTables * OTF_TABLE_ENTRY_SIZE; + var tableOffsets = [offset]; + for (i = 0; i < numTables; i++) { + table = tables[tablesNames[i]]; + var paddedLength = ((table.length + 3) & ~3) >>> 0; + offset += paddedLength; + tableOffsets.push(offset); + } - var type = 0, offset = 0, generation = 0; - for (j = 0; j < typeFieldWidth; ++j) { - type = (type << 8) | stream.getByte(); - } - // if type field is absent, its default value is 1 - if (typeFieldWidth === 0) { - type = 1; - } - for (j = 0; j < offsetFieldWidth; ++j) { - offset = (offset << 8) | stream.getByte(); - } - for (j = 0; j < generationFieldWidth; ++j) { - generation = (generation << 8) | stream.getByte(); - } - var entry = {}; - entry.offset = offset; - entry.gen = generation; - switch (type) { - case 0: - entry.free = true; - break; - case 1: - entry.uncompressed = true; - break; - case 2: - break; - default: - error('Invalid XRef entry type: ' + type); - } - if (!this.entries[first + i]) { - this.entries[first + i] = entry; - } - } + var file = new Uint8Array(offset); + // write the table data first (mostly for checksum) + for (i = 0; i < numTables; i++) { + table = tables[tablesNames[i]]; + writeData(file, tableOffsets[i], table); + } - streamState.entryNum = 0; - streamState.streamPos = stream.pos; - entryRanges.splice(0, 2); + // sfnt version (4 bytes) + if (sfnt === 'true') { + // Windows hates the Mac TrueType sfnt version number + sfnt = string32(0x00010000); } - }, + file[0] = sfnt.charCodeAt(0) & 0xFF; + file[1] = sfnt.charCodeAt(1) & 0xFF; + file[2] = sfnt.charCodeAt(2) & 0xFF; + file[3] = sfnt.charCodeAt(3) & 0xFF; - indexObjects: function XRef_indexObjects() { - // Simple scan through the PDF content to find objects, - // trailers and XRef streams. - var TAB = 0x9, LF = 0xA, CR = 0xD, SPACE = 0x20; - var PERCENT = 0x25, LT = 0x3C; + // numTables (2 bytes) + writeInt16(file, 4, numTables); - function readToken(data, offset) { - var token = '', ch = data[offset]; - while (ch !== LF && ch !== CR && ch !== LT) { - if (++offset >= data.length) { - break; - } - token += String.fromCharCode(ch); - ch = data[offset]; - } - return token; - } - function skipUntil(data, offset, what) { - var length = what.length, dataLength = data.length; - var skipped = 0; - // finding byte sequence - while (offset < dataLength) { - var i = 0; - while (i < length && data[offset + i] === what[i]) { - ++i; - } - if (i >= length) { - break; // sequence found - } - offset++; - skipped++; - } - return skipped; - } - var objRegExp = /^(\d+)\s+(\d+)\s+obj\b/; - var trailerBytes = new Uint8Array([116, 114, 97, 105, 108, 101, 114]); - var startxrefBytes = new Uint8Array([115, 116, 97, 114, 116, 120, 114, - 101, 102]); - var endobjBytes = new Uint8Array([101, 110, 100, 111, 98, 106]); - var xrefBytes = new Uint8Array([47, 88, 82, 101, 102]); + var searchParams = OpenTypeFileBuilder.getSearchParams(numTables, 16); - // Clear out any existing entries, since they may be bogus. - this.entries.length = 0; + // searchRange (2 bytes) + writeInt16(file, 6, searchParams.range); + // entrySelector (2 bytes) + writeInt16(file, 8, searchParams.entry); + // rangeShift (2 bytes) + writeInt16(file, 10, searchParams.rangeShift); - var stream = this.stream; - stream.pos = 0; - var buffer = stream.getBytes(); - var position = stream.start, length = buffer.length; - var trailers = [], xrefStms = []; - while (position < length) { - var ch = buffer[position]; - if (ch === TAB || ch === LF || ch === CR || ch === SPACE) { - ++position; - continue; - } - if (ch === PERCENT) { // %-comment - do { - ++position; - if (position >= length) { - break; - } - ch = buffer[position]; - } while (ch !== LF && ch !== CR); - continue; + offset = OTF_HEADER_SIZE; + // writing table entries + for (i = 0; i < numTables; i++) { + tableName = tablesNames[i]; + file[offset] = tableName.charCodeAt(0) & 0xFF; + file[offset + 1] = tableName.charCodeAt(1) & 0xFF; + file[offset + 2] = tableName.charCodeAt(2) & 0xFF; + file[offset + 3] = tableName.charCodeAt(3) & 0xFF; + + // checksum + var checksum = 0; + for (j = tableOffsets[i], jj = tableOffsets[i + 1]; j < jj; j += 4) { + var quad = (file[j] << 24) + (file[j + 1] << 16) + + (file[j + 2] << 8) + file[j + 3]; + checksum = (checksum + quad) | 0; } - var token = readToken(buffer, position); - var m; - if (token.indexOf('xref') === 0 && - (token.length === 4 || /\s/.test(token[4]))) { - position += skipUntil(buffer, position, trailerBytes); - trailers.push(position); - position += skipUntil(buffer, position, startxrefBytes); - } else if ((m = objRegExp.exec(token))) { - if (typeof this.entries[m[1]] === 'undefined') { - this.entries[m[1]] = { - offset: position - stream.start, - gen: m[2] | 0, - uncompressed: true - }; - } - var contentLength = skipUntil(buffer, position, endobjBytes) + 7; - var content = buffer.subarray(position, position + contentLength); + writeInt32(file, offset + 4, checksum); - // checking XRef stream suspect - // (it shall have '/XRef' and next char is not a letter) - var xrefTagOffset = skipUntil(content, 0, xrefBytes); - if (xrefTagOffset < contentLength && - content[xrefTagOffset + 5] < 64) { - xrefStms.push(position - stream.start); - this.xrefstms[position - stream.start] = 1; // Avoid recursion - } + // offset + writeInt32(file, offset + 8, tableOffsets[i]); + // length + writeInt32(file, offset + 12, tables[tableName].length); - position += contentLength; - } else if (token.indexOf('trailer') === 0 && - (token.length === 7 || /\s/.test(token[7]))) { - trailers.push(position); - position += skipUntil(buffer, position, startxrefBytes); - } else { - position += token.length + 1; - } - } - // reading XRef streams - var i, ii; - for (i = 0, ii = xrefStms.length; i < ii; ++i) { - this.startXRefQueue.push(xrefStms[i]); - this.readXRef(/* recoveryMode */ true); - } - // finding main trailer - var dict; - for (i = 0, ii = trailers.length; i < ii; ++i) { - stream.pos = trailers[i]; - var parser = new Parser(new Lexer(stream), true, this); - var obj = parser.getObj(); - if (!isCmd(obj, 'trailer')) { - continue; - } - // read the trailer dictionary - if (!isDict(dict = parser.getObj())) { - continue; - } - // taking the first one with 'ID' - if (dict.has('ID')) { - return dict; - } - } - // no tailer with 'ID', taking last one (if exists) - if (dict) { - return dict; + offset += OTF_TABLE_ENTRY_SIZE; } - // nothing helps - // calling error() would reject worker with an UnknownErrorException. - throw new InvalidPDFException('Invalid PDF structure'); + return file; }, - readXRef: function XRef_readXRef(recoveryMode) { - var stream = this.stream; + addTable: function OpenTypeFileBuilder_addTable(tag, data) { + if (tag in this.tables) { + throw new Error('Table ' + tag + ' already exists'); + } + this.tables[tag] = data; + } + }; - try { - while (this.startXRefQueue.length) { - var startXRef = this.startXRefQueue[0]; + return OpenTypeFileBuilder; +})(); - stream.pos = startXRef + stream.start; +// Problematic Unicode characters in the fonts that needs to be moved to avoid +// issues when they are painted on the canvas, e.g. complex-script shaping or +// control/whitespace characters. The ranges are listed in pairs: the first item +// is a code of the first problematic code, the second one is the next +// non-problematic code. The ranges must be in sorted order. +var ProblematicCharRanges = new Int32Array([ + // Control characters. + 0x0000, 0x0020, + 0x007F, 0x00A1, + 0x00AD, 0x00AE, + // Chars that is used in complex-script shaping. + 0x0600, 0x0780, + 0x08A0, 0x10A0, + 0x1780, 0x1800, + // General punctuation chars. + 0x2000, 0x2010, + 0x2011, 0x2012, + 0x2028, 0x2030, + 0x205F, 0x2070, + 0x25CC, 0x25CD, + // Chars that is used in complex-script shaping. + 0xAA60, 0xAA80, + // Specials Unicode block. + 0xFFF0, 0x10000 +]); - var parser = new Parser(new Lexer(stream), true, this); - var obj = parser.getObj(); - var dict; +/** + * 'Font' is the class the outside world should use, it encapsulate all the font + * decoding logics whatever type it is (assuming the font type is supported). + * + * For example to read a Type1 font and to attach it to the document: + * var type1Font = new Font("MyFontName", binaryFile, propertiesObject); + * type1Font.bind(); + */ +var Font = (function FontClosure() { + function Font(name, file, properties) { + var charCode, glyphName, fontChar; - // Get dictionary - if (isCmd(obj, 'xref')) { - // Parse end-of-file XRef - dict = this.processXRefTable(parser); - if (!this.topDict) { - this.topDict = dict; - } + this.name = name; + this.loadedName = properties.loadedName; + this.isType3Font = properties.isType3Font; + this.sizes = []; - // Recursively get other XRefs 'XRefStm', if any - obj = dict.get('XRefStm'); - if (isInt(obj)) { - var pos = obj; - // ignore previously loaded xref streams - // (possible infinite recursion) - if (!(pos in this.xrefstms)) { - this.xrefstms[pos] = 1; - this.startXRefQueue.push(pos); - } - } - } else if (isInt(obj)) { - // Parse in-stream XRef - if (!isInt(parser.getObj()) || - !isCmd(parser.getObj(), 'obj') || - !isStream(obj = parser.getObj())) { - error('Invalid XRef stream'); - } - dict = this.processXRefStream(obj); - if (!this.topDict) { - this.topDict = dict; - } - if (!dict) { - error('Failed to read XRef stream'); - } - } else { - error('Invalid XRef stream header'); - } + this.glyphCache = Object.create(null); - // Recursively get previous dictionary, if any - obj = dict.get('Prev'); - if (isInt(obj)) { - this.startXRefQueue.push(obj); - } else if (isRef(obj)) { - // The spec says Prev must not be a reference, i.e. "/Prev NNN" - // This is a fallback for non-compliant PDFs, i.e. "/Prev NNN 0 R" - this.startXRefQueue.push(obj.num); - } + var names = name.split('+'); + names = names.length > 1 ? names[1] : names[0]; + names = names.split(/[-,_]/g)[0]; + this.isSerifFont = !!(properties.flags & FontFlags.Serif); + this.isSymbolicFont = !!(properties.flags & FontFlags.Symbolic); + this.isMonospace = !!(properties.flags & FontFlags.FixedPitch); - this.startXRefQueue.shift(); - } + var type = properties.type; + var subtype = properties.subtype; + this.type = type; - return this.topDict; - } catch (e) { - if (e instanceof MissingDataException) { - throw e; - } - info('(while reading XRef): ' + e); - } + this.fallbackName = (this.isMonospace ? 'monospace' : + (this.isSerifFont ? 'serif' : 'sans-serif')); - if (recoveryMode) { - return; - } - throw new XRefParseException(); - }, + this.differences = properties.differences; + this.widths = properties.widths; + this.defaultWidth = properties.defaultWidth; + this.composite = properties.composite; + this.wideChars = properties.wideChars; + this.cMap = properties.cMap; + this.ascent = properties.ascent / PDF_GLYPH_SPACE_UNITS; + this.descent = properties.descent / PDF_GLYPH_SPACE_UNITS; + this.fontMatrix = properties.fontMatrix; + this.bbox = properties.bbox; - getEntry: function XRef_getEntry(i) { - var xrefEntry = this.entries[i]; - if (xrefEntry && !xrefEntry.free && xrefEntry.offset) { - return xrefEntry; - } - return null; - }, + this.toUnicode = properties.toUnicode = this.buildToUnicode(properties); - fetchIfRef: function XRef_fetchIfRef(obj) { - if (!isRef(obj)) { - return obj; + this.toFontChar = []; + + if (properties.type === 'Type3') { + for (charCode = 0; charCode < 256; charCode++) { + this.toFontChar[charCode] = (this.differences[charCode] || + properties.defaultEncoding[charCode]); } - return this.fetch(obj); - }, + this.fontType = FontType.TYPE3; + return; + } - fetch: function XRef_fetch(ref, suppressEncryption) { - assert(isRef(ref), 'ref object is not a reference'); - var num = ref.num; - if (num in this.cache) { - var cacheEntry = this.cache[num]; - return cacheEntry; + this.cidEncoding = properties.cidEncoding; + this.vertical = properties.vertical; + if (this.vertical) { + this.vmetrics = properties.vmetrics; + this.defaultVMetrics = properties.defaultVMetrics; + } + var glyphsUnicodeMap; + if (!file || file.isEmpty) { + if (file) { + // Some bad PDF generators will include empty font files, + // attempting to recover by assuming that no file exists. + warn('Font file is empty in "' + name + '" (' + this.loadedName + ')'); } - var xrefEntry = this.getEntry(num); + this.missingFile = true; + // The file data is not specified. Trying to fix the font name + // to be used with the canvas.font. + var fontName = name.replace(/[,_]/g, '-'); + var stdFontMap = getStdFontMap(), nonStdFontMap = getNonStdFontMap(); + var isStandardFont = !!stdFontMap[fontName] || + !!(nonStdFontMap[fontName] && stdFontMap[nonStdFontMap[fontName]]); + fontName = stdFontMap[fontName] || nonStdFontMap[fontName] || fontName; - // the referenced entry can be free - if (xrefEntry === null) { - return (this.cache[num] = null); - } + this.bold = (fontName.search(/bold/gi) !== -1); + this.italic = ((fontName.search(/oblique/gi) !== -1) || + (fontName.search(/italic/gi) !== -1)); - if (xrefEntry.uncompressed) { - xrefEntry = this.fetchUncompressed(ref, xrefEntry, suppressEncryption); - } else { - xrefEntry = this.fetchCompressed(xrefEntry, suppressEncryption); - } - if (isDict(xrefEntry)){ - xrefEntry.objId = ref.toString(); - } else if (isStream(xrefEntry)) { - xrefEntry.dict.objId = ref.toString(); - } - return xrefEntry; - }, + // Use 'name' instead of 'fontName' here because the original + // name ArialBlack for example will be replaced by Helvetica. + this.black = (name.search(/Black/g) !== -1); - fetchUncompressed: function XRef_fetchUncompressed(ref, xrefEntry, - suppressEncryption) { - var gen = ref.gen; - var num = ref.num; - if (xrefEntry.gen !== gen) { - error('inconsistent generation in XRef'); - } - var stream = this.stream.makeSubStream(xrefEntry.offset + - this.stream.start); - var parser = new Parser(new Lexer(stream), true, this); - var obj1 = parser.getObj(); - var obj2 = parser.getObj(); - var obj3 = parser.getObj(); - if (!isInt(obj1) || parseInt(obj1, 10) !== num || - !isInt(obj2) || parseInt(obj2, 10) !== gen || - !isCmd(obj3)) { - error('bad XRef entry'); - } - if (!isCmd(obj3, 'obj')) { - // some bad PDFs use "obj1234" and really mean 1234 - if (obj3.cmd.indexOf('obj') === 0) { - num = parseInt(obj3.cmd.substring(3), 10); - if (!isNaN(num)) { - return num; - } + // if at least one width is present, remeasure all chars when exists + this.remeasure = Object.keys(this.widths).length > 0; + if (isStandardFont && type === 'CIDFontType2' && + properties.cidEncoding.indexOf('Identity-') === 0) { + var GlyphMapForStandardFonts = getGlyphMapForStandardFonts(); + // Standard fonts might be embedded as CID font without glyph mapping. + // Building one based on GlyphMapForStandardFonts. + var map = []; + for (charCode in GlyphMapForStandardFonts) { + map[+charCode] = GlyphMapForStandardFonts[charCode]; } - error('bad XRef entry'); - } - if (this.encrypt && !suppressEncryption) { - xrefEntry = parser.getObj(this.encrypt.createCipherTransform(num, gen)); - } else { - xrefEntry = parser.getObj(); - } - if (!isStream(xrefEntry)) { - this.cache[num] = xrefEntry; - } - return xrefEntry; - }, - - fetchCompressed: function XRef_fetchCompressed(xrefEntry, - suppressEncryption) { - var tableOffset = xrefEntry.offset; - var stream = this.fetch(new Ref(tableOffset, 0)); - if (!isStream(stream)) { - error('bad ObjStm stream'); - } - var first = stream.dict.get('First'); - var n = stream.dict.get('N'); - if (!isInt(first) || !isInt(n)) { - error('invalid first and n parameters for ObjStm stream'); - } - var parser = new Parser(new Lexer(stream), false, this); - parser.allowStreams = true; - var i, entries = [], num, nums = []; - // read the object numbers to populate cache - for (i = 0; i < n; ++i) { - num = parser.getObj(); - if (!isInt(num)) { - error('invalid object number in the ObjStm stream: ' + num); + if (/ArialBlack/i.test(name)) { + var SupplementalGlyphMapForArialBlack = + getSupplementalGlyphMapForArialBlack(); + for (charCode in SupplementalGlyphMapForArialBlack) { + map[+charCode] = SupplementalGlyphMapForArialBlack[charCode]; + } } - nums.push(num); - var offset = parser.getObj(); - if (!isInt(offset)) { - error('invalid object offset in the ObjStm stream: ' + offset); + var isIdentityUnicode = this.toUnicode instanceof IdentityToUnicodeMap; + if (!isIdentityUnicode) { + this.toUnicode.forEach(function(charCode, unicodeCharCode) { + map[+charCode] = unicodeCharCode; + }); } - } - // read stream objects for cache - for (i = 0; i < n; ++i) { - entries.push(parser.getObj()); - num = nums[i]; - var entry = this.entries[num]; - if (entry && entry.offset === tableOffset && entry.gen === i) { - this.cache[num] = entries[i]; + this.toFontChar = map; + this.toUnicode = new ToUnicodeMap(map); + } else if (/Symbol/i.test(fontName)) { + var symbols = SymbolSetEncoding; + glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode in symbols) { + fontChar = glyphsUnicodeMap[symbols[charCode]]; + if (!fontChar) { + continue; + } + this.toFontChar[charCode] = fontChar; } - } - xrefEntry = entries[xrefEntry.gen]; - if (xrefEntry === undefined) { - error('bad XRef entry for compressed object'); - } - return xrefEntry; - }, - - fetchIfRefAsync: function XRef_fetchIfRefAsync(obj) { - if (!isRef(obj)) { - return Promise.resolve(obj); - } - return this.fetchAsync(obj); - }, - - fetchAsync: function XRef_fetchAsync(ref, suppressEncryption) { - var streamManager = this.stream.manager; - var xref = this; - return new Promise(function tryFetch(resolve, reject) { - try { - resolve(xref.fetch(ref, suppressEncryption)); - } catch (e) { - if (e instanceof MissingDataException) { - streamManager.requestRange(e.begin, e.end).then(function () { - tryFetch(resolve, reject); - }, reject); - return; + for (charCode in properties.differences) { + fontChar = glyphsUnicodeMap[properties.differences[charCode]]; + if (!fontChar) { + continue; } - reject(e); + this.toFontChar[charCode] = fontChar; } - }); - }, - - getCatalogObj: function XRef_getCatalogObj() { - return this.root; - } - }; - - return XRef; -})(); - -/** - * A NameTree/NumberTree is like a Dict but has some advantageous properties, - * see the specification (7.9.6 and 7.9.7) for additional details. - * TODO: implement all the Dict functions and make this more efficient. - */ -var NameOrNumberTree = (function NameOrNumberTreeClosure() { - function NameOrNumberTree(root, xref) { - throw new Error('Cannot initialize NameOrNumberTree.'); - } - - NameOrNumberTree.prototype = { - getAll: function NameOrNumberTree_getAll() { - var dict = Object.create(null); - if (!this.root) { - return dict; - } - var xref = this.xref; - // Reading Name/Number tree. - var processed = new RefSet(); - processed.put(this.root); - var queue = [this.root]; - while (queue.length > 0) { - var i, n; - var obj = xref.fetchIfRef(queue.shift()); - if (!isDict(obj)) { - continue; + } else if (/Dingbats/i.test(fontName)) { + glyphsUnicodeMap = getDingbatsGlyphsUnicode(); + if (/Wingdings/i.test(name)) { + warn('Wingdings font without embedded font file, ' + + 'falling back to the ZapfDingbats encoding.'); } - if (obj.has('Kids')) { - var kids = obj.get('Kids'); - for (i = 0, n = kids.length; i < n; i++) { - var kid = kids[i]; - assert(!processed.has(kid), - 'Duplicate entry in "' + this._type + '" tree.'); - queue.push(kid); - processed.put(kid); + var dingbats = ZapfDingbatsEncoding; + for (charCode in dingbats) { + fontChar = glyphsUnicodeMap[dingbats[charCode]]; + if (!fontChar) { + continue; } - continue; + this.toFontChar[charCode] = fontChar; } - var entries = obj.get(this._type); - if (isArray(entries)) { - for (i = 0, n = entries.length; i < n; i += 2) { - dict[xref.fetchIfRef(entries[i])] = xref.fetchIfRef(entries[i + 1]); + for (charCode in properties.differences) { + fontChar = glyphsUnicodeMap[properties.differences[charCode]]; + if (!fontChar) { + continue; } + this.toFontChar[charCode] = fontChar; + } + } else if (isStandardFont) { + this.toFontChar = []; + glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode in properties.defaultEncoding) { + glyphName = (properties.differences[charCode] || + properties.defaultEncoding[charCode]); + this.toFontChar[charCode] = glyphsUnicodeMap[glyphName]; } + } else { + var unicodeCharCode, notCidFont = (type.indexOf('CIDFontType') === -1); + glyphsUnicodeMap = getGlyphsUnicode(); + this.toUnicode.forEach(function(charCode, unicodeCharCode) { + if (notCidFont) { + glyphName = (properties.differences[charCode] || + properties.defaultEncoding[charCode]); + unicodeCharCode = (glyphsUnicodeMap[glyphName] || unicodeCharCode); + } + this.toFontChar[charCode] = unicodeCharCode; + }.bind(this)); } - return dict; - }, + this.loadedName = fontName.split('-')[0]; + this.loading = false; + this.fontType = getFontType(type, subtype); + return; + } - get: function NameOrNumberTree_get(key) { - if (!this.root) { - return null; + // Some fonts might use wrong font types for Type1C or CIDFontType0C + if (subtype === 'Type1C' && (type !== 'Type1' && type !== 'MMType1')) { + // Some TrueType fonts by mistake claim Type1C + if (isTrueTypeFile(file)) { + subtype = 'TrueType'; + } else { + type = 'Type1'; + } + } + if (subtype === 'CIDFontType0C' && type !== 'CIDFontType0') { + type = 'CIDFontType0'; + } + if (subtype === 'OpenType') { + type = 'OpenType'; + } + // Some CIDFontType0C fonts by mistake claim CIDFontType0. + if (type === 'CIDFontType0') { + if (isType1File(file)) { + subtype = 'CIDFontType0'; + } else if (isOpenTypeFile(file)) { + // Sometimes the type/subtype can be a complete lie (see issue6782.pdf). + type = subtype = 'OpenType'; + } else { + subtype = 'CIDFontType0C'; } + } - var xref = this.xref; - var kidsOrEntries = xref.fetchIfRef(this.root); - var loopCount = 0; - var MAX_LEVELS = 10; - var l, r, m; + var data; + switch (type) { + case 'MMType1': + info('MMType1 font (' + name + '), falling back to Type1.'); + /* falls through */ + case 'Type1': + case 'CIDFontType0': + this.mimetype = 'font/opentype'; - // Perform a binary search to quickly find the entry that - // contains the key we are looking for. - while (kidsOrEntries.has('Kids')) { - if (++loopCount > MAX_LEVELS) { - warn('Search depth limit reached for "' + this._type + '" tree.'); - return null; - } + var cff = (subtype === 'Type1C' || subtype === 'CIDFontType0C') ? + new CFFFont(file, properties) : new Type1Font(name, file, properties); - var kids = kidsOrEntries.get('Kids'); - if (!isArray(kids)) { - return null; - } + adjustWidths(properties); - l = 0; - r = kids.length - 1; - while (l <= r) { - m = (l + r) >> 1; - var kid = xref.fetchIfRef(kids[m]); - var limits = kid.get('Limits'); + // Wrap the CFF data inside an OTF font file + data = this.convert(name, cff, properties); + break; - if (key < xref.fetchIfRef(limits[0])) { - r = m - 1; - } else if (key > xref.fetchIfRef(limits[1])) { - l = m + 1; - } else { - kidsOrEntries = xref.fetchIfRef(kids[m]); - break; - } - } - if (l > r) { - return null; - } - } + case 'OpenType': + case 'TrueType': + case 'CIDFontType2': + this.mimetype = 'font/opentype'; - // If we get here, then we have found the right entry. Now go through the - // entries in the dictionary until we find the key we're looking for. - var entries = kidsOrEntries.get(this._type); - if (isArray(entries)) { - // Perform a binary search to reduce the lookup time. - l = 0; - r = entries.length - 2; - while (l <= r) { - // Check only even indices (0, 2, 4, ...) because the - // odd indices contain the actual data. - m = (l + r) & ~1; - var currentKey = xref.fetchIfRef(entries[m]); - if (key < currentKey) { - r = m - 2; - } else if (key > currentKey) { - l = m + 2; - } else { - return xref.fetchIfRef(entries[m + 1]); - } + // Repair the TrueType file. It is can be damaged in the point of + // view of the sanitizer + data = this.checkAndRepair(name, file, properties); + if (this.isOpenType) { + adjustWidths(properties); + + type = 'OpenType'; } - } - return null; + break; + + default: + error('Font ' + type + ' is not supported'); + break; } - }; - return NameOrNumberTree; -})(); -var NameTree = (function NameTreeClosure() { - function NameTree(root, xref) { - this.root = root; - this.xref = xref; - this._type = 'Names'; + this.data = data; + this.fontType = getFontType(type, subtype); + + // Transfer some properties again that could change during font conversion + this.fontMatrix = properties.fontMatrix; + this.widths = properties.widths; + this.defaultWidth = properties.defaultWidth; + this.encoding = properties.baseEncoding; + this.seacMap = properties.seacMap; + + this.loading = true; } - Util.inherit(NameTree, NameOrNumberTree, {}); + Font.getFontID = (function () { + var ID = 1; + return function Font_getFontID() { + return String(ID++); + }; + })(); - return NameTree; -})(); + function int16(b0, b1) { + return (b0 << 8) + b1; + } -var NumberTree = (function NumberTreeClosure() { - function NumberTree(root, xref) { - this.root = root; - this.xref = xref; - this._type = 'Nums'; + function int32(b0, b1, b2, b3) { + return (b0 << 24) + (b1 << 16) + (b2 << 8) + b3; } - Util.inherit(NumberTree, NameOrNumberTree, {}); + function string16(value) { + return String.fromCharCode((value >> 8) & 0xff, value & 0xff); + } - return NumberTree; -})(); + function safeString16(value) { + // clamp value to the 16-bit int range + value = (value > 0x7FFF ? 0x7FFF : (value < -0x8000 ? -0x8000 : value)); + return String.fromCharCode((value >> 8) & 0xff, value & 0xff); + } -/** - * "A PDF file can refer to the contents of another file by using a File - * Specification (PDF 1.1)", see the spec (7.11) for more details. - * NOTE: Only embedded files are supported (as part of the attachments support) - * TODO: support the 'URL' file system (with caching if !/V), portable - * collections attributes and related files (/RF) - */ -var FileSpec = (function FileSpecClosure() { - function FileSpec(root, xref) { - if (!root || !isDict(root)) { - return; - } - this.xref = xref; - this.root = root; - if (root.has('FS')) { - this.fs = root.get('FS'); - } - this.description = root.has('Desc') ? - stringToPDFString(root.get('Desc')) : - ''; - if (root.has('RF')) { - warn('Related file specifications are not supported'); - } - this.contentAvailable = true; - if (!root.has('EF')) { - this.contentAvailable = false; - warn('Non-embedded file specifications are not supported'); - } + function isTrueTypeFile(file) { + var header = file.peekBytes(4); + return readUint32(header, 0) === 0x00010000; } - function pickPlatformItem(dict) { - // Look for the filename in this order: - // UF, F, Unix, Mac, DOS - if (dict.has('UF')) { - return dict.get('UF'); - } else if (dict.has('F')) { - return dict.get('F'); - } else if (dict.has('Unix')) { - return dict.get('Unix'); - } else if (dict.has('Mac')) { - return dict.get('Mac'); - } else if (dict.has('DOS')) { - return dict.get('DOS'); - } else { - return null; + function isOpenTypeFile(file) { + var header = file.peekBytes(4); + return bytesToString(header) === 'OTTO'; + } + + function isType1File(file) { + var header = file.peekBytes(2); + // All Type1 font programs must begin with the comment '%!' (0x25 + 0x21). + if (header[0] === 0x25 && header[1] === 0x21) { + return true; + } + // ... obviously some fonts violate that part of the specification, + // please refer to the comment in |Type1Font| below. + if (header[0] === 0x80 && header[1] === 0x01) { // pfb file header. + return true; } + return false; } - FileSpec.prototype = { - get filename() { - if (!this._filename && this.root) { - var filename = pickPlatformItem(this.root) || 'unnamed'; - this._filename = stringToPDFString(filename). - replace(/\\\\/g, '\\'). - replace(/\\\//g, '/'). - replace(/\\/g, '/'); - } - return this._filename; - }, - get content() { - if (!this.contentAvailable) { - return null; - } - if (!this.contentRef && this.root) { - this.contentRef = pickPlatformItem(this.root.get('EF')); - } - var content = null; - if (this.contentRef) { - var xref = this.xref; - var fileObj = xref.fetchIfRef(this.contentRef); - if (fileObj && isStream(fileObj)) { - content = fileObj.getBytes(); - } else { - warn('Embedded file specification points to non-existing/invalid ' + - 'content'); - } + /** + * Helper function for |adjustMapping|. + * @return {boolean} + */ + function isProblematicUnicodeLocation(code) { + // Using binary search to find a range start. + var i = 0, j = ProblematicCharRanges.length - 1; + while (i < j) { + var c = (i + j + 1) >> 1; + if (code < ProblematicCharRanges[c]) { + j = c - 1; } else { - warn('Embedded file specification does not have a content'); + i = c; } - return content; - }, - get serializable() { - return { - filename: this.filename, - content: this.content - }; } - }; - return FileSpec; -})(); + // Even index means code in problematic range. + return !(i & 1); + } -/** - * A helper for loading missing data in object graphs. It traverses the graph - * depth first and queues up any objects that have missing data. Once it has - * has traversed as many objects that are available it attempts to bundle the - * missing data requests and then resume from the nodes that weren't ready. - * - * NOTE: It provides protection from circular references by keeping track of - * of loaded references. However, you must be careful not to load any graphs - * that have references to the catalog or other pages since that will cause the - * entire PDF document object graph to be traversed. - */ -var ObjectLoader = (function() { - function mayHaveChildren(value) { - return isRef(value) || isDict(value) || isArray(value) || isStream(value); - } - - function addChildren(node, nodesToVisit) { - var value; - if (isDict(node) || isStream(node)) { - var map; - if (isDict(node)) { - map = node.map; - } else { - map = node.dict.map; - } - for (var key in map) { - value = map[key]; - if (mayHaveChildren(value)) { - nodesToVisit.push(value); - } - } - } else if (isArray(node)) { - for (var i = 0, ii = node.length; i < ii; i++) { - value = node[i]; - if (mayHaveChildren(value)) { - nodesToVisit.push(value); + /** + * Rebuilds the char code to glyph ID map by trying to replace the char codes + * with their unicode value. It also moves char codes that are in known + * problematic locations. + * @return {Object} Two properties: + * 'toFontChar' - maps original char codes(the value that will be read + * from commands such as show text) to the char codes that will be used in the + * font that we build + * 'charCodeToGlyphId' - maps the new font char codes to glyph ids + */ + function adjustMapping(charCodeToGlyphId, properties) { + var toUnicode = properties.toUnicode; + var isSymbolic = !!(properties.flags & FontFlags.Symbolic); + var isIdentityUnicode = + properties.toUnicode instanceof IdentityToUnicodeMap; + var newMap = Object.create(null); + var toFontChar = []; + var usedFontCharCodes = []; + var nextAvailableFontCharCode = PRIVATE_USE_OFFSET_START; + for (var originalCharCode in charCodeToGlyphId) { + originalCharCode |= 0; + var glyphId = charCodeToGlyphId[originalCharCode]; + var fontCharCode = originalCharCode; + // First try to map the value to a unicode position if a non identity map + // was created. + if (!isIdentityUnicode && toUnicode.has(originalCharCode)) { + var unicode = toUnicode.get(fontCharCode); + // TODO: Try to map ligatures to the correct spot. + if (unicode.length === 1) { + fontCharCode = unicode.charCodeAt(0); } } - } - } - - function ObjectLoader(obj, keys, xref) { - this.obj = obj; - this.keys = keys; - this.xref = xref; - this.refSet = null; - this.capability = null; - } + // Try to move control characters, special characters and already mapped + // characters to the private use area since they will not be drawn by + // canvas if left in their current position. Also, move characters if the + // font was symbolic and there is only an identity unicode map since the + // characters probably aren't in the correct position (fixes an issue + // with firefox and thuluthfont). + if ((usedFontCharCodes[fontCharCode] !== undefined || + isProblematicUnicodeLocation(fontCharCode) || + (isSymbolic && isIdentityUnicode)) && + nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END) { // Room left. + // Loop to try and find a free spot in the private use area. + do { + fontCharCode = nextAvailableFontCharCode++; - ObjectLoader.prototype = { - load: function ObjectLoader_load() { - var keys = this.keys; - this.capability = createPromiseCapability(); - // Don't walk the graph if all the data is already loaded. - if (!(this.xref.stream instanceof ChunkedStream) || - this.xref.stream.getMissingChunks().length === 0) { - this.capability.resolve(); - return this.capability.promise; - } + if (SKIP_PRIVATE_USE_RANGE_F000_TO_F01F && fontCharCode === 0xF000) { + fontCharCode = 0xF020; + nextAvailableFontCharCode = fontCharCode + 1; + } - this.refSet = new RefSet(); - // Setup the initial nodes to visit. - var nodesToVisit = []; - for (var i = 0; i < keys.length; i++) { - nodesToVisit.push(this.obj[keys[i]]); + } while (usedFontCharCodes[fontCharCode] !== undefined && + nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END); } - this._walk(nodesToVisit); - return this.capability.promise; - }, - - _walk: function ObjectLoader_walk(nodesToVisit) { - var nodesToRevisit = []; - var pendingRequests = []; - // DFS walk of the object graph. - while (nodesToVisit.length) { - var currentNode = nodesToVisit.pop(); - - // Only references or chunked streams can cause missing data exceptions. - if (isRef(currentNode)) { - // Skip nodes that have already been visited. - if (this.refSet.has(currentNode)) { - continue; - } - try { - var ref = currentNode; - this.refSet.put(ref); - currentNode = this.xref.fetch(currentNode); - } catch (e) { - if (!(e instanceof MissingDataException)) { - throw e; - } - nodesToRevisit.push(currentNode); - pendingRequests.push({ begin: e.begin, end: e.end }); - } - } - if (currentNode && currentNode.getBaseStreams) { - var baseStreams = currentNode.getBaseStreams(); - var foundMissingData = false; - for (var i = 0; i < baseStreams.length; i++) { - var stream = baseStreams[i]; - if (stream.getMissingChunks && stream.getMissingChunks().length) { - foundMissingData = true; - pendingRequests.push({ - begin: stream.start, - end: stream.end - }); - } - } - if (foundMissingData) { - nodesToRevisit.push(currentNode); - } - } + newMap[fontCharCode] = glyphId; + toFontChar[originalCharCode] = fontCharCode; + usedFontCharCodes[fontCharCode] = true; + } + return { + toFontChar: toFontChar, + charCodeToGlyphId: newMap, + nextAvailableFontCharCode: nextAvailableFontCharCode + }; + } - addChildren(currentNode, nodesToVisit); + function getRanges(glyphs, numGlyphs) { + // Array.sort() sorts by characters, not numerically, so convert to an + // array of characters. + var codes = []; + for (var charCode in glyphs) { + // Remove an invalid glyph ID mappings to make OTS happy. + if (glyphs[charCode] >= numGlyphs) { + continue; } + codes.push({ fontCharCode: charCode | 0, glyphId: glyphs[charCode] }); + } + codes.sort(function fontGetRangesSort(a, b) { + return a.fontCharCode - b.fontCharCode; + }); - if (pendingRequests.length) { - this.xref.stream.manager.requestRanges(pendingRequests).then( - function pendingRequestCallback() { - nodesToVisit = nodesToRevisit; - for (var i = 0; i < nodesToRevisit.length; i++) { - var node = nodesToRevisit[i]; - // Remove any reference nodes from the currrent refset so they - // aren't skipped when we revist them. - if (isRef(node)) { - this.refSet.remove(node); - } - } - this._walk(nodesToVisit); - }.bind(this), this.capability.reject); - return; + // Split the sorted codes into ranges. + var ranges = []; + var length = codes.length; + for (var n = 0; n < length; ) { + var start = codes[n].fontCharCode; + var codeIndices = [codes[n].glyphId]; + ++n; + var end = start; + while (n < length && end + 1 === codes[n].fontCharCode) { + codeIndices.push(codes[n].glyphId); + ++end; + ++n; + if (end === 0xFFFF) { + break; + } } - // Everything is loaded. - this.refSet = null; - this.capability.resolve(); + ranges.push([start, end, codeIndices]); } - }; - return ObjectLoader; -})(); + return ranges; + } -exports.Catalog = Catalog; -exports.ObjectLoader = ObjectLoader; -exports.XRef = XRef; -})); + function createCmapTable(glyphs, numGlyphs) { + var ranges = getRanges(glyphs, numGlyphs); + var numTables = ranges[ranges.length - 1][1] > 0xFFFF ? 2 : 1; + var cmap = '\x00\x00' + // version + string16(numTables) + // numTables + '\x00\x03' + // platformID + '\x00\x01' + // encodingID + string32(4 + numTables * 8); // start of the table record + var i, ii, j, jj; + for (i = ranges.length - 1; i >= 0; --i) { + if (ranges[i][0] <= 0xFFFF) { break; } + } + var bmpLength = i + 1; -(function (root, factory) { - { - factory((root.pdfjsCorePsParser = {}), root.pdfjsSharedUtil, - root.pdfjsCoreParser); - } -}(this, function (exports, sharedUtil, coreParser) { + if (ranges[i][0] < 0xFFFF && ranges[i][1] === 0xFFFF) { + ranges[i][1] = 0xFFFE; + } + var trailingRangesCount = ranges[i][1] < 0xFFFF ? 1 : 0; + var segCount = bmpLength + trailingRangesCount; + var searchParams = OpenTypeFileBuilder.getSearchParams(segCount, 2); -var error = sharedUtil.error; -var EOF = coreParser.EOF; -var Lexer = coreParser.Lexer; + // Fill up the 4 parallel arrays describing the segments. + var startCount = ''; + var endCount = ''; + var idDeltas = ''; + var idRangeOffsets = ''; + var glyphsIds = ''; + var bias = 0; -var PostScriptParser = (function PostScriptParserClosure() { - function PostScriptParser(lexer) { - this.lexer = lexer; - this.operators = []; - this.token = null; - this.prev = null; - } - PostScriptParser.prototype = { - nextToken: function PostScriptParser_nextToken() { - this.prev = this.token; - this.token = this.lexer.getToken(); - }, - accept: function PostScriptParser_accept(type) { - if (this.token.type === type) { - this.nextToken(); - return true; - } - return false; - }, - expect: function PostScriptParser_expect(type) { - if (this.accept(type)) { - return true; - } - error('Unexpected symbol: found ' + this.token.type + ' expected ' + - type + '.'); - }, - parse: function PostScriptParser_parse() { - this.nextToken(); - this.expect(PostScriptTokenTypes.LBRACE); - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - return this.operators; - }, - parseBlock: function PostScriptParser_parseBlock() { - while (true) { - if (this.accept(PostScriptTokenTypes.NUMBER)) { - this.operators.push(this.prev.value); - } else if (this.accept(PostScriptTokenTypes.OPERATOR)) { - this.operators.push(this.prev.value); - } else if (this.accept(PostScriptTokenTypes.LBRACE)) { - this.parseCondition(); - } else { - return; + var range, start, end, codes; + for (i = 0, ii = bmpLength; i < ii; i++) { + range = ranges[i]; + start = range[0]; + end = range[1]; + startCount += string16(start); + endCount += string16(end); + codes = range[2]; + var contiguous = true; + for (j = 1, jj = codes.length; j < jj; ++j) { + if (codes[j] !== codes[j - 1] + 1) { + contiguous = false; + break; } } - }, - parseCondition: function PostScriptParser_parseCondition() { - // Add two place holders that will be updated later - var conditionLocation = this.operators.length; - this.operators.push(null, null); + if (!contiguous) { + var offset = (segCount - i) * 2 + bias * 2; + bias += (end - start + 1); - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - if (this.accept(PostScriptTokenTypes.IF)) { - // The true block is right after the 'if' so it just falls through on - // true else it jumps and skips the true block. - this.operators[conditionLocation] = this.operators.length; - this.operators[conditionLocation + 1] = 'jz'; - } else if (this.accept(PostScriptTokenTypes.LBRACE)) { - var jumpLocation = this.operators.length; - this.operators.push(null, null); - var endOfTrue = this.operators.length; - this.parseBlock(); - this.expect(PostScriptTokenTypes.RBRACE); - this.expect(PostScriptTokenTypes.IFELSE); - // The jump is added at the end of the true block to skip the false - // block. - this.operators[jumpLocation] = this.operators.length; - this.operators[jumpLocation + 1] = 'j'; + idDeltas += string16(0); + idRangeOffsets += string16(offset); - this.operators[conditionLocation] = endOfTrue; - this.operators[conditionLocation + 1] = 'jz'; + for (j = 0, jj = codes.length; j < jj; ++j) { + glyphsIds += string16(codes[j]); + } } else { - error('PS Function: error parsing conditional.'); + var startCode = codes[0]; + + idDeltas += string16((startCode - start) & 0xFFFF); + idRangeOffsets += string16(0); } } - }; - return PostScriptParser; -})(); - -var PostScriptTokenTypes = { - LBRACE: 0, - RBRACE: 1, - NUMBER: 2, - OPERATOR: 3, - IF: 4, - IFELSE: 5 -}; -var PostScriptToken = (function PostScriptTokenClosure() { - function PostScriptToken(type, value) { - this.type = type; - this.value = value; - } + if (trailingRangesCount > 0) { + endCount += '\xFF\xFF'; + startCount += '\xFF\xFF'; + idDeltas += '\x00\x01'; + idRangeOffsets += '\x00\x00'; + } - var opCache = Object.create(null); + var format314 = '\x00\x00' + // language + string16(2 * segCount) + + string16(searchParams.range) + + string16(searchParams.entry) + + string16(searchParams.rangeShift) + + endCount + '\x00\x00' + startCount + + idDeltas + idRangeOffsets + glyphsIds; - PostScriptToken.getOperator = function PostScriptToken_getOperator(op) { - var opValue = opCache[op]; - if (opValue) { - return opValue; + var format31012 = ''; + var header31012 = ''; + if (numTables > 1) { + cmap += '\x00\x03' + // platformID + '\x00\x0A' + // encodingID + string32(4 + numTables * 8 + + 4 + format314.length); // start of the table record + format31012 = ''; + for (i = 0, ii = ranges.length; i < ii; i++) { + range = ranges[i]; + start = range[0]; + codes = range[2]; + var code = codes[0]; + for (j = 1, jj = codes.length; j < jj; ++j) { + if (codes[j] !== codes[j - 1] + 1) { + end = range[0] + j - 1; + format31012 += string32(start) + // startCharCode + string32(end) + // endCharCode + string32(code); // startGlyphID + start = end + 1; + code = codes[j]; + } + } + format31012 += string32(start) + // startCharCode + string32(range[1]) + // endCharCode + string32(code); // startGlyphID + } + header31012 = '\x00\x0C' + // format + '\x00\x00' + // reserved + string32(format31012.length + 16) + // length + '\x00\x00\x00\x00' + // language + string32(format31012.length / 12); // nGroups } - return opCache[op] = new PostScriptToken(PostScriptTokenTypes.OPERATOR, op); - }; - PostScriptToken.LBRACE = new PostScriptToken(PostScriptTokenTypes.LBRACE, - '{'); - PostScriptToken.RBRACE = new PostScriptToken(PostScriptTokenTypes.RBRACE, - '}'); - PostScriptToken.IF = new PostScriptToken(PostScriptTokenTypes.IF, 'IF'); - PostScriptToken.IFELSE = new PostScriptToken(PostScriptTokenTypes.IFELSE, - 'IFELSE'); - return PostScriptToken; -})(); + return cmap + '\x00\x04' + // format + string16(format314.length + 4) + // length + format314 + header31012 + format31012; + } -var PostScriptLexer = (function PostScriptLexerClosure() { - function PostScriptLexer(stream) { - this.stream = stream; - this.nextChar(); + function validateOS2Table(os2) { + var stream = new Stream(os2.data); + var version = stream.getUint16(); + // TODO verify all OS/2 tables fields, but currently we validate only those + // that give us issues + stream.getBytes(60); // skipping type, misc sizes, panose, unicode ranges + var selection = stream.getUint16(); + if (version < 4 && (selection & 0x0300)) { + return false; + } + var firstChar = stream.getUint16(); + var lastChar = stream.getUint16(); + if (firstChar > lastChar) { + return false; + } + stream.getBytes(6); // skipping sTypoAscender/Descender/LineGap + var usWinAscent = stream.getUint16(); + if (usWinAscent === 0) { // makes font unreadable by windows + return false; + } - this.strBuf = []; + // OS/2 appears to be valid, resetting some fields + os2.data[8] = os2.data[9] = 0; // IE rejects fonts if fsType != 0 + return true; } - PostScriptLexer.prototype = { - nextChar: function PostScriptLexer_nextChar() { - return (this.currentChar = this.stream.getByte()); - }, - getToken: function PostScriptLexer_getToken() { - var comment = false; - var ch = this.currentChar; - // skip comments - while (true) { - if (ch < 0) { - return EOF; + function createOS2Table(properties, charstrings, override) { + override = override || { + unitsPerEm: 0, + yMax: 0, + yMin: 0, + ascent: 0, + descent: 0 + }; + + var ulUnicodeRange1 = 0; + var ulUnicodeRange2 = 0; + var ulUnicodeRange3 = 0; + var ulUnicodeRange4 = 0; + + var firstCharIndex = null; + var lastCharIndex = 0; + + if (charstrings) { + for (var code in charstrings) { + code |= 0; + if (firstCharIndex > code || !firstCharIndex) { + firstCharIndex = code; + } + if (lastCharIndex < code) { + lastCharIndex = code; } - if (comment) { - if (ch === 0x0A || ch === 0x0D) { - comment = false; - } - } else if (ch === 0x25) { // '%' - comment = true; - } else if (!Lexer.isSpace(ch)) { - break; + var position = getUnicodeRangeFor(code); + if (position < 32) { + ulUnicodeRange1 |= 1 << position; + } else if (position < 64) { + ulUnicodeRange2 |= 1 << position - 32; + } else if (position < 96) { + ulUnicodeRange3 |= 1 << position - 64; + } else if (position < 123) { + ulUnicodeRange4 |= 1 << position - 96; + } else { + error('Unicode ranges Bits > 123 are reserved for internal usage'); } - ch = this.nextChar(); } - switch (ch | 0) { - case 0x30: case 0x31: case 0x32: case 0x33: case 0x34: // '0'-'4' - case 0x35: case 0x36: case 0x37: case 0x38: case 0x39: // '5'-'9' - case 0x2B: case 0x2D: case 0x2E: // '+', '-', '.' - return new PostScriptToken(PostScriptTokenTypes.NUMBER, - this.getNumber()); - case 0x7B: // '{' - this.nextChar(); - return PostScriptToken.LBRACE; - case 0x7D: // '}' - this.nextChar(); - return PostScriptToken.RBRACE; - } - // operator - var strBuf = this.strBuf; - strBuf.length = 0; - strBuf[0] = String.fromCharCode(ch); - - while ((ch = this.nextChar()) >= 0 && // and 'A'-'Z', 'a'-'z' - ((ch >= 0x41 && ch <= 0x5A) || (ch >= 0x61 && ch <= 0x7A))) { - strBuf.push(String.fromCharCode(ch)); - } - var str = strBuf.join(''); - switch (str.toLowerCase()) { - case 'if': - return PostScriptToken.IF; - case 'ifelse': - return PostScriptToken.IFELSE; - default: - return PostScriptToken.getOperator(str); - } - }, - getNumber: function PostScriptLexer_getNumber() { - var ch = this.currentChar; - var strBuf = this.strBuf; - strBuf.length = 0; - strBuf[0] = String.fromCharCode(ch); - - while ((ch = this.nextChar()) >= 0) { - if ((ch >= 0x30 && ch <= 0x39) || // '0'-'9' - ch === 0x2D || ch === 0x2E) { // '-', '.' - strBuf.push(String.fromCharCode(ch)); - } else { - break; - } - } - var value = parseFloat(strBuf.join('')); - if (isNaN(value)) { - error('Invalid floating point number: ' + value); - } - return value; + } else { + // TODO + firstCharIndex = 0; + lastCharIndex = 255; } - }; - return PostScriptLexer; -})(); - -exports.PostScriptLexer = PostScriptLexer; -exports.PostScriptParser = PostScriptParser; -})); - - -(function (root, factory) { - { - factory((root.pdfjsCoreFonts = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreStream, root.pdfjsCoreParser, - root.pdfjsCoreCMap, root.pdfjsCoreGlyphList, root.pdfjsCoreCharsets, - root.pdfjsCoreFontRenderer, root.pdfjsCoreEncodings, - root.pdfjsCoreStandardFonts, root.pdfjsCoreUnicode); - } -}(this, function (exports, sharedUtil, corePrimitives, coreStream, coreParser, - coreCMap, coreGlyphList, coreCharsets, coreFontRenderer, - coreEncodings, coreStandardFonts, coreUnicode) { - -var FONT_IDENTITY_MATRIX = sharedUtil.FONT_IDENTITY_MATRIX; -var FontType = sharedUtil.FontType; -var Util = sharedUtil.Util; -var assert = sharedUtil.assert; -var bytesToString = sharedUtil.bytesToString; -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isInt = sharedUtil.isInt; -var isNum = sharedUtil.isNum; -var readUint32 = sharedUtil.readUint32; -var shadow = sharedUtil.shadow; -var stringToBytes = sharedUtil.stringToBytes; -var string32 = sharedUtil.string32; -var warn = sharedUtil.warn; -var Name = corePrimitives.Name; -var Stream = coreStream.Stream; -var Lexer = coreParser.Lexer; -var CMapFactory = coreCMap.CMapFactory; -var IdentityCMap = coreCMap.IdentityCMap; -var getGlyphsUnicode = coreGlyphList.getGlyphsUnicode; -var getDingbatsGlyphsUnicode = coreGlyphList.getDingbatsGlyphsUnicode; -var ISOAdobeCharset = coreCharsets.ISOAdobeCharset; -var ExpertCharset = coreCharsets.ExpertCharset; -var ExpertSubsetCharset = coreCharsets.ExpertSubsetCharset; -var FontRendererFactory = coreFontRenderer.FontRendererFactory; -var WinAnsiEncoding = coreEncodings.WinAnsiEncoding; -var StandardEncoding = coreEncodings.StandardEncoding; -var MacRomanEncoding = coreEncodings.MacRomanEncoding; -var SymbolSetEncoding = coreEncodings.SymbolSetEncoding; -var ZapfDingbatsEncoding = coreEncodings.ZapfDingbatsEncoding; -var ExpertEncoding = coreEncodings.ExpertEncoding; -var getEncoding = coreEncodings.getEncoding; -var getStdFontMap = coreStandardFonts.getStdFontMap; -var getNonStdFontMap = coreStandardFonts.getNonStdFontMap; -var getGlyphMapForStandardFonts = coreStandardFonts.getGlyphMapForStandardFonts; -var getSupplementalGlyphMapForArialBlack = - coreStandardFonts.getSupplementalGlyphMapForArialBlack; -var getUnicodeRangeFor = coreUnicode.getUnicodeRangeFor; -var mapSpecialUnicodeValues = coreUnicode.mapSpecialUnicodeValues; - -// Unicode Private Use Area -var PRIVATE_USE_OFFSET_START = 0xE000; -var PRIVATE_USE_OFFSET_END = 0xF8FF; -var SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = false; - -// PDF Glyph Space Units are one Thousandth of a TextSpace Unit -// except for Type 3 fonts -var PDF_GLYPH_SPACE_UNITS = 1000; - -// Hinting is currently disabled due to unknown problems on windows -// in tracemonkey and various other pdfs with type1 fonts. -var HINTING_ENABLED = false; - -// Accented charactars are not displayed properly on windows, using this flag -// to control analysis of seac charstrings. -var SEAC_ANALYSIS_ENABLED = false; - -// Maximum subroutine call depth of type 2 chartrings. Matches OTS. -var MAX_SUBR_NESTING = 10; -var FontFlags = { - FixedPitch: 1, - Serif: 2, - Symbolic: 4, - Script: 8, - Nonsymbolic: 32, - Italic: 64, - AllCap: 65536, - SmallCap: 131072, - ForceBold: 262144 -}; + var bbox = properties.bbox || [0, 0, 0, 0]; + var unitsPerEm = (override.unitsPerEm || + 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]); -var MacStandardGlyphOrdering = [ - '.notdef', '.null', 'nonmarkingreturn', 'space', 'exclam', 'quotedbl', - 'numbersign', 'dollar', 'percent', 'ampersand', 'quotesingle', 'parenleft', - 'parenright', 'asterisk', 'plus', 'comma', 'hyphen', 'period', 'slash', - 'zero', 'one', 'two', 'three', 'four', 'five', 'six', 'seven', 'eight', - 'nine', 'colon', 'semicolon', 'less', 'equal', 'greater', 'question', 'at', - 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', - 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'bracketleft', - 'backslash', 'bracketright', 'asciicircum', 'underscore', 'grave', 'a', 'b', - 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', - 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 'braceleft', 'bar', 'braceright', - 'asciitilde', 'Adieresis', 'Aring', 'Ccedilla', 'Eacute', 'Ntilde', - 'Odieresis', 'Udieresis', 'aacute', 'agrave', 'acircumflex', 'adieresis', - 'atilde', 'aring', 'ccedilla', 'eacute', 'egrave', 'ecircumflex', 'edieresis', - 'iacute', 'igrave', 'icircumflex', 'idieresis', 'ntilde', 'oacute', 'ograve', - 'ocircumflex', 'odieresis', 'otilde', 'uacute', 'ugrave', 'ucircumflex', - 'udieresis', 'dagger', 'degree', 'cent', 'sterling', 'section', 'bullet', - 'paragraph', 'germandbls', 'registered', 'copyright', 'trademark', 'acute', - 'dieresis', 'notequal', 'AE', 'Oslash', 'infinity', 'plusminus', 'lessequal', - 'greaterequal', 'yen', 'mu', 'partialdiff', 'summation', 'product', 'pi', - 'integral', 'ordfeminine', 'ordmasculine', 'Omega', 'ae', 'oslash', - 'questiondown', 'exclamdown', 'logicalnot', 'radical', 'florin', - 'approxequal', 'Delta', 'guillemotleft', 'guillemotright', 'ellipsis', - 'nonbreakingspace', 'Agrave', 'Atilde', 'Otilde', 'OE', 'oe', 'endash', - 'emdash', 'quotedblleft', 'quotedblright', 'quoteleft', 'quoteright', - 'divide', 'lozenge', 'ydieresis', 'Ydieresis', 'fraction', 'currency', - 'guilsinglleft', 'guilsinglright', 'fi', 'fl', 'daggerdbl', 'periodcentered', - 'quotesinglbase', 'quotedblbase', 'perthousand', 'Acircumflex', - 'Ecircumflex', 'Aacute', 'Edieresis', 'Egrave', 'Iacute', 'Icircumflex', - 'Idieresis', 'Igrave', 'Oacute', 'Ocircumflex', 'apple', 'Ograve', 'Uacute', - 'Ucircumflex', 'Ugrave', 'dotlessi', 'circumflex', 'tilde', 'macron', - 'breve', 'dotaccent', 'ring', 'cedilla', 'hungarumlaut', 'ogonek', 'caron', - 'Lslash', 'lslash', 'Scaron', 'scaron', 'Zcaron', 'zcaron', 'brokenbar', - 'Eth', 'eth', 'Yacute', 'yacute', 'Thorn', 'thorn', 'minus', 'multiply', - 'onesuperior', 'twosuperior', 'threesuperior', 'onehalf', 'onequarter', - 'threequarters', 'franc', 'Gbreve', 'gbreve', 'Idotaccent', 'Scedilla', - 'scedilla', 'Cacute', 'cacute', 'Ccaron', 'ccaron', 'dcroat']; + // if the font units differ to the PDF glyph space units + // then scale up the values + var scale = (properties.ascentScaled ? 1.0 : + unitsPerEm / PDF_GLYPH_SPACE_UNITS); -function adjustWidths(properties) { - if (!properties.fontMatrix) { - return; - } - if (properties.fontMatrix[0] === FONT_IDENTITY_MATRIX[0]) { - return; - } - // adjusting width to fontMatrix scale - var scale = 0.001 / properties.fontMatrix[0]; - var glyphsWidths = properties.widths; - for (var glyph in glyphsWidths) { - glyphsWidths[glyph] *= scale; - } - properties.defaultWidth *= scale; -} + var typoAscent = (override.ascent || + Math.round(scale * (properties.ascent || bbox[3]))); + var typoDescent = (override.descent || + Math.round(scale * (properties.descent || bbox[1]))); + if (typoDescent > 0 && properties.descent > 0 && bbox[1] < 0) { + typoDescent = -typoDescent; // fixing incorrect descent + } + var winAscent = override.yMax || typoAscent; + var winDescent = -override.yMin || -typoDescent; -function getFontType(type, subtype) { - switch (type) { - case 'Type1': - return subtype === 'Type1C' ? FontType.TYPE1C : FontType.TYPE1; - case 'CIDFontType0': - return subtype === 'CIDFontType0C' ? FontType.CIDFONTTYPE0C : - FontType.CIDFONTTYPE0; - case 'OpenType': - return FontType.OPENTYPE; - case 'TrueType': - return FontType.TRUETYPE; - case 'CIDFontType2': - return FontType.CIDFONTTYPE2; - case 'MMType1': - return FontType.MMTYPE1; - case 'Type0': - return FontType.TYPE0; - default: - return FontType.UNKNOWN; + return '\x00\x03' + // version + '\x02\x24' + // xAvgCharWidth + '\x01\xF4' + // usWeightClass + '\x00\x05' + // usWidthClass + '\x00\x00' + // fstype (0 to let the font loads via font-face on IE) + '\x02\x8A' + // ySubscriptXSize + '\x02\xBB' + // ySubscriptYSize + '\x00\x00' + // ySubscriptXOffset + '\x00\x8C' + // ySubscriptYOffset + '\x02\x8A' + // ySuperScriptXSize + '\x02\xBB' + // ySuperScriptYSize + '\x00\x00' + // ySuperScriptXOffset + '\x01\xDF' + // ySuperScriptYOffset + '\x00\x31' + // yStrikeOutSize + '\x01\x02' + // yStrikeOutPosition + '\x00\x00' + // sFamilyClass + '\x00\x00\x06' + + String.fromCharCode(properties.fixedPitch ? 0x09 : 0x00) + + '\x00\x00\x00\x00\x00\x00' + // Panose + string32(ulUnicodeRange1) + // ulUnicodeRange1 (Bits 0-31) + string32(ulUnicodeRange2) + // ulUnicodeRange2 (Bits 32-63) + string32(ulUnicodeRange3) + // ulUnicodeRange3 (Bits 64-95) + string32(ulUnicodeRange4) + // ulUnicodeRange4 (Bits 96-127) + '\x2A\x32\x31\x2A' + // achVendID + string16(properties.italicAngle ? 1 : 0) + // fsSelection + string16(firstCharIndex || + properties.firstChar) + // usFirstCharIndex + string16(lastCharIndex || properties.lastChar) + // usLastCharIndex + string16(typoAscent) + // sTypoAscender + string16(typoDescent) + // sTypoDescender + '\x00\x64' + // sTypoLineGap (7%-10% of the unitsPerEM value) + string16(winAscent) + // usWinAscent + string16(winDescent) + // usWinDescent + '\x00\x00\x00\x00' + // ulCodePageRange1 (Bits 0-31) + '\x00\x00\x00\x00' + // ulCodePageRange2 (Bits 32-63) + string16(properties.xHeight) + // sxHeight + string16(properties.capHeight) + // sCapHeight + string16(0) + // usDefaultChar + string16(firstCharIndex || properties.firstChar) + // usBreakChar + '\x00\x03'; // usMaxContext } -} -var Glyph = (function GlyphClosure() { - function Glyph(fontChar, unicode, accent, width, vmetric, operatorListId, - isSpace) { - this.fontChar = fontChar; - this.unicode = unicode; - this.accent = accent; - this.width = width; - this.vmetric = vmetric; - this.operatorListId = operatorListId; - this.isSpace = isSpace; + function createPostTable(properties) { + var angle = Math.floor(properties.italicAngle * (Math.pow(2, 16))); + return ('\x00\x03\x00\x00' + // Version number + string32(angle) + // italicAngle + '\x00\x00' + // underlinePosition + '\x00\x00' + // underlineThickness + string32(properties.fixedPitch) + // isFixedPitch + '\x00\x00\x00\x00' + // minMemType42 + '\x00\x00\x00\x00' + // maxMemType42 + '\x00\x00\x00\x00' + // minMemType1 + '\x00\x00\x00\x00'); // maxMemType1 } - Glyph.prototype.matchesForCache = function(fontChar, unicode, accent, width, - vmetric, operatorListId, isSpace) { - return this.fontChar === fontChar && - this.unicode === unicode && - this.accent === accent && - this.width === width && - this.vmetric === vmetric && - this.operatorListId === operatorListId && - this.isSpace === isSpace; - }; - - return Glyph; -})(); + function createNameTable(name, proto) { + if (!proto) { + proto = [[], []]; // no strings and unicode strings + } -var ToUnicodeMap = (function ToUnicodeMapClosure() { - function ToUnicodeMap(cmap) { - // The elements of this._map can be integers or strings, depending on how - // |cmap| was created. - this._map = cmap; - } + var strings = [ + proto[0][0] || 'Original licence', // 0.Copyright + proto[0][1] || name, // 1.Font family + proto[0][2] || 'Unknown', // 2.Font subfamily (font weight) + proto[0][3] || 'uniqueID', // 3.Unique ID + proto[0][4] || name, // 4.Full font name + proto[0][5] || 'Version 0.11', // 5.Version + proto[0][6] || '', // 6.Postscript name + proto[0][7] || 'Unknown', // 7.Trademark + proto[0][8] || 'Unknown', // 8.Manufacturer + proto[0][9] || 'Unknown' // 9.Designer + ]; - ToUnicodeMap.prototype = { - get length() { - return this._map.length; - }, + // Mac want 1-byte per character strings while Windows want + // 2-bytes per character, so duplicate the names table + var stringsUnicode = []; + var i, ii, j, jj, str; + for (i = 0, ii = strings.length; i < ii; i++) { + str = proto[1][i] || strings[i]; - forEach: function(callback) { - for (var charCode in this._map) { - callback(charCode, this._map[charCode].charCodeAt(0)); + var strBufUnicode = []; + for (j = 0, jj = str.length; j < jj; j++) { + strBufUnicode.push(string16(str.charCodeAt(j))); } - }, + stringsUnicode.push(strBufUnicode.join('')); + } - has: function(i) { - return this._map[i] !== undefined; - }, + var names = [strings, stringsUnicode]; + var platforms = ['\x00\x01', '\x00\x03']; + var encodings = ['\x00\x00', '\x00\x01']; + var languages = ['\x00\x00', '\x04\x09']; - get: function(i) { - return this._map[i]; - }, + var namesRecordCount = strings.length * platforms.length; + var nameTable = + '\x00\x00' + // format + string16(namesRecordCount) + // Number of names Record + string16(namesRecordCount * 12 + 6); // Storage - charCodeOf: function(v) { - return this._map.indexOf(v); + // Build the name records field + var strOffset = 0; + for (i = 0, ii = platforms.length; i < ii; i++) { + var strs = names[i]; + for (j = 0, jj = strs.length; j < jj; j++) { + str = strs[j]; + var nameRecord = + platforms[i] + // platform ID + encodings[i] + // encoding ID + languages[i] + // language ID + string16(j) + // name ID + string16(str.length) + + string16(strOffset); + nameTable += nameRecord; + strOffset += str.length; + } } - }; - - return ToUnicodeMap; -})(); -var IdentityToUnicodeMap = (function IdentityToUnicodeMapClosure() { - function IdentityToUnicodeMap(firstChar, lastChar) { - this.firstChar = firstChar; - this.lastChar = lastChar; + nameTable += strings.join('') + stringsUnicode.join(''); + return nameTable; } - IdentityToUnicodeMap.prototype = { - get length() { - return (this.lastChar + 1) - this.firstChar; - }, - - forEach: function (callback) { - for (var i = this.firstChar, ii = this.lastChar; i <= ii; i++) { - callback(i, i); - } - }, - - has: function (i) { - return this.firstChar <= i && i <= this.lastChar; + Font.prototype = { + name: null, + font: null, + mimetype: null, + encoding: null, + get renderer() { + var renderer = FontRendererFactory.create(this); + return shadow(this, 'renderer', renderer); }, - get: function (i) { - if (this.firstChar <= i && i <= this.lastChar) { - return String.fromCharCode(i); + exportData: function Font_exportData() { + // TODO remove enumerating of the properties, e.g. hardcode exact names. + var data = {}; + for (var i in this) { + if (this.hasOwnProperty(i)) { + data[i] = this[i]; + } } - return undefined; + return data; }, - charCodeOf: function (v) { - return (isInt(v) && v >= this.firstChar && v <= this.lastChar) ? v : -1; - } - }; + checkAndRepair: function Font_checkAndRepair(name, font, properties) { + function readTableEntry(file) { + var tag = bytesToString(file.getBytes(4)); - return IdentityToUnicodeMap; -})(); + var checksum = file.getInt32(); + var offset = file.getInt32() >>> 0; + var length = file.getInt32() >>> 0; -var OpenTypeFileBuilder = (function OpenTypeFileBuilderClosure() { - function writeInt16(dest, offset, num) { - dest[offset] = (num >> 8) & 0xFF; - dest[offset + 1] = num & 0xFF; - } + // Read the table associated data + var previousPosition = file.pos; + file.pos = file.start ? file.start : 0; + file.skip(offset); + var data = file.getBytes(length); + file.pos = previousPosition; - function writeInt32(dest, offset, num) { - dest[offset] = (num >> 24) & 0xFF; - dest[offset + 1] = (num >> 16) & 0xFF; - dest[offset + 2] = (num >> 8) & 0xFF; - dest[offset + 3] = num & 0xFF; - } + if (tag === 'head') { + // clearing checksum adjustment + data[8] = data[9] = data[10] = data[11] = 0; + data[17] |= 0x20; //Set font optimized for cleartype flag + } - function writeData(dest, offset, data) { - var i, ii; - if (data instanceof Uint8Array) { - dest.set(data, offset); - } else if (typeof data === 'string') { - for (i = 0, ii = data.length; i < ii; i++) { - dest[offset++] = data.charCodeAt(i) & 0xFF; + return { + tag: tag, + checksum: checksum, + length: length, + offset: offset, + data: data + }; } - } else { - // treating everything else as array - for (i = 0, ii = data.length; i < ii; i++) { - dest[offset++] = data[i] & 0xFF; + + function readOpenTypeHeader(ttf) { + return { + version: bytesToString(ttf.getBytes(4)), + numTables: ttf.getUint16(), + searchRange: ttf.getUint16(), + entrySelector: ttf.getUint16(), + rangeShift: ttf.getUint16() + }; } - } - } - function OpenTypeFileBuilder(sfnt) { - this.sfnt = sfnt; - this.tables = Object.create(null); - } + /** + * Read the appropriate subtable from the cmap according to 9.6.6.4 from + * PDF spec + */ + function readCmapTable(cmap, font, isSymbolicFont, hasEncoding) { + if (!cmap) { + warn('No cmap table available.'); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } + var segment; + var start = (font.start ? font.start : 0) + cmap.offset; + font.pos = start; - OpenTypeFileBuilder.getSearchParams = - function OpenTypeFileBuilder_getSearchParams(entriesCount, entrySize) { - var maxPower2 = 1, log2 = 0; - while ((maxPower2 ^ entriesCount) > maxPower2) { - maxPower2 <<= 1; - log2++; - } - var searchRange = maxPower2 * entrySize; - return { - range: searchRange, - entry: log2, - rangeShift: entrySize * entriesCount - searchRange - }; - }; - - var OTF_HEADER_SIZE = 12; - var OTF_TABLE_ENTRY_SIZE = 16; - - OpenTypeFileBuilder.prototype = { - toArray: function OpenTypeFileBuilder_toArray() { - var sfnt = this.sfnt; - - // Tables needs to be written by ascendant alphabetic order - var tables = this.tables; - var tablesNames = Object.keys(tables); - tablesNames.sort(); - var numTables = tablesNames.length; - - var i, j, jj, table, tableName; - // layout the tables data - var offset = OTF_HEADER_SIZE + numTables * OTF_TABLE_ENTRY_SIZE; - var tableOffsets = [offset]; - for (i = 0; i < numTables; i++) { - table = tables[tablesNames[i]]; - var paddedLength = ((table.length + 3) & ~3) >>> 0; - offset += paddedLength; - tableOffsets.push(offset); - } - - var file = new Uint8Array(offset); - // write the table data first (mostly for checksum) - for (i = 0; i < numTables; i++) { - table = tables[tablesNames[i]]; - writeData(file, tableOffsets[i], table); - } - - // sfnt version (4 bytes) - if (sfnt === 'true') { - // Windows hates the Mac TrueType sfnt version number - sfnt = string32(0x00010000); - } - file[0] = sfnt.charCodeAt(0) & 0xFF; - file[1] = sfnt.charCodeAt(1) & 0xFF; - file[2] = sfnt.charCodeAt(2) & 0xFF; - file[3] = sfnt.charCodeAt(3) & 0xFF; - - // numTables (2 bytes) - writeInt16(file, 4, numTables); - - var searchParams = OpenTypeFileBuilder.getSearchParams(numTables, 16); + var version = font.getUint16(); + var numTables = font.getUint16(); - // searchRange (2 bytes) - writeInt16(file, 6, searchParams.range); - // entrySelector (2 bytes) - writeInt16(file, 8, searchParams.entry); - // rangeShift (2 bytes) - writeInt16(file, 10, searchParams.rangeShift); + var potentialTable; + var canBreak = false; + // There's an order of preference in terms of which cmap subtable to + // use: + // - non-symbolic fonts the preference is a 3,1 table then a 1,0 table + // - symbolic fonts the preference is a 3,0 table then a 1,0 table + // The following takes advantage of the fact that the tables are sorted + // to work. + for (var i = 0; i < numTables; i++) { + var platformId = font.getUint16(); + var encodingId = font.getUint16(); + var offset = font.getInt32() >>> 0; + var useTable = false; - offset = OTF_HEADER_SIZE; - // writing table entries - for (i = 0; i < numTables; i++) { - tableName = tablesNames[i]; - file[offset] = tableName.charCodeAt(0) & 0xFF; - file[offset + 1] = tableName.charCodeAt(1) & 0xFF; - file[offset + 2] = tableName.charCodeAt(2) & 0xFF; - file[offset + 3] = tableName.charCodeAt(3) & 0xFF; + if (platformId === 0 && encodingId === 0) { + useTable = true; + // Continue the loop since there still may be a higher priority + // table. + } else if (platformId === 1 && encodingId === 0) { + useTable = true; + // Continue the loop since there still may be a higher priority + // table. + } else if (platformId === 3 && encodingId === 1 && + ((!isSymbolicFont && hasEncoding) || !potentialTable)) { + useTable = true; + if (!isSymbolicFont) { + canBreak = true; + } + } else if (isSymbolicFont && platformId === 3 && encodingId === 0) { + useTable = true; + canBreak = true; + } - // checksum - var checksum = 0; - for (j = tableOffsets[i], jj = tableOffsets[i + 1]; j < jj; j += 4) { - var quad = (file[j] << 24) + (file[j + 1] << 16) + - (file[j + 2] << 8) + file[j + 3]; - checksum = (checksum + quad) | 0; + if (useTable) { + potentialTable = { + platformId: platformId, + encodingId: encodingId, + offset: offset + }; + } + if (canBreak) { + break; + } } - writeInt32(file, offset + 4, checksum); - // offset - writeInt32(file, offset + 8, tableOffsets[i]); - // length - writeInt32(file, offset + 12, tables[tableName].length); + if (potentialTable) { + font.pos = start + potentialTable.offset; + } + if (!potentialTable || font.peekByte() === -1) { + warn('Could not find a preferred cmap table.'); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } - offset += OTF_TABLE_ENTRY_SIZE; - } - return file; - }, + var format = font.getUint16(); + var length = font.getUint16(); + var language = font.getUint16(); - addTable: function OpenTypeFileBuilder_addTable(tag, data) { - if (tag in this.tables) { - throw new Error('Table ' + tag + ' already exists'); - } - this.tables[tag] = data; - } - }; + var hasShortCmap = false; + var mappings = []; + var j, glyphId; - return OpenTypeFileBuilder; -})(); + // TODO(mack): refactor this cmap subtable reading logic out + if (format === 0) { + for (j = 0; j < 256; j++) { + var index = font.getByte(); + if (!index) { + continue; + } + mappings.push({ + charCode: j, + glyphId: index + }); + } + hasShortCmap = true; + } else if (format === 4) { + // re-creating the table in format 4 since the encoding + // might be changed + var segCount = (font.getUint16() >> 1); + font.getBytes(6); // skipping range fields + var segIndex, segments = []; + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments.push({ end: font.getUint16() }); + } + font.getUint16(); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments[segIndex].start = font.getUint16(); + } -// Problematic Unicode characters in the fonts that needs to be moved to avoid -// issues when they are painted on the canvas, e.g. complex-script shaping or -// control/whitespace characters. The ranges are listed in pairs: the first item -// is a code of the first problematic code, the second one is the next -// non-problematic code. The ranges must be in sorted order. -var ProblematicCharRanges = new Int32Array([ - // Control characters. - 0x0000, 0x0020, - 0x007F, 0x00A1, - 0x00AD, 0x00AE, - // Chars that is used in complex-script shaping. - 0x0600, 0x0780, - 0x08A0, 0x10A0, - 0x1780, 0x1800, - // General punctuation chars. - 0x2000, 0x2010, - 0x2011, 0x2012, - 0x2028, 0x2030, - 0x205F, 0x2070, - 0x25CC, 0x25CD, - // Chars that is used in complex-script shaping. - 0xAA60, 0xAA80, - // Specials Unicode block. - 0xFFF0, 0x10000 -]); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segments[segIndex].delta = font.getUint16(); + } -/** - * 'Font' is the class the outside world should use, it encapsulate all the font - * decoding logics whatever type it is (assuming the font type is supported). - * - * For example to read a Type1 font and to attach it to the document: - * var type1Font = new Font("MyFontName", binaryFile, propertiesObject); - * type1Font.bind(); - */ -var Font = (function FontClosure() { - function Font(name, file, properties) { - var charCode, glyphName, fontChar; + var offsetsCount = 0; + for (segIndex = 0; segIndex < segCount; segIndex++) { + segment = segments[segIndex]; + var rangeOffset = font.getUint16(); + if (!rangeOffset) { + segment.offsetIndex = -1; + continue; + } - this.name = name; - this.loadedName = properties.loadedName; - this.isType3Font = properties.isType3Font; - this.sizes = []; + var offsetIndex = (rangeOffset >> 1) - (segCount - segIndex); + segment.offsetIndex = offsetIndex; + offsetsCount = Math.max(offsetsCount, offsetIndex + + segment.end - segment.start + 1); + } - this.glyphCache = Object.create(null); + var offsets = []; + for (j = 0; j < offsetsCount; j++) { + offsets.push(font.getUint16()); + } - var names = name.split('+'); - names = names.length > 1 ? names[1] : names[0]; - names = names.split(/[-,_]/g)[0]; - this.isSerifFont = !!(properties.flags & FontFlags.Serif); - this.isSymbolicFont = !!(properties.flags & FontFlags.Symbolic); - this.isMonospace = !!(properties.flags & FontFlags.FixedPitch); + for (segIndex = 0; segIndex < segCount; segIndex++) { + segment = segments[segIndex]; + start = segment.start; + var end = segment.end; + var delta = segment.delta; + offsetIndex = segment.offsetIndex; - var type = properties.type; - var subtype = properties.subtype; - this.type = type; + for (j = start; j <= end; j++) { + if (j === 0xFFFF) { + continue; + } - this.fallbackName = (this.isMonospace ? 'monospace' : - (this.isSerifFont ? 'serif' : 'sans-serif')); + glyphId = (offsetIndex < 0 ? + j : offsets[offsetIndex + j - start]); + glyphId = (glyphId + delta) & 0xFFFF; + if (glyphId === 0) { + continue; + } + mappings.push({ + charCode: j, + glyphId: glyphId + }); + } + } + } else if (format === 6) { + // Format 6 is a 2-bytes dense mapping, which means the font data + // lives glue together even if they are pretty far in the unicode + // table. (This looks weird, so I can have missed something), this + // works on Linux but seems to fails on Mac so let's rewrite the + // cmap table to a 3-1-4 style + var firstCode = font.getUint16(); + var entryCount = font.getUint16(); - this.differences = properties.differences; - this.widths = properties.widths; - this.defaultWidth = properties.defaultWidth; - this.composite = properties.composite; - this.wideChars = properties.wideChars; - this.cMap = properties.cMap; - this.ascent = properties.ascent / PDF_GLYPH_SPACE_UNITS; - this.descent = properties.descent / PDF_GLYPH_SPACE_UNITS; - this.fontMatrix = properties.fontMatrix; - this.bbox = properties.bbox; + for (j = 0; j < entryCount; j++) { + glyphId = font.getUint16(); + var charCode = firstCode + j; - this.toUnicode = properties.toUnicode = this.buildToUnicode(properties); + mappings.push({ + charCode: charCode, + glyphId: glyphId + }); + } + } else { + warn('cmap table has unsupported format: ' + format); + return { + platformId: -1, + encodingId: -1, + mappings: [], + hasShortCmap: false + }; + } - this.toFontChar = []; + // removing duplicate entries + mappings.sort(function (a, b) { + return a.charCode - b.charCode; + }); + for (i = 1; i < mappings.length; i++) { + if (mappings[i - 1].charCode === mappings[i].charCode) { + mappings.splice(i, 1); + i--; + } + } - if (properties.type === 'Type3') { - for (charCode = 0; charCode < 256; charCode++) { - this.toFontChar[charCode] = (this.differences[charCode] || - properties.defaultEncoding[charCode]); + return { + platformId: potentialTable.platformId, + encodingId: potentialTable.encodingId, + mappings: mappings, + hasShortCmap: hasShortCmap + }; } - this.fontType = FontType.TYPE3; - return; - } - this.cidEncoding = properties.cidEncoding; - this.vertical = properties.vertical; - if (this.vertical) { - this.vmetrics = properties.vmetrics; - this.defaultVMetrics = properties.defaultVMetrics; - } - var glyphsUnicodeMap; - if (!file || file.isEmpty) { - if (file) { - // Some bad PDF generators will include empty font files, - // attempting to recover by assuming that no file exists. - warn('Font file is empty in "' + name + '" (' + this.loadedName + ')'); - } + function sanitizeMetrics(font, header, metrics, numGlyphs) { + if (!header) { + if (metrics) { + metrics.data = null; + } + return; + } - this.missingFile = true; - // The file data is not specified. Trying to fix the font name - // to be used with the canvas.font. - var fontName = name.replace(/[,_]/g, '-'); - var stdFontMap = getStdFontMap(), nonStdFontMap = getNonStdFontMap(); - var isStandardFont = !!stdFontMap[fontName] || - !!(nonStdFontMap[fontName] && stdFontMap[nonStdFontMap[fontName]]); - fontName = stdFontMap[fontName] || nonStdFontMap[fontName] || fontName; + font.pos = (font.start ? font.start : 0) + header.offset; + font.pos += header.length - 2; + var numOfMetrics = font.getUint16(); - this.bold = (fontName.search(/bold/gi) !== -1); - this.italic = ((fontName.search(/oblique/gi) !== -1) || - (fontName.search(/italic/gi) !== -1)); + if (numOfMetrics > numGlyphs) { + info('The numOfMetrics (' + numOfMetrics + ') should not be ' + + 'greater than the numGlyphs (' + numGlyphs + ')'); + // Reduce numOfMetrics if it is greater than numGlyphs + numOfMetrics = numGlyphs; + header.data[34] = (numOfMetrics & 0xff00) >> 8; + header.data[35] = numOfMetrics & 0x00ff; + } - // Use 'name' instead of 'fontName' here because the original - // name ArialBlack for example will be replaced by Helvetica. - this.black = (name.search(/Black/g) !== -1); + var numOfSidebearings = numGlyphs - numOfMetrics; + var numMissing = numOfSidebearings - + ((metrics.length - numOfMetrics * 4) >> 1); - // if at least one width is present, remeasure all chars when exists - this.remeasure = Object.keys(this.widths).length > 0; - if (isStandardFont && type === 'CIDFontType2' && - properties.cidEncoding.indexOf('Identity-') === 0) { - var GlyphMapForStandardFonts = getGlyphMapForStandardFonts(); - // Standard fonts might be embedded as CID font without glyph mapping. - // Building one based on GlyphMapForStandardFonts. - var map = []; - for (charCode in GlyphMapForStandardFonts) { - map[+charCode] = GlyphMapForStandardFonts[charCode]; + if (numMissing > 0) { + // For each missing glyph, we set both the width and lsb to 0 (zero). + // Since we need to add two properties for each glyph, this explains + // the use of |numMissing * 2| when initializing the typed array. + var entries = new Uint8Array(metrics.length + numMissing * 2); + entries.set(metrics.data); + metrics.data = entries; } - if (/ArialBlack/i.test(name)) { - var SupplementalGlyphMapForArialBlack = - getSupplementalGlyphMapForArialBlack(); - for (charCode in SupplementalGlyphMapForArialBlack) { - map[+charCode] = SupplementalGlyphMapForArialBlack[charCode]; - } + } + + function sanitizeGlyph(source, sourceStart, sourceEnd, dest, destStart, + hintsValid) { + if (sourceEnd - sourceStart <= 12) { + // glyph with data less than 12 is invalid one + return 0; } - var isIdentityUnicode = this.toUnicode instanceof IdentityToUnicodeMap; - if (!isIdentityUnicode) { - this.toUnicode.forEach(function(charCode, unicodeCharCode) { - map[+charCode] = unicodeCharCode; - }); + var glyf = source.subarray(sourceStart, sourceEnd); + var contoursCount = (glyf[0] << 8) | glyf[1]; + if (contoursCount & 0x8000) { + // complex glyph, writing as is + dest.set(glyf, destStart); + return glyf.length; } - this.toFontChar = map; - this.toUnicode = new ToUnicodeMap(map); - } else if (/Symbol/i.test(fontName)) { - var symbols = SymbolSetEncoding; - glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode in symbols) { - fontChar = glyphsUnicodeMap[symbols[charCode]]; - if (!fontChar) { - continue; - } - this.toFontChar[charCode] = fontChar; + + var i, j = 10, flagsCount = 0; + for (i = 0; i < contoursCount; i++) { + var endPoint = (glyf[j] << 8) | glyf[j + 1]; + flagsCount = endPoint + 1; + j += 2; } - for (charCode in properties.differences) { - fontChar = glyphsUnicodeMap[properties.differences[charCode]]; - if (!fontChar) { - continue; + // skipping instructions + var instructionsStart = j; + var instructionsLength = (glyf[j] << 8) | glyf[j + 1]; + j += 2 + instructionsLength; + var instructionsEnd = j; + // validating flags + var coordinatesLength = 0; + for (i = 0; i < flagsCount; i++) { + var flag = glyf[j++]; + if (flag & 0xC0) { + // reserved flags must be zero, cleaning up + glyf[j - 1] = flag & 0x3F; + } + var xyLength = ((flag & 2) ? 1 : (flag & 16) ? 0 : 2) + + ((flag & 4) ? 1 : (flag & 32) ? 0 : 2); + coordinatesLength += xyLength; + if (flag & 8) { + var repeat = glyf[j++]; + i += repeat; + coordinatesLength += repeat * xyLength; } - this.toFontChar[charCode] = fontChar; } - } else if (/Dingbats/i.test(fontName)) { - glyphsUnicodeMap = getDingbatsGlyphsUnicode(); - if (/Wingdings/i.test(name)) { - warn('Wingdings font without embedded font file, ' + - 'falling back to the ZapfDingbats encoding.'); + // glyph without coordinates will be rejected + if (coordinatesLength === 0) { + return 0; } - var dingbats = ZapfDingbatsEncoding; - for (charCode in dingbats) { - fontChar = glyphsUnicodeMap[dingbats[charCode]]; - if (!fontChar) { - continue; - } - this.toFontChar[charCode] = fontChar; + var glyphDataLength = j + coordinatesLength; + if (glyphDataLength > glyf.length) { + // not enough data for coordinates + return 0; } - for (charCode in properties.differences) { - fontChar = glyphsUnicodeMap[properties.differences[charCode]]; - if (!fontChar) { - continue; + if (!hintsValid && instructionsLength > 0) { + dest.set(glyf.subarray(0, instructionsStart), destStart); + dest.set([0, 0], destStart + instructionsStart); + dest.set(glyf.subarray(instructionsEnd, glyphDataLength), + destStart + instructionsStart + 2); + glyphDataLength -= instructionsLength; + if (glyf.length - glyphDataLength > 3) { + glyphDataLength = (glyphDataLength + 3) & ~3; } - this.toFontChar[charCode] = fontChar; + return glyphDataLength; } - } else if (isStandardFont) { - this.toFontChar = []; - glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode in properties.defaultEncoding) { - glyphName = (properties.differences[charCode] || - properties.defaultEncoding[charCode]); - this.toFontChar[charCode] = glyphsUnicodeMap[glyphName]; + if (glyf.length - glyphDataLength > 3) { + // truncating and aligning to 4 bytes the long glyph data + glyphDataLength = (glyphDataLength + 3) & ~3; + dest.set(glyf.subarray(0, glyphDataLength), destStart); + return glyphDataLength; } - } else { - var unicodeCharCode, notCidFont = (type.indexOf('CIDFontType') === -1); - glyphsUnicodeMap = getGlyphsUnicode(); - this.toUnicode.forEach(function(charCode, unicodeCharCode) { - if (notCidFont) { - glyphName = (properties.differences[charCode] || - properties.defaultEncoding[charCode]); - unicodeCharCode = (glyphsUnicodeMap[glyphName] || unicodeCharCode); - } - this.toFontChar[charCode] = unicodeCharCode; - }.bind(this)); - } - this.loadedName = fontName.split('-')[0]; - this.loading = false; - this.fontType = getFontType(type, subtype); - return; - } - - // Some fonts might use wrong font types for Type1C or CIDFontType0C - if (subtype === 'Type1C' && (type !== 'Type1' && type !== 'MMType1')) { - // Some TrueType fonts by mistake claim Type1C - if (isTrueTypeFile(file)) { - subtype = 'TrueType'; - } else { - type = 'Type1'; - } - } - if (subtype === 'CIDFontType0C' && type !== 'CIDFontType0') { - type = 'CIDFontType0'; - } - if (subtype === 'OpenType') { - type = 'OpenType'; - } - // Some CIDFontType0C fonts by mistake claim CIDFontType0. - if (type === 'CIDFontType0') { - if (isType1File(file)) { - subtype = 'CIDFontType0'; - } else if (isOpenTypeFile(file)) { - // Sometimes the type/subtype can be a complete lie (see issue6782.pdf). - type = subtype = 'OpenType'; - } else { - subtype = 'CIDFontType0C'; + // glyph data is fine + dest.set(glyf, destStart); + return glyf.length; } - } - - var data; - switch (type) { - case 'MMType1': - info('MMType1 font (' + name + '), falling back to Type1.'); - /* falls through */ - case 'Type1': - case 'CIDFontType0': - this.mimetype = 'font/opentype'; - - var cff = (subtype === 'Type1C' || subtype === 'CIDFontType0C') ? - new CFFFont(file, properties) : new Type1Font(name, file, properties); - - adjustWidths(properties); - - // Wrap the CFF data inside an OTF font file - data = this.convert(name, cff, properties); - break; - - case 'OpenType': - case 'TrueType': - case 'CIDFontType2': - this.mimetype = 'font/opentype'; - // Repair the TrueType file. It is can be damaged in the point of - // view of the sanitizer - data = this.checkAndRepair(name, file, properties); - if (this.isOpenType) { - adjustWidths(properties); + function sanitizeHead(head, numGlyphs, locaLength) { + var data = head.data; - type = 'OpenType'; + // Validate version: + // Should always be 0x00010000 + var version = int32(data[0], data[1], data[2], data[3]); + if (version >> 16 !== 1) { + info('Attempting to fix invalid version in head table: ' + version); + data[0] = 0; + data[1] = 1; + data[2] = 0; + data[3] = 0; } - break; - - default: - error('Font ' + type + ' is not supported'); - break; - } - - this.data = data; - this.fontType = getFontType(type, subtype); - - // Transfer some properties again that could change during font conversion - this.fontMatrix = properties.fontMatrix; - this.widths = properties.widths; - this.defaultWidth = properties.defaultWidth; - this.encoding = properties.baseEncoding; - this.seacMap = properties.seacMap; - - this.loading = true; - } - - Font.getFontID = (function () { - var ID = 1; - return function Font_getFontID() { - return String(ID++); - }; - })(); - - function int16(b0, b1) { - return (b0 << 8) + b1; - } - - function int32(b0, b1, b2, b3) { - return (b0 << 24) + (b1 << 16) + (b2 << 8) + b3; - } - - function string16(value) { - return String.fromCharCode((value >> 8) & 0xff, value & 0xff); - } - - function safeString16(value) { - // clamp value to the 16-bit int range - value = (value > 0x7FFF ? 0x7FFF : (value < -0x8000 ? -0x8000 : value)); - return String.fromCharCode((value >> 8) & 0xff, value & 0xff); - } - - function isTrueTypeFile(file) { - var header = file.peekBytes(4); - return readUint32(header, 0) === 0x00010000; - } - - function isOpenTypeFile(file) { - var header = file.peekBytes(4); - return bytesToString(header) === 'OTTO'; - } - function isType1File(file) { - var header = file.peekBytes(2); - // All Type1 font programs must begin with the comment '%!' (0x25 + 0x21). - if (header[0] === 0x25 && header[1] === 0x21) { - return true; - } - // ... obviously some fonts violate that part of the specification, - // please refer to the comment in |Type1Font| below. - if (header[0] === 0x80 && header[1] === 0x01) { // pfb file header. - return true; - } - return false; - } + var indexToLocFormat = int16(data[50], data[51]); + if (indexToLocFormat < 0 || indexToLocFormat > 1) { + info('Attempting to fix invalid indexToLocFormat in head table: ' + + indexToLocFormat); - /** - * Helper function for |adjustMapping|. - * @return {boolean} - */ - function isProblematicUnicodeLocation(code) { - // Using binary search to find a range start. - var i = 0, j = ProblematicCharRanges.length - 1; - while (i < j) { - var c = (i + j + 1) >> 1; - if (code < ProblematicCharRanges[c]) { - j = c - 1; - } else { - i = c; - } - } - // Even index means code in problematic range. - return !(i & 1); - } + // The value of indexToLocFormat should be 0 if the loca table + // consists of short offsets, and should be 1 if the loca table + // consists of long offsets. + // + // The number of entries in the loca table should be numGlyphs + 1. + // + // Using this information, we can work backwards to deduce if the + // size of each offset in the loca table, and thus figure out the + // appropriate value for indexToLocFormat. - /** - * Rebuilds the char code to glyph ID map by trying to replace the char codes - * with their unicode value. It also moves char codes that are in known - * problematic locations. - * @return {Object} Two properties: - * 'toFontChar' - maps original char codes(the value that will be read - * from commands such as show text) to the char codes that will be used in the - * font that we build - * 'charCodeToGlyphId' - maps the new font char codes to glyph ids - */ - function adjustMapping(charCodeToGlyphId, properties) { - var toUnicode = properties.toUnicode; - var isSymbolic = !!(properties.flags & FontFlags.Symbolic); - var isIdentityUnicode = - properties.toUnicode instanceof IdentityToUnicodeMap; - var newMap = Object.create(null); - var toFontChar = []; - var usedFontCharCodes = []; - var nextAvailableFontCharCode = PRIVATE_USE_OFFSET_START; - for (var originalCharCode in charCodeToGlyphId) { - originalCharCode |= 0; - var glyphId = charCodeToGlyphId[originalCharCode]; - var fontCharCode = originalCharCode; - // First try to map the value to a unicode position if a non identity map - // was created. - if (!isIdentityUnicode && toUnicode.has(originalCharCode)) { - var unicode = toUnicode.get(fontCharCode); - // TODO: Try to map ligatures to the correct spot. - if (unicode.length === 1) { - fontCharCode = unicode.charCodeAt(0); + var numGlyphsPlusOne = numGlyphs + 1; + if (locaLength === numGlyphsPlusOne << 1) { + // 0x0000 indicates the loca table consists of short offsets + data[50] = 0; + data[51] = 0; + } else if (locaLength === numGlyphsPlusOne << 2) { + // 0x0001 indicates the loca table consists of long offsets + data[50] = 0; + data[51] = 1; + } else { + warn('Could not fix indexToLocFormat: ' + indexToLocFormat); + } } } - // Try to move control characters, special characters and already mapped - // characters to the private use area since they will not be drawn by - // canvas if left in their current position. Also, move characters if the - // font was symbolic and there is only an identity unicode map since the - // characters probably aren't in the correct position (fixes an issue - // with firefox and thuluthfont). - if ((usedFontCharCodes[fontCharCode] !== undefined || - isProblematicUnicodeLocation(fontCharCode) || - (isSymbolic && isIdentityUnicode)) && - nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END) { // Room left. - // Loop to try and find a free spot in the private use area. - do { - fontCharCode = nextAvailableFontCharCode++; - if (SKIP_PRIVATE_USE_RANGE_F000_TO_F01F && fontCharCode === 0xF000) { - fontCharCode = 0xF020; - nextAvailableFontCharCode = fontCharCode + 1; + function sanitizeGlyphLocations(loca, glyf, numGlyphs, + isGlyphLocationsLong, hintsValid, + dupFirstEntry) { + var itemSize, itemDecode, itemEncode; + if (isGlyphLocationsLong) { + itemSize = 4; + itemDecode = function fontItemDecodeLong(data, offset) { + return (data[offset] << 24) | (data[offset + 1] << 16) | + (data[offset + 2] << 8) | data[offset + 3]; + }; + itemEncode = function fontItemEncodeLong(data, offset, value) { + data[offset] = (value >>> 24) & 0xFF; + data[offset + 1] = (value >> 16) & 0xFF; + data[offset + 2] = (value >> 8) & 0xFF; + data[offset + 3] = value & 0xFF; + }; + } else { + itemSize = 2; + itemDecode = function fontItemDecode(data, offset) { + return (data[offset] << 9) | (data[offset + 1] << 1); + }; + itemEncode = function fontItemEncode(data, offset, value) { + data[offset] = (value >> 9) & 0xFF; + data[offset + 1] = (value >> 1) & 0xFF; + }; + } + var locaData = loca.data; + var locaDataSize = itemSize * (1 + numGlyphs); + // is loca.data too short or long? + if (locaData.length !== locaDataSize) { + locaData = new Uint8Array(locaDataSize); + locaData.set(loca.data.subarray(0, locaDataSize)); + loca.data = locaData; + } + // removing the invalid glyphs + var oldGlyfData = glyf.data; + var oldGlyfDataLength = oldGlyfData.length; + var newGlyfData = new Uint8Array(oldGlyfDataLength); + var startOffset = itemDecode(locaData, 0); + var writeOffset = 0; + var missingGlyphData = Object.create(null); + itemEncode(locaData, 0, writeOffset); + var i, j; + for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { + var endOffset = itemDecode(locaData, j); + if (endOffset > oldGlyfDataLength && + ((oldGlyfDataLength + 3) & ~3) === endOffset) { + // Aspose breaks fonts by aligning the glyphs to the qword, but not + // the glyf table size, which makes last glyph out of range. + endOffset = oldGlyfDataLength; + } + if (endOffset > oldGlyfDataLength) { + // glyph end offset points outside glyf data, rejecting the glyph + itemEncode(locaData, j, writeOffset); + startOffset = endOffset; + continue; } - } while (usedFontCharCodes[fontCharCode] !== undefined && - nextAvailableFontCharCode <= PRIVATE_USE_OFFSET_END); - } + if (startOffset === endOffset) { + missingGlyphData[i] = true; + } - newMap[fontCharCode] = glyphId; - toFontChar[originalCharCode] = fontCharCode; - usedFontCharCodes[fontCharCode] = true; - } - return { - toFontChar: toFontChar, - charCodeToGlyphId: newMap, - nextAvailableFontCharCode: nextAvailableFontCharCode - }; - } + var newLength = sanitizeGlyph(oldGlyfData, startOffset, endOffset, + newGlyfData, writeOffset, hintsValid); + writeOffset += newLength; + itemEncode(locaData, j, writeOffset); + startOffset = endOffset; + } - function getRanges(glyphs, numGlyphs) { - // Array.sort() sorts by characters, not numerically, so convert to an - // array of characters. - var codes = []; - for (var charCode in glyphs) { - // Remove an invalid glyph ID mappings to make OTS happy. - if (glyphs[charCode] >= numGlyphs) { - continue; - } - codes.push({ fontCharCode: charCode | 0, glyphId: glyphs[charCode] }); - } - codes.sort(function fontGetRangesSort(a, b) { - return a.fontCharCode - b.fontCharCode; - }); + if (writeOffset === 0) { + // glyf table cannot be empty -- redoing the glyf and loca tables + // to have single glyph with one point + var simpleGlyph = new Uint8Array( + [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 49, 0]); + for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { + itemEncode(locaData, j, simpleGlyph.length); + } + glyf.data = simpleGlyph; + return missingGlyphData; + } - // Split the sorted codes into ranges. - var ranges = []; - var length = codes.length; - for (var n = 0; n < length; ) { - var start = codes[n].fontCharCode; - var codeIndices = [codes[n].glyphId]; - ++n; - var end = start; - while (n < length && end + 1 === codes[n].fontCharCode) { - codeIndices.push(codes[n].glyphId); - ++end; - ++n; - if (end === 0xFFFF) { - break; + if (dupFirstEntry) { + var firstEntryLength = itemDecode(locaData, itemSize); + if (newGlyfData.length > firstEntryLength + writeOffset) { + glyf.data = newGlyfData.subarray(0, firstEntryLength + writeOffset); + } else { + glyf.data = new Uint8Array(firstEntryLength + writeOffset); + glyf.data.set(newGlyfData.subarray(0, writeOffset)); + } + glyf.data.set(newGlyfData.subarray(0, firstEntryLength), writeOffset); + itemEncode(loca.data, locaData.length - itemSize, + writeOffset + firstEntryLength); + } else { + glyf.data = newGlyfData.subarray(0, writeOffset); } + return missingGlyphData; } - ranges.push([start, end, codeIndices]); - } - - return ranges; - } - - function createCmapTable(glyphs, numGlyphs) { - var ranges = getRanges(glyphs, numGlyphs); - var numTables = ranges[ranges.length - 1][1] > 0xFFFF ? 2 : 1; - var cmap = '\x00\x00' + // version - string16(numTables) + // numTables - '\x00\x03' + // platformID - '\x00\x01' + // encodingID - string32(4 + numTables * 8); // start of the table record - var i, ii, j, jj; - for (i = ranges.length - 1; i >= 0; --i) { - if (ranges[i][0] <= 0xFFFF) { break; } - } - var bmpLength = i + 1; + function readPostScriptTable(post, properties, maxpNumGlyphs) { + var start = (font.start ? font.start : 0) + post.offset; + font.pos = start; - if (ranges[i][0] < 0xFFFF && ranges[i][1] === 0xFFFF) { - ranges[i][1] = 0xFFFE; - } - var trailingRangesCount = ranges[i][1] < 0xFFFF ? 1 : 0; - var segCount = bmpLength + trailingRangesCount; - var searchParams = OpenTypeFileBuilder.getSearchParams(segCount, 2); + var length = post.length, end = start + length; + var version = font.getInt32(); + // skip rest to the tables + font.getBytes(28); - // Fill up the 4 parallel arrays describing the segments. - var startCount = ''; - var endCount = ''; - var idDeltas = ''; - var idRangeOffsets = ''; - var glyphsIds = ''; - var bias = 0; + var glyphNames; + var valid = true; + var i; - var range, start, end, codes; - for (i = 0, ii = bmpLength; i < ii; i++) { - range = ranges[i]; - start = range[0]; - end = range[1]; - startCount += string16(start); - endCount += string16(end); - codes = range[2]; - var contiguous = true; - for (j = 1, jj = codes.length; j < jj; ++j) { - if (codes[j] !== codes[j - 1] + 1) { - contiguous = false; - break; + switch (version) { + case 0x00010000: + glyphNames = MacStandardGlyphOrdering; + break; + case 0x00020000: + var numGlyphs = font.getUint16(); + if (numGlyphs !== maxpNumGlyphs) { + valid = false; + break; + } + var glyphNameIndexes = []; + for (i = 0; i < numGlyphs; ++i) { + var index = font.getUint16(); + if (index >= 32768) { + valid = false; + break; + } + glyphNameIndexes.push(index); + } + if (!valid) { + break; + } + var customNames = []; + var strBuf = []; + while (font.pos < end) { + var stringLength = font.getByte(); + strBuf.length = stringLength; + for (i = 0; i < stringLength; ++i) { + strBuf[i] = String.fromCharCode(font.getByte()); + } + customNames.push(strBuf.join('')); + } + glyphNames = []; + for (i = 0; i < numGlyphs; ++i) { + var j = glyphNameIndexes[i]; + if (j < 258) { + glyphNames.push(MacStandardGlyphOrdering[j]); + continue; + } + glyphNames.push(customNames[j - 258]); + } + break; + case 0x00030000: + break; + default: + warn('Unknown/unsupported post table version ' + version); + valid = false; + if (properties.defaultEncoding) { + glyphNames = properties.defaultEncoding; + } + break; } + properties.glyphNames = glyphNames; + return valid; } - if (!contiguous) { - var offset = (segCount - i) * 2 + bias * 2; - bias += (end - start + 1); - idDeltas += string16(0); - idRangeOffsets += string16(offset); + function readNameTable(nameTable) { + var start = (font.start ? font.start : 0) + nameTable.offset; + font.pos = start; - for (j = 0, jj = codes.length; j < jj; ++j) { - glyphsIds += string16(codes[j]); + var names = [[], []]; + var length = nameTable.length, end = start + length; + var format = font.getUint16(); + var FORMAT_0_HEADER_LENGTH = 6; + if (format !== 0 || length < FORMAT_0_HEADER_LENGTH) { + // unsupported name table format or table "too" small + return names; } - } else { - var startCode = codes[0]; - - idDeltas += string16((startCode - start) & 0xFFFF); - idRangeOffsets += string16(0); - } - } + var numRecords = font.getUint16(); + var stringsStart = font.getUint16(); + var records = []; + var NAME_RECORD_LENGTH = 12; + var i, ii; - if (trailingRangesCount > 0) { - endCount += '\xFF\xFF'; - startCount += '\xFF\xFF'; - idDeltas += '\x00\x01'; - idRangeOffsets += '\x00\x00'; - } - - var format314 = '\x00\x00' + // language - string16(2 * segCount) + - string16(searchParams.range) + - string16(searchParams.entry) + - string16(searchParams.rangeShift) + - endCount + '\x00\x00' + startCount + - idDeltas + idRangeOffsets + glyphsIds; - - var format31012 = ''; - var header31012 = ''; - if (numTables > 1) { - cmap += '\x00\x03' + // platformID - '\x00\x0A' + // encodingID - string32(4 + numTables * 8 + - 4 + format314.length); // start of the table record - format31012 = ''; - for (i = 0, ii = ranges.length; i < ii; i++) { - range = ranges[i]; - start = range[0]; - codes = range[2]; - var code = codes[0]; - for (j = 1, jj = codes.length; j < jj; ++j) { - if (codes[j] !== codes[j - 1] + 1) { - end = range[0] + j - 1; - format31012 += string32(start) + // startCharCode - string32(end) + // endCharCode - string32(code); // startGlyphID - start = end + 1; - code = codes[j]; + for (i = 0; i < numRecords && + font.pos + NAME_RECORD_LENGTH <= end; i++) { + var r = { + platform: font.getUint16(), + encoding: font.getUint16(), + language: font.getUint16(), + name: font.getUint16(), + length: font.getUint16(), + offset: font.getUint16() + }; + // using only Macintosh and Windows platform/encoding names + if ((r.platform === 1 && r.encoding === 0 && r.language === 0) || + (r.platform === 3 && r.encoding === 1 && r.language === 0x409)) { + records.push(r); } } - format31012 += string32(start) + // startCharCode - string32(range[1]) + // endCharCode - string32(code); // startGlyphID + for (i = 0, ii = records.length; i < ii; i++) { + var record = records[i]; + var pos = start + stringsStart + record.offset; + if (pos + record.length > end) { + continue; // outside of name table, ignoring + } + font.pos = pos; + var nameIndex = record.name; + if (record.encoding) { + // unicode + var str = ''; + for (var j = 0, jj = record.length; j < jj; j += 2) { + str += String.fromCharCode(font.getUint16()); + } + names[1][nameIndex] = str; + } else { + names[0][nameIndex] = bytesToString(font.getBytes(record.length)); + } + } + return names; } - header31012 = '\x00\x0C' + // format - '\x00\x00' + // reserved - string32(format31012.length + 16) + // length - '\x00\x00\x00\x00' + // language - string32(format31012.length / 12); // nGroups - } - - return cmap + '\x00\x04' + // format - string16(format314.length + 4) + // length - format314 + header31012 + format31012; - } - - function validateOS2Table(os2) { - var stream = new Stream(os2.data); - var version = stream.getUint16(); - // TODO verify all OS/2 tables fields, but currently we validate only those - // that give us issues - stream.getBytes(60); // skipping type, misc sizes, panose, unicode ranges - var selection = stream.getUint16(); - if (version < 4 && (selection & 0x0300)) { - return false; - } - var firstChar = stream.getUint16(); - var lastChar = stream.getUint16(); - if (firstChar > lastChar) { - return false; - } - stream.getBytes(6); // skipping sTypoAscender/Descender/LineGap - var usWinAscent = stream.getUint16(); - if (usWinAscent === 0) { // makes font unreadable by windows - return false; - } - - // OS/2 appears to be valid, resetting some fields - os2.data[8] = os2.data[9] = 0; // IE rejects fonts if fsType != 0 - return true; - } - - function createOS2Table(properties, charstrings, override) { - override = override || { - unitsPerEm: 0, - yMax: 0, - yMin: 0, - ascent: 0, - descent: 0 - }; - var ulUnicodeRange1 = 0; - var ulUnicodeRange2 = 0; - var ulUnicodeRange3 = 0; - var ulUnicodeRange4 = 0; + var TTOpsStackDeltas = [ + 0, 0, 0, 0, 0, 0, 0, 0, -2, -2, -2, -2, 0, 0, -2, -5, + -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, -1, -1, + 1, -1, -999, 0, 1, 0, -1, -2, 0, -1, -2, -1, -1, 0, -1, -1, + 0, 0, -999, -999, -1, -1, -1, -1, -2, -999, -2, -2, -999, 0, -2, -2, + 0, 0, -2, 0, -2, 0, 0, 0, -2, -1, -1, 1, 1, 0, 0, -1, + -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, 0, -999, -1, -1, + -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + -2, -999, -999, -999, -999, -999, -1, -1, -2, -2, 0, 0, 0, 0, -1, -1, + -999, -2, -2, 0, 0, -1, -2, -2, 0, 0, 0, -1, -1, -1, -2]; + // 0xC0-DF == -1 and 0xE0-FF == -2 - var firstCharIndex = null; - var lastCharIndex = 0; + function sanitizeTTProgram(table, ttContext) { + var data = table.data; + var i = 0, j, n, b, funcId, pc, lastEndf = 0, lastDeff = 0; + var stack = []; + var callstack = []; + var functionsCalled = []; + var tooComplexToFollowFunctions = + ttContext.tooComplexToFollowFunctions; + var inFDEF = false, ifLevel = 0, inELSE = 0; + for (var ii = data.length; i < ii;) { + var op = data[i++]; + // The TrueType instruction set docs can be found at + // https://developer.apple.com/fonts/TTRefMan/RM05/Chap5.html + if (op === 0x40) { // NPUSHB - pushes n bytes + n = data[i++]; + if (inFDEF || inELSE) { + i += n; + } else { + for (j = 0; j < n; j++) { + stack.push(data[i++]); + } + } + } else if (op === 0x41) { // NPUSHW - pushes n words + n = data[i++]; + if (inFDEF || inELSE) { + i += n * 2; + } else { + for (j = 0; j < n; j++) { + b = data[i++]; + stack.push((b << 8) | data[i++]); + } + } + } else if ((op & 0xF8) === 0xB0) { // PUSHB - pushes bytes + n = op - 0xB0 + 1; + if (inFDEF || inELSE) { + i += n; + } else { + for (j = 0; j < n; j++) { + stack.push(data[i++]); + } + } + } else if ((op & 0xF8) === 0xB8) { // PUSHW - pushes words + n = op - 0xB8 + 1; + if (inFDEF || inELSE) { + i += n * 2; + } else { + for (j = 0; j < n; j++) { + b = data[i++]; + stack.push((b << 8) | data[i++]); + } + } + } else if (op === 0x2B && !tooComplexToFollowFunctions) { // CALL + if (!inFDEF && !inELSE) { + // collecting inforamtion about which functions are used + funcId = stack[stack.length - 1]; + ttContext.functionsUsed[funcId] = true; + if (funcId in ttContext.functionsStackDeltas) { + stack.length += ttContext.functionsStackDeltas[funcId]; + } else if (funcId in ttContext.functionsDefined && + functionsCalled.indexOf(funcId) < 0) { + callstack.push({data: data, i: i, stackTop: stack.length - 1}); + functionsCalled.push(funcId); + pc = ttContext.functionsDefined[funcId]; + if (!pc) { + warn('TT: CALL non-existent function'); + ttContext.hintsValid = false; + return; + } + data = pc.data; + i = pc.i; + } + } + } else if (op === 0x2C && !tooComplexToFollowFunctions) { // FDEF + if (inFDEF || inELSE) { + warn('TT: nested FDEFs not allowed'); + tooComplexToFollowFunctions = true; + } + inFDEF = true; + // collecting inforamtion about which functions are defined + lastDeff = i; + funcId = stack.pop(); + ttContext.functionsDefined[funcId] = {data: data, i: i}; + } else if (op === 0x2D) { // ENDF - end of function + if (inFDEF) { + inFDEF = false; + lastEndf = i; + } else { + pc = callstack.pop(); + if (!pc) { + warn('TT: ENDF bad stack'); + ttContext.hintsValid = false; + return; + } + funcId = functionsCalled.pop(); + data = pc.data; + i = pc.i; + ttContext.functionsStackDeltas[funcId] = + stack.length - pc.stackTop; + } + } else if (op === 0x89) { // IDEF - instruction definition + if (inFDEF || inELSE) { + warn('TT: nested IDEFs not allowed'); + tooComplexToFollowFunctions = true; + } + inFDEF = true; + // recording it as a function to track ENDF + lastDeff = i; + } else if (op === 0x58) { // IF + ++ifLevel; + } else if (op === 0x1B) { // ELSE + inELSE = ifLevel; + } else if (op === 0x59) { // EIF + if (inELSE === ifLevel) { + inELSE = 0; + } + --ifLevel; + } else if (op === 0x1C) { // JMPR + if (!inFDEF && !inELSE) { + var offset = stack[stack.length - 1]; + // only jumping forward to prevent infinite loop + if (offset > 0) { + i += offset - 1; + } + } + } + // Adjusting stack not extactly, but just enough to get function id + if (!inFDEF && !inELSE) { + var stackDelta = op <= 0x8E ? TTOpsStackDeltas[op] : + op >= 0xC0 && op <= 0xDF ? -1 : op >= 0xE0 ? -2 : 0; + if (op >= 0x71 && op <= 0x75) { + n = stack.pop(); + if (n === n) { + stackDelta = -n * 2; + } + } + while (stackDelta < 0 && stack.length > 0) { + stack.pop(); + stackDelta++; + } + while (stackDelta > 0) { + stack.push(NaN); // pushing any number into stack + stackDelta--; + } + } + } + ttContext.tooComplexToFollowFunctions = tooComplexToFollowFunctions; + var content = [data]; + if (i > data.length) { + content.push(new Uint8Array(i - data.length)); + } + if (lastDeff > lastEndf) { + warn('TT: complementing a missing function tail'); + // new function definition started, but not finished + // complete function by [CLEAR, ENDF] + content.push(new Uint8Array([0x22, 0x2D])); + } + foldTTTable(table, content); + } - if (charstrings) { - for (var code in charstrings) { - code |= 0; - if (firstCharIndex > code || !firstCharIndex) { - firstCharIndex = code; + function checkInvalidFunctions(ttContext, maxFunctionDefs) { + if (ttContext.tooComplexToFollowFunctions) { + return; } - if (lastCharIndex < code) { - lastCharIndex = code; + if (ttContext.functionsDefined.length > maxFunctionDefs) { + warn('TT: more functions defined than expected'); + ttContext.hintsValid = false; + return; + } + for (var j = 0, jj = ttContext.functionsUsed.length; j < jj; j++) { + if (j > maxFunctionDefs) { + warn('TT: invalid function id: ' + j); + ttContext.hintsValid = false; + return; + } + if (ttContext.functionsUsed[j] && !ttContext.functionsDefined[j]) { + warn('TT: undefined function: ' + j); + ttContext.hintsValid = false; + return; + } } + } - var position = getUnicodeRangeFor(code); - if (position < 32) { - ulUnicodeRange1 |= 1 << position; - } else if (position < 64) { - ulUnicodeRange2 |= 1 << position - 32; - } else if (position < 96) { - ulUnicodeRange3 |= 1 << position - 64; - } else if (position < 123) { - ulUnicodeRange4 |= 1 << position - 96; - } else { - error('Unicode ranges Bits > 123 are reserved for internal usage'); + function foldTTTable(table, content) { + if (content.length > 1) { + // concatenating the content items + var newLength = 0; + var j, jj; + for (j = 0, jj = content.length; j < jj; j++) { + newLength += content[j].length; + } + newLength = (newLength + 3) & ~3; + var result = new Uint8Array(newLength); + var pos = 0; + for (j = 0, jj = content.length; j < jj; j++) { + result.set(content[j], pos); + pos += content[j].length; + } + table.data = result; + table.length = newLength; } } - } else { - // TODO - firstCharIndex = 0; - lastCharIndex = 255; - } - var bbox = properties.bbox || [0, 0, 0, 0]; - var unitsPerEm = (override.unitsPerEm || - 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]); + function sanitizeTTPrograms(fpgm, prep, cvt) { + var ttContext = { + functionsDefined: [], + functionsUsed: [], + functionsStackDeltas: [], + tooComplexToFollowFunctions: false, + hintsValid: true + }; + if (fpgm) { + sanitizeTTProgram(fpgm, ttContext); + } + if (prep) { + sanitizeTTProgram(prep, ttContext); + } + if (fpgm) { + checkInvalidFunctions(ttContext, maxFunctionDefs); + } + if (cvt && (cvt.length & 1)) { + var cvtData = new Uint8Array(cvt.length + 1); + cvtData.set(cvt.data); + cvt.data = cvtData; + } + return ttContext.hintsValid; + } - // if the font units differ to the PDF glyph space units - // then scale up the values - var scale = (properties.ascentScaled ? 1.0 : - unitsPerEm / PDF_GLYPH_SPACE_UNITS); + // The following steps modify the original font data, making copy + font = new Stream(new Uint8Array(font.getBytes())); - var typoAscent = (override.ascent || - Math.round(scale * (properties.ascent || bbox[3]))); - var typoDescent = (override.descent || - Math.round(scale * (properties.descent || bbox[1]))); - if (typoDescent > 0 && properties.descent > 0 && bbox[1] < 0) { - typoDescent = -typoDescent; // fixing incorrect descent - } - var winAscent = override.yMax || typoAscent; - var winDescent = -override.yMin || -typoDescent; + var VALID_TABLES = ['OS/2', 'cmap', 'head', 'hhea', 'hmtx', 'maxp', + 'name', 'post', 'loca', 'glyf', 'fpgm', 'prep', 'cvt ', 'CFF ']; - return '\x00\x03' + // version - '\x02\x24' + // xAvgCharWidth - '\x01\xF4' + // usWeightClass - '\x00\x05' + // usWidthClass - '\x00\x00' + // fstype (0 to let the font loads via font-face on IE) - '\x02\x8A' + // ySubscriptXSize - '\x02\xBB' + // ySubscriptYSize - '\x00\x00' + // ySubscriptXOffset - '\x00\x8C' + // ySubscriptYOffset - '\x02\x8A' + // ySuperScriptXSize - '\x02\xBB' + // ySuperScriptYSize - '\x00\x00' + // ySuperScriptXOffset - '\x01\xDF' + // ySuperScriptYOffset - '\x00\x31' + // yStrikeOutSize - '\x01\x02' + // yStrikeOutPosition - '\x00\x00' + // sFamilyClass - '\x00\x00\x06' + - String.fromCharCode(properties.fixedPitch ? 0x09 : 0x00) + - '\x00\x00\x00\x00\x00\x00' + // Panose - string32(ulUnicodeRange1) + // ulUnicodeRange1 (Bits 0-31) - string32(ulUnicodeRange2) + // ulUnicodeRange2 (Bits 32-63) - string32(ulUnicodeRange3) + // ulUnicodeRange3 (Bits 64-95) - string32(ulUnicodeRange4) + // ulUnicodeRange4 (Bits 96-127) - '\x2A\x32\x31\x2A' + // achVendID - string16(properties.italicAngle ? 1 : 0) + // fsSelection - string16(firstCharIndex || - properties.firstChar) + // usFirstCharIndex - string16(lastCharIndex || properties.lastChar) + // usLastCharIndex - string16(typoAscent) + // sTypoAscender - string16(typoDescent) + // sTypoDescender - '\x00\x64' + // sTypoLineGap (7%-10% of the unitsPerEM value) - string16(winAscent) + // usWinAscent - string16(winDescent) + // usWinDescent - '\x00\x00\x00\x00' + // ulCodePageRange1 (Bits 0-31) - '\x00\x00\x00\x00' + // ulCodePageRange2 (Bits 32-63) - string16(properties.xHeight) + // sxHeight - string16(properties.capHeight) + // sCapHeight - string16(0) + // usDefaultChar - string16(firstCharIndex || properties.firstChar) + // usBreakChar - '\x00\x03'; // usMaxContext - } + var header = readOpenTypeHeader(font); + var numTables = header.numTables; + var cff, cffFile; - function createPostTable(properties) { - var angle = Math.floor(properties.italicAngle * (Math.pow(2, 16))); - return ('\x00\x03\x00\x00' + // Version number - string32(angle) + // italicAngle - '\x00\x00' + // underlinePosition - '\x00\x00' + // underlineThickness - string32(properties.fixedPitch) + // isFixedPitch - '\x00\x00\x00\x00' + // minMemType42 - '\x00\x00\x00\x00' + // maxMemType42 - '\x00\x00\x00\x00' + // minMemType1 - '\x00\x00\x00\x00'); // maxMemType1 - } + var tables = Object.create(null); + tables['OS/2'] = null; + tables['cmap'] = null; + tables['head'] = null; + tables['hhea'] = null; + tables['hmtx'] = null; + tables['maxp'] = null; + tables['name'] = null; + tables['post'] = null; - function createNameTable(name, proto) { - if (!proto) { - proto = [[], []]; // no strings and unicode strings - } + var table; + for (var i = 0; i < numTables; i++) { + table = readTableEntry(font); + if (VALID_TABLES.indexOf(table.tag) < 0) { + continue; // skipping table if it's not a required or optional table + } + if (table.length === 0) { + continue; // skipping empty tables + } + tables[table.tag] = table; + } - var strings = [ - proto[0][0] || 'Original licence', // 0.Copyright - proto[0][1] || name, // 1.Font family - proto[0][2] || 'Unknown', // 2.Font subfamily (font weight) - proto[0][3] || 'uniqueID', // 3.Unique ID - proto[0][4] || name, // 4.Full font name - proto[0][5] || 'Version 0.11', // 5.Version - proto[0][6] || '', // 6.Postscript name - proto[0][7] || 'Unknown', // 7.Trademark - proto[0][8] || 'Unknown', // 8.Manufacturer - proto[0][9] || 'Unknown' // 9.Designer - ]; + var isTrueType = !tables['CFF ']; + if (!isTrueType) { + // OpenType font + if ((header.version === 'OTTO' && properties.type !== 'CIDFontType2') || + !tables['head'] || !tables['hhea'] || !tables['maxp'] || + !tables['post']) { + // no major tables: throwing everything at CFFFont + cffFile = new Stream(tables['CFF '].data); + cff = new CFFFont(cffFile, properties); - // Mac want 1-byte per character strings while Windows want - // 2-bytes per character, so duplicate the names table - var stringsUnicode = []; - var i, ii, j, jj, str; - for (i = 0, ii = strings.length; i < ii; i++) { - str = proto[1][i] || strings[i]; + adjustWidths(properties); - var strBufUnicode = []; - for (j = 0, jj = str.length; j < jj; j++) { - strBufUnicode.push(string16(str.charCodeAt(j))); - } - stringsUnicode.push(strBufUnicode.join('')); - } + return this.convert(name, cff, properties); + } - var names = [strings, stringsUnicode]; - var platforms = ['\x00\x01', '\x00\x03']; - var encodings = ['\x00\x00', '\x00\x01']; - var languages = ['\x00\x00', '\x04\x09']; - - var namesRecordCount = strings.length * platforms.length; - var nameTable = - '\x00\x00' + // format - string16(namesRecordCount) + // Number of names Record - string16(namesRecordCount * 12 + 6); // Storage + delete tables['glyf']; + delete tables['loca']; + delete tables['fpgm']; + delete tables['prep']; + delete tables['cvt ']; + this.isOpenType = true; + } else { + if (!tables['loca']) { + error('Required "loca" table is not found'); + } + if (!tables['glyf']) { + warn('Required "glyf" table is not found -- trying to recover.'); + // Note: We use `sanitizeGlyphLocations` to add dummy glyf data below. + tables['glyf'] = { + tag: 'glyf', + data: new Uint8Array(0), + }; + } + this.isOpenType = false; + } - // Build the name records field - var strOffset = 0; - for (i = 0, ii = platforms.length; i < ii; i++) { - var strs = names[i]; - for (j = 0, jj = strs.length; j < jj; j++) { - str = strs[j]; - var nameRecord = - platforms[i] + // platform ID - encodings[i] + // encoding ID - languages[i] + // language ID - string16(j) + // name ID - string16(str.length) + - string16(strOffset); - nameTable += nameRecord; - strOffset += str.length; + if (!tables['maxp']) { + error('Required "maxp" table is not found'); } - } - nameTable += strings.join('') + stringsUnicode.join(''); - return nameTable; - } + font.pos = (font.start || 0) + tables['maxp'].offset; + var version = font.getInt32(); + var numGlyphs = font.getUint16(); + var maxFunctionDefs = 0; + if (version >= 0x00010000 && tables['maxp'].length >= 22) { + // maxZones can be invalid + font.pos += 8; + var maxZones = font.getUint16(); + if (maxZones > 2) { // reset to 2 if font has invalid maxZones + tables['maxp'].data[14] = 0; + tables['maxp'].data[15] = 2; + } + font.pos += 4; + maxFunctionDefs = font.getUint16(); + } - Font.prototype = { - name: null, - font: null, - mimetype: null, - encoding: null, - get renderer() { - var renderer = FontRendererFactory.create(this); - return shadow(this, 'renderer', renderer); - }, + var dupFirstEntry = false; + if (properties.type === 'CIDFontType2' && properties.toUnicode && + properties.toUnicode.get(0) > '\u0000') { + // oracle's defect (see 3427), duplicating first entry + dupFirstEntry = true; + numGlyphs++; + tables['maxp'].data[4] = numGlyphs >> 8; + tables['maxp'].data[5] = numGlyphs & 255; + } - exportData: function Font_exportData() { - // TODO remove enumerating of the properties, e.g. hardcode exact names. - var data = {}; - for (var i in this) { - if (this.hasOwnProperty(i)) { - data[i] = this[i]; - } + var hintsValid = sanitizeTTPrograms(tables['fpgm'], tables['prep'], + tables['cvt '], maxFunctionDefs); + if (!hintsValid) { + delete tables['fpgm']; + delete tables['prep']; + delete tables['cvt ']; } - return data; - }, - checkAndRepair: function Font_checkAndRepair(name, font, properties) { - function readTableEntry(file) { - var tag = bytesToString(file.getBytes(4)); + // Ensure the hmtx table contains the advance width and + // sidebearings information for numGlyphs in the maxp table + sanitizeMetrics(font, tables['hhea'], tables['hmtx'], numGlyphs); - var checksum = file.getInt32(); - var offset = file.getInt32() >>> 0; - var length = file.getInt32() >>> 0; + if (!tables['head']) { + error('Required "head" table is not found'); + } - // Read the table associated data - var previousPosition = file.pos; - file.pos = file.start ? file.start : 0; - file.skip(offset); - var data = file.getBytes(length); - file.pos = previousPosition; + sanitizeHead(tables['head'], numGlyphs, + isTrueType ? tables['loca'].length : 0); - if (tag === 'head') { - // clearing checksum adjustment - data[8] = data[9] = data[10] = data[11] = 0; - data[17] |= 0x20; //Set font optimized for cleartype flag - } + var missingGlyphs = Object.create(null); + if (isTrueType) { + var isGlyphLocationsLong = int16(tables['head'].data[50], + tables['head'].data[51]); + missingGlyphs = sanitizeGlyphLocations(tables['loca'], tables['glyf'], + numGlyphs, isGlyphLocationsLong, + hintsValid, dupFirstEntry); + } - return { - tag: tag, - checksum: checksum, - length: length, - offset: offset, - data: data - }; + if (!tables['hhea']) { + error('Required "hhea" table is not found'); } - function readOpenTypeHeader(ttf) { - return { - version: bytesToString(ttf.getBytes(4)), - numTables: ttf.getUint16(), - searchRange: ttf.getUint16(), - entrySelector: ttf.getUint16(), - rangeShift: ttf.getUint16() - }; + // Sanitizer reduces the glyph advanceWidth to the maxAdvanceWidth + // Sometimes it's 0. That needs to be fixed + if (tables['hhea'].data[10] === 0 && tables['hhea'].data[11] === 0) { + tables['hhea'].data[10] = 0xFF; + tables['hhea'].data[11] = 0xFF; } - /** - * Read the appropriate subtable from the cmap according to 9.6.6.4 from - * PDF spec - */ - function readCmapTable(cmap, font, isSymbolicFont, hasEncoding) { - if (!cmap) { - warn('No cmap table available.'); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; - } - var segment; - var start = (font.start ? font.start : 0) + cmap.offset; - font.pos = start; + // Extract some more font properties from the OpenType head and + // hhea tables; yMin and descent value are always negative. + var metricsOverride = { + unitsPerEm: int16(tables['head'].data[18], tables['head'].data[19]), + yMax: int16(tables['head'].data[42], tables['head'].data[43]), + yMin: int16(tables['head'].data[38], tables['head'].data[39]) - 0x10000, + ascent: int16(tables['hhea'].data[4], tables['hhea'].data[5]), + descent: int16(tables['hhea'].data[6], tables['hhea'].data[7]) - 0x10000 + }; - var version = font.getUint16(); - var numTables = font.getUint16(); + // PDF FontDescriptor metrics lie -- using data from actual font. + this.ascent = metricsOverride.ascent / metricsOverride.unitsPerEm; + this.descent = metricsOverride.descent / metricsOverride.unitsPerEm; - var potentialTable; - var canBreak = false; - // There's an order of preference in terms of which cmap subtable to - // use: - // - non-symbolic fonts the preference is a 3,1 table then a 1,0 table - // - symbolic fonts the preference is a 3,0 table then a 1,0 table - // The following takes advantage of the fact that the tables are sorted - // to work. - for (var i = 0; i < numTables; i++) { - var platformId = font.getUint16(); - var encodingId = font.getUint16(); - var offset = font.getInt32() >>> 0; - var useTable = false; + // The 'post' table has glyphs names. + if (tables['post']) { + var valid = readPostScriptTable(tables['post'], properties, numGlyphs); + if (!valid) { + tables['post'] = null; + } + } - if (platformId === 0 && encodingId === 0) { - useTable = true; - // Continue the loop since there still may be a higher priority - // table. - } else if (platformId === 1 && encodingId === 0) { - useTable = true; - // Continue the loop since there still may be a higher priority - // table. - } else if (platformId === 3 && encodingId === 1 && - ((!isSymbolicFont && hasEncoding) || !potentialTable)) { - useTable = true; - if (!isSymbolicFont) { - canBreak = true; - } - } else if (isSymbolicFont && platformId === 3 && encodingId === 0) { - useTable = true; - canBreak = true; - } + var charCodeToGlyphId = [], charCode; + var toUnicode = properties.toUnicode, widths = properties.widths; + var skipToUnicode = (toUnicode instanceof IdentityToUnicodeMap || + toUnicode.length === 0x10000); - if (useTable) { - potentialTable = { - platformId: platformId, - encodingId: encodingId, - offset: offset - }; - } - if (canBreak) { - break; - } + // Helper function to try to skip mapping of empty glyphs. + // Note: In some cases, just relying on the glyph data doesn't work, + // hence we also use a few heuristics to fix various PDF files. + function hasGlyph(glyphId, charCode, widthCode) { + if (!missingGlyphs[glyphId]) { + return true; } - - if (potentialTable) { - font.pos = start + potentialTable.offset; + if (!skipToUnicode && charCode >= 0 && toUnicode.has(charCode)) { + return true; } - if (!potentialTable || font.peekByte() === -1) { - warn('Could not find a preferred cmap table.'); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; + if (widths && widthCode >= 0 && isNum(widths[widthCode])) { + return true; } + return false; + } - var format = font.getUint16(); - var length = font.getUint16(); - var language = font.getUint16(); - - var hasShortCmap = false; - var mappings = []; - var j, glyphId; + if (properties.type === 'CIDFontType2') { + var cidToGidMap = properties.cidToGidMap || []; + var isCidToGidMapEmpty = cidToGidMap.length === 0; - // TODO(mack): refactor this cmap subtable reading logic out - if (format === 0) { - for (j = 0; j < 256; j++) { - var index = font.getByte(); - if (!index) { - continue; - } - mappings.push({ - charCode: j, - glyphId: index - }); - } - hasShortCmap = true; - } else if (format === 4) { - // re-creating the table in format 4 since the encoding - // might be changed - var segCount = (font.getUint16() >> 1); - font.getBytes(6); // skipping range fields - var segIndex, segments = []; - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments.push({ end: font.getUint16() }); - } - font.getUint16(); - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments[segIndex].start = font.getUint16(); + properties.cMap.forEach(function(charCode, cid) { + assert(cid <= 0xffff, 'Max size of CID is 65,535'); + var glyphId = -1; + if (isCidToGidMapEmpty) { + glyphId = charCode; + } else if (cidToGidMap[cid] !== undefined) { + glyphId = cidToGidMap[cid]; } - for (segIndex = 0; segIndex < segCount; segIndex++) { - segments[segIndex].delta = font.getUint16(); + if (glyphId >= 0 && glyphId < numGlyphs && + hasGlyph(glyphId, charCode, cid)) { + charCodeToGlyphId[charCode] = glyphId; } + }); + if (dupFirstEntry) { + charCodeToGlyphId[0] = numGlyphs - 1; + } + } else { + // Most of the following logic in this code branch is based on the + // 9.6.6.4 of the PDF spec. + var hasEncoding = + properties.differences.length > 0 || !!properties.baseEncodingName; + var cmapTable = + readCmapTable(tables['cmap'], font, this.isSymbolicFont, hasEncoding); + var cmapPlatformId = cmapTable.platformId; + var cmapEncodingId = cmapTable.encodingId; + var cmapMappings = cmapTable.mappings; + var cmapMappingsLength = cmapMappings.length; - var offsetsCount = 0; - for (segIndex = 0; segIndex < segCount; segIndex++) { - segment = segments[segIndex]; - var rangeOffset = font.getUint16(); - if (!rangeOffset) { - segment.offsetIndex = -1; - continue; - } - - var offsetIndex = (rangeOffset >> 1) - (segCount - segIndex); - segment.offsetIndex = offsetIndex; - offsetsCount = Math.max(offsetsCount, offsetIndex + - segment.end - segment.start + 1); - } + // The spec seems to imply that if the font is symbolic the encoding + // should be ignored, this doesn't appear to work for 'preistabelle.pdf' + // where the the font is symbolic and it has an encoding. + if (hasEncoding && + (cmapPlatformId === 3 && cmapEncodingId === 1 || + cmapPlatformId === 1 && cmapEncodingId === 0) || + (cmapPlatformId === -1 && cmapEncodingId === -1 && // Temporary hack + !!getEncoding(properties.baseEncodingName))) { // Temporary hack + // When no preferred cmap table was found and |baseEncodingName| is + // one of the predefined encodings, we seem to obtain a better + // |charCodeToGlyphId| map from the code below (fixes bug 1057544). + // TODO: Note that this is a hack which should be removed as soon as + // we have proper support for more exotic cmap tables. - var offsets = []; - for (j = 0; j < offsetsCount; j++) { - offsets.push(font.getUint16()); + var baseEncoding = []; + if (properties.baseEncodingName === 'MacRomanEncoding' || + properties.baseEncodingName === 'WinAnsiEncoding') { + baseEncoding = getEncoding(properties.baseEncodingName); } + var glyphsUnicodeMap = getGlyphsUnicode(); + for (charCode = 0; charCode < 256; charCode++) { + var glyphName; + if (this.differences && charCode in this.differences) { + glyphName = this.differences[charCode]; + } else if (charCode in baseEncoding && + baseEncoding[charCode] !== '') { + glyphName = baseEncoding[charCode]; + } else { + glyphName = StandardEncoding[charCode]; + } + if (!glyphName) { + continue; + } + var unicodeOrCharCode, isUnicode = false; + if (cmapPlatformId === 3 && cmapEncodingId === 1) { + unicodeOrCharCode = glyphsUnicodeMap[glyphName]; + isUnicode = true; + } else if (cmapPlatformId === 1 && cmapEncodingId === 0) { + // TODO: the encoding needs to be updated with mac os table. + unicodeOrCharCode = MacRomanEncoding.indexOf(glyphName); + } - for (segIndex = 0; segIndex < segCount; segIndex++) { - segment = segments[segIndex]; - start = segment.start; - var end = segment.end; - var delta = segment.delta; - offsetIndex = segment.offsetIndex; - - for (j = start; j <= end; j++) { - if (j === 0xFFFF) { + var found = false; + for (i = 0; i < cmapMappingsLength; ++i) { + if (cmapMappings[i].charCode !== unicodeOrCharCode) { continue; } - - glyphId = (offsetIndex < 0 ? - j : offsets[offsetIndex + j - start]); - glyphId = (glyphId + delta) & 0xFFFF; - if (glyphId === 0) { - continue; + var code = isUnicode ? charCode : unicodeOrCharCode; + if (hasGlyph(cmapMappings[i].glyphId, code, -1)) { + charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; + found = true; + break; + } + } + if (!found && properties.glyphNames) { + // Try to map using the post table. + var glyphId = properties.glyphNames.indexOf(glyphName); + if (glyphId > 0 && hasGlyph(glyphId, -1, -1)) { + charCodeToGlyphId[charCode] = glyphId; + found = true; } - mappings.push({ - charCode: j, - glyphId: glyphId - }); + } + if (!found) { + charCodeToGlyphId[charCode] = 0; // notdef } } - } else if (format === 6) { - // Format 6 is a 2-bytes dense mapping, which means the font data - // lives glue together even if they are pretty far in the unicode - // table. (This looks weird, so I can have missed something), this - // works on Linux but seems to fails on Mac so let's rewrite the - // cmap table to a 3-1-4 style - var firstCode = font.getUint16(); - var entryCount = font.getUint16(); - - for (j = 0; j < entryCount; j++) { - glyphId = font.getUint16(); - var charCode = firstCode + j; - - mappings.push({ - charCode: charCode, - glyphId: glyphId - }); + } else if (cmapPlatformId === 0 && cmapEncodingId === 0) { + // Default Unicode semantics, use the charcodes as is. + for (i = 0; i < cmapMappingsLength; ++i) { + charCodeToGlyphId[cmapMappings[i].charCode] = + cmapMappings[i].glyphId; } } else { - warn('cmap table has unsupported format: ' + format); - return { - platformId: -1, - encodingId: -1, - mappings: [], - hasShortCmap: false - }; - } - - // removing duplicate entries - mappings.sort(function (a, b) { - return a.charCode - b.charCode; - }); - for (i = 1; i < mappings.length; i++) { - if (mappings[i - 1].charCode === mappings[i].charCode) { - mappings.splice(i, 1); - i--; + // For (3, 0) cmap tables: + // The charcode key being stored in charCodeToGlyphId is the lower + // byte of the two-byte charcodes of the cmap table since according to + // the spec: 'each byte from the string shall be prepended with the + // high byte of the range [of charcodes in the cmap table], to form + // a two-byte character, which shall be used to select the + // associated glyph description from the subtable'. + // + // For (1, 0) cmap tables: + // 'single bytes from the string shall be used to look up the + // associated glyph descriptions from the subtable'. This means + // charcodes in the cmap will be single bytes, so no-op since + // glyph.charCode & 0xFF === glyph.charCode + for (i = 0; i < cmapMappingsLength; ++i) { + charCode = cmapMappings[i].charCode & 0xFF; + charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; } } + } - return { - platformId: potentialTable.platformId, - encodingId: potentialTable.encodingId, - mappings: mappings, - hasShortCmap: hasShortCmap - }; + if (charCodeToGlyphId.length === 0) { + // defines at least one glyph + charCodeToGlyphId[0] = 0; } - function sanitizeMetrics(font, header, metrics, numGlyphs) { - if (!header) { - if (metrics) { - metrics.data = null; - } - return; - } + // Converting glyphs and ids into font's cmap table + var newMapping = adjustMapping(charCodeToGlyphId, properties); + this.toFontChar = newMapping.toFontChar; + tables['cmap'] = { + tag: 'cmap', + data: createCmapTable(newMapping.charCodeToGlyphId, numGlyphs) + }; - font.pos = (font.start ? font.start : 0) + header.offset; - font.pos += header.length - 2; - var numOfMetrics = font.getUint16(); + if (!tables['OS/2'] || !validateOS2Table(tables['OS/2'])) { + tables['OS/2'] = { + tag: 'OS/2', + data: createOS2Table(properties, newMapping.charCodeToGlyphId, + metricsOverride) + }; + } - if (numOfMetrics > numGlyphs) { - info('The numOfMetrics (' + numOfMetrics + ') should not be ' + - 'greater than the numGlyphs (' + numGlyphs + ')'); - // Reduce numOfMetrics if it is greater than numGlyphs - numOfMetrics = numGlyphs; - header.data[34] = (numOfMetrics & 0xff00) >> 8; - header.data[35] = numOfMetrics & 0x00ff; - } - - var numOfSidebearings = numGlyphs - numOfMetrics; - var numMissing = numOfSidebearings - - ((metrics.length - numOfMetrics * 4) >> 1); - - if (numMissing > 0) { - // For each missing glyph, we set both the width and lsb to 0 (zero). - // Since we need to add two properties for each glyph, this explains - // the use of |numMissing * 2| when initializing the typed array. - var entries = new Uint8Array(metrics.length + numMissing * 2); - entries.set(metrics.data); - metrics.data = entries; - } + // Rewrite the 'post' table if needed + if (!tables['post']) { + tables['post'] = { + tag: 'post', + data: createPostTable(properties) + }; } - function sanitizeGlyph(source, sourceStart, sourceEnd, dest, destStart, - hintsValid) { - if (sourceEnd - sourceStart <= 12) { - // glyph with data less than 12 is invalid one - return 0; - } - var glyf = source.subarray(sourceStart, sourceEnd); - var contoursCount = (glyf[0] << 8) | glyf[1]; - if (contoursCount & 0x8000) { - // complex glyph, writing as is - dest.set(glyf, destStart); - return glyf.length; - } - - var i, j = 10, flagsCount = 0; - for (i = 0; i < contoursCount; i++) { - var endPoint = (glyf[j] << 8) | glyf[j + 1]; - flagsCount = endPoint + 1; - j += 2; - } - // skipping instructions - var instructionsStart = j; - var instructionsLength = (glyf[j] << 8) | glyf[j + 1]; - j += 2 + instructionsLength; - var instructionsEnd = j; - // validating flags - var coordinatesLength = 0; - for (i = 0; i < flagsCount; i++) { - var flag = glyf[j++]; - if (flag & 0xC0) { - // reserved flags must be zero, cleaning up - glyf[j - 1] = flag & 0x3F; - } - var xyLength = ((flag & 2) ? 1 : (flag & 16) ? 0 : 2) + - ((flag & 4) ? 1 : (flag & 32) ? 0 : 2); - coordinatesLength += xyLength; - if (flag & 8) { - var repeat = glyf[j++]; - i += repeat; - coordinatesLength += repeat * xyLength; - } - } - // glyph without coordinates will be rejected - if (coordinatesLength === 0) { - return 0; - } - var glyphDataLength = j + coordinatesLength; - if (glyphDataLength > glyf.length) { - // not enough data for coordinates - return 0; - } - if (!hintsValid && instructionsLength > 0) { - dest.set(glyf.subarray(0, instructionsStart), destStart); - dest.set([0, 0], destStart + instructionsStart); - dest.set(glyf.subarray(instructionsEnd, glyphDataLength), - destStart + instructionsStart + 2); - glyphDataLength -= instructionsLength; - if (glyf.length - glyphDataLength > 3) { - glyphDataLength = (glyphDataLength + 3) & ~3; - } - return glyphDataLength; - } - if (glyf.length - glyphDataLength > 3) { - // truncating and aligning to 4 bytes the long glyph data - glyphDataLength = (glyphDataLength + 3) & ~3; - dest.set(glyf.subarray(0, glyphDataLength), destStart); - return glyphDataLength; + if (!isTrueType) { + try { + // Trying to repair CFF file + cffFile = new Stream(tables['CFF '].data); + var parser = new CFFParser(cffFile, properties); + cff = parser.parse(); + var compiler = new CFFCompiler(cff); + tables['CFF '].data = compiler.compile(); + } catch (e) { + warn('Failed to compile font ' + properties.loadedName); } - // glyph data is fine - dest.set(glyf, destStart); - return glyf.length; } - function sanitizeHead(head, numGlyphs, locaLength) { - var data = head.data; + // Re-creating 'name' table + if (!tables['name']) { + tables['name'] = { + tag: 'name', + data: createNameTable(this.name) + }; + } else { + // ... using existing 'name' table as prototype + var namePrototype = readNameTable(tables['name']); + tables['name'].data = createNameTable(name, namePrototype); + } - // Validate version: - // Should always be 0x00010000 - var version = int32(data[0], data[1], data[2], data[3]); - if (version >> 16 !== 1) { - info('Attempting to fix invalid version in head table: ' + version); - data[0] = 0; - data[1] = 1; - data[2] = 0; - data[3] = 0; - } + var builder = new OpenTypeFileBuilder(header.version); + for (var tableTag in tables) { + builder.addTable(tableTag, tables[tableTag].data); + } + return builder.toArray(); + }, - var indexToLocFormat = int16(data[50], data[51]); - if (indexToLocFormat < 0 || indexToLocFormat > 1) { - info('Attempting to fix invalid indexToLocFormat in head table: ' + - indexToLocFormat); + convert: function Font_convert(fontName, font, properties) { + // TODO: Check the charstring widths to determine this. + properties.fixedPitch = false; - // The value of indexToLocFormat should be 0 if the loca table - // consists of short offsets, and should be 1 if the loca table - // consists of long offsets. - // - // The number of entries in the loca table should be numGlyphs + 1. - // - // Using this information, we can work backwards to deduce if the - // size of each offset in the loca table, and thus figure out the - // appropriate value for indexToLocFormat. + var mapping = font.getGlyphMapping(properties); + var newMapping = adjustMapping(mapping, properties); + this.toFontChar = newMapping.toFontChar; + var numGlyphs = font.numGlyphs; - var numGlyphsPlusOne = numGlyphs + 1; - if (locaLength === numGlyphsPlusOne << 1) { - // 0x0000 indicates the loca table consists of short offsets - data[50] = 0; - data[51] = 0; - } else if (locaLength === numGlyphsPlusOne << 2) { - // 0x0001 indicates the loca table consists of long offsets - data[50] = 0; - data[51] = 1; - } else { - warn('Could not fix indexToLocFormat: ' + indexToLocFormat); + function getCharCodes(charCodeToGlyphId, glyphId) { + var charCodes = null; + for (var charCode in charCodeToGlyphId) { + if (glyphId === charCodeToGlyphId[charCode]) { + if (!charCodes) { + charCodes = []; + } + charCodes.push(charCode | 0); } } + return charCodes; } - function sanitizeGlyphLocations(loca, glyf, numGlyphs, - isGlyphLocationsLong, hintsValid, - dupFirstEntry) { - var itemSize, itemDecode, itemEncode; - if (isGlyphLocationsLong) { - itemSize = 4; - itemDecode = function fontItemDecodeLong(data, offset) { - return (data[offset] << 24) | (data[offset + 1] << 16) | - (data[offset + 2] << 8) | data[offset + 3]; - }; - itemEncode = function fontItemEncodeLong(data, offset, value) { - data[offset] = (value >>> 24) & 0xFF; - data[offset + 1] = (value >> 16) & 0xFF; - data[offset + 2] = (value >> 8) & 0xFF; - data[offset + 3] = value & 0xFF; - }; - } else { - itemSize = 2; - itemDecode = function fontItemDecode(data, offset) { - return (data[offset] << 9) | (data[offset + 1] << 1); - }; - itemEncode = function fontItemEncode(data, offset, value) { - data[offset] = (value >> 9) & 0xFF; - data[offset + 1] = (value >> 1) & 0xFF; - }; - } - var locaData = loca.data; - var locaDataSize = itemSize * (1 + numGlyphs); - // is loca.data too short or long? - if (locaData.length !== locaDataSize) { - locaData = new Uint8Array(locaDataSize); - locaData.set(loca.data.subarray(0, locaDataSize)); - loca.data = locaData; - } - // removing the invalid glyphs - var oldGlyfData = glyf.data; - var oldGlyfDataLength = oldGlyfData.length; - var newGlyfData = new Uint8Array(oldGlyfDataLength); - var startOffset = itemDecode(locaData, 0); - var writeOffset = 0; - var missingGlyphData = Object.create(null); - itemEncode(locaData, 0, writeOffset); - var i, j; - for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { - var endOffset = itemDecode(locaData, j); - if (endOffset > oldGlyfDataLength && - ((oldGlyfDataLength + 3) & ~3) === endOffset) { - // Aspose breaks fonts by aligning the glyphs to the qword, but not - // the glyf table size, which makes last glyph out of range. - endOffset = oldGlyfDataLength; - } - if (endOffset > oldGlyfDataLength) { - // glyph end offset points outside glyf data, rejecting the glyph - itemEncode(locaData, j, writeOffset); - startOffset = endOffset; - continue; - } - - if (startOffset === endOffset) { - missingGlyphData[i] = true; + function createCharCode(charCodeToGlyphId, glyphId) { + for (var charCode in charCodeToGlyphId) { + if (glyphId === charCodeToGlyphId[charCode]) { + return charCode | 0; } - - var newLength = sanitizeGlyph(oldGlyfData, startOffset, endOffset, - newGlyfData, writeOffset, hintsValid); - writeOffset += newLength; - itemEncode(locaData, j, writeOffset); - startOffset = endOffset; } + newMapping.charCodeToGlyphId[newMapping.nextAvailableFontCharCode] = + glyphId; + return newMapping.nextAvailableFontCharCode++; + } - if (writeOffset === 0) { - // glyf table cannot be empty -- redoing the glyf and loca tables - // to have single glyph with one point - var simpleGlyph = new Uint8Array( - [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 49, 0]); - for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) { - itemEncode(locaData, j, simpleGlyph.length); + var seacs = font.seacs; + if (SEAC_ANALYSIS_ENABLED && seacs && seacs.length) { + var matrix = properties.fontMatrix || FONT_IDENTITY_MATRIX; + var charset = font.getCharset(); + var seacMap = Object.create(null); + for (var glyphId in seacs) { + glyphId |= 0; + var seac = seacs[glyphId]; + var baseGlyphName = StandardEncoding[seac[2]]; + var accentGlyphName = StandardEncoding[seac[3]]; + var baseGlyphId = charset.indexOf(baseGlyphName); + var accentGlyphId = charset.indexOf(accentGlyphName); + if (baseGlyphId < 0 || accentGlyphId < 0) { + continue; } - glyf.data = simpleGlyph; - return missingGlyphData; - } + var accentOffset = { + x: seac[0] * matrix[0] + seac[1] * matrix[2] + matrix[4], + y: seac[0] * matrix[1] + seac[1] * matrix[3] + matrix[5] + }; - if (dupFirstEntry) { - var firstEntryLength = itemDecode(locaData, itemSize); - if (newGlyfData.length > firstEntryLength + writeOffset) { - glyf.data = newGlyfData.subarray(0, firstEntryLength + writeOffset); - } else { - glyf.data = new Uint8Array(firstEntryLength + writeOffset); - glyf.data.set(newGlyfData.subarray(0, writeOffset)); + var charCodes = getCharCodes(mapping, glyphId); + if (!charCodes) { + // There's no point in mapping it if the char code was never mapped + // to begin with. + continue; + } + for (var i = 0, ii = charCodes.length; i < ii; i++) { + var charCode = charCodes[i]; + // Find a fontCharCode that maps to the base and accent glyphs. + // If one doesn't exists, create it. + var charCodeToGlyphId = newMapping.charCodeToGlyphId; + var baseFontCharCode = createCharCode(charCodeToGlyphId, + baseGlyphId); + var accentFontCharCode = createCharCode(charCodeToGlyphId, + accentGlyphId); + seacMap[charCode] = { + baseFontCharCode: baseFontCharCode, + accentFontCharCode: accentFontCharCode, + accentOffset: accentOffset + }; } - glyf.data.set(newGlyfData.subarray(0, firstEntryLength), writeOffset); - itemEncode(loca.data, locaData.length - itemSize, - writeOffset + firstEntryLength); - } else { - glyf.data = newGlyfData.subarray(0, writeOffset); } - return missingGlyphData; + properties.seacMap = seacMap; } - function readPostScriptTable(post, properties, maxpNumGlyphs) { - var start = (font.start ? font.start : 0) + post.offset; - font.pos = start; + var unitsPerEm = 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]; - var length = post.length, end = start + length; - var version = font.getInt32(); - // skip rest to the tables - font.getBytes(28); + var builder = new OpenTypeFileBuilder('\x4F\x54\x54\x4F'); + // PostScript Font Program + builder.addTable('CFF ', font.data); + // OS/2 and Windows Specific metrics + builder.addTable('OS/2', createOS2Table(properties, + newMapping.charCodeToGlyphId)); + // Character to glyphs mapping + builder.addTable('cmap', createCmapTable(newMapping.charCodeToGlyphId, + numGlyphs)); + // Font header + builder.addTable('head', + '\x00\x01\x00\x00' + // Version number + '\x00\x00\x10\x00' + // fontRevision + '\x00\x00\x00\x00' + // checksumAdjustement + '\x5F\x0F\x3C\xF5' + // magicNumber + '\x00\x00' + // Flags + safeString16(unitsPerEm) + // unitsPerEM + '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // creation date + '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // modifification date + '\x00\x00' + // xMin + safeString16(properties.descent) + // yMin + '\x0F\xFF' + // xMax + safeString16(properties.ascent) + // yMax + string16(properties.italicAngle ? 2 : 0) + // macStyle + '\x00\x11' + // lowestRecPPEM + '\x00\x00' + // fontDirectionHint + '\x00\x00' + // indexToLocFormat + '\x00\x00'); // glyphDataFormat - var glyphNames; - var valid = true; - var i; + // Horizontal header + builder.addTable('hhea', + '\x00\x01\x00\x00' + // Version number + safeString16(properties.ascent) + // Typographic Ascent + safeString16(properties.descent) + // Typographic Descent + '\x00\x00' + // Line Gap + '\xFF\xFF' + // advanceWidthMax + '\x00\x00' + // minLeftSidebearing + '\x00\x00' + // minRightSidebearing + '\x00\x00' + // xMaxExtent + safeString16(properties.capHeight) + // caretSlopeRise + safeString16(Math.tan(properties.italicAngle) * + properties.xHeight) + // caretSlopeRun + '\x00\x00' + // caretOffset + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // -reserved- + '\x00\x00' + // metricDataFormat + string16(numGlyphs)); // Number of HMetrics - switch (version) { - case 0x00010000: - glyphNames = MacStandardGlyphOrdering; - break; - case 0x00020000: - var numGlyphs = font.getUint16(); - if (numGlyphs !== maxpNumGlyphs) { - valid = false; - break; - } - var glyphNameIndexes = []; - for (i = 0; i < numGlyphs; ++i) { - var index = font.getUint16(); - if (index >= 32768) { - valid = false; - break; - } - glyphNameIndexes.push(index); - } - if (!valid) { - break; - } - var customNames = []; - var strBuf = []; - while (font.pos < end) { - var stringLength = font.getByte(); - strBuf.length = stringLength; - for (i = 0; i < stringLength; ++i) { - strBuf[i] = String.fromCharCode(font.getByte()); - } - customNames.push(strBuf.join('')); - } - glyphNames = []; - for (i = 0; i < numGlyphs; ++i) { - var j = glyphNameIndexes[i]; - if (j < 258) { - glyphNames.push(MacStandardGlyphOrdering[j]); - continue; - } - glyphNames.push(customNames[j - 258]); - } - break; - case 0x00030000: - break; - default: - warn('Unknown/unsupported post table version ' + version); - valid = false; - if (properties.defaultEncoding) { - glyphNames = properties.defaultEncoding; + // Horizontal metrics + builder.addTable('hmtx', (function fontFieldsHmtx() { + var charstrings = font.charstrings; + var cffWidths = font.cff ? font.cff.widths : null; + var hmtx = '\x00\x00\x00\x00'; // Fake .notdef + for (var i = 1, ii = numGlyphs; i < ii; i++) { + var width = 0; + if (charstrings) { + var charstring = charstrings[i - 1]; + width = 'width' in charstring ? charstring.width : 0; + } else if (cffWidths) { + width = Math.ceil(cffWidths[i] || 0); } - break; - } - properties.glyphNames = glyphNames; - return valid; - } + hmtx += string16(width) + string16(0); + } + return hmtx; + })()); - function readNameTable(nameTable) { - var start = (font.start ? font.start : 0) + nameTable.offset; - font.pos = start; + // Maximum profile + builder.addTable('maxp', + '\x00\x00\x50\x00' + // Version number + string16(numGlyphs)); // Num of glyphs - var names = [[], []]; - var length = nameTable.length, end = start + length; - var format = font.getUint16(); - var FORMAT_0_HEADER_LENGTH = 6; - if (format !== 0 || length < FORMAT_0_HEADER_LENGTH) { - // unsupported name table format or table "too" small - return names; - } - var numRecords = font.getUint16(); - var stringsStart = font.getUint16(); - var records = []; - var NAME_RECORD_LENGTH = 12; - var i, ii; + // Naming tables + builder.addTable('name', createNameTable(fontName)); - for (i = 0; i < numRecords && - font.pos + NAME_RECORD_LENGTH <= end; i++) { - var r = { - platform: font.getUint16(), - encoding: font.getUint16(), - language: font.getUint16(), - name: font.getUint16(), - length: font.getUint16(), - offset: font.getUint16() - }; - // using only Macintosh and Windows platform/encoding names - if ((r.platform === 1 && r.encoding === 0 && r.language === 0) || - (r.platform === 3 && r.encoding === 1 && r.language === 0x409)) { - records.push(r); - } - } - for (i = 0, ii = records.length; i < ii; i++) { - var record = records[i]; - var pos = start + stringsStart + record.offset; - if (pos + record.length > end) { - continue; // outside of name table, ignoring - } - font.pos = pos; - var nameIndex = record.name; - if (record.encoding) { - // unicode - var str = ''; - for (var j = 0, jj = record.length; j < jj; j += 2) { - str += String.fromCharCode(font.getUint16()); - } - names[1][nameIndex] = str; - } else { - names[0][nameIndex] = bytesToString(font.getBytes(record.length)); - } - } - return names; - } + // PostScript informations + builder.addTable('post', createPostTable(properties)); - var TTOpsStackDeltas = [ - 0, 0, 0, 0, 0, 0, 0, 0, -2, -2, -2, -2, 0, 0, -2, -5, - -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, -1, -1, - 1, -1, -999, 0, 1, 0, -1, -2, 0, -1, -2, -1, -1, 0, -1, -1, - 0, 0, -999, -999, -1, -1, -1, -1, -2, -999, -2, -2, -999, 0, -2, -2, - 0, 0, -2, 0, -2, 0, 0, 0, -2, -1, -1, 1, 1, 0, 0, -1, - -1, -1, -1, -1, -1, -1, 0, 0, -1, 0, -1, -1, 0, -999, -1, -1, - -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - -2, -999, -999, -999, -999, -999, -1, -1, -2, -2, 0, 0, 0, 0, -1, -1, - -999, -2, -2, 0, 0, -1, -2, -2, 0, 0, 0, -1, -1, -1, -2]; - // 0xC0-DF == -1 and 0xE0-FF == -2 + return builder.toArray(); + }, - function sanitizeTTProgram(table, ttContext) { - var data = table.data; - var i = 0, j, n, b, funcId, pc, lastEndf = 0, lastDeff = 0; - var stack = []; - var callstack = []; - var functionsCalled = []; - var tooComplexToFollowFunctions = - ttContext.tooComplexToFollowFunctions; - var inFDEF = false, ifLevel = 0, inELSE = 0; - for (var ii = data.length; i < ii;) { - var op = data[i++]; - // The TrueType instruction set docs can be found at - // https://developer.apple.com/fonts/TTRefMan/RM05/Chap5.html - if (op === 0x40) { // NPUSHB - pushes n bytes - n = data[i++]; - if (inFDEF || inELSE) { - i += n; - } else { - for (j = 0; j < n; j++) { - stack.push(data[i++]); - } - } - } else if (op === 0x41) { // NPUSHW - pushes n words - n = data[i++]; - if (inFDEF || inELSE) { - i += n * 2; - } else { - for (j = 0; j < n; j++) { - b = data[i++]; - stack.push((b << 8) | data[i++]); - } - } - } else if ((op & 0xF8) === 0xB0) { // PUSHB - pushes bytes - n = op - 0xB0 + 1; - if (inFDEF || inELSE) { - i += n; - } else { - for (j = 0; j < n; j++) { - stack.push(data[i++]); - } - } - } else if ((op & 0xF8) === 0xB8) { // PUSHW - pushes words - n = op - 0xB8 + 1; - if (inFDEF || inELSE) { - i += n * 2; - } else { - for (j = 0; j < n; j++) { - b = data[i++]; - stack.push((b << 8) | data[i++]); - } - } - } else if (op === 0x2B && !tooComplexToFollowFunctions) { // CALL - if (!inFDEF && !inELSE) { - // collecting inforamtion about which functions are used - funcId = stack[stack.length - 1]; - ttContext.functionsUsed[funcId] = true; - if (funcId in ttContext.functionsStackDeltas) { - stack.length += ttContext.functionsStackDeltas[funcId]; - } else if (funcId in ttContext.functionsDefined && - functionsCalled.indexOf(funcId) < 0) { - callstack.push({data: data, i: i, stackTop: stack.length - 1}); - functionsCalled.push(funcId); - pc = ttContext.functionsDefined[funcId]; - if (!pc) { - warn('TT: CALL non-existent function'); - ttContext.hintsValid = false; - return; + /** + * Builds a char code to unicode map based on section 9.10 of the spec. + * @param {Object} properties Font properties object. + * @return {Object} A ToUnicodeMap object. + */ + buildToUnicode: function Font_buildToUnicode(properties) { + // Section 9.10.2 Mapping Character Codes to Unicode Values + if (properties.toUnicode && properties.toUnicode.length !== 0) { + return properties.toUnicode; + } + // According to the spec if the font is a simple font we should only map + // to unicode if the base encoding is MacRoman, MacExpert, or WinAnsi or + // the differences array only contains adobe standard or symbol set names, + // in pratice it seems better to always try to create a toUnicode + // map based of the default encoding. + var toUnicode, charcode; + if (!properties.composite /* is simple font */) { + toUnicode = []; + var encoding = properties.defaultEncoding.slice(); + var baseEncodingName = properties.baseEncodingName; + // Merge in the differences array. + var differences = properties.differences; + for (charcode in differences) { + encoding[charcode] = differences[charcode]; + } + var glyphsUnicodeMap = getGlyphsUnicode(); + for (charcode in encoding) { + // a) Map the character code to a character name. + var glyphName = encoding[charcode]; + // b) Look up the character name in the Adobe Glyph List (see the + // Bibliography) to obtain the corresponding Unicode value. + if (glyphName === '') { + continue; + } else if (glyphsUnicodeMap[glyphName] === undefined) { + // (undocumented) c) Few heuristics to recognize unknown glyphs + // NOTE: Adobe Reader does not do this step, but OSX Preview does + var code = 0; + switch (glyphName[0]) { + case 'G': // Gxx glyph + if (glyphName.length === 3) { + code = parseInt(glyphName.substr(1), 16); } - data = pc.data; - i = pc.i; - } - } - } else if (op === 0x2C && !tooComplexToFollowFunctions) { // FDEF - if (inFDEF || inELSE) { - warn('TT: nested FDEFs not allowed'); - tooComplexToFollowFunctions = true; - } - inFDEF = true; - // collecting inforamtion about which functions are defined - lastDeff = i; - funcId = stack.pop(); - ttContext.functionsDefined[funcId] = {data: data, i: i}; - } else if (op === 0x2D) { // ENDF - end of function - if (inFDEF) { - inFDEF = false; - lastEndf = i; - } else { - pc = callstack.pop(); - if (!pc) { - warn('TT: ENDF bad stack'); - ttContext.hintsValid = false; - return; - } - funcId = functionsCalled.pop(); - data = pc.data; - i = pc.i; - ttContext.functionsStackDeltas[funcId] = - stack.length - pc.stackTop; - } - } else if (op === 0x89) { // IDEF - instruction definition - if (inFDEF || inELSE) { - warn('TT: nested IDEFs not allowed'); - tooComplexToFollowFunctions = true; - } - inFDEF = true; - // recording it as a function to track ENDF - lastDeff = i; - } else if (op === 0x58) { // IF - ++ifLevel; - } else if (op === 0x1B) { // ELSE - inELSE = ifLevel; - } else if (op === 0x59) { // EIF - if (inELSE === ifLevel) { - inELSE = 0; - } - --ifLevel; - } else if (op === 0x1C) { // JMPR - if (!inFDEF && !inELSE) { - var offset = stack[stack.length - 1]; - // only jumping forward to prevent infinite loop - if (offset > 0) { - i += offset - 1; - } + break; + case 'g': // g00xx glyph + if (glyphName.length === 5) { + code = parseInt(glyphName.substr(1), 16); + } + break; + case 'C': // Cddd glyph + case 'c': // cddd glyph + if (glyphName.length >= 3) { + code = +glyphName.substr(1); + } + break; } - } - // Adjusting stack not extactly, but just enough to get function id - if (!inFDEF && !inELSE) { - var stackDelta = op <= 0x8E ? TTOpsStackDeltas[op] : - op >= 0xC0 && op <= 0xDF ? -1 : op >= 0xE0 ? -2 : 0; - if (op >= 0x71 && op <= 0x75) { - n = stack.pop(); - if (n === n) { - stackDelta = -n * 2; + if (code) { + // If |baseEncodingName| is one the predefined encodings, + // and |code| equals |charcode|, using the glyph defined in the + // baseEncoding seems to yield a better |toUnicode| mapping + // (fixes issue 5070). + if (baseEncodingName && code === +charcode) { + var baseEncoding = getEncoding(baseEncodingName); + if (baseEncoding && (glyphName = baseEncoding[charcode])) { + toUnicode[charcode] = + String.fromCharCode(glyphsUnicodeMap[glyphName]); + continue; + } } + toUnicode[charcode] = String.fromCharCode(code); } - while (stackDelta < 0 && stack.length > 0) { - stack.pop(); - stackDelta++; - } - while (stackDelta > 0) { - stack.push(NaN); // pushing any number into stack - stackDelta--; - } + continue; } + toUnicode[charcode] = + String.fromCharCode(glyphsUnicodeMap[glyphName]); } - ttContext.tooComplexToFollowFunctions = tooComplexToFollowFunctions; - var content = [data]; - if (i > data.length) { - content.push(new Uint8Array(i - data.length)); - } - if (lastDeff > lastEndf) { - warn('TT: complementing a missing function tail'); - // new function definition started, but not finished - // complete function by [CLEAR, ENDF] - content.push(new Uint8Array([0x22, 0x2D])); - } - foldTTTable(table, content); + return new ToUnicodeMap(toUnicode); } - - function checkInvalidFunctions(ttContext, maxFunctionDefs) { - if (ttContext.tooComplexToFollowFunctions) { - return; - } - if (ttContext.functionsDefined.length > maxFunctionDefs) { - warn('TT: more functions defined than expected'); - ttContext.hintsValid = false; - return; - } - for (var j = 0, jj = ttContext.functionsUsed.length; j < jj; j++) { - if (j > maxFunctionDefs) { - warn('TT: invalid function id: ' + j); - ttContext.hintsValid = false; - return; - } - if (ttContext.functionsUsed[j] && !ttContext.functionsDefined[j]) { - warn('TT: undefined function: ' + j); - ttContext.hintsValid = false; - return; + // If the font is a composite font that uses one of the predefined CMaps + // listed in Table 118 (except Identity–H and Identity–V) or whose + // descendant CIDFont uses the Adobe-GB1, Adobe-CNS1, Adobe-Japan1, or + // Adobe-Korea1 character collection: + if (properties.composite && ( + (properties.cMap.builtInCMap && + !(properties.cMap instanceof IdentityCMap)) || + (properties.cidSystemInfo.registry === 'Adobe' && + (properties.cidSystemInfo.ordering === 'GB1' || + properties.cidSystemInfo.ordering === 'CNS1' || + properties.cidSystemInfo.ordering === 'Japan1' || + properties.cidSystemInfo.ordering === 'Korea1')))) { + // Then: + // a) Map the character code to a character identifier (CID) according + // to the font’s CMap. + // b) Obtain the registry and ordering of the character collection used + // by the font’s CMap (for example, Adobe and Japan1) from its + // CIDSystemInfo dictionary. + var registry = properties.cidSystemInfo.registry; + var ordering = properties.cidSystemInfo.ordering; + // c) Construct a second CMap name by concatenating the registry and + // ordering obtained in step (b) in the format registry–ordering–UCS2 + // (for example, Adobe–Japan1–UCS2). + var ucs2CMapName = new Name(registry + '-' + ordering + '-UCS2'); + // d) Obtain the CMap with the name constructed in step (c) (available + // from the ASN Web site; see the Bibliography). + var ucs2CMap = CMapFactory.create(ucs2CMapName, + { url: PDFJS.cMapUrl, packed: PDFJS.cMapPacked }, null); + var cMap = properties.cMap; + toUnicode = []; + cMap.forEach(function(charcode, cid) { + assert(cid <= 0xffff, 'Max size of CID is 65,535'); + // e) Map the CID obtained in step (a) according to the CMap obtained + // in step (d), producing a Unicode value. + var ucs2 = ucs2CMap.lookup(cid); + if (ucs2) { + toUnicode[charcode] = + String.fromCharCode((ucs2.charCodeAt(0) << 8) + + ucs2.charCodeAt(1)); } - } + }); + return new ToUnicodeMap(toUnicode); } - function foldTTTable(table, content) { - if (content.length > 1) { - // concatenating the content items - var newLength = 0; - var j, jj; - for (j = 0, jj = content.length; j < jj; j++) { - newLength += content[j].length; - } - newLength = (newLength + 3) & ~3; - var result = new Uint8Array(newLength); - var pos = 0; - for (j = 0, jj = content.length; j < jj; j++) { - result.set(content[j], pos); - pos += content[j].length; - } - table.data = result; - table.length = newLength; - } + // The viewer's choice, just use an identity map. + return new IdentityToUnicodeMap(properties.firstChar, + properties.lastChar); + }, + + get spaceWidth() { + if ('_shadowWidth' in this) { + return this._shadowWidth; } - function sanitizeTTPrograms(fpgm, prep, cvt) { - var ttContext = { - functionsDefined: [], - functionsUsed: [], - functionsStackDeltas: [], - tooComplexToFollowFunctions: false, - hintsValid: true - }; - if (fpgm) { - sanitizeTTProgram(fpgm, ttContext); - } - if (prep) { - sanitizeTTProgram(prep, ttContext); + // trying to estimate space character width + var possibleSpaceReplacements = ['space', 'minus', 'one', 'i']; + var width; + for (var i = 0, ii = possibleSpaceReplacements.length; i < ii; i++) { + var glyphName = possibleSpaceReplacements[i]; + // if possible, getting width by glyph name + if (glyphName in this.widths) { + width = this.widths[glyphName]; + break; } - if (fpgm) { - checkInvalidFunctions(ttContext, maxFunctionDefs); + var glyphsUnicodeMap = getGlyphsUnicode(); + var glyphUnicode = glyphsUnicodeMap[glyphName]; + // finding the charcode via unicodeToCID map + var charcode = 0; + if (this.composite) { + if (this.cMap.contains(glyphUnicode)) { + charcode = this.cMap.lookup(glyphUnicode); + } } - if (cvt && (cvt.length & 1)) { - var cvtData = new Uint8Array(cvt.length + 1); - cvtData.set(cvt.data); - cvt.data = cvtData; + // ... via toUnicode map + if (!charcode && this.toUnicode) { + charcode = this.toUnicode.charCodeOf(glyphUnicode); } - return ttContext.hintsValid; - } - - // The following steps modify the original font data, making copy - font = new Stream(new Uint8Array(font.getBytes())); - - var VALID_TABLES = ['OS/2', 'cmap', 'head', 'hhea', 'hmtx', 'maxp', - 'name', 'post', 'loca', 'glyf', 'fpgm', 'prep', 'cvt ', 'CFF ']; - - var header = readOpenTypeHeader(font); - var numTables = header.numTables; - var cff, cffFile; - - var tables = Object.create(null); - tables['OS/2'] = null; - tables['cmap'] = null; - tables['head'] = null; - tables['hhea'] = null; - tables['hmtx'] = null; - tables['maxp'] = null; - tables['name'] = null; - tables['post'] = null; - - var table; - for (var i = 0; i < numTables; i++) { - table = readTableEntry(font); - if (VALID_TABLES.indexOf(table.tag) < 0) { - continue; // skipping table if it's not a required or optional table + // setting it to unicode if negative or undefined + if (charcode <= 0) { + charcode = glyphUnicode; } - if (table.length === 0) { - continue; // skipping empty tables + // trying to get width via charcode + width = this.widths[charcode]; + if (width) { + break; // the non-zero width found } - tables[table.tag] = table; } + width = width || this.defaultWidth; + // Do not shadow the property here. See discussion: + // https://github.com/mozilla/pdf.js/pull/2127#discussion_r1662280 + this._shadowWidth = width; + return width; + }, - var isTrueType = !tables['CFF ']; - if (!isTrueType) { - // OpenType font - if ((header.version === 'OTTO' && properties.type !== 'CIDFontType2') || - !tables['head'] || !tables['hhea'] || !tables['maxp'] || - !tables['post']) { - // no major tables: throwing everything at CFFFont - cffFile = new Stream(tables['CFF '].data); - cff = new CFFFont(cffFile, properties); - - adjustWidths(properties); - - return this.convert(name, cff, properties); - } + charToGlyph: function Font_charToGlyph(charcode, isSpace) { + var fontCharCode, width, operatorListId; - delete tables['glyf']; - delete tables['loca']; - delete tables['fpgm']; - delete tables['prep']; - delete tables['cvt ']; - this.isOpenType = true; - } else { - if (!tables['loca']) { - error('Required "loca" table is not found'); - } - if (!tables['glyf']) { - warn('Required "glyf" table is not found -- trying to recover.'); - // Note: We use `sanitizeGlyphLocations` to add dummy glyf data below. - tables['glyf'] = { - tag: 'glyf', - data: new Uint8Array(0), - }; - } - this.isOpenType = false; + var widthCode = charcode; + if (this.cMap && this.cMap.contains(charcode)) { + widthCode = this.cMap.lookup(charcode); } + width = this.widths[widthCode]; + width = isNum(width) ? width : this.defaultWidth; + var vmetric = this.vmetrics && this.vmetrics[widthCode]; - if (!tables['maxp']) { - error('Required "maxp" table is not found'); + var unicode = this.toUnicode.get(charcode) || charcode; + if (typeof unicode === 'number') { + unicode = String.fromCharCode(unicode); } - font.pos = (font.start || 0) + tables['maxp'].offset; - var version = font.getInt32(); - var numGlyphs = font.getUint16(); - var maxFunctionDefs = 0; - if (version >= 0x00010000 && tables['maxp'].length >= 22) { - // maxZones can be invalid - font.pos += 8; - var maxZones = font.getUint16(); - if (maxZones > 2) { // reset to 2 if font has invalid maxZones - tables['maxp'].data[14] = 0; - tables['maxp'].data[15] = 2; - } - font.pos += 4; - maxFunctionDefs = font.getUint16(); + // First try the toFontChar map, if it's not there then try falling + // back to the char code. + fontCharCode = this.toFontChar[charcode] || charcode; + if (this.missingFile) { + fontCharCode = mapSpecialUnicodeValues(fontCharCode); } - var dupFirstEntry = false; - if (properties.type === 'CIDFontType2' && properties.toUnicode && - properties.toUnicode.get(0) > '\u0000') { - // oracle's defect (see 3427), duplicating first entry - dupFirstEntry = true; - numGlyphs++; - tables['maxp'].data[4] = numGlyphs >> 8; - tables['maxp'].data[5] = numGlyphs & 255; + if (this.isType3Font) { + // Font char code in this case is actually a glyph name. + operatorListId = fontCharCode; } - var hintsValid = sanitizeTTPrograms(tables['fpgm'], tables['prep'], - tables['cvt '], maxFunctionDefs); - if (!hintsValid) { - delete tables['fpgm']; - delete tables['prep']; - delete tables['cvt ']; + var accent = null; + if (this.seacMap && this.seacMap[charcode]) { + var seac = this.seacMap[charcode]; + fontCharCode = seac.baseFontCharCode; + accent = { + fontChar: String.fromCharCode(seac.accentFontCharCode), + offset: seac.accentOffset + }; } - // Ensure the hmtx table contains the advance width and - // sidebearings information for numGlyphs in the maxp table - sanitizeMetrics(font, tables['hhea'], tables['hmtx'], numGlyphs); + var fontChar = String.fromCharCode(fontCharCode); - if (!tables['head']) { - error('Required "head" table is not found'); + var glyph = this.glyphCache[charcode]; + if (!glyph || + !glyph.matchesForCache(fontChar, unicode, accent, width, vmetric, + operatorListId, isSpace)) { + glyph = new Glyph(fontChar, unicode, accent, width, vmetric, + operatorListId, isSpace); + this.glyphCache[charcode] = glyph; } + return glyph; + }, - sanitizeHead(tables['head'], numGlyphs, - isTrueType ? tables['loca'].length : 0); + charsToGlyphs: function Font_charsToGlyphs(chars) { + var charsCache = this.charsCache; + var glyphs, glyph, charcode; - var missingGlyphs = Object.create(null); - if (isTrueType) { - var isGlyphLocationsLong = int16(tables['head'].data[50], - tables['head'].data[51]); - missingGlyphs = sanitizeGlyphLocations(tables['loca'], tables['glyf'], - numGlyphs, isGlyphLocationsLong, - hintsValid, dupFirstEntry); + // if we translated this string before, just grab it from the cache + if (charsCache) { + glyphs = charsCache[chars]; + if (glyphs) { + return glyphs; + } } - if (!tables['hhea']) { - error('Required "hhea" table is not found'); - } - - // Sanitizer reduces the glyph advanceWidth to the maxAdvanceWidth - // Sometimes it's 0. That needs to be fixed - if (tables['hhea'].data[10] === 0 && tables['hhea'].data[11] === 0) { - tables['hhea'].data[10] = 0xFF; - tables['hhea'].data[11] = 0xFF; - } - - // Extract some more font properties from the OpenType head and - // hhea tables; yMin and descent value are always negative. - var metricsOverride = { - unitsPerEm: int16(tables['head'].data[18], tables['head'].data[19]), - yMax: int16(tables['head'].data[42], tables['head'].data[43]), - yMin: int16(tables['head'].data[38], tables['head'].data[39]) - 0x10000, - ascent: int16(tables['hhea'].data[4], tables['hhea'].data[5]), - descent: int16(tables['hhea'].data[6], tables['hhea'].data[7]) - 0x10000 - }; - - // PDF FontDescriptor metrics lie -- using data from actual font. - this.ascent = metricsOverride.ascent / metricsOverride.unitsPerEm; - this.descent = metricsOverride.descent / metricsOverride.unitsPerEm; - - // The 'post' table has glyphs names. - if (tables['post']) { - var valid = readPostScriptTable(tables['post'], properties, numGlyphs); - if (!valid) { - tables['post'] = null; - } - } - - var charCodeToGlyphId = [], charCode; - var toUnicode = properties.toUnicode, widths = properties.widths; - var skipToUnicode = (toUnicode instanceof IdentityToUnicodeMap || - toUnicode.length === 0x10000); - - // Helper function to try to skip mapping of empty glyphs. - // Note: In some cases, just relying on the glyph data doesn't work, - // hence we also use a few heuristics to fix various PDF files. - function hasGlyph(glyphId, charCode, widthCode) { - if (!missingGlyphs[glyphId]) { - return true; - } - if (!skipToUnicode && charCode >= 0 && toUnicode.has(charCode)) { - return true; - } - if (widths && widthCode >= 0 && isNum(widths[widthCode])) { - return true; - } - return false; + // lazily create the translation cache + if (!charsCache) { + charsCache = this.charsCache = Object.create(null); } - if (properties.type === 'CIDFontType2') { - var cidToGidMap = properties.cidToGidMap || []; - var isCidToGidMapEmpty = cidToGidMap.length === 0; - - properties.cMap.forEach(function(charCode, cid) { - assert(cid <= 0xffff, 'Max size of CID is 65,535'); - var glyphId = -1; - if (isCidToGidMapEmpty) { - glyphId = charCode; - } else if (cidToGidMap[cid] !== undefined) { - glyphId = cidToGidMap[cid]; - } + glyphs = []; + var charsCacheKey = chars; + var i = 0, ii; - if (glyphId >= 0 && glyphId < numGlyphs && - hasGlyph(glyphId, charCode, cid)) { - charCodeToGlyphId[charCode] = glyphId; - } - }); - if (dupFirstEntry) { - charCodeToGlyphId[0] = numGlyphs - 1; + if (this.cMap) { + // composite fonts have multi-byte strings convert the string from + // single-byte to multi-byte + var c = Object.create(null); + while (i < chars.length) { + this.cMap.readCharCode(chars, i, c); + charcode = c.charcode; + var length = c.length; + i += length; + // Space is char with code 0x20 and length 1 in multiple-byte codes. + var isSpace = length === 1 && chars.charCodeAt(i - 1) === 0x20; + glyph = this.charToGlyph(charcode, isSpace); + glyphs.push(glyph); } } else { - // Most of the following logic in this code branch is based on the - // 9.6.6.4 of the PDF spec. - var hasEncoding = - properties.differences.length > 0 || !!properties.baseEncodingName; - var cmapTable = - readCmapTable(tables['cmap'], font, this.isSymbolicFont, hasEncoding); - var cmapPlatformId = cmapTable.platformId; - var cmapEncodingId = cmapTable.encodingId; - var cmapMappings = cmapTable.mappings; - var cmapMappingsLength = cmapMappings.length; - - // The spec seems to imply that if the font is symbolic the encoding - // should be ignored, this doesn't appear to work for 'preistabelle.pdf' - // where the the font is symbolic and it has an encoding. - if (hasEncoding && - (cmapPlatformId === 3 && cmapEncodingId === 1 || - cmapPlatformId === 1 && cmapEncodingId === 0) || - (cmapPlatformId === -1 && cmapEncodingId === -1 && // Temporary hack - !!getEncoding(properties.baseEncodingName))) { // Temporary hack - // When no preferred cmap table was found and |baseEncodingName| is - // one of the predefined encodings, we seem to obtain a better - // |charCodeToGlyphId| map from the code below (fixes bug 1057544). - // TODO: Note that this is a hack which should be removed as soon as - // we have proper support for more exotic cmap tables. - - var baseEncoding = []; - if (properties.baseEncodingName === 'MacRomanEncoding' || - properties.baseEncodingName === 'WinAnsiEncoding') { - baseEncoding = getEncoding(properties.baseEncodingName); - } - var glyphsUnicodeMap = getGlyphsUnicode(); - for (charCode = 0; charCode < 256; charCode++) { - var glyphName; - if (this.differences && charCode in this.differences) { - glyphName = this.differences[charCode]; - } else if (charCode in baseEncoding && - baseEncoding[charCode] !== '') { - glyphName = baseEncoding[charCode]; - } else { - glyphName = StandardEncoding[charCode]; - } - if (!glyphName) { - continue; - } - var unicodeOrCharCode, isUnicode = false; - if (cmapPlatformId === 3 && cmapEncodingId === 1) { - unicodeOrCharCode = glyphsUnicodeMap[glyphName]; - isUnicode = true; - } else if (cmapPlatformId === 1 && cmapEncodingId === 0) { - // TODO: the encoding needs to be updated with mac os table. - unicodeOrCharCode = MacRomanEncoding.indexOf(glyphName); - } - - var found = false; - for (i = 0; i < cmapMappingsLength; ++i) { - if (cmapMappings[i].charCode !== unicodeOrCharCode) { - continue; - } - var code = isUnicode ? charCode : unicodeOrCharCode; - if (hasGlyph(cmapMappings[i].glyphId, code, -1)) { - charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; - found = true; - break; - } - } - if (!found && properties.glyphNames) { - // Try to map using the post table. - var glyphId = properties.glyphNames.indexOf(glyphName); - if (glyphId > 0 && hasGlyph(glyphId, -1, -1)) { - charCodeToGlyphId[charCode] = glyphId; - found = true; - } - } - if (!found) { - charCodeToGlyphId[charCode] = 0; // notdef - } - } - } else if (cmapPlatformId === 0 && cmapEncodingId === 0) { - // Default Unicode semantics, use the charcodes as is. - for (i = 0; i < cmapMappingsLength; ++i) { - charCodeToGlyphId[cmapMappings[i].charCode] = - cmapMappings[i].glyphId; - } - } else { - // For (3, 0) cmap tables: - // The charcode key being stored in charCodeToGlyphId is the lower - // byte of the two-byte charcodes of the cmap table since according to - // the spec: 'each byte from the string shall be prepended with the - // high byte of the range [of charcodes in the cmap table], to form - // a two-byte character, which shall be used to select the - // associated glyph description from the subtable'. - // - // For (1, 0) cmap tables: - // 'single bytes from the string shall be used to look up the - // associated glyph descriptions from the subtable'. This means - // charcodes in the cmap will be single bytes, so no-op since - // glyph.charCode & 0xFF === glyph.charCode - for (i = 0; i < cmapMappingsLength; ++i) { - charCode = cmapMappings[i].charCode & 0xFF; - charCodeToGlyphId[charCode] = cmapMappings[i].glyphId; - } + for (i = 0, ii = chars.length; i < ii; ++i) { + charcode = chars.charCodeAt(i); + glyph = this.charToGlyph(charcode, charcode === 0x20); + glyphs.push(glyph); } } - if (charCodeToGlyphId.length === 0) { - // defines at least one glyph - charCodeToGlyphId[0] = 0; - } - - // Converting glyphs and ids into font's cmap table - var newMapping = adjustMapping(charCodeToGlyphId, properties); - this.toFontChar = newMapping.toFontChar; - tables['cmap'] = { - tag: 'cmap', - data: createCmapTable(newMapping.charCodeToGlyphId, numGlyphs) - }; - - if (!tables['OS/2'] || !validateOS2Table(tables['OS/2'])) { - tables['OS/2'] = { - tag: 'OS/2', - data: createOS2Table(properties, newMapping.charCodeToGlyphId, - metricsOverride) - }; - } - - // Rewrite the 'post' table if needed - if (!tables['post']) { - tables['post'] = { - tag: 'post', - data: createPostTable(properties) - }; - } + // Enter the translated string into the cache + return (charsCache[charsCacheKey] = glyphs); + } + }; - if (!isTrueType) { - try { - // Trying to repair CFF file - cffFile = new Stream(tables['CFF '].data); - var parser = new CFFParser(cffFile, properties); - cff = parser.parse(); - var compiler = new CFFCompiler(cff); - tables['CFF '].data = compiler.compile(); - } catch (e) { - warn('Failed to compile font ' + properties.loadedName); - } - } + return Font; +})(); - // Re-creating 'name' table - if (!tables['name']) { - tables['name'] = { - tag: 'name', - data: createNameTable(this.name) - }; - } else { - // ... using existing 'name' table as prototype - var namePrototype = readNameTable(tables['name']); - tables['name'].data = createNameTable(name, namePrototype); - } +var ErrorFont = (function ErrorFontClosure() { + function ErrorFont(error) { + this.error = error; + this.loadedName = 'g_font_error'; + this.loading = false; + } - var builder = new OpenTypeFileBuilder(header.version); - for (var tableTag in tables) { - builder.addTable(tableTag, tables[tableTag].data); - } - return builder.toArray(); + ErrorFont.prototype = { + charsToGlyphs: function ErrorFont_charsToGlyphs() { + return []; }, + exportData: function ErrorFont_exportData() { + return {error: this.error}; + } + }; - convert: function Font_convert(fontName, font, properties) { - // TODO: Check the charstring widths to determine this. - properties.fixedPitch = false; + return ErrorFont; +})(); - var mapping = font.getGlyphMapping(properties); - var newMapping = adjustMapping(mapping, properties); - this.toFontChar = newMapping.toFontChar; - var numGlyphs = font.numGlyphs; +/** + * Shared logic for building a char code to glyph id mapping for Type1 and + * simple CFF fonts. See section 9.6.6.2 of the spec. + * @param {Object} properties Font properties object. + * @param {Object} builtInEncoding The encoding contained within the actual font + * data. + * @param {Array} Array of glyph names where the index is the glyph ID. + * @returns {Object} A char code to glyph ID map. + */ +function type1FontGlyphMapping(properties, builtInEncoding, glyphNames) { + var charCodeToGlyphId = Object.create(null); + var glyphId, charCode, baseEncoding; - function getCharCodes(charCodeToGlyphId, glyphId) { - var charCodes = null; - for (var charCode in charCodeToGlyphId) { - if (glyphId === charCodeToGlyphId[charCode]) { - if (!charCodes) { - charCodes = []; - } - charCodes.push(charCode | 0); - } - } - return charCodes; + if (properties.baseEncodingName) { + // If a valid base encoding name was used, the mapping is initialized with + // that. + baseEncoding = getEncoding(properties.baseEncodingName); + for (charCode = 0; charCode < baseEncoding.length; charCode++) { + glyphId = glyphNames.indexOf(baseEncoding[charCode]); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; + } else { + charCodeToGlyphId[charCode] = 0; // notdef } - - function createCharCode(charCodeToGlyphId, glyphId) { - for (var charCode in charCodeToGlyphId) { - if (glyphId === charCodeToGlyphId[charCode]) { - return charCode | 0; - } - } - newMapping.charCodeToGlyphId[newMapping.nextAvailableFontCharCode] = - glyphId; - return newMapping.nextAvailableFontCharCode++; + } + } else if (!!(properties.flags & FontFlags.Symbolic)) { + // For a symbolic font the encoding should be the fonts built-in + // encoding. + for (charCode in builtInEncoding) { + charCodeToGlyphId[charCode] = builtInEncoding[charCode]; + } + } else { + // For non-symbolic fonts that don't have a base encoding the standard + // encoding should be used. + baseEncoding = StandardEncoding; + for (charCode = 0; charCode < baseEncoding.length; charCode++) { + glyphId = glyphNames.indexOf(baseEncoding[charCode]); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; + } else { + charCodeToGlyphId[charCode] = 0; // notdef } + } + } - var seacs = font.seacs; - if (SEAC_ANALYSIS_ENABLED && seacs && seacs.length) { - var matrix = properties.fontMatrix || FONT_IDENTITY_MATRIX; - var charset = font.getCharset(); - var seacMap = Object.create(null); - for (var glyphId in seacs) { - glyphId |= 0; - var seac = seacs[glyphId]; - var baseGlyphName = StandardEncoding[seac[2]]; - var accentGlyphName = StandardEncoding[seac[3]]; - var baseGlyphId = charset.indexOf(baseGlyphName); - var accentGlyphId = charset.indexOf(accentGlyphName); - if (baseGlyphId < 0 || accentGlyphId < 0) { - continue; - } - var accentOffset = { - x: seac[0] * matrix[0] + seac[1] * matrix[2] + matrix[4], - y: seac[0] * matrix[1] + seac[1] * matrix[3] + matrix[5] - }; - - var charCodes = getCharCodes(mapping, glyphId); - if (!charCodes) { - // There's no point in mapping it if the char code was never mapped - // to begin with. - continue; - } - for (var i = 0, ii = charCodes.length; i < ii; i++) { - var charCode = charCodes[i]; - // Find a fontCharCode that maps to the base and accent glyphs. - // If one doesn't exists, create it. - var charCodeToGlyphId = newMapping.charCodeToGlyphId; - var baseFontCharCode = createCharCode(charCodeToGlyphId, - baseGlyphId); - var accentFontCharCode = createCharCode(charCodeToGlyphId, - accentGlyphId); - seacMap[charCode] = { - baseFontCharCode: baseFontCharCode, - accentFontCharCode: accentFontCharCode, - accentOffset: accentOffset - }; - } - } - properties.seacMap = seacMap; + // Lastly, merge in the differences. + var differences = properties.differences; + if (differences) { + for (charCode in differences) { + var glyphName = differences[charCode]; + glyphId = glyphNames.indexOf(glyphName); + if (glyphId >= 0) { + charCodeToGlyphId[charCode] = glyphId; + } else { + charCodeToGlyphId[charCode] = 0; // notdef } + } + } + return charCodeToGlyphId; +} - var unitsPerEm = 1 / (properties.fontMatrix || FONT_IDENTITY_MATRIX)[0]; - - var builder = new OpenTypeFileBuilder('\x4F\x54\x54\x4F'); - // PostScript Font Program - builder.addTable('CFF ', font.data); - // OS/2 and Windows Specific metrics - builder.addTable('OS/2', createOS2Table(properties, - newMapping.charCodeToGlyphId)); - // Character to glyphs mapping - builder.addTable('cmap', createCmapTable(newMapping.charCodeToGlyphId, - numGlyphs)); - // Font header - builder.addTable('head', - '\x00\x01\x00\x00' + // Version number - '\x00\x00\x10\x00' + // fontRevision - '\x00\x00\x00\x00' + // checksumAdjustement - '\x5F\x0F\x3C\xF5' + // magicNumber - '\x00\x00' + // Flags - safeString16(unitsPerEm) + // unitsPerEM - '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // creation date - '\x00\x00\x00\x00\x9e\x0b\x7e\x27' + // modifification date - '\x00\x00' + // xMin - safeString16(properties.descent) + // yMin - '\x0F\xFF' + // xMax - safeString16(properties.ascent) + // yMax - string16(properties.italicAngle ? 2 : 0) + // macStyle - '\x00\x11' + // lowestRecPPEM - '\x00\x00' + // fontDirectionHint - '\x00\x00' + // indexToLocFormat - '\x00\x00'); // glyphDataFormat +/* + * CharStrings are encoded following the the CharString Encoding sequence + * describe in Chapter 6 of the "Adobe Type1 Font Format" specification. + * The value in a byte indicates a command, a number, or subsequent bytes + * that are to be interpreted in a special way. + * + * CharString Number Encoding: + * A CharString byte containing the values from 32 through 255 inclusive + * indicate an integer. These values are decoded in four ranges. + * + * 1. A CharString byte containing a value, v, between 32 and 246 inclusive, + * indicate the integer v - 139. Thus, the integer values from -107 through + * 107 inclusive may be encoded in single byte. + * + * 2. A CharString byte containing a value, v, between 247 and 250 inclusive, + * indicates an integer involving the next byte, w, according to the formula: + * [(v - 247) x 256] + w + 108 + * + * 3. A CharString byte containing a value, v, between 251 and 254 inclusive, + * indicates an integer involving the next byte, w, according to the formula: + * -[(v - 251) * 256] - w - 108 + * + * 4. A CharString containing the value 255 indicates that the next 4 bytes + * are a two complement signed integer. The first of these bytes contains the + * highest order bits, the second byte contains the next higher order bits + * and the fourth byte contain the lowest order bits. + * + * + * CharString Command Encoding: + * CharStrings commands are encoded in 1 or 2 bytes. + * + * Single byte commands are encoded in 1 byte that contains a value between + * 0 and 31 inclusive. + * If a command byte contains the value 12, then the value in the next byte + * indicates a command. This "escape" mechanism allows many extra commands + * to be encoded and this encoding technique helps to minimize the length of + * the charStrings. + */ +var Type1CharString = (function Type1CharStringClosure() { + var COMMAND_MAP = { + 'hstem': [1], + 'vstem': [3], + 'vmoveto': [4], + 'rlineto': [5], + 'hlineto': [6], + 'vlineto': [7], + 'rrcurveto': [8], + 'callsubr': [10], + 'flex': [12, 35], + 'drop' : [12, 18], + 'endchar': [14], + 'rmoveto': [21], + 'hmoveto': [22], + 'vhcurveto': [30], + 'hvcurveto': [31] + }; - // Horizontal header - builder.addTable('hhea', - '\x00\x01\x00\x00' + // Version number - safeString16(properties.ascent) + // Typographic Ascent - safeString16(properties.descent) + // Typographic Descent - '\x00\x00' + // Line Gap - '\xFF\xFF' + // advanceWidthMax - '\x00\x00' + // minLeftSidebearing - '\x00\x00' + // minRightSidebearing - '\x00\x00' + // xMaxExtent - safeString16(properties.capHeight) + // caretSlopeRise - safeString16(Math.tan(properties.italicAngle) * - properties.xHeight) + // caretSlopeRun - '\x00\x00' + // caretOffset - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // -reserved- - '\x00\x00' + // metricDataFormat - string16(numGlyphs)); // Number of HMetrics - - // Horizontal metrics - builder.addTable('hmtx', (function fontFieldsHmtx() { - var charstrings = font.charstrings; - var cffWidths = font.cff ? font.cff.widths : null; - var hmtx = '\x00\x00\x00\x00'; // Fake .notdef - for (var i = 1, ii = numGlyphs; i < ii; i++) { - var width = 0; - if (charstrings) { - var charstring = charstrings[i - 1]; - width = 'width' in charstring ? charstring.width : 0; - } else if (cffWidths) { - width = Math.ceil(cffWidths[i] || 0); - } - hmtx += string16(width) + string16(0); - } - return hmtx; - })()); - - // Maximum profile - builder.addTable('maxp', - '\x00\x00\x50\x00' + // Version number - string16(numGlyphs)); // Num of glyphs - - // Naming tables - builder.addTable('name', createNameTable(fontName)); - - // PostScript informations - builder.addTable('post', createPostTable(properties)); - - return builder.toArray(); - }, - - /** - * Builds a char code to unicode map based on section 9.10 of the spec. - * @param {Object} properties Font properties object. - * @return {Object} A ToUnicodeMap object. - */ - buildToUnicode: function Font_buildToUnicode(properties) { - // Section 9.10.2 Mapping Character Codes to Unicode Values - if (properties.toUnicode && properties.toUnicode.length !== 0) { - return properties.toUnicode; - } - // According to the spec if the font is a simple font we should only map - // to unicode if the base encoding is MacRoman, MacExpert, or WinAnsi or - // the differences array only contains adobe standard or symbol set names, - // in pratice it seems better to always try to create a toUnicode - // map based of the default encoding. - var toUnicode, charcode; - if (!properties.composite /* is simple font */) { - toUnicode = []; - var encoding = properties.defaultEncoding.slice(); - var baseEncodingName = properties.baseEncodingName; - // Merge in the differences array. - var differences = properties.differences; - for (charcode in differences) { - encoding[charcode] = differences[charcode]; - } - var glyphsUnicodeMap = getGlyphsUnicode(); - for (charcode in encoding) { - // a) Map the character code to a character name. - var glyphName = encoding[charcode]; - // b) Look up the character name in the Adobe Glyph List (see the - // Bibliography) to obtain the corresponding Unicode value. - if (glyphName === '') { - continue; - } else if (glyphsUnicodeMap[glyphName] === undefined) { - // (undocumented) c) Few heuristics to recognize unknown glyphs - // NOTE: Adobe Reader does not do this step, but OSX Preview does - var code = 0; - switch (glyphName[0]) { - case 'G': // Gxx glyph - if (glyphName.length === 3) { - code = parseInt(glyphName.substr(1), 16); - } - break; - case 'g': // g00xx glyph - if (glyphName.length === 5) { - code = parseInt(glyphName.substr(1), 16); - } - break; - case 'C': // Cddd glyph - case 'c': // cddd glyph - if (glyphName.length >= 3) { - code = +glyphName.substr(1); - } - break; - } - if (code) { - // If |baseEncodingName| is one the predefined encodings, - // and |code| equals |charcode|, using the glyph defined in the - // baseEncoding seems to yield a better |toUnicode| mapping - // (fixes issue 5070). - if (baseEncodingName && code === +charcode) { - var baseEncoding = getEncoding(baseEncodingName); - if (baseEncoding && (glyphName = baseEncoding[charcode])) { - toUnicode[charcode] = - String.fromCharCode(glyphsUnicodeMap[glyphName]); - continue; - } - } - toUnicode[charcode] = String.fromCharCode(code); - } - continue; - } - toUnicode[charcode] = - String.fromCharCode(glyphsUnicodeMap[glyphName]); - } - return new ToUnicodeMap(toUnicode); - } - // If the font is a composite font that uses one of the predefined CMaps - // listed in Table 118 (except Identity–H and Identity–V) or whose - // descendant CIDFont uses the Adobe-GB1, Adobe-CNS1, Adobe-Japan1, or - // Adobe-Korea1 character collection: - if (properties.composite && ( - (properties.cMap.builtInCMap && - !(properties.cMap instanceof IdentityCMap)) || - (properties.cidSystemInfo.registry === 'Adobe' && - (properties.cidSystemInfo.ordering === 'GB1' || - properties.cidSystemInfo.ordering === 'CNS1' || - properties.cidSystemInfo.ordering === 'Japan1' || - properties.cidSystemInfo.ordering === 'Korea1')))) { - // Then: - // a) Map the character code to a character identifier (CID) according - // to the font’s CMap. - // b) Obtain the registry and ordering of the character collection used - // by the font’s CMap (for example, Adobe and Japan1) from its - // CIDSystemInfo dictionary. - var registry = properties.cidSystemInfo.registry; - var ordering = properties.cidSystemInfo.ordering; - // c) Construct a second CMap name by concatenating the registry and - // ordering obtained in step (b) in the format registry–ordering–UCS2 - // (for example, Adobe–Japan1–UCS2). - var ucs2CMapName = new Name(registry + '-' + ordering + '-UCS2'); - // d) Obtain the CMap with the name constructed in step (c) (available - // from the ASN Web site; see the Bibliography). - var ucs2CMap = CMapFactory.create(ucs2CMapName, - { url: PDFJS.cMapUrl, packed: PDFJS.cMapPacked }, null); - var cMap = properties.cMap; - toUnicode = []; - cMap.forEach(function(charcode, cid) { - assert(cid <= 0xffff, 'Max size of CID is 65,535'); - // e) Map the CID obtained in step (a) according to the CMap obtained - // in step (d), producing a Unicode value. - var ucs2 = ucs2CMap.lookup(cid); - if (ucs2) { - toUnicode[charcode] = - String.fromCharCode((ucs2.charCodeAt(0) << 8) + - ucs2.charCodeAt(1)); - } - }); - return new ToUnicodeMap(toUnicode); - } - - // The viewer's choice, just use an identity map. - return new IdentityToUnicodeMap(properties.firstChar, - properties.lastChar); - }, - - get spaceWidth() { - if ('_shadowWidth' in this) { - return this._shadowWidth; - } - - // trying to estimate space character width - var possibleSpaceReplacements = ['space', 'minus', 'one', 'i']; - var width; - for (var i = 0, ii = possibleSpaceReplacements.length; i < ii; i++) { - var glyphName = possibleSpaceReplacements[i]; - // if possible, getting width by glyph name - if (glyphName in this.widths) { - width = this.widths[glyphName]; - break; - } - var glyphsUnicodeMap = getGlyphsUnicode(); - var glyphUnicode = glyphsUnicodeMap[glyphName]; - // finding the charcode via unicodeToCID map - var charcode = 0; - if (this.composite) { - if (this.cMap.contains(glyphUnicode)) { - charcode = this.cMap.lookup(glyphUnicode); - } - } - // ... via toUnicode map - if (!charcode && this.toUnicode) { - charcode = this.toUnicode.charCodeOf(glyphUnicode); - } - // setting it to unicode if negative or undefined - if (charcode <= 0) { - charcode = glyphUnicode; - } - // trying to get width via charcode - width = this.widths[charcode]; - if (width) { - break; // the non-zero width found - } - } - width = width || this.defaultWidth; - // Do not shadow the property here. See discussion: - // https://github.com/mozilla/pdf.js/pull/2127#discussion_r1662280 - this._shadowWidth = width; - return width; - }, - - charToGlyph: function Font_charToGlyph(charcode, isSpace) { - var fontCharCode, width, operatorListId; - - var widthCode = charcode; - if (this.cMap && this.cMap.contains(charcode)) { - widthCode = this.cMap.lookup(charcode); - } - width = this.widths[widthCode]; - width = isNum(width) ? width : this.defaultWidth; - var vmetric = this.vmetrics && this.vmetrics[widthCode]; - - var unicode = this.toUnicode.get(charcode) || charcode; - if (typeof unicode === 'number') { - unicode = String.fromCharCode(unicode); - } - - // First try the toFontChar map, if it's not there then try falling - // back to the char code. - fontCharCode = this.toFontChar[charcode] || charcode; - if (this.missingFile) { - fontCharCode = mapSpecialUnicodeValues(fontCharCode); - } - - if (this.isType3Font) { - // Font char code in this case is actually a glyph name. - operatorListId = fontCharCode; - } - - var accent = null; - if (this.seacMap && this.seacMap[charcode]) { - var seac = this.seacMap[charcode]; - fontCharCode = seac.baseFontCharCode; - accent = { - fontChar: String.fromCharCode(seac.accentFontCharCode), - offset: seac.accentOffset - }; - } - - var fontChar = String.fromCharCode(fontCharCode); - - var glyph = this.glyphCache[charcode]; - if (!glyph || - !glyph.matchesForCache(fontChar, unicode, accent, width, vmetric, - operatorListId, isSpace)) { - glyph = new Glyph(fontChar, unicode, accent, width, vmetric, - operatorListId, isSpace); - this.glyphCache[charcode] = glyph; - } - return glyph; - }, - - charsToGlyphs: function Font_charsToGlyphs(chars) { - var charsCache = this.charsCache; - var glyphs, glyph, charcode; - - // if we translated this string before, just grab it from the cache - if (charsCache) { - glyphs = charsCache[chars]; - if (glyphs) { - return glyphs; - } - } - - // lazily create the translation cache - if (!charsCache) { - charsCache = this.charsCache = Object.create(null); - } - - glyphs = []; - var charsCacheKey = chars; - var i = 0, ii; - - if (this.cMap) { - // composite fonts have multi-byte strings convert the string from - // single-byte to multi-byte - var c = Object.create(null); - while (i < chars.length) { - this.cMap.readCharCode(chars, i, c); - charcode = c.charcode; - var length = c.length; - i += length; - // Space is char with code 0x20 and length 1 in multiple-byte codes. - var isSpace = length === 1 && chars.charCodeAt(i - 1) === 0x20; - glyph = this.charToGlyph(charcode, isSpace); - glyphs.push(glyph); - } - } else { - for (i = 0, ii = chars.length; i < ii; ++i) { - charcode = chars.charCodeAt(i); - glyph = this.charToGlyph(charcode, charcode === 0x20); - glyphs.push(glyph); - } - } - - // Enter the translated string into the cache - return (charsCache[charsCacheKey] = glyphs); - } - }; - - return Font; -})(); - -var ErrorFont = (function ErrorFontClosure() { - function ErrorFont(error) { - this.error = error; - this.loadedName = 'g_font_error'; - this.loading = false; - } - - ErrorFont.prototype = { - charsToGlyphs: function ErrorFont_charsToGlyphs() { - return []; - }, - exportData: function ErrorFont_exportData() { - return {error: this.error}; - } - }; - - return ErrorFont; -})(); - -/** - * Shared logic for building a char code to glyph id mapping for Type1 and - * simple CFF fonts. See section 9.6.6.2 of the spec. - * @param {Object} properties Font properties object. - * @param {Object} builtInEncoding The encoding contained within the actual font - * data. - * @param {Array} Array of glyph names where the index is the glyph ID. - * @returns {Object} A char code to glyph ID map. - */ -function type1FontGlyphMapping(properties, builtInEncoding, glyphNames) { - var charCodeToGlyphId = Object.create(null); - var glyphId, charCode, baseEncoding; - - if (properties.baseEncodingName) { - // If a valid base encoding name was used, the mapping is initialized with - // that. - baseEncoding = getEncoding(properties.baseEncodingName); - for (charCode = 0; charCode < baseEncoding.length; charCode++) { - glyphId = glyphNames.indexOf(baseEncoding[charCode]); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef - } - } - } else if (!!(properties.flags & FontFlags.Symbolic)) { - // For a symbolic font the encoding should be the fonts built-in - // encoding. - for (charCode in builtInEncoding) { - charCodeToGlyphId[charCode] = builtInEncoding[charCode]; - } - } else { - // For non-symbolic fonts that don't have a base encoding the standard - // encoding should be used. - baseEncoding = StandardEncoding; - for (charCode = 0; charCode < baseEncoding.length; charCode++) { - glyphId = glyphNames.indexOf(baseEncoding[charCode]); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef - } - } - } - - // Lastly, merge in the differences. - var differences = properties.differences; - if (differences) { - for (charCode in differences) { - var glyphName = differences[charCode]; - glyphId = glyphNames.indexOf(glyphName); - if (glyphId >= 0) { - charCodeToGlyphId[charCode] = glyphId; - } else { - charCodeToGlyphId[charCode] = 0; // notdef - } - } - } - return charCodeToGlyphId; -} - -/* - * CharStrings are encoded following the the CharString Encoding sequence - * describe in Chapter 6 of the "Adobe Type1 Font Format" specification. - * The value in a byte indicates a command, a number, or subsequent bytes - * that are to be interpreted in a special way. - * - * CharString Number Encoding: - * A CharString byte containing the values from 32 through 255 inclusive - * indicate an integer. These values are decoded in four ranges. - * - * 1. A CharString byte containing a value, v, between 32 and 246 inclusive, - * indicate the integer v - 139. Thus, the integer values from -107 through - * 107 inclusive may be encoded in single byte. - * - * 2. A CharString byte containing a value, v, between 247 and 250 inclusive, - * indicates an integer involving the next byte, w, according to the formula: - * [(v - 247) x 256] + w + 108 - * - * 3. A CharString byte containing a value, v, between 251 and 254 inclusive, - * indicates an integer involving the next byte, w, according to the formula: - * -[(v - 251) * 256] - w - 108 - * - * 4. A CharString containing the value 255 indicates that the next 4 bytes - * are a two complement signed integer. The first of these bytes contains the - * highest order bits, the second byte contains the next higher order bits - * and the fourth byte contain the lowest order bits. - * - * - * CharString Command Encoding: - * CharStrings commands are encoded in 1 or 2 bytes. - * - * Single byte commands are encoded in 1 byte that contains a value between - * 0 and 31 inclusive. - * If a command byte contains the value 12, then the value in the next byte - * indicates a command. This "escape" mechanism allows many extra commands - * to be encoded and this encoding technique helps to minimize the length of - * the charStrings. - */ -var Type1CharString = (function Type1CharStringClosure() { - var COMMAND_MAP = { - 'hstem': [1], - 'vstem': [3], - 'vmoveto': [4], - 'rlineto': [5], - 'hlineto': [6], - 'vlineto': [7], - 'rrcurveto': [8], - 'callsubr': [10], - 'flex': [12, 35], - 'drop' : [12, 18], - 'endchar': [14], - 'rmoveto': [21], - 'hmoveto': [22], - 'vhcurveto': [30], - 'hvcurveto': [31] - }; - - function Type1CharString() { - this.width = 0; - this.lsb = 0; - this.flexing = false; - this.output = []; - this.stack = []; - } + function Type1CharString() { + this.width = 0; + this.lsb = 0; + this.flexing = false; + this.output = []; + this.stack = []; + } Type1CharString.prototype = { convert: function Type1CharString_convert(encoded, subrs) { @@ -31376,4796 +29805,6377 @@ Type1Font.prototype = { strings.add('Medium'); // Weight cff.strings = strings; - cff.globalSubrIndex = new CFFIndex(); + cff.globalSubrIndex = new CFFIndex(); + + var count = glyphs.length; + var charsetArray = [0]; + var i, ii; + for (i = 0; i < count; i++) { + var index = CFFStandardStrings.indexOf(charstrings[i].glyphName); + // TODO: Insert the string and correctly map it. Previously it was + // thought mapping names that aren't in the standard strings to .notdef + // was fine, however in issue818 when mapping them all to .notdef the + // adieresis glyph no longer worked. + if (index === -1) { + index = 0; + } + charsetArray.push((index >> 8) & 0xff, index & 0xff); + } + cff.charset = new CFFCharset(false, 0, [], charsetArray); + + var charStringsIndex = new CFFIndex(); + charStringsIndex.add([0x8B, 0x0E]); // .notdef + for (i = 0; i < count; i++) { + charStringsIndex.add(glyphs[i]); + } + cff.charStrings = charStringsIndex; + + var privateDict = new CFFPrivateDict(); + privateDict.setByName('Subrs', null); // placeholder + var fields = [ + 'BlueValues', + 'OtherBlues', + 'FamilyBlues', + 'FamilyOtherBlues', + 'StemSnapH', + 'StemSnapV', + 'BlueShift', + 'BlueFuzz', + 'BlueScale', + 'LanguageGroup', + 'ExpansionFactor', + 'ForceBold', + 'StdHW', + 'StdVW' + ]; + for (i = 0, ii = fields.length; i < ii; i++) { + var field = fields[i]; + if (!(field in properties.privateData)) { + continue; + } + var value = properties.privateData[field]; + if (isArray(value)) { + // All of the private dictionary array data in CFF must be stored as + // "delta-encoded" numbers. + for (var j = value.length - 1; j > 0; j--) { + value[j] -= value[j - 1]; // ... difference from previous value + } + } + privateDict.setByName(field, value); + } + cff.topDict.privateDict = privateDict; + + var subrIndex = new CFFIndex(); + for (i = 0, ii = subrs.length; i < ii; i++) { + subrIndex.add(subrs[i]); + } + privateDict.subrsIndex = subrIndex; + + var compiler = new CFFCompiler(cff); + return compiler.compile(); + } +}; + +var CFFFont = (function CFFFontClosure() { + function CFFFont(file, properties) { + this.properties = properties; + + var parser = new CFFParser(file, properties); + this.cff = parser.parse(); + var compiler = new CFFCompiler(this.cff); + this.seacs = this.cff.seacs; + try { + this.data = compiler.compile(); + } catch (e) { + warn('Failed to compile font ' + properties.loadedName); + // There may have just been an issue with the compiler, set the data + // anyway and hope the font loaded. + this.data = file; + } + } + + CFFFont.prototype = { + get numGlyphs() { + return this.cff.charStrings.count; + }, + getCharset: function CFFFont_getCharset() { + return this.cff.charset.charset; + }, + getGlyphMapping: function CFFFont_getGlyphMapping() { + var cff = this.cff; + var properties = this.properties; + var charsets = cff.charset.charset; + var charCodeToGlyphId; + var glyphId; + + if (properties.composite) { + charCodeToGlyphId = Object.create(null); + if (cff.isCIDFont) { + // If the font is actually a CID font then we should use the charset + // to map CIDs to GIDs. + for (glyphId = 0; glyphId < charsets.length; glyphId++) { + var cid = charsets[glyphId]; + var charCode = properties.cMap.charCodeOf(cid); + charCodeToGlyphId[charCode] = glyphId; + } + } else { + // If it is NOT actually a CID font then CIDs should be mapped + // directly to GIDs. + for (glyphId = 0; glyphId < cff.charStrings.count; glyphId++) { + charCodeToGlyphId[glyphId] = glyphId; + } + } + return charCodeToGlyphId; + } + + var encoding = cff.encoding ? cff.encoding.encoding : null; + charCodeToGlyphId = type1FontGlyphMapping(properties, encoding, charsets); + return charCodeToGlyphId; + } + }; + + return CFFFont; +})(); + +var CFFParser = (function CFFParserClosure() { + var CharstringValidationData = [ + null, + { id: 'hstem', min: 2, stackClearing: true, stem: true }, + null, + { id: 'vstem', min: 2, stackClearing: true, stem: true }, + { id: 'vmoveto', min: 1, stackClearing: true }, + { id: 'rlineto', min: 2, resetStack: true }, + { id: 'hlineto', min: 1, resetStack: true }, + { id: 'vlineto', min: 1, resetStack: true }, + { id: 'rrcurveto', min: 6, resetStack: true }, + null, + { id: 'callsubr', min: 1, undefStack: true }, + { id: 'return', min: 0, undefStack: true }, + null, // 12 + null, + { id: 'endchar', min: 0, stackClearing: true }, + null, + null, + null, + { id: 'hstemhm', min: 2, stackClearing: true, stem: true }, + { id: 'hintmask', min: 0, stackClearing: true }, + { id: 'cntrmask', min: 0, stackClearing: true }, + { id: 'rmoveto', min: 2, stackClearing: true }, + { id: 'hmoveto', min: 1, stackClearing: true }, + { id: 'vstemhm', min: 2, stackClearing: true, stem: true }, + { id: 'rcurveline', min: 8, resetStack: true }, + { id: 'rlinecurve', min: 8, resetStack: true }, + { id: 'vvcurveto', min: 4, resetStack: true }, + { id: 'hhcurveto', min: 4, resetStack: true }, + null, // shortint + { id: 'callgsubr', min: 1, undefStack: true }, + { id: 'vhcurveto', min: 4, resetStack: true }, + { id: 'hvcurveto', min: 4, resetStack: true } + ]; + var CharstringValidationData12 = [ + null, + null, + null, + { id: 'and', min: 2, stackDelta: -1 }, + { id: 'or', min: 2, stackDelta: -1 }, + { id: 'not', min: 1, stackDelta: 0 }, + null, + null, + null, + { id: 'abs', min: 1, stackDelta: 0 }, + { id: 'add', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] + stack[index - 1]; + } + }, + { id: 'sub', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] - stack[index - 1]; + } + }, + { id: 'div', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] / stack[index - 1]; + } + }, + null, + { id: 'neg', min: 1, stackDelta: 0, + stackFn: function stack_div(stack, index) { + stack[index - 1] = -stack[index - 1]; + } + }, + { id: 'eq', min: 2, stackDelta: -1 }, + null, + null, + { id: 'drop', min: 1, stackDelta: -1 }, + null, + { id: 'put', min: 2, stackDelta: -2 }, + { id: 'get', min: 1, stackDelta: 0 }, + { id: 'ifelse', min: 4, stackDelta: -3 }, + { id: 'random', min: 0, stackDelta: 1 }, + { id: 'mul', min: 2, stackDelta: -1, + stackFn: function stack_div(stack, index) { + stack[index - 2] = stack[index - 2] * stack[index - 1]; + } + }, + null, + { id: 'sqrt', min: 1, stackDelta: 0 }, + { id: 'dup', min: 1, stackDelta: 1 }, + { id: 'exch', min: 2, stackDelta: 0 }, + { id: 'index', min: 2, stackDelta: 0 }, + { id: 'roll', min: 3, stackDelta: -2 }, + null, + null, + null, + { id: 'hflex', min: 7, resetStack: true }, + { id: 'flex', min: 13, resetStack: true }, + { id: 'hflex1', min: 9, resetStack: true }, + { id: 'flex1', min: 11, resetStack: true } + ]; + + function CFFParser(file, properties) { + this.bytes = file.getBytes(); + this.properties = properties; + } + CFFParser.prototype = { + parse: function CFFParser_parse() { + var properties = this.properties; + var cff = new CFF(); + this.cff = cff; + + // The first five sections must be in order, all the others are reached + // via offsets contained in one of the below. + var header = this.parseHeader(); + var nameIndex = this.parseIndex(header.endPos); + var topDictIndex = this.parseIndex(nameIndex.endPos); + var stringIndex = this.parseIndex(topDictIndex.endPos); + var globalSubrIndex = this.parseIndex(stringIndex.endPos); + + var topDictParsed = this.parseDict(topDictIndex.obj.get(0)); + var topDict = this.createDict(CFFTopDict, topDictParsed, cff.strings); + + cff.header = header.obj; + cff.names = this.parseNameIndex(nameIndex.obj); + cff.strings = this.parseStringIndex(stringIndex.obj); + cff.topDict = topDict; + cff.globalSubrIndex = globalSubrIndex.obj; + + this.parsePrivateDict(cff.topDict); + + cff.isCIDFont = topDict.hasName('ROS'); + + var charStringOffset = topDict.getByName('CharStrings'); + var charStringIndex = this.parseIndex(charStringOffset).obj; + + var fontMatrix = topDict.getByName('FontMatrix'); + if (fontMatrix) { + properties.fontMatrix = fontMatrix; + } + + var fontBBox = topDict.getByName('FontBBox'); + if (fontBBox) { + // adjusting ascent/descent + properties.ascent = fontBBox[3]; + properties.descent = fontBBox[1]; + properties.ascentScaled = true; + } + + var charset, encoding; + if (cff.isCIDFont) { + var fdArrayIndex = this.parseIndex(topDict.getByName('FDArray')).obj; + for (var i = 0, ii = fdArrayIndex.count; i < ii; ++i) { + var dictRaw = fdArrayIndex.get(i); + var fontDict = this.createDict(CFFTopDict, this.parseDict(dictRaw), + cff.strings); + this.parsePrivateDict(fontDict); + cff.fdArray.push(fontDict); + } + // cid fonts don't have an encoding + encoding = null; + charset = this.parseCharsets(topDict.getByName('charset'), + charStringIndex.count, cff.strings, true); + cff.fdSelect = this.parseFDSelect(topDict.getByName('FDSelect'), + charStringIndex.count); + } else { + charset = this.parseCharsets(topDict.getByName('charset'), + charStringIndex.count, cff.strings, false); + encoding = this.parseEncoding(topDict.getByName('Encoding'), + properties, + cff.strings, charset.charset); + } + + cff.charset = charset; + cff.encoding = encoding; + + var charStringsAndSeacs = this.parseCharStrings( + charStringIndex, + topDict.privateDict.subrsIndex, + globalSubrIndex.obj, + cff.fdSelect, + cff.fdArray); + cff.charStrings = charStringsAndSeacs.charStrings; + cff.seacs = charStringsAndSeacs.seacs; + cff.widths = charStringsAndSeacs.widths; + + return cff; + }, + parseHeader: function CFFParser_parseHeader() { + var bytes = this.bytes; + var bytesLength = bytes.length; + var offset = 0; + + // Prevent an infinite loop, by checking that the offset is within the + // bounds of the bytes array. Necessary in empty, or invalid, font files. + while (offset < bytesLength && bytes[offset] !== 1) { + ++offset; + } + if (offset >= bytesLength) { + error('Invalid CFF header'); + } else if (offset !== 0) { + info('cff data is shifted'); + bytes = bytes.subarray(offset); + this.bytes = bytes; + } + var major = bytes[0]; + var minor = bytes[1]; + var hdrSize = bytes[2]; + var offSize = bytes[3]; + var header = new CFFHeader(major, minor, hdrSize, offSize); + return { obj: header, endPos: hdrSize }; + }, + parseDict: function CFFParser_parseDict(dict) { + var pos = 0; + + function parseOperand() { + var value = dict[pos++]; + if (value === 30) { + return parseFloatOperand(pos); + } else if (value === 28) { + value = dict[pos++]; + value = ((value << 24) | (dict[pos++] << 16)) >> 16; + return value; + } else if (value === 29) { + value = dict[pos++]; + value = (value << 8) | dict[pos++]; + value = (value << 8) | dict[pos++]; + value = (value << 8) | dict[pos++]; + return value; + } else if (value >= 32 && value <= 246) { + return value - 139; + } else if (value >= 247 && value <= 250) { + return ((value - 247) * 256) + dict[pos++] + 108; + } else if (value >= 251 && value <= 254) { + return -((value - 251) * 256) - dict[pos++] - 108; + } else { + error('255 is not a valid DICT command'); + } + return -1; + } + + function parseFloatOperand() { + var str = ''; + var eof = 15; + var lookup = ['0', '1', '2', '3', '4', '5', '6', '7', '8', + '9', '.', 'E', 'E-', null, '-']; + var length = dict.length; + while (pos < length) { + var b = dict[pos++]; + var b1 = b >> 4; + var b2 = b & 15; + + if (b1 === eof) { + break; + } + str += lookup[b1]; + + if (b2 === eof) { + break; + } + str += lookup[b2]; + } + return parseFloat(str); + } + + var operands = []; + var entries = []; + + pos = 0; + var end = dict.length; + while (pos < end) { + var b = dict[pos]; + if (b <= 21) { + if (b === 12) { + b = (b << 8) | dict[++pos]; + } + entries.push([b, operands]); + operands = []; + ++pos; + } else { + operands.push(parseOperand()); + } + } + return entries; + }, + parseIndex: function CFFParser_parseIndex(pos) { + var cffIndex = new CFFIndex(); + var bytes = this.bytes; + var count = (bytes[pos++] << 8) | bytes[pos++]; + var offsets = []; + var end = pos; + var i, ii; + + if (count !== 0) { + var offsetSize = bytes[pos++]; + // add 1 for offset to determine size of last object + var startPos = pos + ((count + 1) * offsetSize) - 1; + + for (i = 0, ii = count + 1; i < ii; ++i) { + var offset = 0; + for (var j = 0; j < offsetSize; ++j) { + offset <<= 8; + offset += bytes[pos++]; + } + offsets.push(startPos + offset); + } + end = offsets[count]; + } + for (i = 0, ii = offsets.length - 1; i < ii; ++i) { + var offsetStart = offsets[i]; + var offsetEnd = offsets[i + 1]; + cffIndex.add(bytes.subarray(offsetStart, offsetEnd)); + } + return {obj: cffIndex, endPos: end}; + }, + parseNameIndex: function CFFParser_parseNameIndex(index) { + var names = []; + for (var i = 0, ii = index.count; i < ii; ++i) { + var name = index.get(i); + // OTS doesn't allow names to be over 127 characters. + var length = Math.min(name.length, 127); + var data = []; + // OTS also only permits certain characters in the name. + for (var j = 0; j < length; ++j) { + var c = name[j]; + if (j === 0 && c === 0) { + data[j] = c; + continue; + } + if ((c < 33 || c > 126) || c === 91 /* [ */ || c === 93 /* ] */ || + c === 40 /* ( */ || c === 41 /* ) */ || c === 123 /* { */ || + c === 125 /* } */ || c === 60 /* < */ || c === 62 /* > */ || + c === 47 /* / */ || c === 37 /* % */ || c === 35 /* # */) { + data[j] = 95; + continue; + } + data[j] = c; + } + names.push(bytesToString(data)); + } + return names; + }, + parseStringIndex: function CFFParser_parseStringIndex(index) { + var strings = new CFFStrings(); + for (var i = 0, ii = index.count; i < ii; ++i) { + var data = index.get(i); + strings.add(bytesToString(data)); + } + return strings; + }, + createDict: function CFFParser_createDict(Type, dict, strings) { + var cffDict = new Type(strings); + for (var i = 0, ii = dict.length; i < ii; ++i) { + var pair = dict[i]; + var key = pair[0]; + var value = pair[1]; + cffDict.setByKey(key, value); + } + return cffDict; + }, + parseCharString: function CFFParser_parseCharString(state, data, + localSubrIndex, + globalSubrIndex) { + if (state.callDepth > MAX_SUBR_NESTING) { + return false; + } + var stackSize = state.stackSize; + var stack = state.stack; + + var length = data.length; + + for (var j = 0; j < length;) { + var value = data[j++]; + var validationCommand = null; + if (value === 12) { + var q = data[j++]; + if (q === 0) { + // The CFF specification state that the 'dotsection' command + // (12, 0) is deprecated and treated as a no-op, but all Type2 + // charstrings processors should support them. Unfortunately + // the font sanitizer don't. As a workaround the sequence (12, 0) + // is replaced by a useless (0, hmoveto). + data[j - 2] = 139; + data[j - 1] = 22; + stackSize = 0; + } else { + validationCommand = CharstringValidationData12[q]; + } + } else if (value === 28) { // number (16 bit) + stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16)) >> 16; + j += 2; + stackSize++; + } else if (value === 14) { + if (stackSize >= 4) { + stackSize -= 4; + if (SEAC_ANALYSIS_ENABLED) { + state.seac = stack.slice(stackSize, stackSize + 4); + return false; + } + } + validationCommand = CharstringValidationData[value]; + } else if (value >= 32 && value <= 246) { // number + stack[stackSize] = value - 139; + stackSize++; + } else if (value >= 247 && value <= 254) { // number (+1 bytes) + stack[stackSize] = (value < 251 ? + ((value - 247) << 8) + data[j] + 108 : + -((value - 251) << 8) - data[j] - 108); + j++; + stackSize++; + } else if (value === 255) { // number (32 bit) + stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16) | + (data[j + 2] << 8) | data[j + 3]) / 65536; + j += 4; + stackSize++; + } else if (value === 19 || value === 20) { + state.hints += stackSize >> 1; + // skipping right amount of hints flag data + j += (state.hints + 7) >> 3; + stackSize %= 2; + validationCommand = CharstringValidationData[value]; + } else if (value === 10 || value === 29) { + var subrsIndex; + if (value === 10) { + subrsIndex = localSubrIndex; + } else { + subrsIndex = globalSubrIndex; + } + if (!subrsIndex) { + validationCommand = CharstringValidationData[value]; + warn('Missing subrsIndex for ' + validationCommand.id); + return false; + } + var bias = 32768; + if (subrsIndex.count < 1240) { + bias = 107; + } else if (subrsIndex.count < 33900) { + bias = 1131; + } + var subrNumber = stack[--stackSize] + bias; + if (subrNumber < 0 || subrNumber >= subrsIndex.count) { + validationCommand = CharstringValidationData[value]; + warn('Out of bounds subrIndex for ' + validationCommand.id); + return false; + } + state.stackSize = stackSize; + state.callDepth++; + var valid = this.parseCharString(state, subrsIndex.get(subrNumber), + localSubrIndex, globalSubrIndex); + if (!valid) { + return false; + } + state.callDepth--; + stackSize = state.stackSize; + continue; + } else if (value === 11) { + state.stackSize = stackSize; + return true; + } else { + validationCommand = CharstringValidationData[value]; + } + if (validationCommand) { + if (validationCommand.stem) { + state.hints += stackSize >> 1; + } + if ('min' in validationCommand) { + if (!state.undefStack && stackSize < validationCommand.min) { + warn('Not enough parameters for ' + validationCommand.id + + '; actual: ' + stackSize + + ', expected: ' + validationCommand.min); + return false; + } + } + if (state.firstStackClearing && validationCommand.stackClearing) { + state.firstStackClearing = false; + // the optional character width can be found before the first + // stack-clearing command arguments + stackSize -= validationCommand.min; + if (stackSize >= 2 && validationCommand.stem) { + // there are even amount of arguments for stem commands + stackSize %= 2; + } else if (stackSize > 1) { + warn('Found too many parameters for stack-clearing command'); + } + if (stackSize > 0 && stack[stackSize - 1] >= 0) { + state.width = stack[stackSize - 1]; + } + } + if ('stackDelta' in validationCommand) { + if ('stackFn' in validationCommand) { + validationCommand.stackFn(stack, stackSize); + } + stackSize += validationCommand.stackDelta; + } else if (validationCommand.stackClearing) { + stackSize = 0; + } else if (validationCommand.resetStack) { + stackSize = 0; + state.undefStack = false; + } else if (validationCommand.undefStack) { + stackSize = 0; + state.undefStack = true; + state.firstStackClearing = false; + } + } + } + state.stackSize = stackSize; + return true; + }, + parseCharStrings: function CFFParser_parseCharStrings(charStrings, + localSubrIndex, + globalSubrIndex, + fdSelect, + fdArray) { + var seacs = []; + var widths = []; + var count = charStrings.count; + for (var i = 0; i < count; i++) { + var charstring = charStrings.get(i); + var state = { + callDepth: 0, + stackSize: 0, + stack: [], + undefStack: true, + hints: 0, + firstStackClearing: true, + seac: null, + width: null + }; + var valid = true; + var localSubrToUse = null; + if (fdSelect && fdArray.length) { + var fdIndex = fdSelect.getFDIndex(i); + if (fdIndex === -1) { + warn('Glyph index is not in fd select.'); + valid = false; + } + if (fdIndex >= fdArray.length) { + warn('Invalid fd index for glyph index.'); + valid = false; + } + if (valid) { + localSubrToUse = fdArray[fdIndex].privateDict.subrsIndex; + } + } else if (localSubrIndex) { + localSubrToUse = localSubrIndex; + } + if (valid) { + valid = this.parseCharString(state, charstring, localSubrToUse, + globalSubrIndex); + } + if (state.width !== null) { + widths[i] = state.width; + } + if (state.seac !== null) { + seacs[i] = state.seac; + } + if (!valid) { + // resetting invalid charstring to single 'endchar' + charStrings.set(i, new Uint8Array([14])); + } + } + return { charStrings: charStrings, seacs: seacs, widths: widths }; + }, + emptyPrivateDictionary: + function CFFParser_emptyPrivateDictionary(parentDict) { + var privateDict = this.createDict(CFFPrivateDict, [], + parentDict.strings); + parentDict.setByKey(18, [0, 0]); + parentDict.privateDict = privateDict; + }, + parsePrivateDict: function CFFParser_parsePrivateDict(parentDict) { + // no private dict, do nothing + if (!parentDict.hasName('Private')) { + this.emptyPrivateDictionary(parentDict); + return; + } + var privateOffset = parentDict.getByName('Private'); + // make sure the params are formatted correctly + if (!isArray(privateOffset) || privateOffset.length !== 2) { + parentDict.removeByName('Private'); + return; + } + var size = privateOffset[0]; + var offset = privateOffset[1]; + // remove empty dicts or ones that refer to invalid location + if (size === 0 || offset >= this.bytes.length) { + this.emptyPrivateDictionary(parentDict); + return; + } + + var privateDictEnd = offset + size; + var dictData = this.bytes.subarray(offset, privateDictEnd); + var dict = this.parseDict(dictData); + var privateDict = this.createDict(CFFPrivateDict, dict, + parentDict.strings); + parentDict.privateDict = privateDict; + + // Parse the Subrs index also since it's relative to the private dict. + if (!privateDict.getByName('Subrs')) { + return; + } + var subrsOffset = privateDict.getByName('Subrs'); + var relativeOffset = offset + subrsOffset; + // Validate the offset. + if (subrsOffset === 0 || relativeOffset >= this.bytes.length) { + this.emptyPrivateDictionary(parentDict); + return; + } + var subrsIndex = this.parseIndex(relativeOffset); + privateDict.subrsIndex = subrsIndex.obj; + }, + parseCharsets: function CFFParser_parseCharsets(pos, length, strings, cid) { + if (pos === 0) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.ISO_ADOBE, + ISOAdobeCharset); + } else if (pos === 1) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT, + ExpertCharset); + } else if (pos === 2) { + return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT_SUBSET, + ExpertSubsetCharset); + } + + var bytes = this.bytes; + var start = pos; + var format = bytes[pos++]; + var charset = ['.notdef']; + var id, count, i; + + // subtract 1 for the .notdef glyph + length -= 1; - var count = glyphs.length; - var charsetArray = [0]; - var i, ii; - for (i = 0; i < count; i++) { - var index = CFFStandardStrings.indexOf(charstrings[i].glyphName); - // TODO: Insert the string and correctly map it. Previously it was - // thought mapping names that aren't in the standard strings to .notdef - // was fine, however in issue818 when mapping them all to .notdef the - // adieresis glyph no longer worked. - if (index === -1) { - index = 0; + switch (format) { + case 0: + for (i = 0; i < length; i++) { + id = (bytes[pos++] << 8) | bytes[pos++]; + charset.push(cid ? id : strings.get(id)); + } + break; + case 1: + while (charset.length <= length) { + id = (bytes[pos++] << 8) | bytes[pos++]; + count = bytes[pos++]; + for (i = 0; i <= count; i++) { + charset.push(cid ? id++ : strings.get(id++)); + } + } + break; + case 2: + while (charset.length <= length) { + id = (bytes[pos++] << 8) | bytes[pos++]; + count = (bytes[pos++] << 8) | bytes[pos++]; + for (i = 0; i <= count; i++) { + charset.push(cid ? id++ : strings.get(id++)); + } + } + break; + default: + error('Unknown charset format'); } - charsetArray.push((index >> 8) & 0xff, index & 0xff); - } - cff.charset = new CFFCharset(false, 0, [], charsetArray); + // Raw won't be needed if we actually compile the charset. + var end = pos; + var raw = bytes.subarray(start, end); - var charStringsIndex = new CFFIndex(); - charStringsIndex.add([0x8B, 0x0E]); // .notdef - for (i = 0; i < count; i++) { - charStringsIndex.add(glyphs[i]); - } - cff.charStrings = charStringsIndex; + return new CFFCharset(false, format, charset, raw); + }, + parseEncoding: function CFFParser_parseEncoding(pos, + properties, + strings, + charset) { + var encoding = Object.create(null); + var bytes = this.bytes; + var predefined = false; + var hasSupplement = false; + var format, i, ii; + var raw = null; - var privateDict = new CFFPrivateDict(); - privateDict.setByName('Subrs', null); // placeholder - var fields = [ - 'BlueValues', - 'OtherBlues', - 'FamilyBlues', - 'FamilyOtherBlues', - 'StemSnapH', - 'StemSnapV', - 'BlueShift', - 'BlueFuzz', - 'BlueScale', - 'LanguageGroup', - 'ExpansionFactor', - 'ForceBold', - 'StdHW', - 'StdVW' - ]; - for (i = 0, ii = fields.length; i < ii; i++) { - var field = fields[i]; - if (!(field in properties.privateData)) { - continue; + function readSupplement() { + var supplementsCount = bytes[pos++]; + for (i = 0; i < supplementsCount; i++) { + var code = bytes[pos++]; + var sid = (bytes[pos++] << 8) + (bytes[pos++] & 0xff); + encoding[code] = charset.indexOf(strings.get(sid)); + } } - var value = properties.privateData[field]; - if (isArray(value)) { - // All of the private dictionary array data in CFF must be stored as - // "delta-encoded" numbers. - for (var j = value.length - 1; j > 0; j--) { - value[j] -= value[j - 1]; // ... difference from previous value + + if (pos === 0 || pos === 1) { + predefined = true; + format = pos; + var baseEncoding = pos ? ExpertEncoding : StandardEncoding; + for (i = 0, ii = charset.length; i < ii; i++) { + var index = baseEncoding.indexOf(charset[i]); + if (index !== -1) { + encoding[index] = i; + } + } + } else { + var dataStart = pos; + format = bytes[pos++]; + switch (format & 0x7f) { + case 0: + var glyphsCount = bytes[pos++]; + for (i = 1; i <= glyphsCount; i++) { + encoding[bytes[pos++]] = i; + } + break; + + case 1: + var rangesCount = bytes[pos++]; + var gid = 1; + for (i = 0; i < rangesCount; i++) { + var start = bytes[pos++]; + var left = bytes[pos++]; + for (var j = start; j <= start + left; j++) { + encoding[j] = gid++; + } + } + break; + + default: + error('Unknow encoding format: ' + format + ' in CFF'); + break; + } + var dataEnd = pos; + if (format & 0x80) { + // The font sanitizer does not support CFF encoding with a + // supplement, since the encoding is not really used to map + // between gid to glyph, let's overwrite what is declared in + // the top dictionary to let the sanitizer think the font use + // StandardEncoding, that's a lie but that's ok. + bytes[dataStart] &= 0x7f; + readSupplement(); + hasSupplement = true; } + raw = bytes.subarray(dataStart, dataEnd); } - privateDict.setByName(field, value); - } - cff.topDict.privateDict = privateDict; + format = format & 0x7f; + return new CFFEncoding(predefined, format, encoding, raw); + }, + parseFDSelect: function CFFParser_parseFDSelect(pos, length) { + var start = pos; + var bytes = this.bytes; + var format = bytes[pos++]; + var fdSelect = []; + var i; - var subrIndex = new CFFIndex(); - for (i = 0, ii = subrs.length; i < ii; i++) { - subrIndex.add(subrs[i]); + switch (format) { + case 0: + for (i = 0; i < length; ++i) { + var id = bytes[pos++]; + fdSelect.push(id); + } + break; + case 3: + var rangesCount = (bytes[pos++] << 8) | bytes[pos++]; + for (i = 0; i < rangesCount; ++i) { + var first = (bytes[pos++] << 8) | bytes[pos++]; + var fdIndex = bytes[pos++]; + var next = (bytes[pos] << 8) | bytes[pos + 1]; + for (var j = first; j < next; ++j) { + fdSelect.push(fdIndex); + } + } + // Advance past the sentinel(next). + pos += 2; + break; + default: + error('Unknown fdselect format ' + format); + break; + } + var end = pos; + return new CFFFDSelect(fdSelect, bytes.subarray(start, end)); } - privateDict.subrsIndex = subrIndex; + }; + return CFFParser; +})(); - var compiler = new CFFCompiler(cff); - return compiler.compile(); +// Compact Font Format +var CFF = (function CFFClosure() { + function CFF() { + this.header = null; + this.names = []; + this.topDict = null; + this.strings = new CFFStrings(); + this.globalSubrIndex = null; + + // The following could really be per font, but since we only have one font + // store them here. + this.encoding = null; + this.charset = null; + this.charStrings = null; + this.fdArray = []; + this.fdSelect = null; + + this.isCIDFont = false; } -}; + return CFF; +})(); -var CFFFont = (function CFFFontClosure() { - function CFFFont(file, properties) { - this.properties = properties; +var CFFHeader = (function CFFHeaderClosure() { + function CFFHeader(major, minor, hdrSize, offSize) { + this.major = major; + this.minor = minor; + this.hdrSize = hdrSize; + this.offSize = offSize; + } + return CFFHeader; +})(); - var parser = new CFFParser(file, properties); - this.cff = parser.parse(); - var compiler = new CFFCompiler(this.cff); - this.seacs = this.cff.seacs; - try { - this.data = compiler.compile(); - } catch (e) { - warn('Failed to compile font ' + properties.loadedName); - // There may have just been an issue with the compiler, set the data - // anyway and hope the font loaded. - this.data = file; - } +var CFFStrings = (function CFFStringsClosure() { + function CFFStrings() { + this.strings = []; } + CFFStrings.prototype = { + get: function CFFStrings_get(index) { + if (index >= 0 && index <= 390) { + return CFFStandardStrings[index]; + } + if (index - 391 <= this.strings.length) { + return this.strings[index - 391]; + } + return CFFStandardStrings[0]; + }, + add: function CFFStrings_add(value) { + this.strings.push(value); + }, + get count() { + return this.strings.length; + } + }; + return CFFStrings; +})(); - CFFFont.prototype = { - get numGlyphs() { - return this.cff.charStrings.count; +var CFFIndex = (function CFFIndexClosure() { + function CFFIndex() { + this.objects = []; + this.length = 0; + } + CFFIndex.prototype = { + add: function CFFIndex_add(data) { + this.length += data.length; + this.objects.push(data); }, - getCharset: function CFFFont_getCharset() { - return this.cff.charset.charset; + set: function CFFIndex_set(index, data) { + this.length += data.length - this.objects[index].length; + this.objects[index] = data; }, - getGlyphMapping: function CFFFont_getGlyphMapping() { - var cff = this.cff; - var properties = this.properties; - var charsets = cff.charset.charset; - var charCodeToGlyphId; - var glyphId; + get: function CFFIndex_get(index) { + return this.objects[index]; + }, + get count() { + return this.objects.length; + } + }; + return CFFIndex; +})(); - if (properties.composite) { - charCodeToGlyphId = Object.create(null); - if (cff.isCIDFont) { - // If the font is actually a CID font then we should use the charset - // to map CIDs to GIDs. - for (glyphId = 0; glyphId < charsets.length; glyphId++) { - var cid = charsets[glyphId]; - var charCode = properties.cMap.charCodeOf(cid); - charCodeToGlyphId[charCode] = glyphId; - } - } else { - // If it is NOT actually a CID font then CIDs should be mapped - // directly to GIDs. - for (glyphId = 0; glyphId < cff.charStrings.count; glyphId++) { - charCodeToGlyphId[glyphId] = glyphId; - } - } - return charCodeToGlyphId; +var CFFDict = (function CFFDictClosure() { + function CFFDict(tables, strings) { + this.keyToNameMap = tables.keyToNameMap; + this.nameToKeyMap = tables.nameToKeyMap; + this.defaults = tables.defaults; + this.types = tables.types; + this.opcodes = tables.opcodes; + this.order = tables.order; + this.strings = strings; + this.values = Object.create(null); + } + CFFDict.prototype = { + // value should always be an array + setByKey: function CFFDict_setByKey(key, value) { + if (!(key in this.keyToNameMap)) { + return false; } - - var encoding = cff.encoding ? cff.encoding.encoding : null; - charCodeToGlyphId = type1FontGlyphMapping(properties, encoding, charsets); - return charCodeToGlyphId; + // ignore empty values + if (value.length === 0) { + return true; + } + var type = this.types[key]; + // remove the array wrapping these types of values + if (type === 'num' || type === 'sid' || type === 'offset') { + value = value[0]; + } + this.values[key] = value; + return true; + }, + setByName: function CFFDict_setByName(name, value) { + if (!(name in this.nameToKeyMap)) { + error('Invalid dictionary name "' + name + '"'); + } + this.values[this.nameToKeyMap[name]] = value; + }, + hasName: function CFFDict_hasName(name) { + return this.nameToKeyMap[name] in this.values; + }, + getByName: function CFFDict_getByName(name) { + if (!(name in this.nameToKeyMap)) { + error('Invalid dictionary name "' + name + '"'); + } + var key = this.nameToKeyMap[name]; + if (!(key in this.values)) { + return this.defaults[key]; + } + return this.values[key]; + }, + removeByName: function CFFDict_removeByName(name) { + delete this.values[this.nameToKeyMap[name]]; + } + }; + CFFDict.createTables = function CFFDict_createTables(layout) { + var tables = { + keyToNameMap: {}, + nameToKeyMap: {}, + defaults: {}, + types: {}, + opcodes: {}, + order: [] + }; + for (var i = 0, ii = layout.length; i < ii; ++i) { + var entry = layout[i]; + var key = isArray(entry[0]) ? (entry[0][0] << 8) + entry[0][1] : entry[0]; + tables.keyToNameMap[key] = entry[1]; + tables.nameToKeyMap[entry[1]] = key; + tables.types[key] = entry[2]; + tables.defaults[key] = entry[3]; + tables.opcodes[key] = isArray(entry[0]) ? entry[0] : [entry[0]]; + tables.order.push(key); } + return tables; }; + return CFFDict; +})(); - return CFFFont; +var CFFTopDict = (function CFFTopDictClosure() { + var layout = [ + [[12, 30], 'ROS', ['sid', 'sid', 'num'], null], + [[12, 20], 'SyntheticBase', 'num', null], + [0, 'version', 'sid', null], + [1, 'Notice', 'sid', null], + [[12, 0], 'Copyright', 'sid', null], + [2, 'FullName', 'sid', null], + [3, 'FamilyName', 'sid', null], + [4, 'Weight', 'sid', null], + [[12, 1], 'isFixedPitch', 'num', 0], + [[12, 2], 'ItalicAngle', 'num', 0], + [[12, 3], 'UnderlinePosition', 'num', -100], + [[12, 4], 'UnderlineThickness', 'num', 50], + [[12, 5], 'PaintType', 'num', 0], + [[12, 6], 'CharstringType', 'num', 2], + [[12, 7], 'FontMatrix', ['num', 'num', 'num', 'num', 'num', 'num'], + [0.001, 0, 0, 0.001, 0, 0]], + [13, 'UniqueID', 'num', null], + [5, 'FontBBox', ['num', 'num', 'num', 'num'], [0, 0, 0, 0]], + [[12, 8], 'StrokeWidth', 'num', 0], + [14, 'XUID', 'array', null], + [15, 'charset', 'offset', 0], + [16, 'Encoding', 'offset', 0], + [17, 'CharStrings', 'offset', 0], + [18, 'Private', ['offset', 'offset'], null], + [[12, 21], 'PostScript', 'sid', null], + [[12, 22], 'BaseFontName', 'sid', null], + [[12, 23], 'BaseFontBlend', 'delta', null], + [[12, 31], 'CIDFontVersion', 'num', 0], + [[12, 32], 'CIDFontRevision', 'num', 0], + [[12, 33], 'CIDFontType', 'num', 0], + [[12, 34], 'CIDCount', 'num', 8720], + [[12, 35], 'UIDBase', 'num', null], + // XXX: CID Fonts on DirectWrite 6.1 only seem to work if FDSelect comes + // before FDArray. + [[12, 37], 'FDSelect', 'offset', null], + [[12, 36], 'FDArray', 'offset', null], + [[12, 38], 'FontName', 'sid', null] + ]; + var tables = null; + function CFFTopDict(strings) { + if (tables === null) { + tables = CFFDict.createTables(layout); + } + CFFDict.call(this, tables, strings); + this.privateDict = null; + } + CFFTopDict.prototype = Object.create(CFFDict.prototype); + return CFFTopDict; })(); -var CFFParser = (function CFFParserClosure() { - var CharstringValidationData = [ - null, - { id: 'hstem', min: 2, stackClearing: true, stem: true }, - null, - { id: 'vstem', min: 2, stackClearing: true, stem: true }, - { id: 'vmoveto', min: 1, stackClearing: true }, - { id: 'rlineto', min: 2, resetStack: true }, - { id: 'hlineto', min: 1, resetStack: true }, - { id: 'vlineto', min: 1, resetStack: true }, - { id: 'rrcurveto', min: 6, resetStack: true }, - null, - { id: 'callsubr', min: 1, undefStack: true }, - { id: 'return', min: 0, undefStack: true }, - null, // 12 - null, - { id: 'endchar', min: 0, stackClearing: true }, - null, - null, - null, - { id: 'hstemhm', min: 2, stackClearing: true, stem: true }, - { id: 'hintmask', min: 0, stackClearing: true }, - { id: 'cntrmask', min: 0, stackClearing: true }, - { id: 'rmoveto', min: 2, stackClearing: true }, - { id: 'hmoveto', min: 1, stackClearing: true }, - { id: 'vstemhm', min: 2, stackClearing: true, stem: true }, - { id: 'rcurveline', min: 8, resetStack: true }, - { id: 'rlinecurve', min: 8, resetStack: true }, - { id: 'vvcurveto', min: 4, resetStack: true }, - { id: 'hhcurveto', min: 4, resetStack: true }, - null, // shortint - { id: 'callgsubr', min: 1, undefStack: true }, - { id: 'vhcurveto', min: 4, resetStack: true }, - { id: 'hvcurveto', min: 4, resetStack: true } +var CFFPrivateDict = (function CFFPrivateDictClosure() { + var layout = [ + [6, 'BlueValues', 'delta', null], + [7, 'OtherBlues', 'delta', null], + [8, 'FamilyBlues', 'delta', null], + [9, 'FamilyOtherBlues', 'delta', null], + [[12, 9], 'BlueScale', 'num', 0.039625], + [[12, 10], 'BlueShift', 'num', 7], + [[12, 11], 'BlueFuzz', 'num', 1], + [10, 'StdHW', 'num', null], + [11, 'StdVW', 'num', null], + [[12, 12], 'StemSnapH', 'delta', null], + [[12, 13], 'StemSnapV', 'delta', null], + [[12, 14], 'ForceBold', 'num', 0], + [[12, 17], 'LanguageGroup', 'num', 0], + [[12, 18], 'ExpansionFactor', 'num', 0.06], + [[12, 19], 'initialRandomSeed', 'num', 0], + [20, 'defaultWidthX', 'num', 0], + [21, 'nominalWidthX', 'num', 0], + [19, 'Subrs', 'offset', null] ]; - var CharstringValidationData12 = [ - null, - null, - null, - { id: 'and', min: 2, stackDelta: -1 }, - { id: 'or', min: 2, stackDelta: -1 }, - { id: 'not', min: 1, stackDelta: 0 }, - null, - null, - null, - { id: 'abs', min: 1, stackDelta: 0 }, - { id: 'add', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] + stack[index - 1]; + var tables = null; + function CFFPrivateDict(strings) { + if (tables === null) { + tables = CFFDict.createTables(layout); + } + CFFDict.call(this, tables, strings); + this.subrsIndex = null; + } + CFFPrivateDict.prototype = Object.create(CFFDict.prototype); + return CFFPrivateDict; +})(); + +var CFFCharsetPredefinedTypes = { + ISO_ADOBE: 0, + EXPERT: 1, + EXPERT_SUBSET: 2 +}; +var CFFCharset = (function CFFCharsetClosure() { + function CFFCharset(predefined, format, charset, raw) { + this.predefined = predefined; + this.format = format; + this.charset = charset; + this.raw = raw; + } + return CFFCharset; +})(); + +var CFFEncoding = (function CFFEncodingClosure() { + function CFFEncoding(predefined, format, encoding, raw) { + this.predefined = predefined; + this.format = format; + this.encoding = encoding; + this.raw = raw; + } + return CFFEncoding; +})(); + +var CFFFDSelect = (function CFFFDSelectClosure() { + function CFFFDSelect(fdSelect, raw) { + this.fdSelect = fdSelect; + this.raw = raw; + } + CFFFDSelect.prototype = { + getFDIndex: function CFFFDSelect_get(glyphIndex) { + if (glyphIndex < 0 || glyphIndex >= this.fdSelect.length) { + return -1; } + return this.fdSelect[glyphIndex]; + } + }; + return CFFFDSelect; +})(); + +// Helper class to keep track of where an offset is within the data and helps +// filling in that offset once it's known. +var CFFOffsetTracker = (function CFFOffsetTrackerClosure() { + function CFFOffsetTracker() { + this.offsets = Object.create(null); + } + CFFOffsetTracker.prototype = { + isTracking: function CFFOffsetTracker_isTracking(key) { + return key in this.offsets; }, - { id: 'sub', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] - stack[index - 1]; + track: function CFFOffsetTracker_track(key, location) { + if (key in this.offsets) { + error('Already tracking location of ' + key); } + this.offsets[key] = location; }, - { id: 'div', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] / stack[index - 1]; + offset: function CFFOffsetTracker_offset(value) { + for (var key in this.offsets) { + this.offsets[key] += value; } }, - null, - { id: 'neg', min: 1, stackDelta: 0, - stackFn: function stack_div(stack, index) { - stack[index - 1] = -stack[index - 1]; + setEntryLocation: function CFFOffsetTracker_setEntryLocation(key, + values, + output) { + if (!(key in this.offsets)) { + error('Not tracking location of ' + key); } - }, - { id: 'eq', min: 2, stackDelta: -1 }, - null, - null, - { id: 'drop', min: 1, stackDelta: -1 }, - null, - { id: 'put', min: 2, stackDelta: -2 }, - { id: 'get', min: 1, stackDelta: 0 }, - { id: 'ifelse', min: 4, stackDelta: -3 }, - { id: 'random', min: 0, stackDelta: 1 }, - { id: 'mul', min: 2, stackDelta: -1, - stackFn: function stack_div(stack, index) { - stack[index - 2] = stack[index - 2] * stack[index - 1]; + var data = output.data; + var dataOffset = this.offsets[key]; + var size = 5; + for (var i = 0, ii = values.length; i < ii; ++i) { + var offset0 = i * size + dataOffset; + var offset1 = offset0 + 1; + var offset2 = offset0 + 2; + var offset3 = offset0 + 3; + var offset4 = offset0 + 4; + // It's easy to screw up offsets so perform this sanity check. + if (data[offset0] !== 0x1d || data[offset1] !== 0 || + data[offset2] !== 0 || data[offset3] !== 0 || data[offset4] !== 0) { + error('writing to an offset that is not empty'); + } + var value = values[i]; + data[offset0] = 0x1d; + data[offset1] = (value >> 24) & 0xFF; + data[offset2] = (value >> 16) & 0xFF; + data[offset3] = (value >> 8) & 0xFF; + data[offset4] = value & 0xFF; } - }, - null, - { id: 'sqrt', min: 1, stackDelta: 0 }, - { id: 'dup', min: 1, stackDelta: 1 }, - { id: 'exch', min: 2, stackDelta: 0 }, - { id: 'index', min: 2, stackDelta: 0 }, - { id: 'roll', min: 3, stackDelta: -2 }, - null, - null, - null, - { id: 'hflex', min: 7, resetStack: true }, - { id: 'flex', min: 13, resetStack: true }, - { id: 'hflex1', min: 9, resetStack: true }, - { id: 'flex1', min: 11, resetStack: true } - ]; + } + }; + return CFFOffsetTracker; +})(); - function CFFParser(file, properties) { - this.bytes = file.getBytes(); - this.properties = properties; +// Takes a CFF and converts it to the binary representation. +var CFFCompiler = (function CFFCompilerClosure() { + function CFFCompiler(cff) { + this.cff = cff; } - CFFParser.prototype = { - parse: function CFFParser_parse() { - var properties = this.properties; - var cff = new CFF(); - this.cff = cff; + CFFCompiler.prototype = { + compile: function CFFCompiler_compile() { + var cff = this.cff; + var output = { + data: [], + length: 0, + add: function CFFCompiler_add(data) { + this.data = this.data.concat(data); + this.length = this.data.length; + } + }; - // The first five sections must be in order, all the others are reached - // via offsets contained in one of the below. - var header = this.parseHeader(); - var nameIndex = this.parseIndex(header.endPos); - var topDictIndex = this.parseIndex(nameIndex.endPos); - var stringIndex = this.parseIndex(topDictIndex.endPos); - var globalSubrIndex = this.parseIndex(stringIndex.endPos); + // Compile the five entries that must be in order. + var header = this.compileHeader(cff.header); + output.add(header); - var topDictParsed = this.parseDict(topDictIndex.obj.get(0)); - var topDict = this.createDict(CFFTopDict, topDictParsed, cff.strings); + var nameIndex = this.compileNameIndex(cff.names); + output.add(nameIndex); - cff.header = header.obj; - cff.names = this.parseNameIndex(nameIndex.obj); - cff.strings = this.parseStringIndex(stringIndex.obj); - cff.topDict = topDict; - cff.globalSubrIndex = globalSubrIndex.obj; + if (cff.isCIDFont) { + // The spec is unclear on how font matrices should relate to each other + // when there is one in the main top dict and the sub top dicts. + // Windows handles this differently than linux and osx so we have to + // normalize to work on all. + // Rules based off of some mailing list discussions: + // - If main font has a matrix and subfont doesn't, use the main matrix. + // - If no main font matrix and there is a subfont matrix, use the + // subfont matrix. + // - If both have matrices, concat together. + // - If neither have matrices, use default. + // To make this work on all platforms we move the top matrix into each + // sub top dict and concat if necessary. + if (cff.topDict.hasName('FontMatrix')) { + var base = cff.topDict.getByName('FontMatrix'); + cff.topDict.removeByName('FontMatrix'); + for (var i = 0, ii = cff.fdArray.length; i < ii; i++) { + var subDict = cff.fdArray[i]; + var matrix = base.slice(0); + if (subDict.hasName('FontMatrix')) { + matrix = Util.transform(matrix, subDict.getByName('FontMatrix')); + } + subDict.setByName('FontMatrix', matrix); + } + } + } - this.parsePrivateDict(cff.topDict); + var compiled = this.compileTopDicts([cff.topDict], + output.length, + cff.isCIDFont); + output.add(compiled.output); + var topDictTracker = compiled.trackers[0]; - cff.isCIDFont = topDict.hasName('ROS'); + var stringIndex = this.compileStringIndex(cff.strings.strings); + output.add(stringIndex); - var charStringOffset = topDict.getByName('CharStrings'); - var charStringIndex = this.parseIndex(charStringOffset).obj; + var globalSubrIndex = this.compileIndex(cff.globalSubrIndex); + output.add(globalSubrIndex); - var fontMatrix = topDict.getByName('FontMatrix'); - if (fontMatrix) { - properties.fontMatrix = fontMatrix; + // Now start on the other entries that have no specfic order. + if (cff.encoding && cff.topDict.hasName('Encoding')) { + if (cff.encoding.predefined) { + topDictTracker.setEntryLocation('Encoding', [cff.encoding.format], + output); + } else { + var encoding = this.compileEncoding(cff.encoding); + topDictTracker.setEntryLocation('Encoding', [output.length], output); + output.add(encoding); + } } - var fontBBox = topDict.getByName('FontBBox'); - if (fontBBox) { - // adjusting ascent/descent - properties.ascent = fontBBox[3]; - properties.descent = fontBBox[1]; - properties.ascentScaled = true; + if (cff.charset && cff.topDict.hasName('charset')) { + if (cff.charset.predefined) { + topDictTracker.setEntryLocation('charset', [cff.charset.format], + output); + } else { + var charset = this.compileCharset(cff.charset); + topDictTracker.setEntryLocation('charset', [output.length], output); + output.add(charset); + } } - var charset, encoding; + var charStrings = this.compileCharStrings(cff.charStrings); + topDictTracker.setEntryLocation('CharStrings', [output.length], output); + output.add(charStrings); + if (cff.isCIDFont) { - var fdArrayIndex = this.parseIndex(topDict.getByName('FDArray')).obj; - for (var i = 0, ii = fdArrayIndex.count; i < ii; ++i) { - var dictRaw = fdArrayIndex.get(i); - var fontDict = this.createDict(CFFTopDict, this.parseDict(dictRaw), - cff.strings); - this.parsePrivateDict(fontDict); - cff.fdArray.push(fontDict); - } - // cid fonts don't have an encoding - encoding = null; - charset = this.parseCharsets(topDict.getByName('charset'), - charStringIndex.count, cff.strings, true); - cff.fdSelect = this.parseFDSelect(topDict.getByName('FDSelect'), - charStringIndex.count); - } else { - charset = this.parseCharsets(topDict.getByName('charset'), - charStringIndex.count, cff.strings, false); - encoding = this.parseEncoding(topDict.getByName('Encoding'), - properties, - cff.strings, charset.charset); + // For some reason FDSelect must be in front of FDArray on windows. OSX + // and linux don't seem to care. + topDictTracker.setEntryLocation('FDSelect', [output.length], output); + var fdSelect = this.compileFDSelect(cff.fdSelect.raw); + output.add(fdSelect); + // It is unclear if the sub font dictionary can have CID related + // dictionary keys, but the sanitizer doesn't like them so remove them. + compiled = this.compileTopDicts(cff.fdArray, output.length, true); + topDictTracker.setEntryLocation('FDArray', [output.length], output); + output.add(compiled.output); + var fontDictTrackers = compiled.trackers; + + this.compilePrivateDicts(cff.fdArray, fontDictTrackers, output); } - cff.charset = charset; - cff.encoding = encoding; + this.compilePrivateDicts([cff.topDict], [topDictTracker], output); - var charStringsAndSeacs = this.parseCharStrings( - charStringIndex, - topDict.privateDict.subrsIndex, - globalSubrIndex.obj, - cff.fdSelect, - cff.fdArray); - cff.charStrings = charStringsAndSeacs.charStrings; - cff.seacs = charStringsAndSeacs.seacs; - cff.widths = charStringsAndSeacs.widths; + // If the font data ends with INDEX whose object data is zero-length, + // the sanitizer will bail out. Add a dummy byte to avoid that. + output.add([0]); - return cff; + return output.data; }, - parseHeader: function CFFParser_parseHeader() { - var bytes = this.bytes; - var bytesLength = bytes.length; - var offset = 0; - - // Prevent an infinite loop, by checking that the offset is within the - // bounds of the bytes array. Necessary in empty, or invalid, font files. - while (offset < bytesLength && bytes[offset] !== 1) { - ++offset; - } - if (offset >= bytesLength) { - error('Invalid CFF header'); - } else if (offset !== 0) { - info('cff data is shifted'); - bytes = bytes.subarray(offset); - this.bytes = bytes; + encodeNumber: function CFFCompiler_encodeNumber(value) { + if (parseFloat(value) === parseInt(value, 10) && !isNaN(value)) { // isInt + return this.encodeInteger(value); + } else { + return this.encodeFloat(value); } - var major = bytes[0]; - var minor = bytes[1]; - var hdrSize = bytes[2]; - var offSize = bytes[3]; - var header = new CFFHeader(major, minor, hdrSize, offSize); - return { obj: header, endPos: hdrSize }; }, - parseDict: function CFFParser_parseDict(dict) { - var pos = 0; + encodeFloat: function CFFCompiler_encodeFloat(num) { + var value = num.toString(); - function parseOperand() { - var value = dict[pos++]; - if (value === 30) { - return parseFloatOperand(pos); - } else if (value === 28) { - value = dict[pos++]; - value = ((value << 24) | (dict[pos++] << 16)) >> 16; - return value; - } else if (value === 29) { - value = dict[pos++]; - value = (value << 8) | dict[pos++]; - value = (value << 8) | dict[pos++]; - value = (value << 8) | dict[pos++]; - return value; - } else if (value >= 32 && value <= 246) { - return value - 139; - } else if (value >= 247 && value <= 250) { - return ((value - 247) * 256) + dict[pos++] + 108; - } else if (value >= 251 && value <= 254) { - return -((value - 251) * 256) - dict[pos++] - 108; + // rounding inaccurate doubles + var m = /\.(\d*?)(?:9{5,20}|0{5,20})\d{0,2}(?:e(.+)|$)/.exec(value); + if (m) { + var epsilon = parseFloat('1e' + ((m[2] ? +m[2] : 0) + m[1].length)); + value = (Math.round(num * epsilon) / epsilon).toString(); + } + + var nibbles = ''; + var i, ii; + for (i = 0, ii = value.length; i < ii; ++i) { + var a = value[i]; + if (a === 'e') { + nibbles += value[++i] === '-' ? 'c' : 'b'; + } else if (a === '.') { + nibbles += 'a'; + } else if (a === '-') { + nibbles += 'e'; } else { - error('255 is not a valid DICT command'); + nibbles += a; } - return -1; } + nibbles += (nibbles.length & 1) ? 'f' : 'ff'; + var out = [30]; + for (i = 0, ii = nibbles.length; i < ii; i += 2) { + out.push(parseInt(nibbles.substr(i, 2), 16)); + } + return out; + }, + encodeInteger: function CFFCompiler_encodeInteger(value) { + var code; + if (value >= -107 && value <= 107) { + code = [value + 139]; + } else if (value >= 108 && value <= 1131) { + value = [value - 108]; + code = [(value >> 8) + 247, value & 0xFF]; + } else if (value >= -1131 && value <= -108) { + value = -value - 108; + code = [(value >> 8) + 251, value & 0xFF]; + } else if (value >= -32768 && value <= 32767) { + code = [0x1c, (value >> 8) & 0xFF, value & 0xFF]; + } else { + code = [0x1d, + (value >> 24) & 0xFF, + (value >> 16) & 0xFF, + (value >> 8) & 0xFF, + value & 0xFF]; + } + return code; + }, + compileHeader: function CFFCompiler_compileHeader(header) { + return [ + header.major, + header.minor, + header.hdrSize, + header.offSize + ]; + }, + compileNameIndex: function CFFCompiler_compileNameIndex(names) { + var nameIndex = new CFFIndex(); + for (var i = 0, ii = names.length; i < ii; ++i) { + nameIndex.add(stringToBytes(names[i])); + } + return this.compileIndex(nameIndex); + }, + compileTopDicts: function CFFCompiler_compileTopDicts(dicts, + length, + removeCidKeys) { + var fontDictTrackers = []; + var fdArrayIndex = new CFFIndex(); + for (var i = 0, ii = dicts.length; i < ii; ++i) { + var fontDict = dicts[i]; + if (removeCidKeys) { + fontDict.removeByName('CIDFontVersion'); + fontDict.removeByName('CIDFontRevision'); + fontDict.removeByName('CIDFontType'); + fontDict.removeByName('CIDCount'); + fontDict.removeByName('UIDBase'); + } + var fontDictTracker = new CFFOffsetTracker(); + var fontDictData = this.compileDict(fontDict, fontDictTracker); + fontDictTrackers.push(fontDictTracker); + fdArrayIndex.add(fontDictData); + fontDictTracker.offset(length); + } + fdArrayIndex = this.compileIndex(fdArrayIndex, fontDictTrackers); + return { + trackers: fontDictTrackers, + output: fdArrayIndex + }; + }, + compilePrivateDicts: function CFFCompiler_compilePrivateDicts(dicts, + trackers, + output) { + for (var i = 0, ii = dicts.length; i < ii; ++i) { + var fontDict = dicts[i]; + assert(fontDict.privateDict && fontDict.hasName('Private'), + 'There must be an private dictionary.'); + var privateDict = fontDict.privateDict; + var privateDictTracker = new CFFOffsetTracker(); + var privateDictData = this.compileDict(privateDict, privateDictTracker); - function parseFloatOperand() { - var str = ''; - var eof = 15; - var lookup = ['0', '1', '2', '3', '4', '5', '6', '7', '8', - '9', '.', 'E', 'E-', null, '-']; - var length = dict.length; - while (pos < length) { - var b = dict[pos++]; - var b1 = b >> 4; - var b2 = b & 15; + var outputLength = output.length; + privateDictTracker.offset(outputLength); + if (!privateDictData.length) { + // The private dictionary was empty, set the output length to zero to + // ensure the offset length isn't out of bounds in the eyes of the + // sanitizer. + outputLength = 0; + } - if (b1 === eof) { - break; - } - str += lookup[b1]; + trackers[i].setEntryLocation('Private', + [privateDictData.length, outputLength], + output); + output.add(privateDictData); - if (b2 === eof) { - break; - } - str += lookup[b2]; + if (privateDict.subrsIndex && privateDict.hasName('Subrs')) { + var subrs = this.compileIndex(privateDict.subrsIndex); + privateDictTracker.setEntryLocation('Subrs', [privateDictData.length], + output); + output.add(subrs); } - return parseFloat(str); } - - var operands = []; - var entries = []; - - pos = 0; - var end = dict.length; - while (pos < end) { - var b = dict[pos]; - if (b <= 21) { - if (b === 12) { - b = (b << 8) | dict[++pos]; - } - entries.push([b, operands]); - operands = []; - ++pos; - } else { - operands.push(parseOperand()); + }, + compileDict: function CFFCompiler_compileDict(dict, offsetTracker) { + var out = []; + // The dictionary keys must be in a certain order. + var order = dict.order; + for (var i = 0; i < order.length; ++i) { + var key = order[i]; + if (!(key in dict.values)) { + continue; + } + var values = dict.values[key]; + var types = dict.types[key]; + if (!isArray(types)) { + types = [types]; + } + if (!isArray(values)) { + values = [values]; } - } - return entries; - }, - parseIndex: function CFFParser_parseIndex(pos) { - var cffIndex = new CFFIndex(); - var bytes = this.bytes; - var count = (bytes[pos++] << 8) | bytes[pos++]; - var offsets = []; - var end = pos; - var i, ii; - if (count !== 0) { - var offsetSize = bytes[pos++]; - // add 1 for offset to determine size of last object - var startPos = pos + ((count + 1) * offsetSize) - 1; + // Remove any empty dict values. + if (values.length === 0) { + continue; + } - for (i = 0, ii = count + 1; i < ii; ++i) { - var offset = 0; - for (var j = 0; j < offsetSize; ++j) { - offset <<= 8; - offset += bytes[pos++]; + for (var j = 0, jj = types.length; j < jj; ++j) { + var type = types[j]; + var value = values[j]; + switch (type) { + case 'num': + case 'sid': + out = out.concat(this.encodeNumber(value)); + break; + case 'offset': + // For offsets we just insert a 32bit integer so we don't have to + // deal with figuring out the length of the offset when it gets + // replaced later on by the compiler. + var name = dict.keyToNameMap[key]; + // Some offsets have the offset and the length, so just record the + // position of the first one. + if (!offsetTracker.isTracking(name)) { + offsetTracker.track(name, out.length); + } + out = out.concat([0x1d, 0, 0, 0, 0]); + break; + case 'array': + case 'delta': + out = out.concat(this.encodeNumber(value)); + for (var k = 1, kk = values.length; k < kk; ++k) { + out = out.concat(this.encodeNumber(values[k])); + } + break; + default: + error('Unknown data type of ' + type); + break; } - offsets.push(startPos + offset); } - end = offsets[count]; - } - for (i = 0, ii = offsets.length - 1; i < ii; ++i) { - var offsetStart = offsets[i]; - var offsetEnd = offsets[i + 1]; - cffIndex.add(bytes.subarray(offsetStart, offsetEnd)); + out = out.concat(dict.opcodes[key]); } - return {obj: cffIndex, endPos: end}; + return out; }, - parseNameIndex: function CFFParser_parseNameIndex(index) { - var names = []; - for (var i = 0, ii = index.count; i < ii; ++i) { - var name = index.get(i); - // OTS doesn't allow names to be over 127 characters. - var length = Math.min(name.length, 127); - var data = []; - // OTS also only permits certain characters in the name. - for (var j = 0; j < length; ++j) { - var c = name[j]; - if (j === 0 && c === 0) { - data[j] = c; - continue; - } - if ((c < 33 || c > 126) || c === 91 /* [ */ || c === 93 /* ] */ || - c === 40 /* ( */ || c === 41 /* ) */ || c === 123 /* { */ || - c === 125 /* } */ || c === 60 /* < */ || c === 62 /* > */ || - c === 47 /* / */ || c === 37 /* % */ || c === 35 /* # */) { - data[j] = 95; - continue; - } - data[j] = c; - } - names.push(bytesToString(data)); + compileStringIndex: function CFFCompiler_compileStringIndex(strings) { + var stringIndex = new CFFIndex(); + for (var i = 0, ii = strings.length; i < ii; ++i) { + stringIndex.add(stringToBytes(strings[i])); } - return names; + return this.compileIndex(stringIndex); }, - parseStringIndex: function CFFParser_parseStringIndex(index) { - var strings = new CFFStrings(); - for (var i = 0, ii = index.count; i < ii; ++i) { - var data = index.get(i); - strings.add(bytesToString(data)); - } - return strings; + compileGlobalSubrIndex: function CFFCompiler_compileGlobalSubrIndex() { + var globalSubrIndex = this.cff.globalSubrIndex; + this.out.writeByteArray(this.compileIndex(globalSubrIndex)); }, - createDict: function CFFParser_createDict(Type, dict, strings) { - var cffDict = new Type(strings); - for (var i = 0, ii = dict.length; i < ii; ++i) { - var pair = dict[i]; - var key = pair[0]; - var value = pair[1]; - cffDict.setByKey(key, value); + compileCharStrings: function CFFCompiler_compileCharStrings(charStrings) { + return this.compileIndex(charStrings); + }, + compileCharset: function CFFCompiler_compileCharset(charset) { + return this.compileTypedArray(charset.raw); + }, + compileEncoding: function CFFCompiler_compileEncoding(encoding) { + return this.compileTypedArray(encoding.raw); + }, + compileFDSelect: function CFFCompiler_compileFDSelect(fdSelect) { + return this.compileTypedArray(fdSelect); + }, + compileTypedArray: function CFFCompiler_compileTypedArray(data) { + var out = []; + for (var i = 0, ii = data.length; i < ii; ++i) { + out[i] = data[i]; } - return cffDict; + return out; }, - parseCharString: function CFFParser_parseCharString(state, data, - localSubrIndex, - globalSubrIndex) { - if (state.callDepth > MAX_SUBR_NESTING) { - return false; + compileIndex: function CFFCompiler_compileIndex(index, trackers) { + trackers = trackers || []; + var objects = index.objects; + // First 2 bytes contains the number of objects contained into this index + var count = objects.length; + + // If there is no object, just create an index. This technically + // should just be [0, 0] but OTS has an issue with that. + if (count === 0) { + return [0, 0, 0]; } - var stackSize = state.stackSize; - var stack = state.stack; - var length = data.length; + var data = [(count >> 8) & 0xFF, count & 0xff]; - for (var j = 0; j < length;) { - var value = data[j++]; - var validationCommand = null; - if (value === 12) { - var q = data[j++]; - if (q === 0) { - // The CFF specification state that the 'dotsection' command - // (12, 0) is deprecated and treated as a no-op, but all Type2 - // charstrings processors should support them. Unfortunately - // the font sanitizer don't. As a workaround the sequence (12, 0) - // is replaced by a useless (0, hmoveto). - data[j - 2] = 139; - data[j - 1] = 22; - stackSize = 0; - } else { - validationCommand = CharstringValidationData12[q]; - } - } else if (value === 28) { // number (16 bit) - stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16)) >> 16; - j += 2; - stackSize++; - } else if (value === 14) { - if (stackSize >= 4) { - stackSize -= 4; - if (SEAC_ANALYSIS_ENABLED) { - state.seac = stack.slice(stackSize, stackSize + 4); - return false; - } - } - validationCommand = CharstringValidationData[value]; - } else if (value >= 32 && value <= 246) { // number - stack[stackSize] = value - 139; - stackSize++; - } else if (value >= 247 && value <= 254) { // number (+1 bytes) - stack[stackSize] = (value < 251 ? - ((value - 247) << 8) + data[j] + 108 : - -((value - 251) << 8) - data[j] - 108); - j++; - stackSize++; - } else if (value === 255) { // number (32 bit) - stack[stackSize] = ((data[j] << 24) | (data[j + 1] << 16) | - (data[j + 2] << 8) | data[j + 3]) / 65536; - j += 4; - stackSize++; - } else if (value === 19 || value === 20) { - state.hints += stackSize >> 1; - // skipping right amount of hints flag data - j += (state.hints + 7) >> 3; - stackSize %= 2; - validationCommand = CharstringValidationData[value]; - } else if (value === 10 || value === 29) { - var subrsIndex; - if (value === 10) { - subrsIndex = localSubrIndex; - } else { - subrsIndex = globalSubrIndex; - } - if (!subrsIndex) { - validationCommand = CharstringValidationData[value]; - warn('Missing subrsIndex for ' + validationCommand.id); - return false; - } - var bias = 32768; - if (subrsIndex.count < 1240) { - bias = 107; - } else if (subrsIndex.count < 33900) { - bias = 1131; - } - var subrNumber = stack[--stackSize] + bias; - if (subrNumber < 0 || subrNumber >= subrsIndex.count) { - validationCommand = CharstringValidationData[value]; - warn('Out of bounds subrIndex for ' + validationCommand.id); - return false; - } - state.stackSize = stackSize; - state.callDepth++; - var valid = this.parseCharString(state, subrsIndex.get(subrNumber), - localSubrIndex, globalSubrIndex); - if (!valid) { - return false; - } - state.callDepth--; - stackSize = state.stackSize; - continue; - } else if (value === 11) { - state.stackSize = stackSize; - return true; + var lastOffset = 1, i; + for (i = 0; i < count; ++i) { + lastOffset += objects[i].length; + } + + var offsetSize; + if (lastOffset < 0x100) { + offsetSize = 1; + } else if (lastOffset < 0x10000) { + offsetSize = 2; + } else if (lastOffset < 0x1000000) { + offsetSize = 3; + } else { + offsetSize = 4; + } + + // Next byte contains the offset size use to reference object in the file + data.push(offsetSize); + + // Add another offset after this one because we need a new offset + var relativeOffset = 1; + for (i = 0; i < count + 1; i++) { + if (offsetSize === 1) { + data.push(relativeOffset & 0xFF); + } else if (offsetSize === 2) { + data.push((relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); + } else if (offsetSize === 3) { + data.push((relativeOffset >> 16) & 0xFF, + (relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); } else { - validationCommand = CharstringValidationData[value]; + data.push((relativeOffset >>> 24) & 0xFF, + (relativeOffset >> 16) & 0xFF, + (relativeOffset >> 8) & 0xFF, + relativeOffset & 0xFF); } - if (validationCommand) { - if (validationCommand.stem) { - state.hints += stackSize >> 1; - } - if ('min' in validationCommand) { - if (!state.undefStack && stackSize < validationCommand.min) { - warn('Not enough parameters for ' + validationCommand.id + - '; actual: ' + stackSize + - ', expected: ' + validationCommand.min); - return false; - } - } - if (state.firstStackClearing && validationCommand.stackClearing) { - state.firstStackClearing = false; - // the optional character width can be found before the first - // stack-clearing command arguments - stackSize -= validationCommand.min; - if (stackSize >= 2 && validationCommand.stem) { - // there are even amount of arguments for stem commands - stackSize %= 2; - } else if (stackSize > 1) { - warn('Found too many parameters for stack-clearing command'); - } - if (stackSize > 0 && stack[stackSize - 1] >= 0) { - state.width = stack[stackSize - 1]; - } - } - if ('stackDelta' in validationCommand) { - if ('stackFn' in validationCommand) { - validationCommand.stackFn(stack, stackSize); - } - stackSize += validationCommand.stackDelta; - } else if (validationCommand.stackClearing) { - stackSize = 0; - } else if (validationCommand.resetStack) { - stackSize = 0; - state.undefStack = false; - } else if (validationCommand.undefStack) { - stackSize = 0; - state.undefStack = true; - state.firstStackClearing = false; - } + + if (objects[i]) { + relativeOffset += objects[i].length; } } - state.stackSize = stackSize; - return true; - }, - parseCharStrings: function CFFParser_parseCharStrings(charStrings, - localSubrIndex, - globalSubrIndex, - fdSelect, - fdArray) { - var seacs = []; - var widths = []; - var count = charStrings.count; - for (var i = 0; i < count; i++) { - var charstring = charStrings.get(i); - var state = { - callDepth: 0, - stackSize: 0, - stack: [], - undefStack: true, - hints: 0, - firstStackClearing: true, - seac: null, - width: null - }; - var valid = true; - var localSubrToUse = null; - if (fdSelect && fdArray.length) { - var fdIndex = fdSelect.getFDIndex(i); - if (fdIndex === -1) { - warn('Glyph index is not in fd select.'); - valid = false; - } - if (fdIndex >= fdArray.length) { - warn('Invalid fd index for glyph index.'); - valid = false; - } - if (valid) { - localSubrToUse = fdArray[fdIndex].privateDict.subrsIndex; - } - } else if (localSubrIndex) { - localSubrToUse = localSubrIndex; - } - if (valid) { - valid = this.parseCharString(state, charstring, localSubrToUse, - globalSubrIndex); - } - if (state.width !== null) { - widths[i] = state.width; + + for (i = 0; i < count; i++) { + // Notify the tracker where the object will be offset in the data. + if (trackers[i]) { + trackers[i].offset(data.length); } - if (state.seac !== null) { - seacs[i] = state.seac; + for (var j = 0, jj = objects[i].length; j < jj; j++) { + data.push(objects[i][j]); } - if (!valid) { - // resetting invalid charstring to single 'endchar' - charStrings.set(i, new Uint8Array([14])); + } + return data; + } + }; + return CFFCompiler; +})(); + +function _enableSeacAnalysis(enabled) { + exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED = enabled; +} + +// Workaround for seac on Windows. +(function checkSeacSupport() { + if (typeof navigator !== 'undefined' && /Windows/.test(navigator.userAgent)) { + SEAC_ANALYSIS_ENABLED = true; + } +})(); + +// Workaround for Private Use Area characters in Chrome on Windows +// http://code.google.com/p/chromium/issues/detail?id=122465 +// https://github.com/mozilla/pdf.js/issues/1689 +(function checkChromeWindows() { + if (typeof navigator !== 'undefined' && + /Windows.*Chrome/.test(navigator.userAgent)) { + SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = true; + } +})(); + +exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED; +exports.CFFCompiler = CFFCompiler; +exports.CFFIndex = CFFIndex; +exports.CFFParser = CFFParser; +exports.CFFStrings = CFFStrings; +exports.ErrorFont = ErrorFont; +exports.FontFlags = FontFlags; +exports.Font = Font; +exports.IdentityToUnicodeMap = IdentityToUnicodeMap; +exports.ToUnicodeMap = ToUnicodeMap; +exports.Type1Parser = Type1Parser; +exports.getFontType = getFontType; +exports._enableSeacAnalysis = _enableSeacAnalysis; + +// TODO refactor to remove cyclic dependency on font_renderer.js +coreFontRenderer._setCoreFonts(exports); +})); + + +(function (root, factory) { + { + factory((root.pdfjsCoreFunction = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCorePsParser); + } +}(this, function (exports, sharedUtil, corePrimitives, corePsParser) { + +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isBool = sharedUtil.isBool; +var isDict = corePrimitives.isDict; +var isStream = corePrimitives.isStream; +var PostScriptLexer = corePsParser.PostScriptLexer; +var PostScriptParser = corePsParser.PostScriptParser; + +var PDFFunction = (function PDFFunctionClosure() { + var CONSTRUCT_SAMPLED = 0; + var CONSTRUCT_INTERPOLATED = 2; + var CONSTRUCT_STICHED = 3; + var CONSTRUCT_POSTSCRIPT = 4; + + return { + getSampleArray: function PDFFunction_getSampleArray(size, outputSize, bps, + str) { + var i, ii; + var length = 1; + for (i = 0, ii = size.length; i < ii; i++) { + length *= size[i]; + } + length *= outputSize; + + var array = new Array(length); + var codeSize = 0; + var codeBuf = 0; + // 32 is a valid bps so shifting won't work + var sampleMul = 1.0 / (Math.pow(2.0, bps) - 1); + + var strBytes = str.getBytes((length * bps + 7) / 8); + var strIdx = 0; + for (i = 0; i < length; i++) { + while (codeSize < bps) { + codeBuf <<= 8; + codeBuf |= strBytes[strIdx++]; + codeSize += 8; } + codeSize -= bps; + array[i] = (codeBuf >> codeSize) * sampleMul; + codeBuf &= (1 << codeSize) - 1; } - return { charStrings: charStrings, seacs: seacs, widths: widths }; + return array; }, - emptyPrivateDictionary: - function CFFParser_emptyPrivateDictionary(parentDict) { - var privateDict = this.createDict(CFFPrivateDict, [], - parentDict.strings); - parentDict.setByKey(18, [0, 0]); - parentDict.privateDict = privateDict; + + getIR: function PDFFunction_getIR(xref, fn) { + var dict = fn.dict; + if (!dict) { + dict = fn; + } + + var types = [this.constructSampled, + null, + this.constructInterpolated, + this.constructStiched, + this.constructPostScript]; + + var typeNum = dict.get('FunctionType'); + var typeFn = types[typeNum]; + if (!typeFn) { + error('Unknown type of function'); + } + + return typeFn.call(this, fn, dict, xref); + }, + + fromIR: function PDFFunction_fromIR(IR) { + var type = IR[0]; + switch (type) { + case CONSTRUCT_SAMPLED: + return this.constructSampledFromIR(IR); + case CONSTRUCT_INTERPOLATED: + return this.constructInterpolatedFromIR(IR); + case CONSTRUCT_STICHED: + return this.constructStichedFromIR(IR); + //case CONSTRUCT_POSTSCRIPT: + default: + return this.constructPostScriptFromIR(IR); + } }, - parsePrivateDict: function CFFParser_parsePrivateDict(parentDict) { - // no private dict, do nothing - if (!parentDict.hasName('Private')) { - this.emptyPrivateDictionary(parentDict); - return; + + parse: function PDFFunction_parse(xref, fn) { + var IR = this.getIR(xref, fn); + return this.fromIR(IR); + }, + + parseArray: function PDFFunction_parseArray(xref, fnObj) { + if (!isArray(fnObj)) { + // not an array -- parsing as regular function + return this.parse(xref, fnObj); } - var privateOffset = parentDict.getByName('Private'); - // make sure the params are formatted correctly - if (!isArray(privateOffset) || privateOffset.length !== 2) { - parentDict.removeByName('Private'); - return; + + var fnArray = []; + for (var j = 0, jj = fnObj.length; j < jj; j++) { + var obj = xref.fetchIfRef(fnObj[j]); + fnArray.push(PDFFunction.parse(xref, obj)); } - var size = privateOffset[0]; - var offset = privateOffset[1]; - // remove empty dicts or ones that refer to invalid location - if (size === 0 || offset >= this.bytes.length) { - this.emptyPrivateDictionary(parentDict); - return; + return function (src, srcOffset, dest, destOffset) { + for (var i = 0, ii = fnArray.length; i < ii; i++) { + fnArray[i](src, srcOffset, dest, destOffset + i); + } + }; + }, + + constructSampled: function PDFFunction_constructSampled(str, dict) { + function toMultiArray(arr) { + var inputLength = arr.length; + var out = []; + var index = 0; + for (var i = 0; i < inputLength; i += 2) { + out[index] = [arr[i], arr[i + 1]]; + ++index; + } + return out; } + var domain = dict.get('Domain'); + var range = dict.get('Range'); - var privateDictEnd = offset + size; - var dictData = this.bytes.subarray(offset, privateDictEnd); - var dict = this.parseDict(dictData); - var privateDict = this.createDict(CFFPrivateDict, dict, - parentDict.strings); - parentDict.privateDict = privateDict; + if (!domain || !range) { + error('No domain or range'); + } - // Parse the Subrs index also since it's relative to the private dict. - if (!privateDict.getByName('Subrs')) { - return; + var inputSize = domain.length / 2; + var outputSize = range.length / 2; + + domain = toMultiArray(domain); + range = toMultiArray(range); + + var size = dict.get('Size'); + var bps = dict.get('BitsPerSample'); + var order = dict.get('Order') || 1; + if (order !== 1) { + // No description how cubic spline interpolation works in PDF32000:2008 + // As in poppler, ignoring order, linear interpolation may work as good + info('No support for cubic spline interpolation: ' + order); } - var subrsOffset = privateDict.getByName('Subrs'); - var relativeOffset = offset + subrsOffset; - // Validate the offset. - if (subrsOffset === 0 || relativeOffset >= this.bytes.length) { - this.emptyPrivateDictionary(parentDict); - return; + + var encode = dict.get('Encode'); + if (!encode) { + encode = []; + for (var i = 0; i < inputSize; ++i) { + encode.push(0); + encode.push(size[i] - 1); + } } - var subrsIndex = this.parseIndex(relativeOffset); - privateDict.subrsIndex = subrsIndex.obj; + encode = toMultiArray(encode); + + var decode = dict.get('Decode'); + if (!decode) { + decode = range; + } else { + decode = toMultiArray(decode); + } + + var samples = this.getSampleArray(size, outputSize, bps, str); + + return [ + CONSTRUCT_SAMPLED, inputSize, domain, encode, decode, samples, size, + outputSize, Math.pow(2, bps) - 1, range + ]; }, - parseCharsets: function CFFParser_parseCharsets(pos, length, strings, cid) { - if (pos === 0) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.ISO_ADOBE, - ISOAdobeCharset); - } else if (pos === 1) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT, - ExpertCharset); - } else if (pos === 2) { - return new CFFCharset(true, CFFCharsetPredefinedTypes.EXPERT_SUBSET, - ExpertSubsetCharset); + + constructSampledFromIR: function PDFFunction_constructSampledFromIR(IR) { + // See chapter 3, page 109 of the PDF reference + function interpolate(x, xmin, xmax, ymin, ymax) { + return ymin + ((x - xmin) * ((ymax - ymin) / (xmax - xmin))); } - var bytes = this.bytes; - var start = pos; - var format = bytes[pos++]; - var charset = ['.notdef']; - var id, count, i; + return function constructSampledFromIRResult(src, srcOffset, + dest, destOffset) { + // See chapter 3, page 110 of the PDF reference. + var m = IR[1]; + var domain = IR[2]; + var encode = IR[3]; + var decode = IR[4]; + var samples = IR[5]; + var size = IR[6]; + var n = IR[7]; + //var mask = IR[8]; + var range = IR[9]; - // subtract 1 for the .notdef glyph - length -= 1; + // Building the cube vertices: its part and sample index + // http://rjwagner49.com/Mathematics/Interpolation.pdf + var cubeVertices = 1 << m; + var cubeN = new Float64Array(cubeVertices); + var cubeVertex = new Uint32Array(cubeVertices); + var i, j; + for (j = 0; j < cubeVertices; j++) { + cubeN[j] = 1; + } - switch (format) { - case 0: - for (i = 0; i < length; i++) { - id = (bytes[pos++] << 8) | bytes[pos++]; - charset.push(cid ? id : strings.get(id)); - } - break; - case 1: - while (charset.length <= length) { - id = (bytes[pos++] << 8) | bytes[pos++]; - count = bytes[pos++]; - for (i = 0; i <= count; i++) { - charset.push(cid ? id++ : strings.get(id++)); + var k = n, pos = 1; + // Map x_i to y_j for 0 <= i < m using the sampled function. + for (i = 0; i < m; ++i) { + // x_i' = min(max(x_i, Domain_2i), Domain_2i+1) + var domain_2i = domain[i][0]; + var domain_2i_1 = domain[i][1]; + var xi = Math.min(Math.max(src[srcOffset +i], domain_2i), + domain_2i_1); + + // e_i = Interpolate(x_i', Domain_2i, Domain_2i+1, + // Encode_2i, Encode_2i+1) + var e = interpolate(xi, domain_2i, domain_2i_1, + encode[i][0], encode[i][1]); + + // e_i' = min(max(e_i, 0), Size_i - 1) + var size_i = size[i]; + e = Math.min(Math.max(e, 0), size_i - 1); + + // Adjusting the cube: N and vertex sample index + var e0 = e < size_i - 1 ? Math.floor(e) : e - 1; // e1 = e0 + 1; + var n0 = e0 + 1 - e; // (e1 - e) / (e1 - e0); + var n1 = e - e0; // (e - e0) / (e1 - e0); + var offset0 = e0 * k; + var offset1 = offset0 + k; // e1 * k + for (j = 0; j < cubeVertices; j++) { + if (j & pos) { + cubeN[j] *= n1; + cubeVertex[j] += offset1; + } else { + cubeN[j] *= n0; + cubeVertex[j] += offset0; } } - break; - case 2: - while (charset.length <= length) { - id = (bytes[pos++] << 8) | bytes[pos++]; - count = (bytes[pos++] << 8) | bytes[pos++]; - for (i = 0; i <= count; i++) { - charset.push(cid ? id++ : strings.get(id++)); - } + + k *= size_i; + pos <<= 1; + } + + for (j = 0; j < n; ++j) { + // Sum all cube vertices' samples portions + var rj = 0; + for (i = 0; i < cubeVertices; i++) { + rj += samples[cubeVertex[i] + j] * cubeN[i]; } - break; - default: - error('Unknown charset format'); + + // r_j' = Interpolate(r_j, 0, 2^BitsPerSample - 1, + // Decode_2j, Decode_2j+1) + rj = interpolate(rj, 0, 1, decode[j][0], decode[j][1]); + + // y_j = min(max(r_j, range_2j), range_2j+1) + dest[destOffset + j] = Math.min(Math.max(rj, range[j][0]), + range[j][1]); + } + }; + }, + + constructInterpolated: function PDFFunction_constructInterpolated(str, + dict) { + var c0 = dict.get('C0') || [0]; + var c1 = dict.get('C1') || [1]; + var n = dict.get('N'); + + if (!isArray(c0) || !isArray(c1)) { + error('Illegal dictionary for interpolated function'); } - // Raw won't be needed if we actually compile the charset. - var end = pos; - var raw = bytes.subarray(start, end); - return new CFFCharset(false, format, charset, raw); + var length = c0.length; + var diff = []; + for (var i = 0; i < length; ++i) { + diff.push(c1[i] - c0[i]); + } + + return [CONSTRUCT_INTERPOLATED, c0, diff, n]; }, - parseEncoding: function CFFParser_parseEncoding(pos, - properties, - strings, - charset) { - var encoding = Object.create(null); - var bytes = this.bytes; - var predefined = false; - var hasSupplement = false; - var format, i, ii; - var raw = null; - function readSupplement() { - var supplementsCount = bytes[pos++]; - for (i = 0; i < supplementsCount; i++) { - var code = bytes[pos++]; - var sid = (bytes[pos++] << 8) + (bytes[pos++] & 0xff); - encoding[code] = charset.indexOf(strings.get(sid)); + constructInterpolatedFromIR: + function PDFFunction_constructInterpolatedFromIR(IR) { + var c0 = IR[1]; + var diff = IR[2]; + var n = IR[3]; + + var length = diff.length; + + return function constructInterpolatedFromIRResult(src, srcOffset, + dest, destOffset) { + var x = n === 1 ? src[srcOffset] : Math.pow(src[srcOffset], n); + + for (var j = 0; j < length; ++j) { + dest[destOffset + j] = c0[j] + (x * diff[j]); } + }; + }, + + constructStiched: function PDFFunction_constructStiched(fn, dict, xref) { + var domain = dict.get('Domain'); + + if (!domain) { + error('No domain'); } - if (pos === 0 || pos === 1) { - predefined = true; - format = pos; - var baseEncoding = pos ? ExpertEncoding : StandardEncoding; - for (i = 0, ii = charset.length; i < ii; i++) { - var index = baseEncoding.indexOf(charset[i]); - if (index !== -1) { - encoding[index] = i; + var inputSize = domain.length / 2; + if (inputSize !== 1) { + error('Bad domain for stiched function'); + } + + var fnRefs = dict.get('Functions'); + var fns = []; + for (var i = 0, ii = fnRefs.length; i < ii; ++i) { + fns.push(PDFFunction.getIR(xref, xref.fetchIfRef(fnRefs[i]))); + } + + var bounds = dict.get('Bounds'); + var encode = dict.get('Encode'); + + return [CONSTRUCT_STICHED, domain, bounds, encode, fns]; + }, + + constructStichedFromIR: function PDFFunction_constructStichedFromIR(IR) { + var domain = IR[1]; + var bounds = IR[2]; + var encode = IR[3]; + var fnsIR = IR[4]; + var fns = []; + var tmpBuf = new Float32Array(1); + + for (var i = 0, ii = fnsIR.length; i < ii; i++) { + fns.push(PDFFunction.fromIR(fnsIR[i])); + } + + return function constructStichedFromIRResult(src, srcOffset, + dest, destOffset) { + var clip = function constructStichedFromIRClip(v, min, max) { + if (v > max) { + v = max; + } else if (v < min) { + v = min; } - } - } else { - var dataStart = pos; - format = bytes[pos++]; - switch (format & 0x7f) { - case 0: - var glyphsCount = bytes[pos++]; - for (i = 1; i <= glyphsCount; i++) { - encoding[bytes[pos++]] = i; - } - break; + return v; + }; - case 1: - var rangesCount = bytes[pos++]; - var gid = 1; - for (i = 0; i < rangesCount; i++) { - var start = bytes[pos++]; - var left = bytes[pos++]; - for (var j = start; j <= start + left; j++) { - encoding[j] = gid++; - } - } + // clip to domain + var v = clip(src[srcOffset], domain[0], domain[1]); + // calulate which bound the value is in + for (var i = 0, ii = bounds.length; i < ii; ++i) { + if (v < bounds[i]) { break; + } + } - default: - error('Unknow encoding format: ' + format + ' in CFF'); - break; + // encode value into domain of function + var dmin = domain[0]; + if (i > 0) { + dmin = bounds[i - 1]; } - var dataEnd = pos; - if (format & 0x80) { - // The font sanitizer does not support CFF encoding with a - // supplement, since the encoding is not really used to map - // between gid to glyph, let's overwrite what is declared in - // the top dictionary to let the sanitizer think the font use - // StandardEncoding, that's a lie but that's ok. - bytes[dataStart] &= 0x7f; - readSupplement(); - hasSupplement = true; + var dmax = domain[1]; + if (i < bounds.length) { + dmax = bounds[i]; } - raw = bytes.subarray(dataStart, dataEnd); + + var rmin = encode[2 * i]; + var rmax = encode[2 * i + 1]; + + // Prevent the value from becoming NaN as a result + // of division by zero (fixes issue6113.pdf). + tmpBuf[0] = dmin === dmax ? rmin : + rmin + (v - dmin) * (rmax - rmin) / (dmax - dmin); + + // call the appropriate function + fns[i](tmpBuf, 0, dest, destOffset); + }; + }, + + constructPostScript: function PDFFunction_constructPostScript(fn, dict, + xref) { + var domain = dict.get('Domain'); + var range = dict.get('Range'); + + if (!domain) { + error('No domain.'); } - format = format & 0x7f; - return new CFFEncoding(predefined, format, encoding, raw); + + if (!range) { + error('No range.'); + } + + var lexer = new PostScriptLexer(fn); + var parser = new PostScriptParser(lexer); + var code = parser.parse(); + + return [CONSTRUCT_POSTSCRIPT, domain, range, code]; }, - parseFDSelect: function CFFParser_parseFDSelect(pos, length) { - var start = pos; - var bytes = this.bytes; - var format = bytes[pos++]; - var fdSelect = []; - var i; - switch (format) { - case 0: - for (i = 0; i < length; ++i) { - var id = bytes[pos++]; - fdSelect.push(id); - } - break; - case 3: - var rangesCount = (bytes[pos++] << 8) | bytes[pos++]; - for (i = 0; i < rangesCount; ++i) { - var first = (bytes[pos++] << 8) | bytes[pos++]; - var fdIndex = bytes[pos++]; - var next = (bytes[pos] << 8) | bytes[pos + 1]; - for (var j = first; j < next; ++j) { - fdSelect.push(fdIndex); - } - } - // Advance past the sentinel(next). - pos += 2; - break; - default: - error('Unknown fdselect format ' + format); - break; + constructPostScriptFromIR: function PDFFunction_constructPostScriptFromIR( + IR) { + var domain = IR[1]; + var range = IR[2]; + var code = IR[3]; + + var compiled = (new PostScriptCompiler()).compile(code, domain, range); + if (compiled) { + // Compiled function consists of simple expressions such as addition, + // subtraction, Math.max, and also contains 'var' and 'return' + // statements. See the generation in the PostScriptCompiler below. + /*jshint -W054 */ + return new Function('src', 'srcOffset', 'dest', 'destOffset', compiled); } - var end = pos; - return new CFFFDSelect(fdSelect, bytes.subarray(start, end)); - } - }; - return CFFParser; -})(); -// Compact Font Format -var CFF = (function CFFClosure() { - function CFF() { - this.header = null; - this.names = []; - this.topDict = null; - this.strings = new CFFStrings(); - this.globalSubrIndex = null; + info('Unable to compile PS function'); - // The following could really be per font, but since we only have one font - // store them here. - this.encoding = null; - this.charset = null; - this.charStrings = null; - this.fdArray = []; - this.fdSelect = null; + var numOutputs = range.length >> 1; + var numInputs = domain.length >> 1; + var evaluator = new PostScriptEvaluator(code); + // Cache the values for a big speed up, the cache size is limited though + // since the number of possible values can be huge from a PS function. + var cache = Object.create(null); + // The MAX_CACHE_SIZE is set to ~4x the maximum number of distinct values + // seen in our tests. + var MAX_CACHE_SIZE = 2048 * 4; + var cache_available = MAX_CACHE_SIZE; + var tmpBuf = new Float32Array(numInputs); - this.isCIDFont = false; - } - return CFF; -})(); + return function constructPostScriptFromIRResult(src, srcOffset, + dest, destOffset) { + var i, value; + var key = ''; + var input = tmpBuf; + for (i = 0; i < numInputs; i++) { + value = src[srcOffset + i]; + input[i] = value; + key += value + '_'; + } -var CFFHeader = (function CFFHeaderClosure() { - function CFFHeader(major, minor, hdrSize, offSize) { - this.major = major; - this.minor = minor; - this.hdrSize = hdrSize; - this.offSize = offSize; - } - return CFFHeader; -})(); + var cachedValue = cache[key]; + if (cachedValue !== undefined) { + dest.set(cachedValue, destOffset); + return; + } -var CFFStrings = (function CFFStringsClosure() { - function CFFStrings() { - this.strings = []; - } - CFFStrings.prototype = { - get: function CFFStrings_get(index) { - if (index >= 0 && index <= 390) { - return CFFStandardStrings[index]; - } - if (index - 391 <= this.strings.length) { - return this.strings[index - 391]; - } - return CFFStandardStrings[0]; - }, - add: function CFFStrings_add(value) { - this.strings.push(value); - }, - get count() { - return this.strings.length; + var output = new Float32Array(numOutputs); + var stack = evaluator.execute(input); + var stackIndex = stack.length - numOutputs; + for (i = 0; i < numOutputs; i++) { + value = stack[stackIndex + i]; + var bound = range[i * 2]; + if (value < bound) { + value = bound; + } else { + bound = range[i * 2 +1]; + if (value > bound) { + value = bound; + } + } + output[i] = value; + } + if (cache_available > 0) { + cache_available--; + cache[key] = output; + } + dest.set(output, destOffset); + }; } }; - return CFFStrings; })(); -var CFFIndex = (function CFFIndexClosure() { - function CFFIndex() { - this.objects = []; - this.length = 0; +function isPDFFunction(v) { + var fnDict; + if (typeof v !== 'object') { + return false; + } else if (isDict(v)) { + fnDict = v; + } else if (isStream(v)) { + fnDict = v.dict; + } else { + return false; } - CFFIndex.prototype = { - add: function CFFIndex_add(data) { - this.length += data.length; - this.objects.push(data); - }, - set: function CFFIndex_set(index, data) { - this.length += data.length - this.objects[index].length; - this.objects[index] = data; - }, - get: function CFFIndex_get(index) { - return this.objects[index]; - }, - get count() { - return this.objects.length; - } - }; - return CFFIndex; -})(); + return fnDict.has('FunctionType'); +} -var CFFDict = (function CFFDictClosure() { - function CFFDict(tables, strings) { - this.keyToNameMap = tables.keyToNameMap; - this.nameToKeyMap = tables.nameToKeyMap; - this.defaults = tables.defaults; - this.types = tables.types; - this.opcodes = tables.opcodes; - this.order = tables.order; - this.strings = strings; - this.values = Object.create(null); +var PostScriptStack = (function PostScriptStackClosure() { + var MAX_STACK_SIZE = 100; + function PostScriptStack(initialStack) { + this.stack = !initialStack ? [] : + Array.prototype.slice.call(initialStack, 0); } - CFFDict.prototype = { - // value should always be an array - setByKey: function CFFDict_setByKey(key, value) { - if (!(key in this.keyToNameMap)) { - return false; - } - // ignore empty values - if (value.length === 0) { - return true; + + PostScriptStack.prototype = { + push: function PostScriptStack_push(value) { + if (this.stack.length >= MAX_STACK_SIZE) { + error('PostScript function stack overflow.'); } - var type = this.types[key]; - // remove the array wrapping these types of values - if (type === 'num' || type === 'sid' || type === 'offset') { - value = value[0]; + this.stack.push(value); + }, + pop: function PostScriptStack_pop() { + if (this.stack.length <= 0) { + error('PostScript function stack underflow.'); } - this.values[key] = value; - return true; + return this.stack.pop(); }, - setByName: function CFFDict_setByName(name, value) { - if (!(name in this.nameToKeyMap)) { - error('Invalid dictionary name "' + name + '"'); + copy: function PostScriptStack_copy(n) { + if (this.stack.length + n >= MAX_STACK_SIZE) { + error('PostScript function stack overflow.'); + } + var stack = this.stack; + for (var i = stack.length - n, j = n - 1; j >= 0; j--, i++) { + stack.push(stack[i]); } - this.values[this.nameToKeyMap[name]] = value; }, - hasName: function CFFDict_hasName(name) { - return this.nameToKeyMap[name] in this.values; + index: function PostScriptStack_index(n) { + this.push(this.stack[this.stack.length - n - 1]); }, - getByName: function CFFDict_getByName(name) { - if (!(name in this.nameToKeyMap)) { - error('Invalid dictionary name "' + name + '"'); + // rotate the last n stack elements p times + roll: function PostScriptStack_roll(n, p) { + var stack = this.stack; + var l = stack.length - n; + var r = stack.length - 1, c = l + (p - Math.floor(p / n) * n), i, j, t; + for (i = l, j = r; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; } - var key = this.nameToKeyMap[name]; - if (!(key in this.values)) { - return this.defaults[key]; + for (i = l, j = c - 1; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; + } + for (i = c, j = r; i < j; i++, j--) { + t = stack[i]; stack[i] = stack[j]; stack[j] = t; } - return this.values[key]; - }, - removeByName: function CFFDict_removeByName(name) { - delete this.values[this.nameToKeyMap[name]]; - } - }; - CFFDict.createTables = function CFFDict_createTables(layout) { - var tables = { - keyToNameMap: {}, - nameToKeyMap: {}, - defaults: {}, - types: {}, - opcodes: {}, - order: [] - }; - for (var i = 0, ii = layout.length; i < ii; ++i) { - var entry = layout[i]; - var key = isArray(entry[0]) ? (entry[0][0] << 8) + entry[0][1] : entry[0]; - tables.keyToNameMap[key] = entry[1]; - tables.nameToKeyMap[entry[1]] = key; - tables.types[key] = entry[2]; - tables.defaults[key] = entry[3]; - tables.opcodes[key] = isArray(entry[0]) ? entry[0] : [entry[0]]; - tables.order.push(key); } - return tables; }; - return CFFDict; -})(); - -var CFFTopDict = (function CFFTopDictClosure() { - var layout = [ - [[12, 30], 'ROS', ['sid', 'sid', 'num'], null], - [[12, 20], 'SyntheticBase', 'num', null], - [0, 'version', 'sid', null], - [1, 'Notice', 'sid', null], - [[12, 0], 'Copyright', 'sid', null], - [2, 'FullName', 'sid', null], - [3, 'FamilyName', 'sid', null], - [4, 'Weight', 'sid', null], - [[12, 1], 'isFixedPitch', 'num', 0], - [[12, 2], 'ItalicAngle', 'num', 0], - [[12, 3], 'UnderlinePosition', 'num', -100], - [[12, 4], 'UnderlineThickness', 'num', 50], - [[12, 5], 'PaintType', 'num', 0], - [[12, 6], 'CharstringType', 'num', 2], - [[12, 7], 'FontMatrix', ['num', 'num', 'num', 'num', 'num', 'num'], - [0.001, 0, 0, 0.001, 0, 0]], - [13, 'UniqueID', 'num', null], - [5, 'FontBBox', ['num', 'num', 'num', 'num'], [0, 0, 0, 0]], - [[12, 8], 'StrokeWidth', 'num', 0], - [14, 'XUID', 'array', null], - [15, 'charset', 'offset', 0], - [16, 'Encoding', 'offset', 0], - [17, 'CharStrings', 'offset', 0], - [18, 'Private', ['offset', 'offset'], null], - [[12, 21], 'PostScript', 'sid', null], - [[12, 22], 'BaseFontName', 'sid', null], - [[12, 23], 'BaseFontBlend', 'delta', null], - [[12, 31], 'CIDFontVersion', 'num', 0], - [[12, 32], 'CIDFontRevision', 'num', 0], - [[12, 33], 'CIDFontType', 'num', 0], - [[12, 34], 'CIDCount', 'num', 8720], - [[12, 35], 'UIDBase', 'num', null], - // XXX: CID Fonts on DirectWrite 6.1 only seem to work if FDSelect comes - // before FDArray. - [[12, 37], 'FDSelect', 'offset', null], - [[12, 36], 'FDArray', 'offset', null], - [[12, 38], 'FontName', 'sid', null] - ]; - var tables = null; - function CFFTopDict(strings) { - if (tables === null) { - tables = CFFDict.createTables(layout); - } - CFFDict.call(this, tables, strings); - this.privateDict = null; - } - CFFTopDict.prototype = Object.create(CFFDict.prototype); - return CFFTopDict; -})(); - -var CFFPrivateDict = (function CFFPrivateDictClosure() { - var layout = [ - [6, 'BlueValues', 'delta', null], - [7, 'OtherBlues', 'delta', null], - [8, 'FamilyBlues', 'delta', null], - [9, 'FamilyOtherBlues', 'delta', null], - [[12, 9], 'BlueScale', 'num', 0.039625], - [[12, 10], 'BlueShift', 'num', 7], - [[12, 11], 'BlueFuzz', 'num', 1], - [10, 'StdHW', 'num', null], - [11, 'StdVW', 'num', null], - [[12, 12], 'StemSnapH', 'delta', null], - [[12, 13], 'StemSnapV', 'delta', null], - [[12, 14], 'ForceBold', 'num', 0], - [[12, 17], 'LanguageGroup', 'num', 0], - [[12, 18], 'ExpansionFactor', 'num', 0.06], - [[12, 19], 'initialRandomSeed', 'num', 0], - [20, 'defaultWidthX', 'num', 0], - [21, 'nominalWidthX', 'num', 0], - [19, 'Subrs', 'offset', null] - ]; - var tables = null; - function CFFPrivateDict(strings) { - if (tables === null) { - tables = CFFDict.createTables(layout); - } - CFFDict.call(this, tables, strings); - this.subrsIndex = null; - } - CFFPrivateDict.prototype = Object.create(CFFDict.prototype); - return CFFPrivateDict; -})(); - -var CFFCharsetPredefinedTypes = { - ISO_ADOBE: 0, - EXPERT: 1, - EXPERT_SUBSET: 2 -}; -var CFFCharset = (function CFFCharsetClosure() { - function CFFCharset(predefined, format, charset, raw) { - this.predefined = predefined; - this.format = format; - this.charset = charset; - this.raw = raw; - } - return CFFCharset; + return PostScriptStack; })(); - -var CFFEncoding = (function CFFEncodingClosure() { - function CFFEncoding(predefined, format, encoding, raw) { - this.predefined = predefined; - this.format = format; - this.encoding = encoding; - this.raw = raw; +var PostScriptEvaluator = (function PostScriptEvaluatorClosure() { + function PostScriptEvaluator(operators) { + this.operators = operators; } - return CFFEncoding; -})(); + PostScriptEvaluator.prototype = { + execute: function PostScriptEvaluator_execute(initialStack) { + var stack = new PostScriptStack(initialStack); + var counter = 0; + var operators = this.operators; + var length = operators.length; + var operator, a, b; + while (counter < length) { + operator = operators[counter++]; + if (typeof operator === 'number') { + // Operator is really an operand and should be pushed to the stack. + stack.push(operator); + continue; + } + switch (operator) { + // non standard ps operators + case 'jz': // jump if false + b = stack.pop(); + a = stack.pop(); + if (!a) { + counter = b; + } + break; + case 'j': // jump + a = stack.pop(); + counter = a; + break; -var CFFFDSelect = (function CFFFDSelectClosure() { - function CFFFDSelect(fdSelect, raw) { - this.fdSelect = fdSelect; - this.raw = raw; - } - CFFFDSelect.prototype = { - getFDIndex: function CFFFDSelect_get(glyphIndex) { - if (glyphIndex < 0 || glyphIndex >= this.fdSelect.length) { - return -1; + // all ps operators in alphabetical order (excluding if/ifelse) + case 'abs': + a = stack.pop(); + stack.push(Math.abs(a)); + break; + case 'add': + b = stack.pop(); + a = stack.pop(); + stack.push(a + b); + break; + case 'and': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a && b); + } else { + stack.push(a & b); + } + break; + case 'atan': + a = stack.pop(); + stack.push(Math.atan(a)); + break; + case 'bitshift': + b = stack.pop(); + a = stack.pop(); + if (a > 0) { + stack.push(a << b); + } else { + stack.push(a >> b); + } + break; + case 'ceiling': + a = stack.pop(); + stack.push(Math.ceil(a)); + break; + case 'copy': + a = stack.pop(); + stack.copy(a); + break; + case 'cos': + a = stack.pop(); + stack.push(Math.cos(a)); + break; + case 'cvi': + a = stack.pop() | 0; + stack.push(a); + break; + case 'cvr': + // noop + break; + case 'div': + b = stack.pop(); + a = stack.pop(); + stack.push(a / b); + break; + case 'dup': + stack.copy(1); + break; + case 'eq': + b = stack.pop(); + a = stack.pop(); + stack.push(a === b); + break; + case 'exch': + stack.roll(2, 1); + break; + case 'exp': + b = stack.pop(); + a = stack.pop(); + stack.push(Math.pow(a, b)); + break; + case 'false': + stack.push(false); + break; + case 'floor': + a = stack.pop(); + stack.push(Math.floor(a)); + break; + case 'ge': + b = stack.pop(); + a = stack.pop(); + stack.push(a >= b); + break; + case 'gt': + b = stack.pop(); + a = stack.pop(); + stack.push(a > b); + break; + case 'idiv': + b = stack.pop(); + a = stack.pop(); + stack.push((a / b) | 0); + break; + case 'index': + a = stack.pop(); + stack.index(a); + break; + case 'le': + b = stack.pop(); + a = stack.pop(); + stack.push(a <= b); + break; + case 'ln': + a = stack.pop(); + stack.push(Math.log(a)); + break; + case 'log': + a = stack.pop(); + stack.push(Math.log(a) / Math.LN10); + break; + case 'lt': + b = stack.pop(); + a = stack.pop(); + stack.push(a < b); + break; + case 'mod': + b = stack.pop(); + a = stack.pop(); + stack.push(a % b); + break; + case 'mul': + b = stack.pop(); + a = stack.pop(); + stack.push(a * b); + break; + case 'ne': + b = stack.pop(); + a = stack.pop(); + stack.push(a !== b); + break; + case 'neg': + a = stack.pop(); + stack.push(-a); + break; + case 'not': + a = stack.pop(); + if (isBool(a)) { + stack.push(!a); + } else { + stack.push(~a); + } + break; + case 'or': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a || b); + } else { + stack.push(a | b); + } + break; + case 'pop': + stack.pop(); + break; + case 'roll': + b = stack.pop(); + a = stack.pop(); + stack.roll(a, b); + break; + case 'round': + a = stack.pop(); + stack.push(Math.round(a)); + break; + case 'sin': + a = stack.pop(); + stack.push(Math.sin(a)); + break; + case 'sqrt': + a = stack.pop(); + stack.push(Math.sqrt(a)); + break; + case 'sub': + b = stack.pop(); + a = stack.pop(); + stack.push(a - b); + break; + case 'true': + stack.push(true); + break; + case 'truncate': + a = stack.pop(); + a = a < 0 ? Math.ceil(a) : Math.floor(a); + stack.push(a); + break; + case 'xor': + b = stack.pop(); + a = stack.pop(); + if (isBool(a) && isBool(b)) { + stack.push(a !== b); + } else { + stack.push(a ^ b); + } + break; + default: + error('Unknown operator ' + operator); + break; + } } - return this.fdSelect[glyphIndex]; + return stack.stack; } }; - return CFFFDSelect; + return PostScriptEvaluator; })(); -// Helper class to keep track of where an offset is within the data and helps -// filling in that offset once it's known. -var CFFOffsetTracker = (function CFFOffsetTrackerClosure() { - function CFFOffsetTracker() { - this.offsets = Object.create(null); +// Most of the PDFs functions consist of simple operations such as: +// roll, exch, sub, cvr, pop, index, dup, mul, if, gt, add. +// +// We can compile most of such programs, and at the same moment, we can +// optimize some expressions using basic math properties. Keeping track of +// min/max values will allow us to avoid extra Math.min/Math.max calls. +var PostScriptCompiler = (function PostScriptCompilerClosure() { + function AstNode(type) { + this.type = type; } - CFFOffsetTracker.prototype = { - isTracking: function CFFOffsetTracker_isTracking(key) { - return key in this.offsets; - }, - track: function CFFOffsetTracker_track(key, location) { - if (key in this.offsets) { - error('Already tracking location of ' + key); - } - this.offsets[key] = location; - }, - offset: function CFFOffsetTracker_offset(value) { - for (var key in this.offsets) { - this.offsets[key] += value; - } - }, - setEntryLocation: function CFFOffsetTracker_setEntryLocation(key, - values, - output) { - if (!(key in this.offsets)) { - error('Not tracking location of ' + key); - } - var data = output.data; - var dataOffset = this.offsets[key]; - var size = 5; - for (var i = 0, ii = values.length; i < ii; ++i) { - var offset0 = i * size + dataOffset; - var offset1 = offset0 + 1; - var offset2 = offset0 + 2; - var offset3 = offset0 + 3; - var offset4 = offset0 + 4; - // It's easy to screw up offsets so perform this sanity check. - if (data[offset0] !== 0x1d || data[offset1] !== 0 || - data[offset2] !== 0 || data[offset3] !== 0 || data[offset4] !== 0) { - error('writing to an offset that is not empty'); - } - var value = values[i]; - data[offset0] = 0x1d; - data[offset1] = (value >> 24) & 0xFF; - data[offset2] = (value >> 16) & 0xFF; - data[offset3] = (value >> 8) & 0xFF; - data[offset4] = value & 0xFF; - } - } + AstNode.prototype.visit = function (visitor) { + throw new Error('abstract method'); }; - return CFFOffsetTracker; -})(); -// Takes a CFF and converts it to the binary representation. -var CFFCompiler = (function CFFCompilerClosure() { - function CFFCompiler(cff) { - this.cff = cff; + function AstArgument(index, min, max) { + AstNode.call(this, 'args'); + this.index = index; + this.min = min; + this.max = max; } - CFFCompiler.prototype = { - compile: function CFFCompiler_compile() { - var cff = this.cff; - var output = { - data: [], - length: 0, - add: function CFFCompiler_add(data) { - this.data = this.data.concat(data); - this.length = this.data.length; - } - }; - - // Compile the five entries that must be in order. - var header = this.compileHeader(cff.header); - output.add(header); - - var nameIndex = this.compileNameIndex(cff.names); - output.add(nameIndex); - - if (cff.isCIDFont) { - // The spec is unclear on how font matrices should relate to each other - // when there is one in the main top dict and the sub top dicts. - // Windows handles this differently than linux and osx so we have to - // normalize to work on all. - // Rules based off of some mailing list discussions: - // - If main font has a matrix and subfont doesn't, use the main matrix. - // - If no main font matrix and there is a subfont matrix, use the - // subfont matrix. - // - If both have matrices, concat together. - // - If neither have matrices, use default. - // To make this work on all platforms we move the top matrix into each - // sub top dict and concat if necessary. - if (cff.topDict.hasName('FontMatrix')) { - var base = cff.topDict.getByName('FontMatrix'); - cff.topDict.removeByName('FontMatrix'); - for (var i = 0, ii = cff.fdArray.length; i < ii; i++) { - var subDict = cff.fdArray[i]; - var matrix = base.slice(0); - if (subDict.hasName('FontMatrix')) { - matrix = Util.transform(matrix, subDict.getByName('FontMatrix')); - } - subDict.setByName('FontMatrix', matrix); - } - } - } - - var compiled = this.compileTopDicts([cff.topDict], - output.length, - cff.isCIDFont); - output.add(compiled.output); - var topDictTracker = compiled.trackers[0]; - - var stringIndex = this.compileStringIndex(cff.strings.strings); - output.add(stringIndex); - - var globalSubrIndex = this.compileIndex(cff.globalSubrIndex); - output.add(globalSubrIndex); - - // Now start on the other entries that have no specfic order. - if (cff.encoding && cff.topDict.hasName('Encoding')) { - if (cff.encoding.predefined) { - topDictTracker.setEntryLocation('Encoding', [cff.encoding.format], - output); - } else { - var encoding = this.compileEncoding(cff.encoding); - topDictTracker.setEntryLocation('Encoding', [output.length], output); - output.add(encoding); - } - } - - if (cff.charset && cff.topDict.hasName('charset')) { - if (cff.charset.predefined) { - topDictTracker.setEntryLocation('charset', [cff.charset.format], - output); - } else { - var charset = this.compileCharset(cff.charset); - topDictTracker.setEntryLocation('charset', [output.length], output); - output.add(charset); - } - } - - var charStrings = this.compileCharStrings(cff.charStrings); - topDictTracker.setEntryLocation('CharStrings', [output.length], output); - output.add(charStrings); - - if (cff.isCIDFont) { - // For some reason FDSelect must be in front of FDArray on windows. OSX - // and linux don't seem to care. - topDictTracker.setEntryLocation('FDSelect', [output.length], output); - var fdSelect = this.compileFDSelect(cff.fdSelect.raw); - output.add(fdSelect); - // It is unclear if the sub font dictionary can have CID related - // dictionary keys, but the sanitizer doesn't like them so remove them. - compiled = this.compileTopDicts(cff.fdArray, output.length, true); - topDictTracker.setEntryLocation('FDArray', [output.length], output); - output.add(compiled.output); - var fontDictTrackers = compiled.trackers; + AstArgument.prototype = Object.create(AstNode.prototype); + AstArgument.prototype.visit = function (visitor) { + visitor.visitArgument(this); + }; - this.compilePrivateDicts(cff.fdArray, fontDictTrackers, output); - } + function AstLiteral(number) { + AstNode.call(this, 'literal'); + this.number = number; + this.min = number; + this.max = number; + } + AstLiteral.prototype = Object.create(AstNode.prototype); + AstLiteral.prototype.visit = function (visitor) { + visitor.visitLiteral(this); + }; - this.compilePrivateDicts([cff.topDict], [topDictTracker], output); + function AstBinaryOperation(op, arg1, arg2, min, max) { + AstNode.call(this, 'binary'); + this.op = op; + this.arg1 = arg1; + this.arg2 = arg2; + this.min = min; + this.max = max; + } + AstBinaryOperation.prototype = Object.create(AstNode.prototype); + AstBinaryOperation.prototype.visit = function (visitor) { + visitor.visitBinaryOperation(this); + }; - // If the font data ends with INDEX whose object data is zero-length, - // the sanitizer will bail out. Add a dummy byte to avoid that. - output.add([0]); + function AstMin(arg, max) { + AstNode.call(this, 'max'); + this.arg = arg; + this.min = arg.min; + this.max = max; + } + AstMin.prototype = Object.create(AstNode.prototype); + AstMin.prototype.visit = function (visitor) { + visitor.visitMin(this); + }; - return output.data; - }, - encodeNumber: function CFFCompiler_encodeNumber(value) { - if (parseFloat(value) === parseInt(value, 10) && !isNaN(value)) { // isInt - return this.encodeInteger(value); - } else { - return this.encodeFloat(value); - } - }, - encodeFloat: function CFFCompiler_encodeFloat(num) { - var value = num.toString(); + function AstVariable(index, min, max) { + AstNode.call(this, 'var'); + this.index = index; + this.min = min; + this.max = max; + } + AstVariable.prototype = Object.create(AstNode.prototype); + AstVariable.prototype.visit = function (visitor) { + visitor.visitVariable(this); + }; - // rounding inaccurate doubles - var m = /\.(\d*?)(?:9{5,20}|0{5,20})\d{0,2}(?:e(.+)|$)/.exec(value); - if (m) { - var epsilon = parseFloat('1e' + ((m[2] ? +m[2] : 0) + m[1].length)); - value = (Math.round(num * epsilon) / epsilon).toString(); - } + function AstVariableDefinition(variable, arg) { + AstNode.call(this, 'definition'); + this.variable = variable; + this.arg = arg; + } + AstVariableDefinition.prototype = Object.create(AstNode.prototype); + AstVariableDefinition.prototype.visit = function (visitor) { + visitor.visitVariableDefinition(this); + }; - var nibbles = ''; - var i, ii; - for (i = 0, ii = value.length; i < ii; ++i) { - var a = value[i]; - if (a === 'e') { - nibbles += value[++i] === '-' ? 'c' : 'b'; - } else if (a === '.') { - nibbles += 'a'; - } else if (a === '-') { - nibbles += 'e'; - } else { - nibbles += a; - } - } - nibbles += (nibbles.length & 1) ? 'f' : 'ff'; - var out = [30]; - for (i = 0, ii = nibbles.length; i < ii; i += 2) { - out.push(parseInt(nibbles.substr(i, 2), 16)); - } - return out; + function ExpressionBuilderVisitor() { + this.parts = []; + } + ExpressionBuilderVisitor.prototype = { + visitArgument: function (arg) { + this.parts.push('Math.max(', arg.min, ', Math.min(', + arg.max, ', src[srcOffset + ', arg.index, ']))'); }, - encodeInteger: function CFFCompiler_encodeInteger(value) { - var code; - if (value >= -107 && value <= 107) { - code = [value + 139]; - } else if (value >= 108 && value <= 1131) { - value = [value - 108]; - code = [(value >> 8) + 247, value & 0xFF]; - } else if (value >= -1131 && value <= -108) { - value = -value - 108; - code = [(value >> 8) + 251, value & 0xFF]; - } else if (value >= -32768 && value <= 32767) { - code = [0x1c, (value >> 8) & 0xFF, value & 0xFF]; - } else { - code = [0x1d, - (value >> 24) & 0xFF, - (value >> 16) & 0xFF, - (value >> 8) & 0xFF, - value & 0xFF]; - } - return code; + visitVariable: function (variable) { + this.parts.push('v', variable.index); }, - compileHeader: function CFFCompiler_compileHeader(header) { - return [ - header.major, - header.minor, - header.hdrSize, - header.offSize - ]; + visitLiteral: function (literal) { + this.parts.push(literal.number); }, - compileNameIndex: function CFFCompiler_compileNameIndex(names) { - var nameIndex = new CFFIndex(); - for (var i = 0, ii = names.length; i < ii; ++i) { - nameIndex.add(stringToBytes(names[i])); - } - return this.compileIndex(nameIndex); + visitBinaryOperation: function (operation) { + this.parts.push('('); + operation.arg1.visit(this); + this.parts.push(' ', operation.op, ' '); + operation.arg2.visit(this); + this.parts.push(')'); }, - compileTopDicts: function CFFCompiler_compileTopDicts(dicts, - length, - removeCidKeys) { - var fontDictTrackers = []; - var fdArrayIndex = new CFFIndex(); - for (var i = 0, ii = dicts.length; i < ii; ++i) { - var fontDict = dicts[i]; - if (removeCidKeys) { - fontDict.removeByName('CIDFontVersion'); - fontDict.removeByName('CIDFontRevision'); - fontDict.removeByName('CIDFontType'); - fontDict.removeByName('CIDCount'); - fontDict.removeByName('UIDBase'); - } - var fontDictTracker = new CFFOffsetTracker(); - var fontDictData = this.compileDict(fontDict, fontDictTracker); - fontDictTrackers.push(fontDictTracker); - fdArrayIndex.add(fontDictData); - fontDictTracker.offset(length); - } - fdArrayIndex = this.compileIndex(fdArrayIndex, fontDictTrackers); - return { - trackers: fontDictTrackers, - output: fdArrayIndex - }; + visitVariableDefinition: function (definition) { + this.parts.push('var '); + definition.variable.visit(this); + this.parts.push(' = '); + definition.arg.visit(this); + this.parts.push(';'); }, - compilePrivateDicts: function CFFCompiler_compilePrivateDicts(dicts, - trackers, - output) { - for (var i = 0, ii = dicts.length; i < ii; ++i) { - var fontDict = dicts[i]; - assert(fontDict.privateDict && fontDict.hasName('Private'), - 'There must be an private dictionary.'); - var privateDict = fontDict.privateDict; - var privateDictTracker = new CFFOffsetTracker(); - var privateDictData = this.compileDict(privateDict, privateDictTracker); - - var outputLength = output.length; - privateDictTracker.offset(outputLength); - if (!privateDictData.length) { - // The private dictionary was empty, set the output length to zero to - // ensure the offset length isn't out of bounds in the eyes of the - // sanitizer. - outputLength = 0; - } - - trackers[i].setEntryLocation('Private', - [privateDictData.length, outputLength], - output); - output.add(privateDictData); - - if (privateDict.subrsIndex && privateDict.hasName('Subrs')) { - var subrs = this.compileIndex(privateDict.subrsIndex); - privateDictTracker.setEntryLocation('Subrs', [privateDictData.length], - output); - output.add(subrs); - } - } + visitMin: function (max) { + this.parts.push('Math.min('); + max.arg.visit(this); + this.parts.push(', ', max.max, ')'); }, - compileDict: function CFFCompiler_compileDict(dict, offsetTracker) { - var out = []; - // The dictionary keys must be in a certain order. - var order = dict.order; - for (var i = 0; i < order.length; ++i) { - var key = order[i]; - if (!(key in dict.values)) { - continue; - } - var values = dict.values[key]; - var types = dict.types[key]; - if (!isArray(types)) { - types = [types]; - } - if (!isArray(values)) { - values = [values]; - } - - // Remove any empty dict values. - if (values.length === 0) { - continue; - } + toString: function () { + return this.parts.join(''); + } + }; - for (var j = 0, jj = types.length; j < jj; ++j) { - var type = types[j]; - var value = values[j]; - switch (type) { - case 'num': - case 'sid': - out = out.concat(this.encodeNumber(value)); - break; - case 'offset': - // For offsets we just insert a 32bit integer so we don't have to - // deal with figuring out the length of the offset when it gets - // replaced later on by the compiler. - var name = dict.keyToNameMap[key]; - // Some offsets have the offset and the length, so just record the - // position of the first one. - if (!offsetTracker.isTracking(name)) { - offsetTracker.track(name, out.length); - } - out = out.concat([0x1d, 0, 0, 0, 0]); - break; - case 'array': - case 'delta': - out = out.concat(this.encodeNumber(value)); - for (var k = 1, kk = values.length; k < kk; ++k) { - out = out.concat(this.encodeNumber(values[k])); - } - break; - default: - error('Unknown data type of ' + type); - break; - } - } - out = out.concat(dict.opcodes[key]); - } - return out; - }, - compileStringIndex: function CFFCompiler_compileStringIndex(strings) { - var stringIndex = new CFFIndex(); - for (var i = 0, ii = strings.length; i < ii; ++i) { - stringIndex.add(stringToBytes(strings[i])); - } - return this.compileIndex(stringIndex); - }, - compileGlobalSubrIndex: function CFFCompiler_compileGlobalSubrIndex() { - var globalSubrIndex = this.cff.globalSubrIndex; - this.out.writeByteArray(this.compileIndex(globalSubrIndex)); - }, - compileCharStrings: function CFFCompiler_compileCharStrings(charStrings) { - return this.compileIndex(charStrings); - }, - compileCharset: function CFFCompiler_compileCharset(charset) { - return this.compileTypedArray(charset.raw); - }, - compileEncoding: function CFFCompiler_compileEncoding(encoding) { - return this.compileTypedArray(encoding.raw); - }, - compileFDSelect: function CFFCompiler_compileFDSelect(fdSelect) { - return this.compileTypedArray(fdSelect); - }, - compileTypedArray: function CFFCompiler_compileTypedArray(data) { - var out = []; - for (var i = 0, ii = data.length; i < ii; ++i) { - out[i] = data[i]; - } - return out; - }, - compileIndex: function CFFCompiler_compileIndex(index, trackers) { - trackers = trackers || []; - var objects = index.objects; - // First 2 bytes contains the number of objects contained into this index - var count = objects.length; + function buildAddOperation(num1, num2) { + if (num2.type === 'literal' && num2.number === 0) { + // optimization: second operand is 0 + return num1; + } + if (num1.type === 'literal' && num1.number === 0) { + // optimization: first operand is 0 + return num2; + } + if (num2.type === 'literal' && num1.type === 'literal') { + // optimization: operands operand are literals + return new AstLiteral(num1.number + num2.number); + } + return new AstBinaryOperation('+', num1, num2, + num1.min + num2.min, num1.max + num2.max); + } - // If there is no object, just create an index. This technically - // should just be [0, 0] but OTS has an issue with that. - if (count === 0) { - return [0, 0, 0]; + function buildMulOperation(num1, num2) { + if (num2.type === 'literal') { + // optimization: second operands is a literal... + if (num2.number === 0) { + return new AstLiteral(0); // and it's 0 + } else if (num2.number === 1) { + return num1; // and it's 1 + } else if (num1.type === 'literal') { + // ... and first operands is a literal too + return new AstLiteral(num1.number * num2.number); } - - var data = [(count >> 8) & 0xFF, count & 0xff]; - - var lastOffset = 1, i; - for (i = 0; i < count; ++i) { - lastOffset += objects[i].length; + } + if (num1.type === 'literal') { + // optimization: first operands is a literal... + if (num1.number === 0) { + return new AstLiteral(0); // and it's 0 + } else if (num1.number === 1) { + return num2; // and it's 1 } + } + var min = Math.min(num1.min * num2.min, num1.min * num2.max, + num1.max * num2.min, num1.max * num2.max); + var max = Math.max(num1.min * num2.min, num1.min * num2.max, + num1.max * num2.min, num1.max * num2.max); + return new AstBinaryOperation('*', num1, num2, min, max); + } - var offsetSize; - if (lastOffset < 0x100) { - offsetSize = 1; - } else if (lastOffset < 0x10000) { - offsetSize = 2; - } else if (lastOffset < 0x1000000) { - offsetSize = 3; - } else { - offsetSize = 4; + function buildSubOperation(num1, num2) { + if (num2.type === 'literal') { + // optimization: second operands is a literal... + if (num2.number === 0) { + return num1; // ... and it's 0 + } else if (num1.type === 'literal') { + // ... and first operands is a literal too + return new AstLiteral(num1.number - num2.number); } + } + if (num2.type === 'binary' && num2.op === '-' && + num1.type === 'literal' && num1.number === 1 && + num2.arg1.type === 'literal' && num2.arg1.number === 1) { + // optimization for case: 1 - (1 - x) + return num2.arg2; + } + return new AstBinaryOperation('-', num1, num2, + num1.min - num2.max, num1.max - num2.min); + } - // Next byte contains the offset size use to reference object in the file - data.push(offsetSize); + function buildMinOperation(num1, max) { + if (num1.min >= max) { + // optimization: num1 min value is not less than required max + return new AstLiteral(max); // just returning max + } else if (num1.max <= max) { + // optimization: num1 max value is not greater than required max + return num1; // just returning an argument + } + return new AstMin(num1, max); + } - // Add another offset after this one because we need a new offset - var relativeOffset = 1; - for (i = 0; i < count + 1; i++) { - if (offsetSize === 1) { - data.push(relativeOffset & 0xFF); - } else if (offsetSize === 2) { - data.push((relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); - } else if (offsetSize === 3) { - data.push((relativeOffset >> 16) & 0xFF, - (relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); - } else { - data.push((relativeOffset >>> 24) & 0xFF, - (relativeOffset >> 16) & 0xFF, - (relativeOffset >> 8) & 0xFF, - relativeOffset & 0xFF); + function PostScriptCompiler() {} + PostScriptCompiler.prototype = { + compile: function PostScriptCompiler_compile(code, domain, range) { + var stack = []; + var i, ii; + var instructions = []; + var inputSize = domain.length >> 1, outputSize = range.length >> 1; + var lastRegister = 0; + var n, j; + var num1, num2, ast1, ast2, tmpVar, item; + for (i = 0; i < inputSize; i++) { + stack.push(new AstArgument(i, domain[i * 2], domain[i * 2 + 1])); + } + + for (i = 0, ii = code.length; i < ii; i++) { + item = code[i]; + if (typeof item === 'number') { + stack.push(new AstLiteral(item)); + continue; } - if (objects[i]) { - relativeOffset += objects[i].length; + switch (item) { + case 'add': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildAddOperation(num1, num2)); + break; + case 'cvr': + if (stack.length < 1) { + return null; + } + break; + case 'mul': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildMulOperation(num1, num2)); + break; + case 'sub': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + stack.push(buildSubOperation(num1, num2)); + break; + case 'exch': + if (stack.length < 2) { + return null; + } + ast1 = stack.pop(); ast2 = stack.pop(); + stack.push(ast1, ast2); + break; + case 'pop': + if (stack.length < 1) { + return null; + } + stack.pop(); + break; + case 'index': + if (stack.length < 1) { + return null; + } + num1 = stack.pop(); + if (num1.type !== 'literal') { + return null; + } + n = num1.number; + if (n < 0 || (n|0) !== n || stack.length < n) { + return null; + } + ast1 = stack[stack.length - n - 1]; + if (ast1.type === 'literal' || ast1.type === 'var') { + stack.push(ast1); + break; + } + tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); + stack[stack.length - n - 1] = tmpVar; + stack.push(tmpVar); + instructions.push(new AstVariableDefinition(tmpVar, ast1)); + break; + case 'dup': + if (stack.length < 1) { + return null; + } + if (typeof code[i + 1] === 'number' && code[i + 2] === 'gt' && + code[i + 3] === i + 7 && code[i + 4] === 'jz' && + code[i + 5] === 'pop' && code[i + 6] === code[i + 1]) { + // special case of the commands sequence for the min operation + num1 = stack.pop(); + stack.push(buildMinOperation(num1, code[i + 1])); + i += 6; + break; + } + ast1 = stack[stack.length - 1]; + if (ast1.type === 'literal' || ast1.type === 'var') { + // we don't have to save into intermediate variable a literal or + // variable. + stack.push(ast1); + break; + } + tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); + stack[stack.length - 1] = tmpVar; + stack.push(tmpVar); + instructions.push(new AstVariableDefinition(tmpVar, ast1)); + break; + case 'roll': + if (stack.length < 2) { + return null; + } + num2 = stack.pop(); + num1 = stack.pop(); + if (num2.type !== 'literal' || num1.type !== 'literal') { + // both roll operands must be numbers + return null; + } + j = num2.number; + n = num1.number; + if (n <= 0 || (n|0) !== n || (j|0) !== j || stack.length < n) { + // ... and integers + return null; + } + j = ((j % n) + n) % n; + if (j === 0) { + break; // just skipping -- there are nothing to rotate + } + Array.prototype.push.apply(stack, + stack.splice(stack.length - n, n - j)); + break; + default: + return null; // unsupported operator } } - for (i = 0; i < count; i++) { - // Notify the tracker where the object will be offset in the data. - if (trackers[i]) { - trackers[i].offset(data.length); + if (stack.length !== outputSize) { + return null; + } + + var result = []; + instructions.forEach(function (instruction) { + var statementBuilder = new ExpressionBuilderVisitor(); + instruction.visit(statementBuilder); + result.push(statementBuilder.toString()); + }); + stack.forEach(function (expr, i) { + var statementBuilder = new ExpressionBuilderVisitor(); + expr.visit(statementBuilder); + var min = range[i * 2], max = range[i * 2 + 1]; + var out = [statementBuilder.toString()]; + if (min > expr.min) { + out.unshift('Math.max(', min, ', '); + out.push(')'); } - for (var j = 0, jj = objects[i].length; j < jj; j++) { - data.push(objects[i][j]); + if (max < expr.max) { + out.unshift('Math.min(', max, ', '); + out.push(')'); } - } - return data; + out.unshift('dest[destOffset + ', i, '] = '); + out.push(';'); + result.push(out.join('')); + }); + return result.join('\n'); } }; - return CFFCompiler; -})(); - -function _enableSeacAnalysis(enabled) { - exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED = enabled; -} - -// Workaround for seac on Windows. -(function checkSeacSupport() { - if (typeof navigator !== 'undefined' && /Windows/.test(navigator.userAgent)) { - SEAC_ANALYSIS_ENABLED = true; - } -})(); -// Workaround for Private Use Area characters in Chrome on Windows -// http://code.google.com/p/chromium/issues/detail?id=122465 -// https://github.com/mozilla/pdf.js/issues/1689 -(function checkChromeWindows() { - if (typeof navigator !== 'undefined' && - /Windows.*Chrome/.test(navigator.userAgent)) { - SKIP_PRIVATE_USE_RANGE_F000_TO_F01F = true; - } + return PostScriptCompiler; })(); -exports.SEAC_ANALYSIS_ENABLED = SEAC_ANALYSIS_ENABLED; -exports.CFFCompiler = CFFCompiler; -exports.CFFIndex = CFFIndex; -exports.CFFParser = CFFParser; -exports.CFFStrings = CFFStrings; -exports.ErrorFont = ErrorFont; -exports.FontFlags = FontFlags; -exports.Font = Font; -exports.IdentityToUnicodeMap = IdentityToUnicodeMap; -exports.ToUnicodeMap = ToUnicodeMap; -exports.Type1Parser = Type1Parser; -exports.getFontType = getFontType; -exports._enableSeacAnalysis = _enableSeacAnalysis; - -// TODO refactor to remove cyclic dependency on font_renderer.js -coreFontRenderer._setCoreFonts(exports); +exports.isPDFFunction = isPDFFunction; +exports.PDFFunction = PDFFunction; +exports.PostScriptEvaluator = PostScriptEvaluator; +exports.PostScriptCompiler = PostScriptCompiler; })); (function (root, factory) { { - factory((root.pdfjsCoreFunction = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCorePsParser); + factory((root.pdfjsCoreColorSpace = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreFunction, root.pdfjsCoreStream); } -}(this, function (exports, sharedUtil, corePrimitives, corePsParser) { +}(this, function (exports, sharedUtil, corePrimitives, coreFunction, + coreStream) { var error = sharedUtil.error; var info = sharedUtil.info; var isArray = sharedUtil.isArray; -var isBool = sharedUtil.isBool; +var isString = sharedUtil.isString; +var shadow = sharedUtil.shadow; +var warn = sharedUtil.warn; var isDict = corePrimitives.isDict; +var isName = corePrimitives.isName; var isStream = corePrimitives.isStream; -var PostScriptLexer = corePsParser.PostScriptLexer; -var PostScriptParser = corePsParser.PostScriptParser; - -var PDFFunction = (function PDFFunctionClosure() { - var CONSTRUCT_SAMPLED = 0; - var CONSTRUCT_INTERPOLATED = 2; - var CONSTRUCT_STICHED = 3; - var CONSTRUCT_POSTSCRIPT = 4; - - return { - getSampleArray: function PDFFunction_getSampleArray(size, outputSize, bps, - str) { - var i, ii; - var length = 1; - for (i = 0, ii = size.length; i < ii; i++) { - length *= size[i]; - } - length *= outputSize; - - var array = new Array(length); - var codeSize = 0; - var codeBuf = 0; - // 32 is a valid bps so shifting won't work - var sampleMul = 1.0 / (Math.pow(2.0, bps) - 1); - - var strBytes = str.getBytes((length * bps + 7) / 8); - var strIdx = 0; - for (i = 0; i < length; i++) { - while (codeSize < bps) { - codeBuf <<= 8; - codeBuf |= strBytes[strIdx++]; - codeSize += 8; - } - codeSize -= bps; - array[i] = (codeBuf >> codeSize) * sampleMul; - codeBuf &= (1 << codeSize) - 1; - } - return array; - }, - - getIR: function PDFFunction_getIR(xref, fn) { - var dict = fn.dict; - if (!dict) { - dict = fn; - } - - var types = [this.constructSampled, - null, - this.constructInterpolated, - this.constructStiched, - this.constructPostScript]; - - var typeNum = dict.get('FunctionType'); - var typeFn = types[typeNum]; - if (!typeFn) { - error('Unknown type of function'); - } - - return typeFn.call(this, fn, dict, xref); - }, - - fromIR: function PDFFunction_fromIR(IR) { - var type = IR[0]; - switch (type) { - case CONSTRUCT_SAMPLED: - return this.constructSampledFromIR(IR); - case CONSTRUCT_INTERPOLATED: - return this.constructInterpolatedFromIR(IR); - case CONSTRUCT_STICHED: - return this.constructStichedFromIR(IR); - //case CONSTRUCT_POSTSCRIPT: - default: - return this.constructPostScriptFromIR(IR); - } - }, - - parse: function PDFFunction_parse(xref, fn) { - var IR = this.getIR(xref, fn); - return this.fromIR(IR); - }, - - parseArray: function PDFFunction_parseArray(xref, fnObj) { - if (!isArray(fnObj)) { - // not an array -- parsing as regular function - return this.parse(xref, fnObj); - } - - var fnArray = []; - for (var j = 0, jj = fnObj.length; j < jj; j++) { - var obj = xref.fetchIfRef(fnObj[j]); - fnArray.push(PDFFunction.parse(xref, obj)); - } - return function (src, srcOffset, dest, destOffset) { - for (var i = 0, ii = fnArray.length; i < ii; i++) { - fnArray[i](src, srcOffset, dest, destOffset + i); - } - }; - }, - - constructSampled: function PDFFunction_constructSampled(str, dict) { - function toMultiArray(arr) { - var inputLength = arr.length; - var out = []; - var index = 0; - for (var i = 0; i < inputLength; i += 2) { - out[index] = [arr[i], arr[i + 1]]; - ++index; - } - return out; - } - var domain = dict.get('Domain'); - var range = dict.get('Range'); - - if (!domain || !range) { - error('No domain or range'); - } - - var inputSize = domain.length / 2; - var outputSize = range.length / 2; - - domain = toMultiArray(domain); - range = toMultiArray(range); - - var size = dict.get('Size'); - var bps = dict.get('BitsPerSample'); - var order = dict.get('Order') || 1; - if (order !== 1) { - // No description how cubic spline interpolation works in PDF32000:2008 - // As in poppler, ignoring order, linear interpolation may work as good - info('No support for cubic spline interpolation: ' + order); - } - - var encode = dict.get('Encode'); - if (!encode) { - encode = []; - for (var i = 0; i < inputSize; ++i) { - encode.push(0); - encode.push(size[i] - 1); - } - } - encode = toMultiArray(encode); - - var decode = dict.get('Decode'); - if (!decode) { - decode = range; - } else { - decode = toMultiArray(decode); - } - - var samples = this.getSampleArray(size, outputSize, bps, str); - - return [ - CONSTRUCT_SAMPLED, inputSize, domain, encode, decode, samples, size, - outputSize, Math.pow(2, bps) - 1, range - ]; - }, - - constructSampledFromIR: function PDFFunction_constructSampledFromIR(IR) { - // See chapter 3, page 109 of the PDF reference - function interpolate(x, xmin, xmax, ymin, ymax) { - return ymin + ((x - xmin) * ((ymax - ymin) / (xmax - xmin))); - } - - return function constructSampledFromIRResult(src, srcOffset, - dest, destOffset) { - // See chapter 3, page 110 of the PDF reference. - var m = IR[1]; - var domain = IR[2]; - var encode = IR[3]; - var decode = IR[4]; - var samples = IR[5]; - var size = IR[6]; - var n = IR[7]; - //var mask = IR[8]; - var range = IR[9]; - - // Building the cube vertices: its part and sample index - // http://rjwagner49.com/Mathematics/Interpolation.pdf - var cubeVertices = 1 << m; - var cubeN = new Float64Array(cubeVertices); - var cubeVertex = new Uint32Array(cubeVertices); - var i, j; - for (j = 0; j < cubeVertices; j++) { - cubeN[j] = 1; - } - - var k = n, pos = 1; - // Map x_i to y_j for 0 <= i < m using the sampled function. - for (i = 0; i < m; ++i) { - // x_i' = min(max(x_i, Domain_2i), Domain_2i+1) - var domain_2i = domain[i][0]; - var domain_2i_1 = domain[i][1]; - var xi = Math.min(Math.max(src[srcOffset +i], domain_2i), - domain_2i_1); +var PDFFunction = coreFunction.PDFFunction; - // e_i = Interpolate(x_i', Domain_2i, Domain_2i+1, - // Encode_2i, Encode_2i+1) - var e = interpolate(xi, domain_2i, domain_2i_1, - encode[i][0], encode[i][1]); +var coreImage; // see _setCoreImage below +var PDFImage; // = coreImage.PDFImage; - // e_i' = min(max(e_i, 0), Size_i - 1) - var size_i = size[i]; - e = Math.min(Math.max(e, 0), size_i - 1); +var ColorSpace = (function ColorSpaceClosure() { + // Constructor should define this.numComps, this.defaultColor, this.name + function ColorSpace() { + error('should not call ColorSpace constructor'); + } - // Adjusting the cube: N and vertex sample index - var e0 = e < size_i - 1 ? Math.floor(e) : e - 1; // e1 = e0 + 1; - var n0 = e0 + 1 - e; // (e1 - e) / (e1 - e0); - var n1 = e - e0; // (e - e0) / (e1 - e0); - var offset0 = e0 * k; - var offset1 = offset0 + k; // e1 * k - for (j = 0; j < cubeVertices; j++) { - if (j & pos) { - cubeN[j] *= n1; - cubeVertex[j] += offset1; - } else { - cubeN[j] *= n0; - cubeVertex[j] += offset0; - } - } + ColorSpace.prototype = { + /** + * Converts the color value to the RGB color. The color components are + * located in the src array starting from the srcOffset. Returns the array + * of the rgb components, each value ranging from [0,255]. + */ + getRgb: function ColorSpace_getRgb(src, srcOffset) { + var rgb = new Uint8Array(3); + this.getRgbItem(src, srcOffset, rgb, 0); + return rgb; + }, + /** + * Converts the color value to the RGB color, similar to the getRgb method. + * The result placed into the dest array starting from the destOffset. + */ + getRgbItem: function ColorSpace_getRgbItem(src, srcOffset, + dest, destOffset) { + error('Should not call ColorSpace.getRgbItem'); + }, + /** + * Converts the specified number of the color values to the RGB colors. + * The colors are located in the src array starting from the srcOffset. + * The result is placed into the dest array starting from the destOffset. + * The src array items shall be in [0,2^bits) range, the dest array items + * will be in [0,255] range. alpha01 indicates how many alpha components + * there are in the dest array; it will be either 0 (RGB array) or 1 (RGBA + * array). + */ + getRgbBuffer: function ColorSpace_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + error('Should not call ColorSpace.getRgbBuffer'); + }, + /** + * Determines the number of bytes required to store the result of the + * conversion done by the getRgbBuffer method. As in getRgbBuffer, + * |alpha01| is either 0 (RGB output) or 1 (RGBA output). + */ + getOutputLength: function ColorSpace_getOutputLength(inputLength, + alpha01) { + error('Should not call ColorSpace.getOutputLength'); + }, + /** + * Returns true if source data will be equal the result/output data. + */ + isPassthrough: function ColorSpace_isPassthrough(bits) { + return false; + }, + /** + * Fills in the RGB colors in the destination buffer. alpha01 indicates + * how many alpha components there are in the dest array; it will be either + * 0 (RGB array) or 1 (RGBA array). + */ + fillRgb: function ColorSpace_fillRgb(dest, originalWidth, + originalHeight, width, height, + actualHeight, bpc, comps, alpha01) { + var count = originalWidth * originalHeight; + var rgbBuf = null; + var numComponentColors = 1 << bpc; + var needsResizing = originalHeight !== height || originalWidth !== width; + var i, ii; - k *= size_i; - pos <<= 1; + if (this.isPassthrough(bpc)) { + rgbBuf = comps; + } else if (this.numComps === 1 && count > numComponentColors && + this.name !== 'DeviceGray' && this.name !== 'DeviceRGB') { + // Optimization: create a color map when there is just one component and + // we are converting more colors than the size of the color map. We + // don't build the map if the colorspace is gray or rgb since those + // methods are faster than building a map. This mainly offers big speed + // ups for indexed and alternate colorspaces. + // + // TODO it may be worth while to cache the color map. While running + // testing I never hit a cache so I will leave that out for now (perhaps + // we are reparsing colorspaces too much?). + var allColors = bpc <= 8 ? new Uint8Array(numComponentColors) : + new Uint16Array(numComponentColors); + var key; + for (i = 0; i < numComponentColors; i++) { + allColors[i] = i; } + var colorMap = new Uint8Array(numComponentColors * 3); + this.getRgbBuffer(allColors, 0, numComponentColors, colorMap, 0, bpc, + /* alpha01 = */ 0); - for (j = 0; j < n; ++j) { - // Sum all cube vertices' samples portions - var rj = 0; - for (i = 0; i < cubeVertices; i++) { - rj += samples[cubeVertex[i] + j] * cubeN[i]; + var destPos, rgbPos; + if (!needsResizing) { + // Fill in the RGB values directly into |dest|. + destPos = 0; + for (i = 0; i < count; ++i) { + key = comps[i] * 3; + dest[destPos++] = colorMap[key]; + dest[destPos++] = colorMap[key + 1]; + dest[destPos++] = colorMap[key + 2]; + destPos += alpha01; + } + } else { + rgbBuf = new Uint8Array(count * 3); + rgbPos = 0; + for (i = 0; i < count; ++i) { + key = comps[i] * 3; + rgbBuf[rgbPos++] = colorMap[key]; + rgbBuf[rgbPos++] = colorMap[key + 1]; + rgbBuf[rgbPos++] = colorMap[key + 2]; } - - // r_j' = Interpolate(r_j, 0, 2^BitsPerSample - 1, - // Decode_2j, Decode_2j+1) - rj = interpolate(rj, 0, 1, decode[j][0], decode[j][1]); - - // y_j = min(max(r_j, range_2j), range_2j+1) - dest[destOffset + j] = Math.min(Math.max(rj, range[j][0]), - range[j][1]); } - }; - }, - - constructInterpolated: function PDFFunction_constructInterpolated(str, - dict) { - var c0 = dict.get('C0') || [0]; - var c1 = dict.get('C1') || [1]; - var n = dict.get('N'); - - if (!isArray(c0) || !isArray(c1)) { - error('Illegal dictionary for interpolated function'); + } else { + if (!needsResizing) { + // Fill in the RGB values directly into |dest|. + this.getRgbBuffer(comps, 0, width * actualHeight, dest, 0, bpc, + alpha01); + } else { + rgbBuf = new Uint8Array(count * 3); + this.getRgbBuffer(comps, 0, count, rgbBuf, 0, bpc, + /* alpha01 = */ 0); + } } - var length = c0.length; - var diff = []; - for (var i = 0; i < length; ++i) { - diff.push(c1[i] - c0[i]); + if (rgbBuf) { + if (needsResizing) { + PDFImage.resize(rgbBuf, bpc, 3, originalWidth, originalHeight, width, + height, dest, alpha01); + } else { + rgbPos = 0; + destPos = 0; + for (i = 0, ii = width * actualHeight; i < ii; i++) { + dest[destPos++] = rgbBuf[rgbPos++]; + dest[destPos++] = rgbBuf[rgbPos++]; + dest[destPos++] = rgbBuf[rgbPos++]; + destPos += alpha01; + } + } } - - return [CONSTRUCT_INTERPOLATED, c0, diff, n]; }, + /** + * True if the colorspace has components in the default range of [0, 1]. + * This should be true for all colorspaces except for lab color spaces + * which are [0,100], [-128, 127], [-128, 127]. + */ + usesZeroToOneRange: true + }; - constructInterpolatedFromIR: - function PDFFunction_constructInterpolatedFromIR(IR) { - var c0 = IR[1]; - var diff = IR[2]; - var n = IR[3]; - - var length = diff.length; + ColorSpace.parse = function ColorSpace_parse(cs, xref, res) { + var IR = ColorSpace.parseToIR(cs, xref, res); + if (IR instanceof AlternateCS) { + return IR; + } + return ColorSpace.fromIR(IR); + }; - return function constructInterpolatedFromIRResult(src, srcOffset, - dest, destOffset) { - var x = n === 1 ? src[srcOffset] : Math.pow(src[srcOffset], n); + ColorSpace.fromIR = function ColorSpace_fromIR(IR) { + var name = isArray(IR) ? IR[0] : IR; + var whitePoint, blackPoint, gamma; - for (var j = 0; j < length; ++j) { - dest[destOffset + j] = c0[j] + (x * diff[j]); + switch (name) { + case 'DeviceGrayCS': + return this.singletons.gray; + case 'DeviceRgbCS': + return this.singletons.rgb; + case 'DeviceCmykCS': + return this.singletons.cmyk; + case 'CalGrayCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + gamma = IR[3]; + return new CalGrayCS(whitePoint, blackPoint, gamma); + case 'CalRGBCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + gamma = IR[3]; + var matrix = IR[4]; + return new CalRGBCS(whitePoint, blackPoint, gamma, matrix); + case 'PatternCS': + var basePatternCS = IR[1]; + if (basePatternCS) { + basePatternCS = ColorSpace.fromIR(basePatternCS); } - }; - }, + return new PatternCS(basePatternCS); + case 'IndexedCS': + var baseIndexedCS = IR[1]; + var hiVal = IR[2]; + var lookup = IR[3]; + return new IndexedCS(ColorSpace.fromIR(baseIndexedCS), hiVal, lookup); + case 'AlternateCS': + var numComps = IR[1]; + var alt = IR[2]; + var tintFnIR = IR[3]; - constructStiched: function PDFFunction_constructStiched(fn, dict, xref) { - var domain = dict.get('Domain'); + return new AlternateCS(numComps, ColorSpace.fromIR(alt), + PDFFunction.fromIR(tintFnIR)); + case 'LabCS': + whitePoint = IR[1]; + blackPoint = IR[2]; + var range = IR[3]; + return new LabCS(whitePoint, blackPoint, range); + default: + error('Unknown name ' + name); + } + return null; + }; - if (!domain) { - error('No domain'); + ColorSpace.parseToIR = function ColorSpace_parseToIR(cs, xref, res) { + if (isName(cs)) { + var colorSpaces = res.get('ColorSpace'); + if (isDict(colorSpaces)) { + var refcs = colorSpaces.get(cs.name); + if (refcs) { + cs = refcs; + } } + } - var inputSize = domain.length / 2; - if (inputSize !== 1) { - error('Bad domain for stiched function'); + cs = xref.fetchIfRef(cs); + var mode; + + if (isName(cs)) { + mode = cs.name; + this.mode = mode; + + switch (mode) { + case 'DeviceGray': + case 'G': + return 'DeviceGrayCS'; + case 'DeviceRGB': + case 'RGB': + return 'DeviceRgbCS'; + case 'DeviceCMYK': + case 'CMYK': + return 'DeviceCmykCS'; + case 'Pattern': + return ['PatternCS', null]; + default: + error('unrecognized colorspace ' + mode); } + } else if (isArray(cs)) { + mode = xref.fetchIfRef(cs[0]).name; + this.mode = mode; + var numComps, params, alt, whitePoint, blackPoint, gamma; - var fnRefs = dict.get('Functions'); - var fns = []; - for (var i = 0, ii = fnRefs.length; i < ii; ++i) { - fns.push(PDFFunction.getIR(xref, xref.fetchIfRef(fnRefs[i]))); + switch (mode) { + case 'DeviceGray': + case 'G': + return 'DeviceGrayCS'; + case 'DeviceRGB': + case 'RGB': + return 'DeviceRgbCS'; + case 'DeviceCMYK': + case 'CMYK': + return 'DeviceCmykCS'; + case 'CalGray': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + gamma = params.get('Gamma'); + return ['CalGrayCS', whitePoint, blackPoint, gamma]; + case 'CalRGB': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + gamma = params.get('Gamma'); + var matrix = params.get('Matrix'); + return ['CalRGBCS', whitePoint, blackPoint, gamma, matrix]; + case 'ICCBased': + var stream = xref.fetchIfRef(cs[1]); + var dict = stream.dict; + numComps = dict.get('N'); + alt = dict.get('Alternate'); + if (alt) { + var altIR = ColorSpace.parseToIR(alt, xref, res); + // Parse the /Alternate CS to ensure that the number of components + // are correct, and also (indirectly) that it is not a PatternCS. + var altCS = ColorSpace.fromIR(altIR); + if (altCS.numComps === numComps) { + return altIR; + } + warn('ICCBased color space: Ignoring incorrect /Alternate entry.'); + } + if (numComps === 1) { + return 'DeviceGrayCS'; + } else if (numComps === 3) { + return 'DeviceRgbCS'; + } else if (numComps === 4) { + return 'DeviceCmykCS'; + } + break; + case 'Pattern': + var basePatternCS = cs[1] || null; + if (basePatternCS) { + basePatternCS = ColorSpace.parseToIR(basePatternCS, xref, res); + } + return ['PatternCS', basePatternCS]; + case 'Indexed': + case 'I': + var baseIndexedCS = ColorSpace.parseToIR(cs[1], xref, res); + var hiVal = xref.fetchIfRef(cs[2]) + 1; + var lookup = xref.fetchIfRef(cs[3]); + if (isStream(lookup)) { + lookup = lookup.getBytes(); + } + return ['IndexedCS', baseIndexedCS, hiVal, lookup]; + case 'Separation': + case 'DeviceN': + var name = xref.fetchIfRef(cs[1]); + numComps = 1; + if (isName(name)) { + numComps = 1; + } else if (isArray(name)) { + numComps = name.length; + } + alt = ColorSpace.parseToIR(cs[2], xref, res); + var tintFnIR = PDFFunction.getIR(xref, xref.fetchIfRef(cs[3])); + return ['AlternateCS', numComps, alt, tintFnIR]; + case 'Lab': + params = xref.fetchIfRef(cs[1]); + whitePoint = params.get('WhitePoint'); + blackPoint = params.get('BlackPoint'); + var range = params.get('Range'); + return ['LabCS', whitePoint, blackPoint, range]; + default: + error('unimplemented color space object "' + mode + '"'); } + } else { + error('unrecognized color space object: "' + cs + '"'); + } + return null; + }; + /** + * Checks if a decode map matches the default decode map for a color space. + * This handles the general decode maps where there are two values per + * component. e.g. [0, 1, 0, 1, 0, 1] for a RGB color. + * This does not handle Lab, Indexed, or Pattern decode maps since they are + * slightly different. + * @param {Array} decode Decode map (usually from an image). + * @param {Number} n Number of components the color space has. + */ + ColorSpace.isDefaultDecode = function ColorSpace_isDefaultDecode(decode, n) { + if (!isArray(decode)) { + return true; + } - var bounds = dict.get('Bounds'); - var encode = dict.get('Encode'); + if (n * 2 !== decode.length) { + warn('The decode map is not the correct length'); + return true; + } + for (var i = 0, ii = decode.length; i < ii; i += 2) { + if (decode[i] !== 0 || decode[i + 1] !== 1) { + return false; + } + } + return true; + }; - return [CONSTRUCT_STICHED, domain, bounds, encode, fns]; + ColorSpace.singletons = { + get gray() { + return shadow(this, 'gray', new DeviceGrayCS()); + }, + get rgb() { + return shadow(this, 'rgb', new DeviceRgbCS()); }, + get cmyk() { + return shadow(this, 'cmyk', new DeviceCmykCS()); + } + }; - constructStichedFromIR: function PDFFunction_constructStichedFromIR(IR) { - var domain = IR[1]; - var bounds = IR[2]; - var encode = IR[3]; - var fnsIR = IR[4]; - var fns = []; - var tmpBuf = new Float32Array(1); + return ColorSpace; +})(); - for (var i = 0, ii = fnsIR.length; i < ii; i++) { - fns.push(PDFFunction.fromIR(fnsIR[i])); - } +/** + * Alternate color space handles both Separation and DeviceN color spaces. A + * Separation color space is actually just a DeviceN with one color component. + * Both color spaces use a tinting function to convert colors to a base color + * space. + */ +var AlternateCS = (function AlternateCSClosure() { + function AlternateCS(numComps, base, tintFn) { + this.name = 'Alternate'; + this.numComps = numComps; + this.defaultColor = new Float32Array(numComps); + for (var i = 0; i < numComps; ++i) { + this.defaultColor[i] = 1; + } + this.base = base; + this.tintFn = tintFn; + this.tmpBuf = new Float32Array(base.numComps); + } - return function constructStichedFromIRResult(src, srcOffset, - dest, destOffset) { - var clip = function constructStichedFromIRClip(v, min, max) { - if (v > max) { - v = max; - } else if (v < min) { - v = min; - } - return v; - }; + AlternateCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function AlternateCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var tmpBuf = this.tmpBuf; + this.tintFn(src, srcOffset, tmpBuf, 0); + this.base.getRgbItem(tmpBuf, 0, dest, destOffset); + }, + getRgbBuffer: function AlternateCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var tintFn = this.tintFn; + var base = this.base; + var scale = 1 / ((1 << bits) - 1); + var baseNumComps = base.numComps; + var usesZeroToOneRange = base.usesZeroToOneRange; + var isPassthrough = (base.isPassthrough(8) || !usesZeroToOneRange) && + alpha01 === 0; + var pos = isPassthrough ? destOffset : 0; + var baseBuf = isPassthrough ? dest : new Uint8Array(baseNumComps * count); + var numComps = this.numComps; - // clip to domain - var v = clip(src[srcOffset], domain[0], domain[1]); - // calulate which bound the value is in - for (var i = 0, ii = bounds.length; i < ii; ++i) { - if (v < bounds[i]) { - break; + var scaled = new Float32Array(numComps); + var tinted = new Float32Array(baseNumComps); + var i, j; + if (usesZeroToOneRange) { + for (i = 0; i < count; i++) { + for (j = 0; j < numComps; j++) { + scaled[j] = src[srcOffset++] * scale; + } + tintFn(scaled, 0, tinted, 0); + for (j = 0; j < baseNumComps; j++) { + baseBuf[pos++] = tinted[j] * 255; } } - - // encode value into domain of function - var dmin = domain[0]; - if (i > 0) { - dmin = bounds[i - 1]; - } - var dmax = domain[1]; - if (i < bounds.length) { - dmax = bounds[i]; + } else { + for (i = 0; i < count; i++) { + for (j = 0; j < numComps; j++) { + scaled[j] = src[srcOffset++] * scale; + } + tintFn(scaled, 0, tinted, 0); + base.getRgbItem(tinted, 0, baseBuf, pos); + pos += baseNumComps; } + } + if (!isPassthrough) { + base.getRgbBuffer(baseBuf, 0, count, dest, destOffset, 8, alpha01); + } + }, + getOutputLength: function AlternateCS_getOutputLength(inputLength, + alpha01) { + return this.base.getOutputLength(inputLength * + this.base.numComps / this.numComps, + alpha01); + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function AlternateCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true + }; - var rmin = encode[2 * i]; - var rmax = encode[2 * i + 1]; + return AlternateCS; +})(); - // Prevent the value from becoming NaN as a result - // of division by zero (fixes issue6113.pdf). - tmpBuf[0] = dmin === dmax ? rmin : - rmin + (v - dmin) * (rmax - rmin) / (dmax - dmin); +var PatternCS = (function PatternCSClosure() { + function PatternCS(baseCS) { + this.name = 'Pattern'; + this.base = baseCS; + } + PatternCS.prototype = {}; - // call the appropriate function - fns[i](tmpBuf, 0, dest, destOffset); - }; - }, + return PatternCS; +})(); - constructPostScript: function PDFFunction_constructPostScript(fn, dict, - xref) { - var domain = dict.get('Domain'); - var range = dict.get('Range'); +var IndexedCS = (function IndexedCSClosure() { + function IndexedCS(base, highVal, lookup) { + this.name = 'Indexed'; + this.numComps = 1; + this.defaultColor = new Uint8Array([0]); + this.base = base; + this.highVal = highVal; - if (!domain) { - error('No domain.'); - } + var baseNumComps = base.numComps; + var length = baseNumComps * highVal; + var lookupArray; - if (!range) { - error('No range.'); + if (isStream(lookup)) { + lookupArray = new Uint8Array(length); + var bytes = lookup.getBytes(length); + lookupArray.set(bytes); + } else if (isString(lookup)) { + lookupArray = new Uint8Array(length); + for (var i = 0; i < length; ++i) { + lookupArray[i] = lookup.charCodeAt(i); } + } else if (lookup instanceof Uint8Array || lookup instanceof Array) { + lookupArray = lookup; + } else { + error('Unrecognized lookup table: ' + lookup); + } + this.lookup = lookupArray; + } - var lexer = new PostScriptLexer(fn); - var parser = new PostScriptParser(lexer); - var code = parser.parse(); - - return [CONSTRUCT_POSTSCRIPT, domain, range, code]; + IndexedCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function IndexedCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var numComps = this.base.numComps; + var start = src[srcOffset] * numComps; + this.base.getRgbItem(this.lookup, start, dest, destOffset); }, + getRgbBuffer: function IndexedCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var base = this.base; + var numComps = base.numComps; + var outputDelta = base.getOutputLength(numComps, alpha01); + var lookup = this.lookup; - constructPostScriptFromIR: function PDFFunction_constructPostScriptFromIR( - IR) { - var domain = IR[1]; - var range = IR[2]; - var code = IR[3]; - - var compiled = (new PostScriptCompiler()).compile(code, domain, range); - if (compiled) { - // Compiled function consists of simple expressions such as addition, - // subtraction, Math.max, and also contains 'var' and 'return' - // statements. See the generation in the PostScriptCompiler below. - /*jshint -W054 */ - return new Function('src', 'srcOffset', 'dest', 'destOffset', compiled); + for (var i = 0; i < count; ++i) { + var lookupPos = src[srcOffset++] * numComps; + base.getRgbBuffer(lookup, lookupPos, 1, dest, destOffset, 8, alpha01); + destOffset += outputDelta; } - - info('Unable to compile PS function'); - - var numOutputs = range.length >> 1; - var numInputs = domain.length >> 1; - var evaluator = new PostScriptEvaluator(code); - // Cache the values for a big speed up, the cache size is limited though - // since the number of possible values can be huge from a PS function. - var cache = Object.create(null); - // The MAX_CACHE_SIZE is set to ~4x the maximum number of distinct values - // seen in our tests. - var MAX_CACHE_SIZE = 2048 * 4; - var cache_available = MAX_CACHE_SIZE; - var tmpBuf = new Float32Array(numInputs); - - return function constructPostScriptFromIRResult(src, srcOffset, - dest, destOffset) { - var i, value; - var key = ''; - var input = tmpBuf; - for (i = 0; i < numInputs; i++) { - value = src[srcOffset + i]; - input[i] = value; - key += value + '_'; - } - - var cachedValue = cache[key]; - if (cachedValue !== undefined) { - dest.set(cachedValue, destOffset); - return; - } - - var output = new Float32Array(numOutputs); - var stack = evaluator.execute(input); - var stackIndex = stack.length - numOutputs; - for (i = 0; i < numOutputs; i++) { - value = stack[stackIndex + i]; - var bound = range[i * 2]; - if (value < bound) { - value = bound; - } else { - bound = range[i * 2 +1]; - if (value > bound) { - value = bound; - } - } - output[i] = value; - } - if (cache_available > 0) { - cache_available--; - cache[key] = output; - } - dest.set(output, destOffset); - }; - } + }, + getOutputLength: function IndexedCS_getOutputLength(inputLength, alpha01) { + return this.base.getOutputLength(inputLength * this.base.numComps, + alpha01); + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function IndexedCS_isDefaultDecode(decodeMap) { + // indexed color maps shouldn't be changed + return true; + }, + usesZeroToOneRange: true }; + return IndexedCS; })(); -function isPDFFunction(v) { - var fnDict; - if (typeof v !== 'object') { - return false; - } else if (isDict(v)) { - fnDict = v; - } else if (isStream(v)) { - fnDict = v.dict; - } else { - return false; - } - return fnDict.has('FunctionType'); -} - -var PostScriptStack = (function PostScriptStackClosure() { - var MAX_STACK_SIZE = 100; - function PostScriptStack(initialStack) { - this.stack = !initialStack ? [] : - Array.prototype.slice.call(initialStack, 0); +var DeviceGrayCS = (function DeviceGrayCSClosure() { + function DeviceGrayCS() { + this.name = 'DeviceGray'; + this.numComps = 1; + this.defaultColor = new Float32Array([0]); } - PostScriptStack.prototype = { - push: function PostScriptStack_push(value) { - if (this.stack.length >= MAX_STACK_SIZE) { - error('PostScript function stack overflow.'); - } - this.stack.push(value); + DeviceGrayCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceGrayCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var c = (src[srcOffset] * 255) | 0; + c = c < 0 ? 0 : c > 255 ? 255 : c; + dest[destOffset] = dest[destOffset + 1] = dest[destOffset + 2] = c; }, - pop: function PostScriptStack_pop() { - if (this.stack.length <= 0) { - error('PostScript function stack underflow.'); + getRgbBuffer: function DeviceGrayCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 255 / ((1 << bits) - 1); + var j = srcOffset, q = destOffset; + for (var i = 0; i < count; ++i) { + var c = (scale * src[j++]) | 0; + dest[q++] = c; + dest[q++] = c; + dest[q++] = c; + q += alpha01; } - return this.stack.pop(); }, - copy: function PostScriptStack_copy(n) { - if (this.stack.length + n >= MAX_STACK_SIZE) { - error('PostScript function stack overflow.'); - } - var stack = this.stack; - for (var i = stack.length - n, j = n - 1; j >= 0; j--, i++) { - stack.push(stack[i]); - } + getOutputLength: function DeviceGrayCS_getOutputLength(inputLength, + alpha01) { + return inputLength * (3 + alpha01); }, - index: function PostScriptStack_index(n) { - this.push(this.stack[this.stack.length - n - 1]); + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceGrayCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - // rotate the last n stack elements p times - roll: function PostScriptStack_roll(n, p) { - var stack = this.stack; - var l = stack.length - n; - var r = stack.length - 1, c = l + (p - Math.floor(p / n) * n), i, j, t; - for (i = l, j = r; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - for (i = l, j = c - 1; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - for (i = c, j = r; i < j; i++, j--) { - t = stack[i]; stack[i] = stack[j]; stack[j] = t; - } - } + usesZeroToOneRange: true }; - return PostScriptStack; + return DeviceGrayCS; })(); -var PostScriptEvaluator = (function PostScriptEvaluatorClosure() { - function PostScriptEvaluator(operators) { - this.operators = operators; - } - PostScriptEvaluator.prototype = { - execute: function PostScriptEvaluator_execute(initialStack) { - var stack = new PostScriptStack(initialStack); - var counter = 0; - var operators = this.operators; - var length = operators.length; - var operator, a, b; - while (counter < length) { - operator = operators[counter++]; - if (typeof operator === 'number') { - // Operator is really an operand and should be pushed to the stack. - stack.push(operator); - continue; - } - switch (operator) { - // non standard ps operators - case 'jz': // jump if false - b = stack.pop(); - a = stack.pop(); - if (!a) { - counter = b; - } - break; - case 'j': // jump - a = stack.pop(); - counter = a; - break; - // all ps operators in alphabetical order (excluding if/ifelse) - case 'abs': - a = stack.pop(); - stack.push(Math.abs(a)); - break; - case 'add': - b = stack.pop(); - a = stack.pop(); - stack.push(a + b); - break; - case 'and': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a && b); - } else { - stack.push(a & b); - } - break; - case 'atan': - a = stack.pop(); - stack.push(Math.atan(a)); - break; - case 'bitshift': - b = stack.pop(); - a = stack.pop(); - if (a > 0) { - stack.push(a << b); - } else { - stack.push(a >> b); - } - break; - case 'ceiling': - a = stack.pop(); - stack.push(Math.ceil(a)); - break; - case 'copy': - a = stack.pop(); - stack.copy(a); - break; - case 'cos': - a = stack.pop(); - stack.push(Math.cos(a)); - break; - case 'cvi': - a = stack.pop() | 0; - stack.push(a); - break; - case 'cvr': - // noop - break; - case 'div': - b = stack.pop(); - a = stack.pop(); - stack.push(a / b); - break; - case 'dup': - stack.copy(1); - break; - case 'eq': - b = stack.pop(); - a = stack.pop(); - stack.push(a === b); - break; - case 'exch': - stack.roll(2, 1); - break; - case 'exp': - b = stack.pop(); - a = stack.pop(); - stack.push(Math.pow(a, b)); - break; - case 'false': - stack.push(false); - break; - case 'floor': - a = stack.pop(); - stack.push(Math.floor(a)); - break; - case 'ge': - b = stack.pop(); - a = stack.pop(); - stack.push(a >= b); - break; - case 'gt': - b = stack.pop(); - a = stack.pop(); - stack.push(a > b); - break; - case 'idiv': - b = stack.pop(); - a = stack.pop(); - stack.push((a / b) | 0); - break; - case 'index': - a = stack.pop(); - stack.index(a); - break; - case 'le': - b = stack.pop(); - a = stack.pop(); - stack.push(a <= b); - break; - case 'ln': - a = stack.pop(); - stack.push(Math.log(a)); - break; - case 'log': - a = stack.pop(); - stack.push(Math.log(a) / Math.LN10); - break; - case 'lt': - b = stack.pop(); - a = stack.pop(); - stack.push(a < b); - break; - case 'mod': - b = stack.pop(); - a = stack.pop(); - stack.push(a % b); - break; - case 'mul': - b = stack.pop(); - a = stack.pop(); - stack.push(a * b); - break; - case 'ne': - b = stack.pop(); - a = stack.pop(); - stack.push(a !== b); - break; - case 'neg': - a = stack.pop(); - stack.push(-a); - break; - case 'not': - a = stack.pop(); - if (isBool(a)) { - stack.push(!a); - } else { - stack.push(~a); - } - break; - case 'or': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a || b); - } else { - stack.push(a | b); - } - break; - case 'pop': - stack.pop(); - break; - case 'roll': - b = stack.pop(); - a = stack.pop(); - stack.roll(a, b); - break; - case 'round': - a = stack.pop(); - stack.push(Math.round(a)); - break; - case 'sin': - a = stack.pop(); - stack.push(Math.sin(a)); - break; - case 'sqrt': - a = stack.pop(); - stack.push(Math.sqrt(a)); - break; - case 'sub': - b = stack.pop(); - a = stack.pop(); - stack.push(a - b); - break; - case 'true': - stack.push(true); - break; - case 'truncate': - a = stack.pop(); - a = a < 0 ? Math.ceil(a) : Math.floor(a); - stack.push(a); - break; - case 'xor': - b = stack.pop(); - a = stack.pop(); - if (isBool(a) && isBool(b)) { - stack.push(a !== b); - } else { - stack.push(a ^ b); - } - break; - default: - error('Unknown operator ' + operator); - break; - } +var DeviceRgbCS = (function DeviceRgbCSClosure() { + function DeviceRgbCS() { + this.name = 'DeviceRGB'; + this.numComps = 3; + this.defaultColor = new Float32Array([0, 0, 0]); + } + DeviceRgbCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceRgbCS_getRgbItem(src, srcOffset, + dest, destOffset) { + var r = (src[srcOffset] * 255) | 0; + var g = (src[srcOffset + 1] * 255) | 0; + var b = (src[srcOffset + 2] * 255) | 0; + dest[destOffset] = r < 0 ? 0 : r > 255 ? 255 : r; + dest[destOffset + 1] = g < 0 ? 0 : g > 255 ? 255 : g; + dest[destOffset + 2] = b < 0 ? 0 : b > 255 ? 255 : b; + }, + getRgbBuffer: function DeviceRgbCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + if (bits === 8 && alpha01 === 0) { + dest.set(src.subarray(srcOffset, srcOffset + count * 3), destOffset); + return; } - return stack.stack; - } + var scale = 255 / ((1 << bits) - 1); + var j = srcOffset, q = destOffset; + for (var i = 0; i < count; ++i) { + dest[q++] = (scale * src[j++]) | 0; + dest[q++] = (scale * src[j++]) | 0; + dest[q++] = (scale * src[j++]) | 0; + q += alpha01; + } + }, + getOutputLength: function DeviceRgbCS_getOutputLength(inputLength, + alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; + }, + isPassthrough: function DeviceRgbCS_isPassthrough(bits) { + return bits === 8; + }, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceRgbCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true }; - return PostScriptEvaluator; + return DeviceRgbCS; })(); -// Most of the PDFs functions consist of simple operations such as: -// roll, exch, sub, cvr, pop, index, dup, mul, if, gt, add. -// -// We can compile most of such programs, and at the same moment, we can -// optimize some expressions using basic math properties. Keeping track of -// min/max values will allow us to avoid extra Math.min/Math.max calls. -var PostScriptCompiler = (function PostScriptCompilerClosure() { - function AstNode(type) { - this.type = type; - } - AstNode.prototype.visit = function (visitor) { - throw new Error('abstract method'); - }; +var DeviceCmykCS = (function DeviceCmykCSClosure() { + // The coefficients below was found using numerical analysis: the method of + // steepest descent for the sum((f_i - color_value_i)^2) for r/g/b colors, + // where color_value is the tabular value from the table of sampled RGB colors + // from CMYK US Web Coated (SWOP) colorspace, and f_i is the corresponding + // CMYK color conversion using the estimation below: + // f(A, B,.. N) = Acc+Bcm+Ccy+Dck+c+Fmm+Gmy+Hmk+Im+Jyy+Kyk+Ly+Mkk+Nk+255 + function convertToRgb(src, srcOffset, srcScale, dest, destOffset) { + var c = src[srcOffset + 0] * srcScale; + var m = src[srcOffset + 1] * srcScale; + var y = src[srcOffset + 2] * srcScale; + var k = src[srcOffset + 3] * srcScale; - function AstArgument(index, min, max) { - AstNode.call(this, 'args'); - this.index = index; - this.min = min; - this.max = max; - } - AstArgument.prototype = Object.create(AstNode.prototype); - AstArgument.prototype.visit = function (visitor) { - visitor.visitArgument(this); - }; + var r = + (c * (-4.387332384609988 * c + 54.48615194189176 * m + + 18.82290502165302 * y + 212.25662451639585 * k + + -285.2331026137004) + + m * (1.7149763477362134 * m - 5.6096736904047315 * y + + -17.873870861415444 * k - 5.497006427196366) + + y * (-2.5217340131683033 * y - 21.248923337353073 * k + + 17.5119270841813) + + k * (-21.86122147463605 * k - 189.48180835922747) + 255) | 0; + var g = + (c * (8.841041422036149 * c + 60.118027045597366 * m + + 6.871425592049007 * y + 31.159100130055922 * k + + -79.2970844816548) + + m * (-15.310361306967817 * m + 17.575251261109482 * y + + 131.35250912493976 * k - 190.9453302588951) + + y * (4.444339102852739 * y + 9.8632861493405 * k - 24.86741582555878) + + k * (-20.737325471181034 * k - 187.80453709719578) + 255) | 0; + var b = + (c * (0.8842522430003296 * c + 8.078677503112928 * m + + 30.89978309703729 * y - 0.23883238689178934 * k + + -14.183576799673286) + + m * (10.49593273432072 * m + 63.02378494754052 * y + + 50.606957656360734 * k - 112.23884253719248) + + y * (0.03296041114873217 * y + 115.60384449646641 * k + + -193.58209356861505) + + k * (-22.33816807309886 * k - 180.12613974708367) + 255) | 0; - function AstLiteral(number) { - AstNode.call(this, 'literal'); - this.number = number; - this.min = number; - this.max = number; + dest[destOffset] = r > 255 ? 255 : r < 0 ? 0 : r; + dest[destOffset + 1] = g > 255 ? 255 : g < 0 ? 0 : g; + dest[destOffset + 2] = b > 255 ? 255 : b < 0 ? 0 : b; } - AstLiteral.prototype = Object.create(AstNode.prototype); - AstLiteral.prototype.visit = function (visitor) { - visitor.visitLiteral(this); - }; - function AstBinaryOperation(op, arg1, arg2, min, max) { - AstNode.call(this, 'binary'); - this.op = op; - this.arg1 = arg1; - this.arg2 = arg2; - this.min = min; - this.max = max; + function DeviceCmykCS() { + this.name = 'DeviceCMYK'; + this.numComps = 4; + this.defaultColor = new Float32Array([0, 0, 0, 1]); } - AstBinaryOperation.prototype = Object.create(AstNode.prototype); - AstBinaryOperation.prototype.visit = function (visitor) { - visitor.visitBinaryOperation(this); + DeviceCmykCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function DeviceCmykCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(src, srcOffset, 1, dest, destOffset); + }, + getRgbBuffer: function DeviceCmykCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + for (var i = 0; i < count; i++) { + convertToRgb(src, srcOffset, scale, dest, destOffset); + srcOffset += 4; + destOffset += 3 + alpha01; + } + }, + getOutputLength: function DeviceCmykCS_getOutputLength(inputLength, + alpha01) { + return (inputLength / 4 * (3 + alpha01)) | 0; + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function DeviceCmykCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); + }, + usesZeroToOneRange: true }; - function AstMin(arg, max) { - AstNode.call(this, 'max'); - this.arg = arg; - this.min = arg.min; - this.max = max; - } - AstMin.prototype = Object.create(AstNode.prototype); - AstMin.prototype.visit = function (visitor) { - visitor.visitMin(this); - }; + return DeviceCmykCS; +})(); - function AstVariable(index, min, max) { - AstNode.call(this, 'var'); - this.index = index; - this.min = min; - this.max = max; - } - AstVariable.prototype = Object.create(AstNode.prototype); - AstVariable.prototype.visit = function (visitor) { - visitor.visitVariable(this); - }; +// +// CalGrayCS: Based on "PDF Reference, Sixth Ed", p.245 +// +var CalGrayCS = (function CalGrayCSClosure() { + function CalGrayCS(whitePoint, blackPoint, gamma) { + this.name = 'CalGray'; + this.numComps = 1; + this.defaultColor = new Float32Array([0]); - function AstVariableDefinition(variable, arg) { - AstNode.call(this, 'definition'); - this.variable = variable; - this.arg = arg; + if (!whitePoint) { + error('WhitePoint missing - required for color space CalGray'); + } + blackPoint = blackPoint || [0, 0, 0]; + gamma = gamma || 1; + + // Translate arguments to spec variables. + this.XW = whitePoint[0]; + this.YW = whitePoint[1]; + this.ZW = whitePoint[2]; + + this.XB = blackPoint[0]; + this.YB = blackPoint[1]; + this.ZB = blackPoint[2]; + + this.G = gamma; + + // Validate variables as per spec. + if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { + error('Invalid WhitePoint components for ' + this.name + + ', no fallback available'); + } + + if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { + info('Invalid BlackPoint for ' + this.name + ', falling back to default'); + this.XB = this.YB = this.ZB = 0; + } + + if (this.XB !== 0 || this.YB !== 0 || this.ZB !== 0) { + warn(this.name + ', BlackPoint: XB: ' + this.XB + ', YB: ' + this.YB + + ', ZB: ' + this.ZB + ', only default values are supported.'); + } + + if (this.G < 1) { + info('Invalid Gamma: ' + this.G + ' for ' + this.name + + ', falling back to default'); + this.G = 1; + } } - AstVariableDefinition.prototype = Object.create(AstNode.prototype); - AstVariableDefinition.prototype.visit = function (visitor) { - visitor.visitVariableDefinition(this); - }; - function ExpressionBuilderVisitor() { - this.parts = []; + function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { + // A represents a gray component of a calibrated gray space. + // A <---> AG in the spec + var A = src[srcOffset] * scale; + var AG = Math.pow(A, cs.G); + + // Computes L as per spec. ( = cs.YW * AG ) + // Except if other than default BlackPoint values are used. + var L = cs.YW * AG; + // http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html, Ch 4. + // Convert values to rgb range [0, 255]. + var val = Math.max(295.8 * Math.pow(L, 0.333333333333333333) - 40.8, 0) | 0; + dest[destOffset] = val; + dest[destOffset + 1] = val; + dest[destOffset + 2] = val; } - ExpressionBuilderVisitor.prototype = { - visitArgument: function (arg) { - this.parts.push('Math.max(', arg.min, ', Math.min(', - arg.max, ', src[srcOffset + ', arg.index, ']))'); - }, - visitVariable: function (variable) { - this.parts.push('v', variable.index); - }, - visitLiteral: function (literal) { - this.parts.push(literal.number); + + CalGrayCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function CalGrayCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(this, src, srcOffset, dest, destOffset, 1); }, - visitBinaryOperation: function (operation) { - this.parts.push('('); - operation.arg1.visit(this); - this.parts.push(' ', operation.op, ' '); - operation.arg2.visit(this); - this.parts.push(')'); + getRgbBuffer: function CalGrayCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + + for (var i = 0; i < count; ++i) { + convertToRgb(this, src, srcOffset, dest, destOffset, scale); + srcOffset += 1; + destOffset += 3 + alpha01; + } }, - visitVariableDefinition: function (definition) { - this.parts.push('var '); - definition.variable.visit(this); - this.parts.push(' = '); - definition.arg.visit(this); - this.parts.push(';'); + getOutputLength: function CalGrayCS_getOutputLength(inputLength, alpha01) { + return inputLength * (3 + alpha01); }, - visitMin: function (max) { - this.parts.push('Math.min('); - max.arg.visit(this); - this.parts.push(', ', max.max, ')'); + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function CalGrayCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - toString: function () { - return this.parts.join(''); - } + usesZeroToOneRange: true }; + return CalGrayCS; +})(); - function buildAddOperation(num1, num2) { - if (num2.type === 'literal' && num2.number === 0) { - // optimization: second operand is 0 - return num1; +// +// CalRGBCS: Based on "PDF Reference, Sixth Ed", p.247 +// +var CalRGBCS = (function CalRGBCSClosure() { + + // See http://www.brucelindbloom.com/index.html?Eqn_ChromAdapt.html for these + // matrices. + var BRADFORD_SCALE_MATRIX = new Float32Array([ + 0.8951, 0.2664, -0.1614, + -0.7502, 1.7135, 0.0367, + 0.0389, -0.0685, 1.0296]); + + var BRADFORD_SCALE_INVERSE_MATRIX = new Float32Array([ + 0.9869929, -0.1470543, 0.1599627, + 0.4323053, 0.5183603, 0.0492912, + -0.0085287, 0.0400428, 0.9684867]); + + // See http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html. + var SRGB_D65_XYZ_TO_RGB_MATRIX = new Float32Array([ + 3.2404542, -1.5371385, -0.4985314, + -0.9692660, 1.8760108, 0.0415560, + 0.0556434, -0.2040259, 1.0572252]); + + var FLAT_WHITEPOINT_MATRIX = new Float32Array([1, 1, 1]); + + var tempNormalizeMatrix = new Float32Array(3); + var tempConvertMatrix1 = new Float32Array(3); + var tempConvertMatrix2 = new Float32Array(3); + + var DECODE_L_CONSTANT = Math.pow(((8 + 16) / 116), 3) / 8.0; + + function CalRGBCS(whitePoint, blackPoint, gamma, matrix) { + this.name = 'CalRGB'; + this.numComps = 3; + this.defaultColor = new Float32Array(3); + + if (!whitePoint) { + error('WhitePoint missing - required for color space CalRGB'); } - if (num1.type === 'literal' && num1.number === 0) { - // optimization: first operand is 0 - return num2; + blackPoint = blackPoint || new Float32Array(3); + gamma = gamma || new Float32Array([1, 1, 1]); + matrix = matrix || new Float32Array([1, 0, 0, 0, 1, 0, 0, 0, 1]); + + // Translate arguments to spec variables. + var XW = whitePoint[0]; + var YW = whitePoint[1]; + var ZW = whitePoint[2]; + this.whitePoint = whitePoint; + + var XB = blackPoint[0]; + var YB = blackPoint[1]; + var ZB = blackPoint[2]; + this.blackPoint = blackPoint; + + this.GR = gamma[0]; + this.GG = gamma[1]; + this.GB = gamma[2]; + + this.MXA = matrix[0]; + this.MYA = matrix[1]; + this.MZA = matrix[2]; + this.MXB = matrix[3]; + this.MYB = matrix[4]; + this.MZB = matrix[5]; + this.MXC = matrix[6]; + this.MYC = matrix[7]; + this.MZC = matrix[8]; + + // Validate variables as per spec. + if (XW < 0 || ZW < 0 || YW !== 1) { + error('Invalid WhitePoint components for ' + this.name + + ', no fallback available'); } - if (num2.type === 'literal' && num1.type === 'literal') { - // optimization: operands operand are literals - return new AstLiteral(num1.number + num2.number); + + if (XB < 0 || YB < 0 || ZB < 0) { + info('Invalid BlackPoint for ' + this.name + ' [' + XB + ', ' + YB + + ', ' + ZB + '], falling back to default'); + this.blackPoint = new Float32Array(3); } - return new AstBinaryOperation('+', num1, num2, - num1.min + num2.min, num1.max + num2.max); - } - function buildMulOperation(num1, num2) { - if (num2.type === 'literal') { - // optimization: second operands is a literal... - if (num2.number === 0) { - return new AstLiteral(0); // and it's 0 - } else if (num2.number === 1) { - return num1; // and it's 1 - } else if (num1.type === 'literal') { - // ... and first operands is a literal too - return new AstLiteral(num1.number * num2.number); - } + if (this.GR < 0 || this.GG < 0 || this.GB < 0) { + info('Invalid Gamma [' + this.GR + ', ' + this.GG + ', ' + this.GB + + '] for ' + this.name + ', falling back to default'); + this.GR = this.GG = this.GB = 1; } - if (num1.type === 'literal') { - // optimization: first operands is a literal... - if (num1.number === 0) { - return new AstLiteral(0); // and it's 0 - } else if (num1.number === 1) { - return num2; // and it's 1 - } + + if (this.MXA < 0 || this.MYA < 0 || this.MZA < 0 || + this.MXB < 0 || this.MYB < 0 || this.MZB < 0 || + this.MXC < 0 || this.MYC < 0 || this.MZC < 0) { + info('Invalid Matrix for ' + this.name + ' [' + + this.MXA + ', ' + this.MYA + ', ' + this.MZA + + this.MXB + ', ' + this.MYB + ', ' + this.MZB + + this.MXC + ', ' + this.MYC + ', ' + this.MZC + + '], falling back to default'); + this.MXA = this.MYB = this.MZC = 1; + this.MXB = this.MYA = this.MZA = this.MXC = this.MYC = this.MZB = 0; } - var min = Math.min(num1.min * num2.min, num1.min * num2.max, - num1.max * num2.min, num1.max * num2.max); - var max = Math.max(num1.min * num2.min, num1.min * num2.max, - num1.max * num2.min, num1.max * num2.max); - return new AstBinaryOperation('*', num1, num2, min, max); } - function buildSubOperation(num1, num2) { - if (num2.type === 'literal') { - // optimization: second operands is a literal... - if (num2.number === 0) { - return num1; // ... and it's 0 - } else if (num1.type === 'literal') { - // ... and first operands is a literal too - return new AstLiteral(num1.number - num2.number); - } + function matrixProduct(a, b, result) { + result[0] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; + result[1] = a[3] * b[0] + a[4] * b[1] + a[5] * b[2]; + result[2] = a[6] * b[0] + a[7] * b[1] + a[8] * b[2]; + } + + function convertToFlat(sourceWhitePoint, LMS, result) { + result[0] = LMS[0] * 1 / sourceWhitePoint[0]; + result[1] = LMS[1] * 1 / sourceWhitePoint[1]; + result[2] = LMS[2] * 1 / sourceWhitePoint[2]; + } + + function convertToD65(sourceWhitePoint, LMS, result) { + var D65X = 0.95047; + var D65Y = 1; + var D65Z = 1.08883; + + result[0] = LMS[0] * D65X / sourceWhitePoint[0]; + result[1] = LMS[1] * D65Y / sourceWhitePoint[1]; + result[2] = LMS[2] * D65Z / sourceWhitePoint[2]; + } + + function sRGBTransferFunction(color) { + // See http://en.wikipedia.org/wiki/SRGB. + if (color <= 0.0031308){ + return adjustToRange(0, 1, 12.92 * color); } - if (num2.type === 'binary' && num2.op === '-' && - num1.type === 'literal' && num1.number === 1 && - num2.arg1.type === 'literal' && num2.arg1.number === 1) { - // optimization for case: 1 - (1 - x) - return num2.arg2; + + return adjustToRange(0, 1, (1 + 0.055) * Math.pow(color, 1 / 2.4) - 0.055); + } + + function adjustToRange(min, max, value) { + return Math.max(min, Math.min(max, value)); + } + + function decodeL(L) { + if (L < 0) { + return -decodeL(-L); } - return new AstBinaryOperation('-', num1, num2, - num1.min - num2.max, num1.max - num2.min); + + if (L > 8.0) { + return Math.pow(((L + 16) / 116), 3); + } + + return L * DECODE_L_CONSTANT; + } + + function compensateBlackPoint(sourceBlackPoint, XYZ_Flat, result) { + + // In case the blackPoint is already the default blackPoint then there is + // no need to do compensation. + if (sourceBlackPoint[0] === 0 && + sourceBlackPoint[1] === 0 && + sourceBlackPoint[2] === 0) { + result[0] = XYZ_Flat[0]; + result[1] = XYZ_Flat[1]; + result[2] = XYZ_Flat[2]; + return; + } + + // For the blackPoint calculation details, please see + // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ + // AdobeBPC.pdf. + // The destination blackPoint is the default blackPoint [0, 0, 0]. + var zeroDecodeL = decodeL(0); + + var X_DST = zeroDecodeL; + var X_SRC = decodeL(sourceBlackPoint[0]); + + var Y_DST = zeroDecodeL; + var Y_SRC = decodeL(sourceBlackPoint[1]); + + var Z_DST = zeroDecodeL; + var Z_SRC = decodeL(sourceBlackPoint[2]); + + var X_Scale = (1 - X_DST) / (1 - X_SRC); + var X_Offset = 1 - X_Scale; + + var Y_Scale = (1 - Y_DST) / (1 - Y_SRC); + var Y_Offset = 1 - Y_Scale; + + var Z_Scale = (1 - Z_DST) / (1 - Z_SRC); + var Z_Offset = 1 - Z_Scale; + + result[0] = XYZ_Flat[0] * X_Scale + X_Offset; + result[1] = XYZ_Flat[1] * Y_Scale + Y_Offset; + result[2] = XYZ_Flat[2] * Z_Scale + Z_Offset; } - function buildMinOperation(num1, max) { - if (num1.min >= max) { - // optimization: num1 min value is not less than required max - return new AstLiteral(max); // just returning max - } else if (num1.max <= max) { - // optimization: num1 max value is not greater than required max - return num1; // just returning an argument + function normalizeWhitePointToFlat(sourceWhitePoint, XYZ_In, result) { + + // In case the whitePoint is already flat then there is no need to do + // normalization. + if (sourceWhitePoint[0] === 1 && sourceWhitePoint[2] === 1) { + result[0] = XYZ_In[0]; + result[1] = XYZ_In[1]; + result[2] = XYZ_In[2]; + return; } - return new AstMin(num1, max); + + var LMS = result; + matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + + var LMS_Flat = tempNormalizeMatrix; + convertToFlat(sourceWhitePoint, LMS, LMS_Flat); + + matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_Flat, result); } - function PostScriptCompiler() {} - PostScriptCompiler.prototype = { - compile: function PostScriptCompiler_compile(code, domain, range) { - var stack = []; - var i, ii; - var instructions = []; - var inputSize = domain.length >> 1, outputSize = range.length >> 1; - var lastRegister = 0; - var n, j; - var num1, num2, ast1, ast2, tmpVar, item; - for (i = 0; i < inputSize; i++) { - stack.push(new AstArgument(i, domain[i * 2], domain[i * 2 + 1])); - } + function normalizeWhitePointToD65(sourceWhitePoint, XYZ_In, result) { - for (i = 0, ii = code.length; i < ii; i++) { - item = code[i]; - if (typeof item === 'number') { - stack.push(new AstLiteral(item)); - continue; - } + var LMS = result; + matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); - switch (item) { - case 'add': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildAddOperation(num1, num2)); - break; - case 'cvr': - if (stack.length < 1) { - return null; - } - break; - case 'mul': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildMulOperation(num1, num2)); - break; - case 'sub': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - stack.push(buildSubOperation(num1, num2)); - break; - case 'exch': - if (stack.length < 2) { - return null; - } - ast1 = stack.pop(); ast2 = stack.pop(); - stack.push(ast1, ast2); - break; - case 'pop': - if (stack.length < 1) { - return null; - } - stack.pop(); - break; - case 'index': - if (stack.length < 1) { - return null; - } - num1 = stack.pop(); - if (num1.type !== 'literal') { - return null; - } - n = num1.number; - if (n < 0 || (n|0) !== n || stack.length < n) { - return null; - } - ast1 = stack[stack.length - n - 1]; - if (ast1.type === 'literal' || ast1.type === 'var') { - stack.push(ast1); - break; - } - tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); - stack[stack.length - n - 1] = tmpVar; - stack.push(tmpVar); - instructions.push(new AstVariableDefinition(tmpVar, ast1)); - break; - case 'dup': - if (stack.length < 1) { - return null; - } - if (typeof code[i + 1] === 'number' && code[i + 2] === 'gt' && - code[i + 3] === i + 7 && code[i + 4] === 'jz' && - code[i + 5] === 'pop' && code[i + 6] === code[i + 1]) { - // special case of the commands sequence for the min operation - num1 = stack.pop(); - stack.push(buildMinOperation(num1, code[i + 1])); - i += 6; - break; - } - ast1 = stack[stack.length - 1]; - if (ast1.type === 'literal' || ast1.type === 'var') { - // we don't have to save into intermediate variable a literal or - // variable. - stack.push(ast1); - break; - } - tmpVar = new AstVariable(lastRegister++, ast1.min, ast1.max); - stack[stack.length - 1] = tmpVar; - stack.push(tmpVar); - instructions.push(new AstVariableDefinition(tmpVar, ast1)); - break; - case 'roll': - if (stack.length < 2) { - return null; - } - num2 = stack.pop(); - num1 = stack.pop(); - if (num2.type !== 'literal' || num1.type !== 'literal') { - // both roll operands must be numbers - return null; - } - j = num2.number; - n = num1.number; - if (n <= 0 || (n|0) !== n || (j|0) !== j || stack.length < n) { - // ... and integers - return null; - } - j = ((j % n) + n) % n; - if (j === 0) { - break; // just skipping -- there are nothing to rotate - } - Array.prototype.push.apply(stack, - stack.splice(stack.length - n, n - j)); - break; - default: - return null; // unsupported operator - } - } + var LMS_D65 = tempNormalizeMatrix; + convertToD65(sourceWhitePoint, LMS, LMS_D65); - if (stack.length !== outputSize) { - return null; - } + matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_D65, result); + } - var result = []; - instructions.forEach(function (instruction) { - var statementBuilder = new ExpressionBuilderVisitor(); - instruction.visit(statementBuilder); - result.push(statementBuilder.toString()); - }); - stack.forEach(function (expr, i) { - var statementBuilder = new ExpressionBuilderVisitor(); - expr.visit(statementBuilder); - var min = range[i * 2], max = range[i * 2 + 1]; - var out = [statementBuilder.toString()]; - if (min > expr.min) { - out.unshift('Math.max(', min, ', '); - out.push(')'); - } - if (max < expr.max) { - out.unshift('Math.min(', max, ', '); - out.push(')'); - } - out.unshift('dest[destOffset + ', i, '] = '); - out.push(';'); - result.push(out.join('')); - }); - return result.join('\n'); - } - }; + function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { + // A, B and C represent a red, green and blue components of a calibrated + // rgb space. + var A = adjustToRange(0, 1, src[srcOffset] * scale); + var B = adjustToRange(0, 1, src[srcOffset + 1] * scale); + var C = adjustToRange(0, 1, src[srcOffset + 2] * scale); - return PostScriptCompiler; -})(); + // A <---> AGR in the spec + // B <---> BGG in the spec + // C <---> CGB in the spec + var AGR = Math.pow(A, cs.GR); + var BGG = Math.pow(B, cs.GG); + var CGB = Math.pow(C, cs.GB); -exports.isPDFFunction = isPDFFunction; -exports.PDFFunction = PDFFunction; -exports.PostScriptEvaluator = PostScriptEvaluator; -exports.PostScriptCompiler = PostScriptCompiler; -})); + // Computes intermediate variables L, M, N as per spec. + // To decode X, Y, Z values map L, M, N directly to them. + var X = cs.MXA * AGR + cs.MXB * BGG + cs.MXC * CGB; + var Y = cs.MYA * AGR + cs.MYB * BGG + cs.MYC * CGB; + var Z = cs.MZA * AGR + cs.MZB * BGG + cs.MZC * CGB; + // The following calculations are based on this document: + // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ + // AdobeBPC.pdf. + var XYZ = tempConvertMatrix1; + XYZ[0] = X; + XYZ[1] = Y; + XYZ[2] = Z; + var XYZ_Flat = tempConvertMatrix2; -(function (root, factory) { - { - factory((root.pdfjsCoreColorSpace = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreFunction, root.pdfjsCoreStream); - } -}(this, function (exports, sharedUtil, corePrimitives, coreFunction, - coreStream) { + normalizeWhitePointToFlat(cs.whitePoint, XYZ, XYZ_Flat); -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var isString = sharedUtil.isString; -var shadow = sharedUtil.shadow; -var warn = sharedUtil.warn; -var isDict = corePrimitives.isDict; -var isName = corePrimitives.isName; -var isStream = corePrimitives.isStream; -var PDFFunction = coreFunction.PDFFunction; + var XYZ_Black = tempConvertMatrix1; + compensateBlackPoint(cs.blackPoint, XYZ_Flat, XYZ_Black); -var coreImage; // see _setCoreImage below -var PDFImage; // = coreImage.PDFImage; + var XYZ_D65 = tempConvertMatrix2; + normalizeWhitePointToD65(FLAT_WHITEPOINT_MATRIX, XYZ_Black, XYZ_D65); -var ColorSpace = (function ColorSpaceClosure() { - // Constructor should define this.numComps, this.defaultColor, this.name - function ColorSpace() { - error('should not call ColorSpace constructor'); + var SRGB = tempConvertMatrix1; + matrixProduct(SRGB_D65_XYZ_TO_RGB_MATRIX, XYZ_D65, SRGB); + + var sR = sRGBTransferFunction(SRGB[0]); + var sG = sRGBTransferFunction(SRGB[1]); + var sB = sRGBTransferFunction(SRGB[2]); + + // Convert the values to rgb range [0, 255]. + dest[destOffset] = Math.round(sR * 255); + dest[destOffset + 1] = Math.round(sG * 255); + dest[destOffset + 2] = Math.round(sB * 255); } - ColorSpace.prototype = { - /** - * Converts the color value to the RGB color. The color components are - * located in the src array starting from the srcOffset. Returns the array - * of the rgb components, each value ranging from [0,255]. - */ - getRgb: function ColorSpace_getRgb(src, srcOffset) { + CalRGBCS.prototype = { + getRgb: function CalRGBCS_getRgb(src, srcOffset) { var rgb = new Uint8Array(3); this.getRgbItem(src, srcOffset, rgb, 0); return rgb; }, - /** - * Converts the color value to the RGB color, similar to the getRgb method. - * The result placed into the dest array starting from the destOffset. - */ - getRgbItem: function ColorSpace_getRgbItem(src, srcOffset, - dest, destOffset) { - error('Should not call ColorSpace.getRgbItem'); + getRgbItem: function CalRGBCS_getRgbItem(src, srcOffset, + dest, destOffset) { + convertToRgb(this, src, srcOffset, dest, destOffset, 1); }, - /** - * Converts the specified number of the color values to the RGB colors. - * The colors are located in the src array starting from the srcOffset. - * The result is placed into the dest array starting from the destOffset. - * The src array items shall be in [0,2^bits) range, the dest array items - * will be in [0,255] range. alpha01 indicates how many alpha components - * there are in the dest array; it will be either 0 (RGB array) or 1 (RGBA - * array). - */ - getRgbBuffer: function ColorSpace_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - error('Should not call ColorSpace.getRgbBuffer'); + getRgbBuffer: function CalRGBCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var scale = 1 / ((1 << bits) - 1); + + for (var i = 0; i < count; ++i) { + convertToRgb(this, src, srcOffset, dest, destOffset, scale); + srcOffset += 3; + destOffset += 3 + alpha01; + } }, - /** - * Determines the number of bytes required to store the result of the - * conversion done by the getRgbBuffer method. As in getRgbBuffer, - * |alpha01| is either 0 (RGB output) or 1 (RGBA output). - */ - getOutputLength: function ColorSpace_getOutputLength(inputLength, - alpha01) { - error('Should not call ColorSpace.getOutputLength'); + getOutputLength: function CalRGBCS_getOutputLength(inputLength, alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; }, - /** - * Returns true if source data will be equal the result/output data. - */ - isPassthrough: function ColorSpace_isPassthrough(bits) { - return false; + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function CalRGBCS_isDefaultDecode(decodeMap) { + return ColorSpace.isDefaultDecode(decodeMap, this.numComps); }, - /** - * Fills in the RGB colors in the destination buffer. alpha01 indicates - * how many alpha components there are in the dest array; it will be either - * 0 (RGB array) or 1 (RGBA array). - */ - fillRgb: function ColorSpace_fillRgb(dest, originalWidth, - originalHeight, width, height, - actualHeight, bpc, comps, alpha01) { - var count = originalWidth * originalHeight; - var rgbBuf = null; - var numComponentColors = 1 << bpc; - var needsResizing = originalHeight !== height || originalWidth !== width; - var i, ii; + usesZeroToOneRange: true + }; + return CalRGBCS; +})(); - if (this.isPassthrough(bpc)) { - rgbBuf = comps; - } else if (this.numComps === 1 && count > numComponentColors && - this.name !== 'DeviceGray' && this.name !== 'DeviceRGB') { - // Optimization: create a color map when there is just one component and - // we are converting more colors than the size of the color map. We - // don't build the map if the colorspace is gray or rgb since those - // methods are faster than building a map. This mainly offers big speed - // ups for indexed and alternate colorspaces. - // - // TODO it may be worth while to cache the color map. While running - // testing I never hit a cache so I will leave that out for now (perhaps - // we are reparsing colorspaces too much?). - var allColors = bpc <= 8 ? new Uint8Array(numComponentColors) : - new Uint16Array(numComponentColors); - var key; - for (i = 0; i < numComponentColors; i++) { - allColors[i] = i; - } - var colorMap = new Uint8Array(numComponentColors * 3); - this.getRgbBuffer(allColors, 0, numComponentColors, colorMap, 0, bpc, - /* alpha01 = */ 0); +// +// LabCS: Based on "PDF Reference, Sixth Ed", p.250 +// +var LabCS = (function LabCSClosure() { + function LabCS(whitePoint, blackPoint, range) { + this.name = 'Lab'; + this.numComps = 3; + this.defaultColor = new Float32Array([0, 0, 0]); - var destPos, rgbPos; - if (!needsResizing) { - // Fill in the RGB values directly into |dest|. - destPos = 0; - for (i = 0; i < count; ++i) { - key = comps[i] * 3; - dest[destPos++] = colorMap[key]; - dest[destPos++] = colorMap[key + 1]; - dest[destPos++] = colorMap[key + 2]; - destPos += alpha01; - } - } else { - rgbBuf = new Uint8Array(count * 3); - rgbPos = 0; - for (i = 0; i < count; ++i) { - key = comps[i] * 3; - rgbBuf[rgbPos++] = colorMap[key]; - rgbBuf[rgbPos++] = colorMap[key + 1]; - rgbBuf[rgbPos++] = colorMap[key + 2]; - } - } - } else { - if (!needsResizing) { - // Fill in the RGB values directly into |dest|. - this.getRgbBuffer(comps, 0, width * actualHeight, dest, 0, bpc, - alpha01); - } else { - rgbBuf = new Uint8Array(count * 3); - this.getRgbBuffer(comps, 0, count, rgbBuf, 0, bpc, - /* alpha01 = */ 0); - } - } + if (!whitePoint) { + error('WhitePoint missing - required for color space Lab'); + } + blackPoint = blackPoint || [0, 0, 0]; + range = range || [-100, 100, -100, 100]; + + // Translate args to spec variables + this.XW = whitePoint[0]; + this.YW = whitePoint[1]; + this.ZW = whitePoint[2]; + this.amin = range[0]; + this.amax = range[1]; + this.bmin = range[2]; + this.bmax = range[3]; + + // These are here just for completeness - the spec doesn't offer any + // formulas that use BlackPoint in Lab + this.XB = blackPoint[0]; + this.YB = blackPoint[1]; + this.ZB = blackPoint[2]; + + // Validate vars as per spec + if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { + error('Invalid WhitePoint components, no fallback available'); + } + + if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { + info('Invalid BlackPoint, falling back to default'); + this.XB = this.YB = this.ZB = 0; + } + + if (this.amin > this.amax || this.bmin > this.bmax) { + info('Invalid Range, falling back to defaults'); + this.amin = -100; + this.amax = 100; + this.bmin = -100; + this.bmax = 100; + } + } + + // Function g(x) from spec + function fn_g(x) { + if (x >= 6 / 29) { + return x * x * x; + } else { + return (108 / 841) * (x - 4 / 29); + } + } + + function decode(value, high1, low2, high2) { + return low2 + (value) * (high2 - low2) / (high1); + } + + // If decoding is needed maxVal should be 2^bits per component - 1. + function convertToRgb(cs, src, srcOffset, maxVal, dest, destOffset) { + // XXX: Lab input is in the range of [0, 100], [amin, amax], [bmin, bmax] + // not the usual [0, 1]. If a command like setFillColor is used the src + // values will already be within the correct range. However, if we are + // converting an image we have to map the values to the correct range given + // above. + // Ls,as,bs <---> L*,a*,b* in the spec + var Ls = src[srcOffset]; + var as = src[srcOffset + 1]; + var bs = src[srcOffset + 2]; + if (maxVal !== false) { + Ls = decode(Ls, maxVal, 0, 100); + as = decode(as, maxVal, cs.amin, cs.amax); + bs = decode(bs, maxVal, cs.bmin, cs.bmax); + } + + // Adjust limits of 'as' and 'bs' + as = as > cs.amax ? cs.amax : as < cs.amin ? cs.amin : as; + bs = bs > cs.bmax ? cs.bmax : bs < cs.bmin ? cs.bmin : bs; + + // Computes intermediate variables X,Y,Z as per spec + var M = (Ls + 16) / 116; + var L = M + (as / 500); + var N = M - (bs / 200); + + var X = cs.XW * fn_g(L); + var Y = cs.YW * fn_g(M); + var Z = cs.ZW * fn_g(N); + + var r, g, b; + // Using different conversions for D50 and D65 white points, + // per http://www.color.org/srgb.pdf + if (cs.ZW < 1) { + // Assuming D50 (X=0.9642, Y=1.00, Z=0.8249) + r = X * 3.1339 + Y * -1.6170 + Z * -0.4906; + g = X * -0.9785 + Y * 1.9160 + Z * 0.0333; + b = X * 0.0720 + Y * -0.2290 + Z * 1.4057; + } else { + // Assuming D65 (X=0.9505, Y=1.00, Z=1.0888) + r = X * 3.2406 + Y * -1.5372 + Z * -0.4986; + g = X * -0.9689 + Y * 1.8758 + Z * 0.0415; + b = X * 0.0557 + Y * -0.2040 + Z * 1.0570; + } + // clamp color values to [0,1] range then convert to [0,255] range. + dest[destOffset] = r <= 0 ? 0 : r >= 1 ? 255 : Math.sqrt(r) * 255 | 0; + dest[destOffset + 1] = g <= 0 ? 0 : g >= 1 ? 255 : Math.sqrt(g) * 255 | 0; + dest[destOffset + 2] = b <= 0 ? 0 : b >= 1 ? 255 : Math.sqrt(b) * 255 | 0; + } - if (rgbBuf) { - if (needsResizing) { - PDFImage.resize(rgbBuf, bpc, 3, originalWidth, originalHeight, width, - height, dest, alpha01); - } else { - rgbPos = 0; - destPos = 0; - for (i = 0, ii = width * actualHeight; i < ii; i++) { - dest[destPos++] = rgbBuf[rgbPos++]; - dest[destPos++] = rgbBuf[rgbPos++]; - dest[destPos++] = rgbBuf[rgbPos++]; - destPos += alpha01; - } - } + LabCS.prototype = { + getRgb: ColorSpace.prototype.getRgb, + getRgbItem: function LabCS_getRgbItem(src, srcOffset, dest, destOffset) { + convertToRgb(this, src, srcOffset, false, dest, destOffset); + }, + getRgbBuffer: function LabCS_getRgbBuffer(src, srcOffset, count, + dest, destOffset, bits, + alpha01) { + var maxVal = (1 << bits) - 1; + for (var i = 0; i < count; i++) { + convertToRgb(this, src, srcOffset, maxVal, dest, destOffset); + srcOffset += 3; + destOffset += 3 + alpha01; } }, - /** - * True if the colorspace has components in the default range of [0, 1]. - * This should be true for all colorspaces except for lab color spaces - * which are [0,100], [-128, 127], [-128, 127]. - */ - usesZeroToOneRange: true + getOutputLength: function LabCS_getOutputLength(inputLength, alpha01) { + return (inputLength * (3 + alpha01) / 3) | 0; + }, + isPassthrough: ColorSpace.prototype.isPassthrough, + fillRgb: ColorSpace.prototype.fillRgb, + isDefaultDecode: function LabCS_isDefaultDecode(decodeMap) { + // XXX: Decoding is handled with the lab conversion because of the strange + // ranges that are used. + return true; + }, + usesZeroToOneRange: false }; + return LabCS; +})(); - ColorSpace.parse = function ColorSpace_parse(cs, xref, res) { - var IR = ColorSpace.parseToIR(cs, xref, res); - if (IR instanceof AlternateCS) { - return IR; - } - return ColorSpace.fromIR(IR); - }; +// TODO refactor to remove dependency on image.js +function _setCoreImage(coreImage_) { + coreImage = coreImage_; + PDFImage = coreImage_.PDFImage; +} +exports._setCoreImage = _setCoreImage; - ColorSpace.fromIR = function ColorSpace_fromIR(IR) { - var name = isArray(IR) ? IR[0] : IR; - var whitePoint, blackPoint, gamma; +exports.ColorSpace = ColorSpace; - switch (name) { - case 'DeviceGrayCS': - return this.singletons.gray; - case 'DeviceRgbCS': - return this.singletons.rgb; - case 'DeviceCmykCS': - return this.singletons.cmyk; - case 'CalGrayCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - gamma = IR[3]; - return new CalGrayCS(whitePoint, blackPoint, gamma); - case 'CalRGBCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - gamma = IR[3]; - var matrix = IR[4]; - return new CalRGBCS(whitePoint, blackPoint, gamma, matrix); - case 'PatternCS': - var basePatternCS = IR[1]; - if (basePatternCS) { - basePatternCS = ColorSpace.fromIR(basePatternCS); - } - return new PatternCS(basePatternCS); - case 'IndexedCS': - var baseIndexedCS = IR[1]; - var hiVal = IR[2]; - var lookup = IR[3]; - return new IndexedCS(ColorSpace.fromIR(baseIndexedCS), hiVal, lookup); - case 'AlternateCS': - var numComps = IR[1]; - var alt = IR[2]; - var tintFnIR = IR[3]; +// TODO refactor to remove dependency on colorspace.js +coreStream._setCoreColorSpace(exports); +})); - return new AlternateCS(numComps, ColorSpace.fromIR(alt), - PDFFunction.fromIR(tintFnIR)); - case 'LabCS': - whitePoint = IR[1]; - blackPoint = IR[2]; - var range = IR[3]; - return new LabCS(whitePoint, blackPoint, range); - default: - error('Unknown name ' + name); + +(function (root, factory) { + { + factory((root.pdfjsCoreImage = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreColorSpace, root.pdfjsCoreStream, + root.pdfjsCoreJpx); + } +}(this, function (exports, sharedUtil, corePrimitives, coreColorSpace, + coreStream, coreJpx) { + +var ImageKind = sharedUtil.ImageKind; +var assert = sharedUtil.assert; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var warn = sharedUtil.warn; +var Name = corePrimitives.Name; +var isStream = corePrimitives.isStream; +var ColorSpace = coreColorSpace.ColorSpace; +var DecodeStream = coreStream.DecodeStream; +var Stream = coreStream.Stream; +var JpegStream = coreStream.JpegStream; +var JpxImage = coreJpx.JpxImage; + +var PDFImage = (function PDFImageClosure() { + /** + * Decode the image in the main thread if it supported. Resovles the promise + * when the image data is ready. + */ + function handleImageData(handler, xref, res, image) { + if (image instanceof JpegStream && image.isNativelyDecodable(xref, res)) { + // For natively supported jpegs send them to the main thread for decoding. + var dict = image.dict; + var colorSpace = dict.get('ColorSpace', 'CS'); + colorSpace = ColorSpace.parse(colorSpace, xref, res); + var numComps = colorSpace.numComps; + var decodePromise = handler.sendWithPromise('JpegDecode', + [image.getIR(), numComps]); + return decodePromise.then(function (message) { + var data = message.data; + return new Stream(data, 0, data.length, image.dict); + }); + } else { + return Promise.resolve(image); } - return null; - }; + } - ColorSpace.parseToIR = function ColorSpace_parseToIR(cs, xref, res) { - if (isName(cs)) { - var colorSpaces = res.get('ColorSpace'); - if (isDict(colorSpaces)) { - var refcs = colorSpaces.get(cs.name); - if (refcs) { - cs = refcs; - } + /** + * Decode and clamp a value. The formula is different from the spec because we + * don't decode to float range [0,1], we decode it in the [0,max] range. + */ + function decodeAndClamp(value, addend, coefficient, max) { + value = addend + value * coefficient; + // Clamp the value to the range + return (value < 0 ? 0 : (value > max ? max : value)); + } + + function PDFImage(xref, res, image, inline, smask, mask, isMask) { + this.image = image; + var dict = image.dict; + if (dict.has('Filter')) { + var filter = dict.get('Filter').name; + if (filter === 'JPXDecode') { + var jpxImage = new JpxImage(); + jpxImage.parseImageProperties(image.stream); + image.stream.reset(); + image.bitsPerComponent = jpxImage.bitsPerComponent; + image.numComps = jpxImage.componentsCount; + } else if (filter === 'JBIG2Decode') { + image.bitsPerComponent = 1; + image.numComps = 1; } } + // TODO cache rendered images? - cs = xref.fetchIfRef(cs); - var mode; + this.width = dict.get('Width', 'W'); + this.height = dict.get('Height', 'H'); - if (isName(cs)) { - mode = cs.name; - this.mode = mode; + if (this.width < 1 || this.height < 1) { + error('Invalid image width: ' + this.width + ' or height: ' + + this.height); + } - switch (mode) { - case 'DeviceGray': - case 'G': - return 'DeviceGrayCS'; - case 'DeviceRGB': - case 'RGB': - return 'DeviceRgbCS'; - case 'DeviceCMYK': - case 'CMYK': - return 'DeviceCmykCS'; - case 'Pattern': - return ['PatternCS', null]; - default: - error('unrecognized colorspace ' + mode); - } - } else if (isArray(cs)) { - mode = xref.fetchIfRef(cs[0]).name; - this.mode = mode; - var numComps, params, alt, whitePoint, blackPoint, gamma; + this.interpolate = dict.get('Interpolate', 'I') || false; + this.imageMask = dict.get('ImageMask', 'IM') || false; + this.matte = dict.get('Matte') || false; - switch (mode) { - case 'DeviceGray': - case 'G': - return 'DeviceGrayCS'; - case 'DeviceRGB': - case 'RGB': - return 'DeviceRgbCS'; - case 'DeviceCMYK': - case 'CMYK': - return 'DeviceCmykCS'; - case 'CalGray': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - gamma = params.get('Gamma'); - return ['CalGrayCS', whitePoint, blackPoint, gamma]; - case 'CalRGB': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - gamma = params.get('Gamma'); - var matrix = params.get('Matrix'); - return ['CalRGBCS', whitePoint, blackPoint, gamma, matrix]; - case 'ICCBased': - var stream = xref.fetchIfRef(cs[1]); - var dict = stream.dict; - numComps = dict.get('N'); - alt = dict.get('Alternate'); - if (alt) { - var altIR = ColorSpace.parseToIR(alt, xref, res); - // Parse the /Alternate CS to ensure that the number of components - // are correct, and also (indirectly) that it is not a PatternCS. - var altCS = ColorSpace.fromIR(altIR); - if (altCS.numComps === numComps) { - return altIR; - } - warn('ICCBased color space: Ignoring incorrect /Alternate entry.'); - } - if (numComps === 1) { - return 'DeviceGrayCS'; - } else if (numComps === 3) { - return 'DeviceRgbCS'; - } else if (numComps === 4) { - return 'DeviceCmykCS'; - } - break; - case 'Pattern': - var basePatternCS = cs[1] || null; - if (basePatternCS) { - basePatternCS = ColorSpace.parseToIR(basePatternCS, xref, res); - } - return ['PatternCS', basePatternCS]; - case 'Indexed': - case 'I': - var baseIndexedCS = ColorSpace.parseToIR(cs[1], xref, res); - var hiVal = xref.fetchIfRef(cs[2]) + 1; - var lookup = xref.fetchIfRef(cs[3]); - if (isStream(lookup)) { - lookup = lookup.getBytes(); - } - return ['IndexedCS', baseIndexedCS, hiVal, lookup]; - case 'Separation': - case 'DeviceN': - var name = xref.fetchIfRef(cs[1]); - numComps = 1; - if (isName(name)) { - numComps = 1; - } else if (isArray(name)) { - numComps = name.length; - } - alt = ColorSpace.parseToIR(cs[2], xref, res); - var tintFnIR = PDFFunction.getIR(xref, xref.fetchIfRef(cs[3])); - return ['AlternateCS', numComps, alt, tintFnIR]; - case 'Lab': - params = xref.fetchIfRef(cs[1]); - whitePoint = params.get('WhitePoint'); - blackPoint = params.get('BlackPoint'); - var range = params.get('Range'); - return ['LabCS', whitePoint, blackPoint, range]; - default: - error('unimplemented color space object "' + mode + '"'); + var bitsPerComponent = image.bitsPerComponent; + if (!bitsPerComponent) { + bitsPerComponent = dict.get('BitsPerComponent', 'BPC'); + if (!bitsPerComponent) { + if (this.imageMask) { + bitsPerComponent = 1; + } else { + error('Bits per component missing in image: ' + this.imageMask); + } } - } else { - error('unrecognized color space object: "' + cs + '"'); } - return null; - }; - /** - * Checks if a decode map matches the default decode map for a color space. - * This handles the general decode maps where there are two values per - * component. e.g. [0, 1, 0, 1, 0, 1] for a RGB color. - * This does not handle Lab, Indexed, or Pattern decode maps since they are - * slightly different. - * @param {Array} decode Decode map (usually from an image). - * @param {Number} n Number of components the color space has. - */ - ColorSpace.isDefaultDecode = function ColorSpace_isDefaultDecode(decode, n) { - if (!isArray(decode)) { - return true; + this.bpc = bitsPerComponent; + + if (!this.imageMask) { + var colorSpace = dict.get('ColorSpace', 'CS'); + if (!colorSpace) { + info('JPX images (which do not require color spaces)'); + switch (image.numComps) { + case 1: + colorSpace = Name.get('DeviceGray'); + break; + case 3: + colorSpace = Name.get('DeviceRGB'); + break; + case 4: + colorSpace = Name.get('DeviceCMYK'); + break; + default: + error('JPX images with ' + this.numComps + + ' color components not supported.'); + } + } + this.colorSpace = ColorSpace.parse(colorSpace, xref, res); + this.numComps = this.colorSpace.numComps; } - if (n * 2 !== decode.length) { - warn('The decode map is not the correct length'); - return true; + this.decode = dict.get('Decode', 'D'); + this.needsDecode = false; + if (this.decode && + ((this.colorSpace && !this.colorSpace.isDefaultDecode(this.decode)) || + (isMask && !ColorSpace.isDefaultDecode(this.decode, 1)))) { + this.needsDecode = true; + // Do some preprocessing to avoid more math. + var max = (1 << bitsPerComponent) - 1; + this.decodeCoefficients = []; + this.decodeAddends = []; + for (var i = 0, j = 0; i < this.decode.length; i += 2, ++j) { + var dmin = this.decode[i]; + var dmax = this.decode[i + 1]; + this.decodeCoefficients[j] = dmax - dmin; + this.decodeAddends[j] = max * dmin; + } } - for (var i = 0, ii = decode.length; i < ii; i += 2) { - if (decode[i] !== 0 || decode[i + 1] !== 1) { - return false; + + if (smask) { + this.smask = new PDFImage(xref, res, smask, false); + } else if (mask) { + if (isStream(mask)) { + var maskDict = mask.dict, imageMask = maskDict.get('ImageMask', 'IM'); + if (!imageMask) { + warn('Ignoring /Mask in image without /ImageMask.'); + } else { + this.mask = new PDFImage(xref, res, mask, false, null, null, true); + } + } else { + // Color key mask (just an array). + this.mask = mask; } } - return true; - }; + } + /** + * Handles processing of image data and returns the Promise that is resolved + * with a PDFImage when the image is ready to be used. + */ + PDFImage.buildImage = function PDFImage_buildImage(handler, xref, + res, image, inline) { + var imagePromise = handleImageData(handler, xref, res, image); + var smaskPromise; + var maskPromise; - ColorSpace.singletons = { - get gray() { - return shadow(this, 'gray', new DeviceGrayCS()); - }, - get rgb() { - return shadow(this, 'rgb', new DeviceRgbCS()); - }, - get cmyk() { - return shadow(this, 'cmyk', new DeviceCmykCS()); + var smask = image.dict.get('SMask'); + var mask = image.dict.get('Mask'); + + if (smask) { + smaskPromise = handleImageData(handler, xref, res, smask); + maskPromise = Promise.resolve(null); + } else { + smaskPromise = Promise.resolve(null); + if (mask) { + if (isStream(mask)) { + maskPromise = handleImageData(handler, xref, res, mask); + } else if (isArray(mask)) { + maskPromise = Promise.resolve(mask); + } else { + warn('Unsupported mask format.'); + maskPromise = Promise.resolve(null); + } + } else { + maskPromise = Promise.resolve(null); + } } + return Promise.all([imagePromise, smaskPromise, maskPromise]).then( + function(results) { + var imageData = results[0]; + var smaskData = results[1]; + var maskData = results[2]; + return new PDFImage(xref, res, imageData, inline, smaskData, maskData); + }); }; - return ColorSpace; -})(); + /** + * Resize an image using the nearest neighbor algorithm. Currently only + * supports one and three component images. + * @param {TypedArray} pixels The original image with one component. + * @param {Number} bpc Number of bits per component. + * @param {Number} components Number of color components, 1 or 3 is supported. + * @param {Number} w1 Original width. + * @param {Number} h1 Original height. + * @param {Number} w2 New width. + * @param {Number} h2 New height. + * @param {TypedArray} dest (Optional) The destination buffer. + * @param {Number} alpha01 (Optional) Size reserved for the alpha channel. + * @return {TypedArray} Resized image data. + */ + PDFImage.resize = function PDFImage_resize(pixels, bpc, components, + w1, h1, w2, h2, dest, alpha01) { -/** - * Alternate color space handles both Separation and DeviceN color spaces. A - * Separation color space is actually just a DeviceN with one color component. - * Both color spaces use a tinting function to convert colors to a base color - * space. - */ -var AlternateCS = (function AlternateCSClosure() { - function AlternateCS(numComps, base, tintFn) { - this.name = 'Alternate'; - this.numComps = numComps; - this.defaultColor = new Float32Array(numComps); - for (var i = 0; i < numComps; ++i) { - this.defaultColor[i] = 1; + if (components !== 1 && components !== 3) { + error('Unsupported component count for resizing.'); } - this.base = base; - this.tintFn = tintFn; - this.tmpBuf = new Float32Array(base.numComps); - } - AlternateCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function AlternateCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var tmpBuf = this.tmpBuf; - this.tintFn(src, srcOffset, tmpBuf, 0); - this.base.getRgbItem(tmpBuf, 0, dest, destOffset); - }, - getRgbBuffer: function AlternateCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var tintFn = this.tintFn; - var base = this.base; - var scale = 1 / ((1 << bits) - 1); - var baseNumComps = base.numComps; - var usesZeroToOneRange = base.usesZeroToOneRange; - var isPassthrough = (base.isPassthrough(8) || !usesZeroToOneRange) && - alpha01 === 0; - var pos = isPassthrough ? destOffset : 0; - var baseBuf = isPassthrough ? dest : new Uint8Array(baseNumComps * count); - var numComps = this.numComps; + var length = w2 * h2 * components; + var temp = dest ? dest : (bpc <= 8 ? new Uint8Array(length) : + (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); + var xRatio = w1 / w2; + var yRatio = h1 / h2; + var i, j, py, newIndex = 0, oldIndex; + var xScaled = new Uint16Array(w2); + var w1Scanline = w1 * components; + if (alpha01 !== 1) { + alpha01 = 0; + } - var scaled = new Float32Array(numComps); - var tinted = new Float32Array(baseNumComps); - var i, j; - if (usesZeroToOneRange) { - for (i = 0; i < count; i++) { - for (j = 0; j < numComps; j++) { - scaled[j] = src[srcOffset++] * scale; - } - tintFn(scaled, 0, tinted, 0); - for (j = 0; j < baseNumComps; j++) { - baseBuf[pos++] = tinted[j] * 255; - } - } - } else { - for (i = 0; i < count; i++) { - for (j = 0; j < numComps; j++) { - scaled[j] = src[srcOffset++] * scale; - } - tintFn(scaled, 0, tinted, 0); - base.getRgbItem(tinted, 0, baseBuf, pos); - pos += baseNumComps; + for (j = 0; j < w2; j++) { + xScaled[j] = Math.floor(j * xRatio) * components; + } + + if (components === 1) { + for (i = 0; i < h2; i++) { + py = Math.floor(i * yRatio) * w1Scanline; + for (j = 0; j < w2; j++) { + oldIndex = py + xScaled[j]; + temp[newIndex++] = pixels[oldIndex]; } } - if (!isPassthrough) { - base.getRgbBuffer(baseBuf, 0, count, dest, destOffset, 8, alpha01); + } else if (components === 3) { + for (i = 0; i < h2; i++) { + py = Math.floor(i * yRatio) * w1Scanline; + for (j = 0; j < w2; j++) { + oldIndex = py + xScaled[j]; + temp[newIndex++] = pixels[oldIndex++]; + temp[newIndex++] = pixels[oldIndex++]; + temp[newIndex++] = pixels[oldIndex++]; + newIndex += alpha01; + } } - }, - getOutputLength: function AlternateCS_getOutputLength(inputLength, - alpha01) { - return this.base.getOutputLength(inputLength * - this.base.numComps / this.numComps, - alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function AlternateCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true + } + return temp; }; - return AlternateCS; -})(); - -var PatternCS = (function PatternCSClosure() { - function PatternCS(baseCS) { - this.name = 'Pattern'; - this.base = baseCS; - } - PatternCS.prototype = {}; + PDFImage.createMask = + function PDFImage_createMask(imgArray, width, height, + imageIsFromDecodeStream, inverseDecode) { - return PatternCS; -})(); + // |imgArray| might not contain full data for every pixel of the mask, so + // we need to distinguish between |computedLength| and |actualLength|. + // In particular, if inverseDecode is true, then the array we return must + // have a length of |computedLength|. -var IndexedCS = (function IndexedCSClosure() { - function IndexedCS(base, highVal, lookup) { - this.name = 'Indexed'; - this.numComps = 1; - this.defaultColor = new Uint8Array([0]); - this.base = base; - this.highVal = highVal; + var computedLength = ((width + 7) >> 3) * height; + var actualLength = imgArray.byteLength; + var haveFullData = computedLength === actualLength; + var data, i; - var baseNumComps = base.numComps; - var length = baseNumComps * highVal; - var lookupArray; + if (imageIsFromDecodeStream && (!inverseDecode || haveFullData)) { + // imgArray came from a DecodeStream and its data is in an appropriate + // form, so we can just transfer it. + data = imgArray; + } else if (!inverseDecode) { + data = new Uint8Array(actualLength); + data.set(imgArray); + } else { + data = new Uint8Array(computedLength); + data.set(imgArray); + for (i = actualLength; i < computedLength; i++) { + data[i] = 0xff; + } + } - if (isStream(lookup)) { - lookupArray = new Uint8Array(length); - var bytes = lookup.getBytes(length); - lookupArray.set(bytes); - } else if (isString(lookup)) { - lookupArray = new Uint8Array(length); - for (var i = 0; i < length; ++i) { - lookupArray[i] = lookup.charCodeAt(i); + // If necessary, invert the original mask data (but not any extra we might + // have added above). It's safe to modify the array -- whether it's the + // original or a copy, we're about to transfer it anyway, so nothing else + // in this thread can be relying on its contents. + if (inverseDecode) { + for (i = 0; i < actualLength; i++) { + data[i] = ~data[i]; } - } else if (lookup instanceof Uint8Array || lookup instanceof Array) { - lookupArray = lookup; - } else { - error('Unrecognized lookup table: ' + lookup); } - this.lookup = lookupArray; - } - IndexedCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function IndexedCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var numComps = this.base.numComps; - var start = src[srcOffset] * numComps; - this.base.getRgbItem(this.lookup, start, dest, destOffset); - }, - getRgbBuffer: function IndexedCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var base = this.base; - var numComps = base.numComps; - var outputDelta = base.getOutputLength(numComps, alpha01); - var lookup = this.lookup; + return {data: data, width: width, height: height}; + }; - for (var i = 0; i < count; ++i) { - var lookupPos = src[srcOffset++] * numComps; - base.getRgbBuffer(lookup, lookupPos, 1, dest, destOffset, 8, alpha01); - destOffset += outputDelta; - } - }, - getOutputLength: function IndexedCS_getOutputLength(inputLength, alpha01) { - return this.base.getOutputLength(inputLength * this.base.numComps, - alpha01); + PDFImage.prototype = { + get drawWidth() { + return Math.max(this.width, + this.smask && this.smask.width || 0, + this.mask && this.mask.width || 0); }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function IndexedCS_isDefaultDecode(decodeMap) { - // indexed color maps shouldn't be changed - return true; + + get drawHeight() { + return Math.max(this.height, + this.smask && this.smask.height || 0, + this.mask && this.mask.height || 0); }, - usesZeroToOneRange: true - }; - return IndexedCS; -})(); -var DeviceGrayCS = (function DeviceGrayCSClosure() { - function DeviceGrayCS() { - this.name = 'DeviceGray'; - this.numComps = 1; - this.defaultColor = new Float32Array([0]); - } + decodeBuffer: function PDFImage_decodeBuffer(buffer) { + var bpc = this.bpc; + var numComps = this.numComps; - DeviceGrayCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceGrayCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var c = (src[srcOffset] * 255) | 0; - c = c < 0 ? 0 : c > 255 ? 255 : c; - dest[destOffset] = dest[destOffset + 1] = dest[destOffset + 2] = c; - }, - getRgbBuffer: function DeviceGrayCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 255 / ((1 << bits) - 1); - var j = srcOffset, q = destOffset; - for (var i = 0; i < count; ++i) { - var c = (scale * src[j++]) | 0; - dest[q++] = c; - dest[q++] = c; - dest[q++] = c; - q += alpha01; - } - }, - getOutputLength: function DeviceGrayCS_getOutputLength(inputLength, - alpha01) { - return inputLength * (3 + alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceGrayCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceGrayCS; -})(); + var decodeAddends = this.decodeAddends; + var decodeCoefficients = this.decodeCoefficients; + var max = (1 << bpc) - 1; + var i, ii; -var DeviceRgbCS = (function DeviceRgbCSClosure() { - function DeviceRgbCS() { - this.name = 'DeviceRGB'; - this.numComps = 3; - this.defaultColor = new Float32Array([0, 0, 0]); - } - DeviceRgbCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceRgbCS_getRgbItem(src, srcOffset, - dest, destOffset) { - var r = (src[srcOffset] * 255) | 0; - var g = (src[srcOffset + 1] * 255) | 0; - var b = (src[srcOffset + 2] * 255) | 0; - dest[destOffset] = r < 0 ? 0 : r > 255 ? 255 : r; - dest[destOffset + 1] = g < 0 ? 0 : g > 255 ? 255 : g; - dest[destOffset + 2] = b < 0 ? 0 : b > 255 ? 255 : b; - }, - getRgbBuffer: function DeviceRgbCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - if (bits === 8 && alpha01 === 0) { - dest.set(src.subarray(srcOffset, srcOffset + count * 3), destOffset); + if (bpc === 1) { + // If the buffer needed decode that means it just needs to be inverted. + for (i = 0, ii = buffer.length; i < ii; i++) { + buffer[i] = +!(buffer[i]); + } return; } - var scale = 255 / ((1 << bits) - 1); - var j = srcOffset, q = destOffset; - for (var i = 0; i < count; ++i) { - dest[q++] = (scale * src[j++]) | 0; - dest[q++] = (scale * src[j++]) | 0; - dest[q++] = (scale * src[j++]) | 0; - q += alpha01; + var index = 0; + for (i = 0, ii = this.width * this.height; i < ii; i++) { + for (var j = 0; j < numComps; j++) { + buffer[index] = decodeAndClamp(buffer[index], decodeAddends[j], + decodeCoefficients[j], max); + index++; + } } }, - getOutputLength: function DeviceRgbCS_getOutputLength(inputLength, - alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: function DeviceRgbCS_isPassthrough(bits) { - return bits === 8; - }, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceRgbCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceRgbCS; -})(); - -var DeviceCmykCS = (function DeviceCmykCSClosure() { - // The coefficients below was found using numerical analysis: the method of - // steepest descent for the sum((f_i - color_value_i)^2) for r/g/b colors, - // where color_value is the tabular value from the table of sampled RGB colors - // from CMYK US Web Coated (SWOP) colorspace, and f_i is the corresponding - // CMYK color conversion using the estimation below: - // f(A, B,.. N) = Acc+Bcm+Ccy+Dck+c+Fmm+Gmy+Hmk+Im+Jyy+Kyk+Ly+Mkk+Nk+255 - function convertToRgb(src, srcOffset, srcScale, dest, destOffset) { - var c = src[srcOffset + 0] * srcScale; - var m = src[srcOffset + 1] * srcScale; - var y = src[srcOffset + 2] * srcScale; - var k = src[srcOffset + 3] * srcScale; - - var r = - (c * (-4.387332384609988 * c + 54.48615194189176 * m + - 18.82290502165302 * y + 212.25662451639585 * k + - -285.2331026137004) + - m * (1.7149763477362134 * m - 5.6096736904047315 * y + - -17.873870861415444 * k - 5.497006427196366) + - y * (-2.5217340131683033 * y - 21.248923337353073 * k + - 17.5119270841813) + - k * (-21.86122147463605 * k - 189.48180835922747) + 255) | 0; - var g = - (c * (8.841041422036149 * c + 60.118027045597366 * m + - 6.871425592049007 * y + 31.159100130055922 * k + - -79.2970844816548) + - m * (-15.310361306967817 * m + 17.575251261109482 * y + - 131.35250912493976 * k - 190.9453302588951) + - y * (4.444339102852739 * y + 9.8632861493405 * k - 24.86741582555878) + - k * (-20.737325471181034 * k - 187.80453709719578) + 255) | 0; - var b = - (c * (0.8842522430003296 * c + 8.078677503112928 * m + - 30.89978309703729 * y - 0.23883238689178934 * k + - -14.183576799673286) + - m * (10.49593273432072 * m + 63.02378494754052 * y + - 50.606957656360734 * k - 112.23884253719248) + - y * (0.03296041114873217 * y + 115.60384449646641 * k + - -193.58209356861505) + - k * (-22.33816807309886 * k - 180.12613974708367) + 255) | 0; - dest[destOffset] = r > 255 ? 255 : r < 0 ? 0 : r; - dest[destOffset + 1] = g > 255 ? 255 : g < 0 ? 0 : g; - dest[destOffset + 2] = b > 255 ? 255 : b < 0 ? 0 : b; - } + getComponents: function PDFImage_getComponents(buffer) { + var bpc = this.bpc; - function DeviceCmykCS() { - this.name = 'DeviceCMYK'; - this.numComps = 4; - this.defaultColor = new Float32Array([0, 0, 0, 1]); - } - DeviceCmykCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function DeviceCmykCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(src, srcOffset, 1, dest, destOffset); - }, - getRgbBuffer: function DeviceCmykCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; i++) { - convertToRgb(src, srcOffset, scale, dest, destOffset); - srcOffset += 4; - destOffset += 3 + alpha01; + // This image doesn't require any extra work. + if (bpc === 8) { + return buffer; } - }, - getOutputLength: function DeviceCmykCS_getOutputLength(inputLength, - alpha01) { - return (inputLength / 4 * (3 + alpha01)) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function DeviceCmykCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return DeviceCmykCS; -})(); + var width = this.width; + var height = this.height; + var numComps = this.numComps; -// -// CalGrayCS: Based on "PDF Reference, Sixth Ed", p.245 -// -var CalGrayCS = (function CalGrayCSClosure() { - function CalGrayCS(whitePoint, blackPoint, gamma) { - this.name = 'CalGray'; - this.numComps = 1; - this.defaultColor = new Float32Array([0]); + var length = width * height * numComps; + var bufferPos = 0; + var output = (bpc <= 8 ? new Uint8Array(length) : + (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); + var rowComps = width * numComps; - if (!whitePoint) { - error('WhitePoint missing - required for color space CalGray'); - } - blackPoint = blackPoint || [0, 0, 0]; - gamma = gamma || 1; + var max = (1 << bpc) - 1; + var i = 0, ii, buf; - // Translate arguments to spec variables. - this.XW = whitePoint[0]; - this.YW = whitePoint[1]; - this.ZW = whitePoint[2]; + if (bpc === 1) { + // Optimization for reading 1 bpc images. + var mask, loop1End, loop2End; + for (var j = 0; j < height; j++) { + loop1End = i + (rowComps & ~7); + loop2End = i + rowComps; - this.XB = blackPoint[0]; - this.YB = blackPoint[1]; - this.ZB = blackPoint[2]; + // unroll loop for all full bytes + while (i < loop1End) { + buf = buffer[bufferPos++]; + output[i] = (buf >> 7) & 1; + output[i + 1] = (buf >> 6) & 1; + output[i + 2] = (buf >> 5) & 1; + output[i + 3] = (buf >> 4) & 1; + output[i + 4] = (buf >> 3) & 1; + output[i + 5] = (buf >> 2) & 1; + output[i + 6] = (buf >> 1) & 1; + output[i + 7] = buf & 1; + i += 8; + } - this.G = gamma; + // handle remaing bits + if (i < loop2End) { + buf = buffer[bufferPos++]; + mask = 128; + while (i < loop2End) { + output[i++] = +!!(buf & mask); + mask >>= 1; + } + } + } + } else { + // The general case that handles all other bpc values. + var bits = 0; + buf = 0; + for (i = 0, ii = length; i < ii; ++i) { + if (i % rowComps === 0) { + buf = 0; + bits = 0; + } - // Validate variables as per spec. - if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { - error('Invalid WhitePoint components for ' + this.name + - ', no fallback available'); - } + while (bits < bpc) { + buf = (buf << 8) | buffer[bufferPos++]; + bits += 8; + } - if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { - info('Invalid BlackPoint for ' + this.name + ', falling back to default'); - this.XB = this.YB = this.ZB = 0; - } + var remainingBits = bits - bpc; + var value = buf >> remainingBits; + output[i] = (value < 0 ? 0 : (value > max ? max : value)); + buf = buf & ((1 << remainingBits) - 1); + bits = remainingBits; + } + } + return output; + }, - if (this.XB !== 0 || this.YB !== 0 || this.ZB !== 0) { - warn(this.name + ', BlackPoint: XB: ' + this.XB + ', YB: ' + this.YB + - ', ZB: ' + this.ZB + ', only default values are supported.'); - } + fillOpacity: function PDFImage_fillOpacity(rgbaBuf, width, height, + actualHeight, image) { + var smask = this.smask; + var mask = this.mask; + var alphaBuf, sw, sh, i, ii, j; - if (this.G < 1) { - info('Invalid Gamma: ' + this.G + ' for ' + this.name + - ', falling back to default'); - this.G = 1; - } - } + if (smask) { + sw = smask.width; + sh = smask.height; + alphaBuf = new Uint8Array(sw * sh); + smask.fillGrayBuffer(alphaBuf); + if (sw !== width || sh !== height) { + alphaBuf = PDFImage.resize(alphaBuf, smask.bpc, 1, sw, sh, width, + height); + } + } else if (mask) { + if (mask instanceof PDFImage) { + sw = mask.width; + sh = mask.height; + alphaBuf = new Uint8Array(sw * sh); + mask.numComps = 1; + mask.fillGrayBuffer(alphaBuf); - function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { - // A represents a gray component of a calibrated gray space. - // A <---> AG in the spec - var A = src[srcOffset] * scale; - var AG = Math.pow(A, cs.G); + // Need to invert values in rgbaBuf + for (i = 0, ii = sw * sh; i < ii; ++i) { + alphaBuf[i] = 255 - alphaBuf[i]; + } - // Computes L as per spec. ( = cs.YW * AG ) - // Except if other than default BlackPoint values are used. - var L = cs.YW * AG; - // http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html, Ch 4. - // Convert values to rgb range [0, 255]. - var val = Math.max(295.8 * Math.pow(L, 0.333333333333333333) - 40.8, 0) | 0; - dest[destOffset] = val; - dest[destOffset + 1] = val; - dest[destOffset + 2] = val; - } + if (sw !== width || sh !== height) { + alphaBuf = PDFImage.resize(alphaBuf, mask.bpc, 1, sw, sh, width, + height); + } + } else if (isArray(mask)) { + // Color key mask: if any of the compontents are outside the range + // then they should be painted. + alphaBuf = new Uint8Array(width * height); + var numComps = this.numComps; + for (i = 0, ii = width * height; i < ii; ++i) { + var opacity = 0; + var imageOffset = i * numComps; + for (j = 0; j < numComps; ++j) { + var color = image[imageOffset + j]; + var maskOffset = j * 2; + if (color < mask[maskOffset] || color > mask[maskOffset + 1]) { + opacity = 255; + break; + } + } + alphaBuf[i] = opacity; + } + } else { + error('Unknown mask format.'); + } + } - CalGrayCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function CalGrayCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(this, src, srcOffset, dest, destOffset, 1); + if (alphaBuf) { + for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { + rgbaBuf[j] = alphaBuf[i]; + } + } else { + // No mask. + for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { + rgbaBuf[j] = 255; + } + } }, - getRgbBuffer: function CalGrayCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; ++i) { - convertToRgb(this, src, srcOffset, dest, destOffset, scale); - srcOffset += 1; - destOffset += 3 + alpha01; + undoPreblend: function PDFImage_undoPreblend(buffer, width, height) { + var matte = this.smask && this.smask.matte; + if (!matte) { + return; + } + var matteRgb = this.colorSpace.getRgb(matte, 0); + var matteR = matteRgb[0]; + var matteG = matteRgb[1]; + var matteB = matteRgb[2]; + var length = width * height * 4; + var r, g, b; + for (var i = 0; i < length; i += 4) { + var alpha = buffer[i + 3]; + if (alpha === 0) { + // according formula we have to get Infinity in all components + // making it white (typical paper color) should be okay + buffer[i] = 255; + buffer[i + 1] = 255; + buffer[i + 2] = 255; + continue; + } + var k = 255 / alpha; + r = (buffer[i] - matteR) * k + matteR; + g = (buffer[i + 1] - matteG) * k + matteG; + b = (buffer[i + 2] - matteB) * k + matteB; + buffer[i] = r <= 0 ? 0 : r >= 255 ? 255 : r | 0; + buffer[i + 1] = g <= 0 ? 0 : g >= 255 ? 255 : g | 0; + buffer[i + 2] = b <= 0 ? 0 : b >= 255 ? 255 : b | 0; } }, - getOutputLength: function CalGrayCS_getOutputLength(inputLength, alpha01) { - return inputLength * (3 + alpha01); - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function CalGrayCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return CalGrayCS; -})(); - -// -// CalRGBCS: Based on "PDF Reference, Sixth Ed", p.247 -// -var CalRGBCS = (function CalRGBCSClosure() { - - // See http://www.brucelindbloom.com/index.html?Eqn_ChromAdapt.html for these - // matrices. - var BRADFORD_SCALE_MATRIX = new Float32Array([ - 0.8951, 0.2664, -0.1614, - -0.7502, 1.7135, 0.0367, - 0.0389, -0.0685, 1.0296]); - - var BRADFORD_SCALE_INVERSE_MATRIX = new Float32Array([ - 0.9869929, -0.1470543, 0.1599627, - 0.4323053, 0.5183603, 0.0492912, - -0.0085287, 0.0400428, 0.9684867]); - - // See http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html. - var SRGB_D65_XYZ_TO_RGB_MATRIX = new Float32Array([ - 3.2404542, -1.5371385, -0.4985314, - -0.9692660, 1.8760108, 0.0415560, - 0.0556434, -0.2040259, 1.0572252]); - var FLAT_WHITEPOINT_MATRIX = new Float32Array([1, 1, 1]); + createImageData: function PDFImage_createImageData(forceRGBA) { + var drawWidth = this.drawWidth; + var drawHeight = this.drawHeight; + var imgData = { // other fields are filled in below + width: drawWidth, + height: drawHeight + }; - var tempNormalizeMatrix = new Float32Array(3); - var tempConvertMatrix1 = new Float32Array(3); - var tempConvertMatrix2 = new Float32Array(3); + var numComps = this.numComps; + var originalWidth = this.width; + var originalHeight = this.height; + var bpc = this.bpc; - var DECODE_L_CONSTANT = Math.pow(((8 + 16) / 116), 3) / 8.0; + // Rows start at byte boundary. + var rowBytes = (originalWidth * numComps * bpc + 7) >> 3; + var imgArray; - function CalRGBCS(whitePoint, blackPoint, gamma, matrix) { - this.name = 'CalRGB'; - this.numComps = 3; - this.defaultColor = new Float32Array(3); + if (!forceRGBA) { + // If it is a 1-bit-per-pixel grayscale (i.e. black-and-white) image + // without any complications, we pass a same-sized copy to the main + // thread rather than expanding by 32x to RGBA form. This saves *lots* + // of memory for many scanned documents. It's also much faster. + // + // Similarly, if it is a 24-bit-per pixel RGB image without any + // complications, we avoid expanding by 1.333x to RGBA form. + var kind; + if (this.colorSpace.name === 'DeviceGray' && bpc === 1) { + kind = ImageKind.GRAYSCALE_1BPP; + } else if (this.colorSpace.name === 'DeviceRGB' && bpc === 8 && + !this.needsDecode) { + kind = ImageKind.RGB_24BPP; + } + if (kind && !this.smask && !this.mask && + drawWidth === originalWidth && drawHeight === originalHeight) { + imgData.kind = kind; - if (!whitePoint) { - error('WhitePoint missing - required for color space CalRGB'); - } - blackPoint = blackPoint || new Float32Array(3); - gamma = gamma || new Float32Array([1, 1, 1]); - matrix = matrix || new Float32Array([1, 0, 0, 0, 1, 0, 0, 0, 1]); + imgArray = this.getImageBytes(originalHeight * rowBytes); + // If imgArray came from a DecodeStream, we're safe to transfer it + // (and thus detach its underlying buffer) because it will constitute + // the entire DecodeStream's data. But if it came from a Stream, we + // need to copy it because it'll only be a portion of the Stream's + // data, and the rest will be read later on. + if (this.image instanceof DecodeStream) { + imgData.data = imgArray; + } else { + var newArray = new Uint8Array(imgArray.length); + newArray.set(imgArray); + imgData.data = newArray; + } + if (this.needsDecode) { + // Invert the buffer (which must be grayscale if we reached here). + assert(kind === ImageKind.GRAYSCALE_1BPP); + var buffer = imgData.data; + for (var i = 0, ii = buffer.length; i < ii; i++) { + buffer[i] ^= 0xff; + } + } + return imgData; + } + if (this.image instanceof JpegStream && !this.smask && !this.mask && + (this.colorSpace.name === 'DeviceGray' || + this.colorSpace.name === 'DeviceRGB' || + this.colorSpace.name === 'DeviceCMYK')) { + imgData.kind = ImageKind.RGB_24BPP; + imgData.data = this.getImageBytes(originalHeight * rowBytes, + drawWidth, drawHeight, true); + return imgData; + } + } - // Translate arguments to spec variables. - var XW = whitePoint[0]; - var YW = whitePoint[1]; - var ZW = whitePoint[2]; - this.whitePoint = whitePoint; + imgArray = this.getImageBytes(originalHeight * rowBytes); + // imgArray can be incomplete (e.g. after CCITT fax encoding). + var actualHeight = 0 | (imgArray.length / rowBytes * + drawHeight / originalHeight); - var XB = blackPoint[0]; - var YB = blackPoint[1]; - var ZB = blackPoint[2]; - this.blackPoint = blackPoint; + var comps = this.getComponents(imgArray); - this.GR = gamma[0]; - this.GG = gamma[1]; - this.GB = gamma[2]; + // If opacity data is present, use RGBA_32BPP form. Otherwise, use the + // more compact RGB_24BPP form if allowable. + var alpha01, maybeUndoPreblend; + if (!forceRGBA && !this.smask && !this.mask) { + imgData.kind = ImageKind.RGB_24BPP; + imgData.data = new Uint8Array(drawWidth * drawHeight * 3); + alpha01 = 0; + maybeUndoPreblend = false; + } else { + imgData.kind = ImageKind.RGBA_32BPP; + imgData.data = new Uint8Array(drawWidth * drawHeight * 4); + alpha01 = 1; + maybeUndoPreblend = true; - this.MXA = matrix[0]; - this.MYA = matrix[1]; - this.MZA = matrix[2]; - this.MXB = matrix[3]; - this.MYB = matrix[4]; - this.MZB = matrix[5]; - this.MXC = matrix[6]; - this.MYC = matrix[7]; - this.MZC = matrix[8]; + // Color key masking (opacity) must be performed before decoding. + this.fillOpacity(imgData.data, drawWidth, drawHeight, actualHeight, + comps); + } - // Validate variables as per spec. - if (XW < 0 || ZW < 0 || YW !== 1) { - error('Invalid WhitePoint components for ' + this.name + - ', no fallback available'); - } + if (this.needsDecode) { + this.decodeBuffer(comps); + } + this.colorSpace.fillRgb(imgData.data, originalWidth, originalHeight, + drawWidth, drawHeight, actualHeight, bpc, comps, + alpha01); + if (maybeUndoPreblend) { + this.undoPreblend(imgData.data, drawWidth, actualHeight); + } - if (XB < 0 || YB < 0 || ZB < 0) { - info('Invalid BlackPoint for ' + this.name + ' [' + XB + ', ' + YB + - ', ' + ZB + '], falling back to default'); - this.blackPoint = new Float32Array(3); - } + return imgData; + }, - if (this.GR < 0 || this.GG < 0 || this.GB < 0) { - info('Invalid Gamma [' + this.GR + ', ' + this.GG + ', ' + this.GB + - '] for ' + this.name + ', falling back to default'); - this.GR = this.GG = this.GB = 1; - } + fillGrayBuffer: function PDFImage_fillGrayBuffer(buffer) { + var numComps = this.numComps; + if (numComps !== 1) { + error('Reading gray scale from a color image: ' + numComps); + } - if (this.MXA < 0 || this.MYA < 0 || this.MZA < 0 || - this.MXB < 0 || this.MYB < 0 || this.MZB < 0 || - this.MXC < 0 || this.MYC < 0 || this.MZC < 0) { - info('Invalid Matrix for ' + this.name + ' [' + - this.MXA + ', ' + this.MYA + ', ' + this.MZA + - this.MXB + ', ' + this.MYB + ', ' + this.MZB + - this.MXC + ', ' + this.MYC + ', ' + this.MZC + - '], falling back to default'); - this.MXA = this.MYB = this.MZC = 1; - this.MXB = this.MYA = this.MZA = this.MXC = this.MYC = this.MZB = 0; - } - } + var width = this.width; + var height = this.height; + var bpc = this.bpc; - function matrixProduct(a, b, result) { - result[0] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; - result[1] = a[3] * b[0] + a[4] * b[1] + a[5] * b[2]; - result[2] = a[6] * b[0] + a[7] * b[1] + a[8] * b[2]; - } + // rows start at byte boundary + var rowBytes = (width * numComps * bpc + 7) >> 3; + var imgArray = this.getImageBytes(height * rowBytes); - function convertToFlat(sourceWhitePoint, LMS, result) { - result[0] = LMS[0] * 1 / sourceWhitePoint[0]; - result[1] = LMS[1] * 1 / sourceWhitePoint[1]; - result[2] = LMS[2] * 1 / sourceWhitePoint[2]; - } + var comps = this.getComponents(imgArray); + var i, length; - function convertToD65(sourceWhitePoint, LMS, result) { - var D65X = 0.95047; - var D65Y = 1; - var D65Z = 1.08883; + if (bpc === 1) { + // inline decoding (= inversion) for 1 bpc images + length = width * height; + if (this.needsDecode) { + // invert and scale to {0, 255} + for (i = 0; i < length; ++i) { + buffer[i] = (comps[i] - 1) & 255; + } + } else { + // scale to {0, 255} + for (i = 0; i < length; ++i) { + buffer[i] = (-comps[i]) & 255; + } + } + return; + } - result[0] = LMS[0] * D65X / sourceWhitePoint[0]; - result[1] = LMS[1] * D65Y / sourceWhitePoint[1]; - result[2] = LMS[2] * D65Z / sourceWhitePoint[2]; - } + if (this.needsDecode) { + this.decodeBuffer(comps); + } + length = width * height; + // we aren't using a colorspace so we need to scale the value + var scale = 255 / ((1 << bpc) - 1); + for (i = 0; i < length; ++i) { + buffer[i] = (scale * comps[i]) | 0; + } + }, - function sRGBTransferFunction(color) { - // See http://en.wikipedia.org/wiki/SRGB. - if (color <= 0.0031308){ - return adjustToRange(0, 1, 12.92 * color); + getImageBytes: function PDFImage_getImageBytes(length, + drawWidth, drawHeight, + forceRGB) { + this.image.reset(); + this.image.drawWidth = drawWidth || this.width; + this.image.drawHeight = drawHeight || this.height; + this.image.forceRGB = !!forceRGB; + return this.image.getBytes(length); } + }; + return PDFImage; +})(); - return adjustToRange(0, 1, (1 + 0.055) * Math.pow(color, 1 / 2.4) - 0.055); - } - - function adjustToRange(min, max, value) { - return Math.max(min, Math.min(max, value)); - } +exports.PDFImage = PDFImage; - function decodeL(L) { - if (L < 0) { - return -decodeL(-L); - } +// TODO refactor to remove dependency on colorspace.js +coreColorSpace._setCoreImage(exports); +})); - if (L > 8.0) { - return Math.pow(((L + 16) / 116), 3); - } - return L * DECODE_L_CONSTANT; +(function (root, factory) { + { + factory((root.pdfjsCoreObj = {}), root.pdfjsSharedUtil, + root.pdfjsCorePrimitives, root.pdfjsCoreCrypto, root.pdfjsCoreParser, + root.pdfjsCoreChunkedStream, root.pdfjsCoreColorSpace); } +}(this, function (exports, sharedUtil, corePrimitives, coreCrypto, coreParser, + coreChunkedStream, coreColorSpace) { - function compensateBlackPoint(sourceBlackPoint, XYZ_Flat, result) { - - // In case the blackPoint is already the default blackPoint then there is - // no need to do compensation. - if (sourceBlackPoint[0] === 0 && - sourceBlackPoint[1] === 0 && - sourceBlackPoint[2] === 0) { - result[0] = XYZ_Flat[0]; - result[1] = XYZ_Flat[1]; - result[2] = XYZ_Flat[2]; - return; - } +var InvalidPDFException = sharedUtil.InvalidPDFException; +var MissingDataException = sharedUtil.MissingDataException; +var XRefParseException = sharedUtil.XRefParseException; +var assert = sharedUtil.assert; +var bytesToString = sharedUtil.bytesToString; +var createPromiseCapability = sharedUtil.createPromiseCapability; +var error = sharedUtil.error; +var info = sharedUtil.info; +var isArray = sharedUtil.isArray; +var isInt = sharedUtil.isInt; +var isString = sharedUtil.isString; +var shadow = sharedUtil.shadow; +var stringToPDFString = sharedUtil.stringToPDFString; +var stringToUTF8String = sharedUtil.stringToUTF8String; +var warn = sharedUtil.warn; +var isValidUrl = sharedUtil.isValidUrl; +var Util = sharedUtil.Util; +var Ref = corePrimitives.Ref; +var RefSet = corePrimitives.RefSet; +var RefSetCache = corePrimitives.RefSetCache; +var isName = corePrimitives.isName; +var isCmd = corePrimitives.isCmd; +var isDict = corePrimitives.isDict; +var isRef = corePrimitives.isRef; +var isStream = corePrimitives.isStream; +var CipherTransformFactory = coreCrypto.CipherTransformFactory; +var Lexer = coreParser.Lexer; +var Parser = coreParser.Parser; +var ChunkedStream = coreChunkedStream.ChunkedStream; +var ColorSpace = coreColorSpace.ColorSpace; - // For the blackPoint calculation details, please see - // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ - // AdobeBPC.pdf. - // The destination blackPoint is the default blackPoint [0, 0, 0]. - var zeroDecodeL = decodeL(0); +var Catalog = (function CatalogClosure() { + function Catalog(pdfManager, xref, pageFactory) { + this.pdfManager = pdfManager; + this.xref = xref; + this.catDict = xref.getCatalogObj(); + this.fontCache = new RefSetCache(); + assert(isDict(this.catDict), + 'catalog object is not a dictionary'); - var X_DST = zeroDecodeL; - var X_SRC = decodeL(sourceBlackPoint[0]); + // TODO refactor to move getPage() to the PDFDocument. + this.pageFactory = pageFactory; + this.pagePromises = []; + } - var Y_DST = zeroDecodeL; - var Y_SRC = decodeL(sourceBlackPoint[1]); + Catalog.prototype = { + get metadata() { + var streamRef = this.catDict.getRaw('Metadata'); + if (!isRef(streamRef)) { + return shadow(this, 'metadata', null); + } - var Z_DST = zeroDecodeL; - var Z_SRC = decodeL(sourceBlackPoint[2]); + var encryptMetadata = (!this.xref.encrypt ? false : + this.xref.encrypt.encryptMetadata); - var X_Scale = (1 - X_DST) / (1 - X_SRC); - var X_Offset = 1 - X_Scale; + var stream = this.xref.fetch(streamRef, !encryptMetadata); + var metadata; + if (stream && isDict(stream.dict)) { + var type = stream.dict.get('Type'); + var subtype = stream.dict.get('Subtype'); - var Y_Scale = (1 - Y_DST) / (1 - Y_SRC); - var Y_Offset = 1 - Y_Scale; + if (isName(type) && isName(subtype) && + type.name === 'Metadata' && subtype.name === 'XML') { + // XXX: This should examine the charset the XML document defines, + // however since there are currently no real means to decode + // arbitrary charsets, let's just hope that the author of the PDF + // was reasonable enough to stick with the XML default charset, + // which is UTF-8. + try { + metadata = stringToUTF8String(bytesToString(stream.getBytes())); + } catch (e) { + info('Skipping invalid metadata.'); + } + } + } - var Z_Scale = (1 - Z_DST) / (1 - Z_SRC); - var Z_Offset = 1 - Z_Scale; + return shadow(this, 'metadata', metadata); + }, + get toplevelPagesDict() { + var pagesObj = this.catDict.get('Pages'); + assert(isDict(pagesObj), 'invalid top-level pages dictionary'); + // shadow the prototype getter + return shadow(this, 'toplevelPagesDict', pagesObj); + }, + get documentOutline() { + var obj = null; + try { + obj = this.readDocumentOutline(); + } catch (ex) { + if (ex instanceof MissingDataException) { + throw ex; + } + warn('Unable to read document outline'); + } + return shadow(this, 'documentOutline', obj); + }, + readDocumentOutline: function Catalog_readDocumentOutline() { + var obj = this.catDict.get('Outlines'); + if (!isDict(obj)) { + return null; + } + obj = obj.getRaw('First'); + if (!isRef(obj)) { + return null; + } + var root = { items: [] }; + var queue = [{obj: obj, parent: root}]; + // To avoid recursion, keep track of the already processed items. + var processed = new RefSet(); + processed.put(obj); + var xref = this.xref, blackColor = new Uint8Array(3); - result[0] = XYZ_Flat[0] * X_Scale + X_Offset; - result[1] = XYZ_Flat[1] * Y_Scale + Y_Offset; - result[2] = XYZ_Flat[2] * Z_Scale + Z_Offset; - } + while (queue.length > 0) { + var i = queue.shift(); + var outlineDict = xref.fetchIfRef(i.obj); + if (outlineDict === null) { + continue; + } + assert(outlineDict.has('Title'), 'Invalid outline item'); - function normalizeWhitePointToFlat(sourceWhitePoint, XYZ_In, result) { + var actionDict = outlineDict.get('A'), dest = null, url = null; + if (actionDict) { + var destEntry = actionDict.get('D'); + if (destEntry) { + dest = destEntry; + } else { + var uriEntry = actionDict.get('URI'); + if (isString(uriEntry) && isValidUrl(uriEntry, false)) { + url = uriEntry; + } + } + } else if (outlineDict.has('Dest')) { + dest = outlineDict.getRaw('Dest'); + if (isName(dest)) { + dest = dest.name; + } + } + var title = outlineDict.get('Title'); + var flags = outlineDict.get('F') || 0; + + var color = outlineDict.get('C'), rgbColor = blackColor; + // We only need to parse the color when it's valid, and non-default. + if (isArray(color) && color.length === 3 && + (color[0] !== 0 || color[1] !== 0 || color[2] !== 0)) { + rgbColor = ColorSpace.singletons.rgb.getRgb(color, 0); + } + var outlineItem = { + dest: dest, + url: url, + title: stringToPDFString(title), + color: rgbColor, + count: outlineDict.get('Count'), + bold: !!(flags & 2), + italic: !!(flags & 1), + items: [] + }; + i.parent.items.push(outlineItem); + obj = outlineDict.getRaw('First'); + if (isRef(obj) && !processed.has(obj)) { + queue.push({obj: obj, parent: outlineItem}); + processed.put(obj); + } + obj = outlineDict.getRaw('Next'); + if (isRef(obj) && !processed.has(obj)) { + queue.push({obj: obj, parent: i.parent}); + processed.put(obj); + } + } + return (root.items.length > 0 ? root.items : null); + }, + get numPages() { + var obj = this.toplevelPagesDict.get('Count'); + assert( + isInt(obj), + 'page count in top level pages object is not an integer' + ); + // shadow the prototype getter + return shadow(this, 'num', obj); + }, + get destinations() { + function fetchDestination(dest) { + return isDict(dest) ? dest.get('D') : dest; + } - // In case the whitePoint is already flat then there is no need to do - // normalization. - if (sourceWhitePoint[0] === 1 && sourceWhitePoint[2] === 1) { - result[0] = XYZ_In[0]; - result[1] = XYZ_In[1]; - result[2] = XYZ_In[2]; - return; - } + var xref = this.xref; + var dests = {}, nameTreeRef, nameDictionaryRef; + var obj = this.catDict.get('Names'); + if (obj && obj.has('Dests')) { + nameTreeRef = obj.getRaw('Dests'); + } else if (this.catDict.has('Dests')) { + nameDictionaryRef = this.catDict.get('Dests'); + } - var LMS = result; - matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + if (nameDictionaryRef) { + // reading simple destination dictionary + obj = nameDictionaryRef; + obj.forEach(function catalogForEach(key, value) { + if (!value) { + return; + } + dests[key] = fetchDestination(value); + }); + } + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + var names = nameTree.getAll(); + for (var name in names) { + dests[name] = fetchDestination(names[name]); + } + } + return shadow(this, 'destinations', dests); + }, + getDestination: function Catalog_getDestination(destinationId) { + function fetchDestination(dest) { + return isDict(dest) ? dest.get('D') : dest; + } - var LMS_Flat = tempNormalizeMatrix; - convertToFlat(sourceWhitePoint, LMS, LMS_Flat); + var xref = this.xref; + var dest = null, nameTreeRef, nameDictionaryRef; + var obj = this.catDict.get('Names'); + if (obj && obj.has('Dests')) { + nameTreeRef = obj.getRaw('Dests'); + } else if (this.catDict.has('Dests')) { + nameDictionaryRef = this.catDict.get('Dests'); + } - matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_Flat, result); - } + if (nameDictionaryRef) { // Simple destination dictionary. + var value = nameDictionaryRef.get(destinationId); + if (value) { + dest = fetchDestination(value); + } + } + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + dest = fetchDestination(nameTree.get(destinationId)); + } + return dest; + }, - function normalizeWhitePointToD65(sourceWhitePoint, XYZ_In, result) { + get pageLabels() { + var obj = null; + try { + obj = this.readPageLabels(); + } catch (ex) { + if (ex instanceof MissingDataException) { + throw ex; + } + warn('Unable to read page labels.'); + } + return shadow(this, 'pageLabels', obj); + }, + readPageLabels: function Catalog_readPageLabels() { + var obj = this.catDict.getRaw('PageLabels'); + if (!obj) { + return null; + } + var pageLabels = new Array(this.numPages); + var style = null; + var prefix = ''; + var start = 1; - var LMS = result; - matrixProduct(BRADFORD_SCALE_MATRIX, XYZ_In, LMS); + var numberTree = new NumberTree(obj, this.xref); + var nums = numberTree.getAll(); + var currentLabel = '', currentIndex = 1; - var LMS_D65 = tempNormalizeMatrix; - convertToD65(sourceWhitePoint, LMS, LMS_D65); + for (var i = 0, ii = this.numPages; i < ii; i++) { + if (i in nums) { + var labelDict = nums[i]; + assert(isDict(labelDict), 'The PageLabel is not a dictionary.'); - matrixProduct(BRADFORD_SCALE_INVERSE_MATRIX, LMS_D65, result); - } + var type = labelDict.get('Type'); + assert(!type || (isName(type) && type.name === 'PageLabel'), + 'Invalid type in PageLabel dictionary.'); - function convertToRgb(cs, src, srcOffset, dest, destOffset, scale) { - // A, B and C represent a red, green and blue components of a calibrated - // rgb space. - var A = adjustToRange(0, 1, src[srcOffset] * scale); - var B = adjustToRange(0, 1, src[srcOffset + 1] * scale); - var C = adjustToRange(0, 1, src[srcOffset + 2] * scale); + var s = labelDict.get('S'); + assert(!s || isName(s), 'Invalid style in PageLabel dictionary.'); + style = (s ? s.name : null); - // A <---> AGR in the spec - // B <---> BGG in the spec - // C <---> CGB in the spec - var AGR = Math.pow(A, cs.GR); - var BGG = Math.pow(B, cs.GG); - var CGB = Math.pow(C, cs.GB); + prefix = labelDict.get('P') || ''; + assert(isString(prefix), 'Invalid prefix in PageLabel dictionary.'); - // Computes intermediate variables L, M, N as per spec. - // To decode X, Y, Z values map L, M, N directly to them. - var X = cs.MXA * AGR + cs.MXB * BGG + cs.MXC * CGB; - var Y = cs.MYA * AGR + cs.MYB * BGG + cs.MYC * CGB; - var Z = cs.MZA * AGR + cs.MZB * BGG + cs.MZC * CGB; + start = labelDict.get('St') || 1; + assert(isInt(start), 'Invalid start in PageLabel dictionary.'); + currentIndex = start; + } - // The following calculations are based on this document: - // http://www.adobe.com/content/dam/Adobe/en/devnet/photoshop/sdk/ - // AdobeBPC.pdf. - var XYZ = tempConvertMatrix1; - XYZ[0] = X; - XYZ[1] = Y; - XYZ[2] = Z; - var XYZ_Flat = tempConvertMatrix2; + switch (style) { + case 'D': + currentLabel = currentIndex; + break; + case 'R': + case 'r': + currentLabel = Util.toRoman(currentIndex, style === 'r'); + break; + case 'A': + case 'a': + var LIMIT = 26; // Use only the characters A--Z, or a--z. + var A_UPPER_CASE = 0x41, A_LOWER_CASE = 0x61; - normalizeWhitePointToFlat(cs.whitePoint, XYZ, XYZ_Flat); + var baseCharCode = (style === 'a' ? A_LOWER_CASE : A_UPPER_CASE); + var letterIndex = currentIndex - 1; + var character = String.fromCharCode(baseCharCode + + (letterIndex % LIMIT)); + var charBuf = []; + for (var j = 0, jj = (letterIndex / LIMIT) | 0; j <= jj; j++) { + charBuf.push(character); + } + currentLabel = charBuf.join(''); + break; + default: + assert(!style, + 'Invalid style "' + style + '" in PageLabel dictionary.'); + } + pageLabels[i] = prefix + currentLabel; - var XYZ_Black = tempConvertMatrix1; - compensateBlackPoint(cs.blackPoint, XYZ_Flat, XYZ_Black); + currentLabel = ''; + currentIndex++; + } + return pageLabels; + }, - var XYZ_D65 = tempConvertMatrix2; - normalizeWhitePointToD65(FLAT_WHITEPOINT_MATRIX, XYZ_Black, XYZ_D65); + get attachments() { + var xref = this.xref; + var attachments = null, nameTreeRef; + var obj = this.catDict.get('Names'); + if (obj) { + nameTreeRef = obj.getRaw('EmbeddedFiles'); + } - var SRGB = tempConvertMatrix1; - matrixProduct(SRGB_D65_XYZ_TO_RGB_MATRIX, XYZ_D65, SRGB); + if (nameTreeRef) { + var nameTree = new NameTree(nameTreeRef, xref); + var names = nameTree.getAll(); + for (var name in names) { + var fs = new FileSpec(names[name], xref); + if (!attachments) { + attachments = Object.create(null); + } + attachments[stringToPDFString(name)] = fs.serializable; + } + } + return shadow(this, 'attachments', attachments); + }, + get javaScript() { + var xref = this.xref; + var obj = this.catDict.get('Names'); - var sR = sRGBTransferFunction(SRGB[0]); - var sG = sRGBTransferFunction(SRGB[1]); - var sB = sRGBTransferFunction(SRGB[2]); + var javaScript = []; + function appendIfJavaScriptDict(jsDict) { + var type = jsDict.get('S'); + if (!isName(type) || type.name !== 'JavaScript') { + return; + } + var js = jsDict.get('JS'); + if (isStream(js)) { + js = bytesToString(js.getBytes()); + } else if (!isString(js)) { + return; + } + javaScript.push(stringToPDFString(js)); + } + if (obj && obj.has('JavaScript')) { + var nameTree = new NameTree(obj.getRaw('JavaScript'), xref); + var names = nameTree.getAll(); + for (var name in names) { + // We don't really use the JavaScript right now. This code is + // defensive so we don't cause errors on document load. + var jsDict = names[name]; + if (isDict(jsDict)) { + appendIfJavaScriptDict(jsDict); + } + } + } - // Convert the values to rgb range [0, 255]. - dest[destOffset] = Math.round(sR * 255); - dest[destOffset + 1] = Math.round(sG * 255); - dest[destOffset + 2] = Math.round(sB * 255); - } + // Append OpenAction actions to javaScript array + var openactionDict = this.catDict.get('OpenAction'); + if (isDict(openactionDict, 'Action')) { + var actionType = openactionDict.get('S'); + if (isName(actionType) && actionType.name === 'Named') { + // The named Print action is not a part of the PDF 1.7 specification, + // but is supported by many PDF readers/writers (including Adobe's). + var action = openactionDict.get('N'); + if (isName(action) && action.name === 'Print') { + javaScript.push('print({});'); + } + } else { + appendIfJavaScriptDict(openactionDict); + } + } - CalRGBCS.prototype = { - getRgb: function CalRGBCS_getRgb(src, srcOffset) { - var rgb = new Uint8Array(3); - this.getRgbItem(src, srcOffset, rgb, 0); - return rgb; + return shadow(this, 'javaScript', javaScript); }, - getRgbItem: function CalRGBCS_getRgbItem(src, srcOffset, - dest, destOffset) { - convertToRgb(this, src, srcOffset, dest, destOffset, 1); + + cleanup: function Catalog_cleanup() { + var promises = []; + this.fontCache.forEach(function (promise) { + promises.push(promise); + }); + return Promise.all(promises).then(function (translatedFonts) { + for (var i = 0, ii = translatedFonts.length; i < ii; i++) { + var font = translatedFonts[i].dict; + delete font.translated; + } + this.fontCache.clear(); + }.bind(this)); }, - getRgbBuffer: function CalRGBCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var scale = 1 / ((1 << bits) - 1); - for (var i = 0; i < count; ++i) { - convertToRgb(this, src, srcOffset, dest, destOffset, scale); - srcOffset += 3; - destOffset += 3 + alpha01; + getPage: function Catalog_getPage(pageIndex) { + if (!(pageIndex in this.pagePromises)) { + this.pagePromises[pageIndex] = this.getPageDict(pageIndex).then( + function (a) { + var dict = a[0]; + var ref = a[1]; + return this.pageFactory.createPage(pageIndex, dict, ref, + this.fontCache); + }.bind(this) + ); } + return this.pagePromises[pageIndex]; }, - getOutputLength: function CalRGBCS_getOutputLength(inputLength, alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function CalRGBCS_isDefaultDecode(decodeMap) { - return ColorSpace.isDefaultDecode(decodeMap, this.numComps); - }, - usesZeroToOneRange: true - }; - return CalRGBCS; -})(); - -// -// LabCS: Based on "PDF Reference, Sixth Ed", p.250 -// -var LabCS = (function LabCSClosure() { - function LabCS(whitePoint, blackPoint, range) { - this.name = 'Lab'; - this.numComps = 3; - this.defaultColor = new Float32Array([0, 0, 0]); - - if (!whitePoint) { - error('WhitePoint missing - required for color space Lab'); - } - blackPoint = blackPoint || [0, 0, 0]; - range = range || [-100, 100, -100, 100]; - - // Translate args to spec variables - this.XW = whitePoint[0]; - this.YW = whitePoint[1]; - this.ZW = whitePoint[2]; - this.amin = range[0]; - this.amax = range[1]; - this.bmin = range[2]; - this.bmax = range[3]; - - // These are here just for completeness - the spec doesn't offer any - // formulas that use BlackPoint in Lab - this.XB = blackPoint[0]; - this.YB = blackPoint[1]; - this.ZB = blackPoint[2]; - // Validate vars as per spec - if (this.XW < 0 || this.ZW < 0 || this.YW !== 1) { - error('Invalid WhitePoint components, no fallback available'); - } - - if (this.XB < 0 || this.YB < 0 || this.ZB < 0) { - info('Invalid BlackPoint, falling back to default'); - this.XB = this.YB = this.ZB = 0; - } - - if (this.amin > this.amax || this.bmin > this.bmax) { - info('Invalid Range, falling back to defaults'); - this.amin = -100; - this.amax = 100; - this.bmin = -100; - this.bmax = 100; - } - } + getPageDict: function Catalog_getPageDict(pageIndex) { + var capability = createPromiseCapability(); + var nodesToVisit = [this.catDict.getRaw('Pages')]; + var currentPageIndex = 0; + var xref = this.xref; + var checkAllKids = false; - // Function g(x) from spec - function fn_g(x) { - if (x >= 6 / 29) { - return x * x * x; - } else { - return (108 / 841) * (x - 4 / 29); - } - } + function next() { + while (nodesToVisit.length) { + var currentNode = nodesToVisit.pop(); - function decode(value, high1, low2, high2) { - return low2 + (value) * (high2 - low2) / (high1); - } + if (isRef(currentNode)) { + xref.fetchAsync(currentNode).then(function (obj) { + if (isDict(obj, 'Page') || (isDict(obj) && !obj.has('Kids'))) { + if (pageIndex === currentPageIndex) { + capability.resolve([obj, currentNode]); + } else { + currentPageIndex++; + next(); + } + return; + } + nodesToVisit.push(obj); + next(); + }, capability.reject); + return; + } - // If decoding is needed maxVal should be 2^bits per component - 1. - function convertToRgb(cs, src, srcOffset, maxVal, dest, destOffset) { - // XXX: Lab input is in the range of [0, 100], [amin, amax], [bmin, bmax] - // not the usual [0, 1]. If a command like setFillColor is used the src - // values will already be within the correct range. However, if we are - // converting an image we have to map the values to the correct range given - // above. - // Ls,as,bs <---> L*,a*,b* in the spec - var Ls = src[srcOffset]; - var as = src[srcOffset + 1]; - var bs = src[srcOffset + 2]; - if (maxVal !== false) { - Ls = decode(Ls, maxVal, 0, 100); - as = decode(as, maxVal, cs.amin, cs.amax); - bs = decode(bs, maxVal, cs.bmin, cs.bmax); - } + // Must be a child page dictionary. + assert( + isDict(currentNode), + 'page dictionary kid reference points to wrong type of object' + ); + var count = currentNode.get('Count'); + // If the current node doesn't have any children, avoid getting stuck + // in an empty node further down in the tree (see issue5644.pdf). + if (count === 0) { + checkAllKids = true; + } + // Skip nodes where the page can't be. + if (currentPageIndex + count <= pageIndex) { + currentPageIndex += count; + continue; + } - // Adjust limits of 'as' and 'bs' - as = as > cs.amax ? cs.amax : as < cs.amin ? cs.amin : as; - bs = bs > cs.bmax ? cs.bmax : bs < cs.bmin ? cs.bmin : bs; + var kids = currentNode.get('Kids'); + assert(isArray(kids), 'page dictionary kids object is not an array'); + if (!checkAllKids && count === kids.length) { + // Nodes that don't have the page have been skipped and this is the + // bottom of the tree which means the page requested must be a + // descendant of this pages node. Ideally we would just resolve the + // promise with the page ref here, but there is the case where more + // pages nodes could link to single a page (see issue 3666 pdf). To + // handle this push it back on the queue so if it is a pages node it + // will be descended into. + nodesToVisit = [kids[pageIndex - currentPageIndex]]; + currentPageIndex = pageIndex; + continue; + } else { + for (var last = kids.length - 1; last >= 0; last--) { + nodesToVisit.push(kids[last]); + } + } + } + capability.reject('Page index ' + pageIndex + ' not found.'); + } + next(); + return capability.promise; + }, - // Computes intermediate variables X,Y,Z as per spec - var M = (Ls + 16) / 116; - var L = M + (as / 500); - var N = M - (bs / 200); + getPageIndex: function Catalog_getPageIndex(ref) { + // The page tree nodes have the count of all the leaves below them. To get + // how many pages are before we just have to walk up the tree and keep + // adding the count of siblings to the left of the node. + var xref = this.xref; + function pagesBeforeRef(kidRef) { + var total = 0; + var parentRef; + return xref.fetchAsync(kidRef).then(function (node) { + if (!node) { + return null; + } + parentRef = node.getRaw('Parent'); + return node.getAsync('Parent'); + }).then(function (parent) { + if (!parent) { + return null; + } + return parent.getAsync('Kids'); + }).then(function (kids) { + if (!kids) { + return null; + } + var kidPromises = []; + var found = false; + for (var i = 0; i < kids.length; i++) { + var kid = kids[i]; + assert(isRef(kid), 'kids must be a ref'); + if (kid.num === kidRef.num) { + found = true; + break; + } + kidPromises.push(xref.fetchAsync(kid).then(function (kid) { + if (kid.has('Count')) { + var count = kid.get('Count'); + total += count; + } else { // page leaf node + total++; + } + })); + } + if (!found) { + error('kid ref not found in parents kids'); + } + return Promise.all(kidPromises).then(function () { + return [total, parentRef]; + }); + }); + } - var X = cs.XW * fn_g(L); - var Y = cs.YW * fn_g(M); - var Z = cs.ZW * fn_g(N); + var total = 0; + function next(ref) { + return pagesBeforeRef(ref).then(function (args) { + if (!args) { + return total; + } + var count = args[0]; + var parentRef = args[1]; + total += count; + return next(parentRef); + }); + } - var r, g, b; - // Using different conversions for D50 and D65 white points, - // per http://www.color.org/srgb.pdf - if (cs.ZW < 1) { - // Assuming D50 (X=0.9642, Y=1.00, Z=0.8249) - r = X * 3.1339 + Y * -1.6170 + Z * -0.4906; - g = X * -0.9785 + Y * 1.9160 + Z * 0.0333; - b = X * 0.0720 + Y * -0.2290 + Z * 1.4057; - } else { - // Assuming D65 (X=0.9505, Y=1.00, Z=1.0888) - r = X * 3.2406 + Y * -1.5372 + Z * -0.4986; - g = X * -0.9689 + Y * 1.8758 + Z * 0.0415; - b = X * 0.0557 + Y * -0.2040 + Z * 1.0570; + return next(ref); } - // clamp color values to [0,1] range then convert to [0,255] range. - dest[destOffset] = r <= 0 ? 0 : r >= 1 ? 255 : Math.sqrt(r) * 255 | 0; - dest[destOffset + 1] = g <= 0 ? 0 : g >= 1 ? 255 : Math.sqrt(g) * 255 | 0; - dest[destOffset + 2] = b <= 0 ? 0 : b >= 1 ? 255 : Math.sqrt(b) * 255 | 0; + }; + + return Catalog; +})(); + +var XRef = (function XRefClosure() { + function XRef(stream, password) { + this.stream = stream; + this.entries = []; + this.xrefstms = Object.create(null); + // prepare the XRef cache + this.cache = []; + this.password = password; + this.stats = { + streamTypes: [], + fontTypes: [] + }; } - LabCS.prototype = { - getRgb: ColorSpace.prototype.getRgb, - getRgbItem: function LabCS_getRgbItem(src, srcOffset, dest, destOffset) { - convertToRgb(this, src, srcOffset, false, dest, destOffset); + XRef.prototype = { + setStartXRef: function XRef_setStartXRef(startXRef) { + // Store the starting positions of xref tables as we process them + // so we can recover from missing data errors + this.startXRefQueue = [startXRef]; }, - getRgbBuffer: function LabCS_getRgbBuffer(src, srcOffset, count, - dest, destOffset, bits, - alpha01) { - var maxVal = (1 << bits) - 1; - for (var i = 0; i < count; i++) { - convertToRgb(this, src, srcOffset, maxVal, dest, destOffset); - srcOffset += 3; - destOffset += 3 + alpha01; + + parse: function XRef_parse(recoveryMode) { + var trailerDict; + if (!recoveryMode) { + trailerDict = this.readXRef(); + } else { + warn('Indexing all PDF objects'); + trailerDict = this.indexObjects(); + } + trailerDict.assignXref(this); + this.trailer = trailerDict; + var encrypt = trailerDict.get('Encrypt'); + if (encrypt) { + var ids = trailerDict.get('ID'); + var fileId = (ids && ids.length) ? ids[0] : ''; + this.encrypt = new CipherTransformFactory(encrypt, fileId, + this.password); + } + + // get the root dictionary (catalog) object + if (!(this.root = trailerDict.get('Root'))) { + error('Invalid root reference'); } }, - getOutputLength: function LabCS_getOutputLength(inputLength, alpha01) { - return (inputLength * (3 + alpha01) / 3) | 0; - }, - isPassthrough: ColorSpace.prototype.isPassthrough, - fillRgb: ColorSpace.prototype.fillRgb, - isDefaultDecode: function LabCS_isDefaultDecode(decodeMap) { - // XXX: Decoding is handled with the lab conversion because of the strange - // ranges that are used. - return true; - }, - usesZeroToOneRange: false - }; - return LabCS; -})(); -// TODO refactor to remove dependency on image.js -function _setCoreImage(coreImage_) { - coreImage = coreImage_; - PDFImage = coreImage_.PDFImage; -} -exports._setCoreImage = _setCoreImage; + processXRefTable: function XRef_processXRefTable(parser) { + if (!('tableState' in this)) { + // Stores state of the table as we process it so we can resume + // from middle of table in case of missing data error + this.tableState = { + entryNum: 0, + streamPos: parser.lexer.stream.pos, + parserBuf1: parser.buf1, + parserBuf2: parser.buf2 + }; + } -exports.ColorSpace = ColorSpace; + var obj = this.readXRefTable(parser); -// TODO refactor to remove dependency on colorspace.js -coreStream._setCoreColorSpace(exports); -})); + // Sanity check + if (!isCmd(obj, 'trailer')) { + error('Invalid XRef table: could not find trailer dictionary'); + } + // Read trailer dictionary, e.g. + // trailer + // << /Size 22 + // /Root 20R + // /Info 10R + // /ID [ <81b14aafa313db63dbd6f981e49f94f4> ] + // >> + // The parser goes through the entire stream << ... >> and provides + // a getter interface for the key-value table + var dict = parser.getObj(); + // The pdflib PDF generator can generate a nested trailer dictionary + if (!isDict(dict) && dict.dict) { + dict = dict.dict; + } + if (!isDict(dict)) { + error('Invalid XRef table: could not parse trailer dictionary'); + } + delete this.tableState; -(function (root, factory) { - { - factory((root.pdfjsCoreImage = {}), root.pdfjsSharedUtil, - root.pdfjsCorePrimitives, root.pdfjsCoreColorSpace, root.pdfjsCoreStream, - root.pdfjsCoreJpx); - } -}(this, function (exports, sharedUtil, corePrimitives, coreColorSpace, - coreStream, coreJpx) { + return dict; + }, -var ImageKind = sharedUtil.ImageKind; -var assert = sharedUtil.assert; -var error = sharedUtil.error; -var info = sharedUtil.info; -var isArray = sharedUtil.isArray; -var warn = sharedUtil.warn; -var Name = corePrimitives.Name; -var isStream = corePrimitives.isStream; -var ColorSpace = coreColorSpace.ColorSpace; -var DecodeStream = coreStream.DecodeStream; -var Stream = coreStream.Stream; -var JpegStream = coreStream.JpegStream; -var JpxImage = coreJpx.JpxImage; + readXRefTable: function XRef_readXRefTable(parser) { + // Example of cross-reference table: + // xref + // 0 1 <-- subsection header (first obj #, obj count) + // 0000000000 65535 f <-- actual object (offset, generation #, f/n) + // 23 2 <-- subsection header ... and so on ... + // 0000025518 00002 n + // 0000025635 00000 n + // trailer + // ... -var PDFImage = (function PDFImageClosure() { - /** - * Decode the image in the main thread if it supported. Resovles the promise - * when the image data is ready. - */ - function handleImageData(handler, xref, res, image) { - if (image instanceof JpegStream && image.isNativelyDecodable(xref, res)) { - // For natively supported jpegs send them to the main thread for decoding. - var dict = image.dict; - var colorSpace = dict.get('ColorSpace', 'CS'); - colorSpace = ColorSpace.parse(colorSpace, xref, res); - var numComps = colorSpace.numComps; - var decodePromise = handler.sendWithPromise('JpegDecode', - [image.getIR(), numComps]); - return decodePromise.then(function (message) { - var data = message.data; - return new Stream(data, 0, data.length, image.dict); - }); - } else { - return Promise.resolve(image); - } - } + var stream = parser.lexer.stream; + var tableState = this.tableState; + stream.pos = tableState.streamPos; + parser.buf1 = tableState.parserBuf1; + parser.buf2 = tableState.parserBuf2; - /** - * Decode and clamp a value. The formula is different from the spec because we - * don't decode to float range [0,1], we decode it in the [0,max] range. - */ - function decodeAndClamp(value, addend, coefficient, max) { - value = addend + value * coefficient; - // Clamp the value to the range - return (value < 0 ? 0 : (value > max ? max : value)); - } + // Outer loop is over subsection headers + var obj; - function PDFImage(xref, res, image, inline, smask, mask, isMask) { - this.image = image; - var dict = image.dict; - if (dict.has('Filter')) { - var filter = dict.get('Filter').name; - if (filter === 'JPXDecode') { - var jpxImage = new JpxImage(); - jpxImage.parseImageProperties(image.stream); - image.stream.reset(); - image.bitsPerComponent = jpxImage.bitsPerComponent; - image.numComps = jpxImage.componentsCount; - } else if (filter === 'JBIG2Decode') { - image.bitsPerComponent = 1; - image.numComps = 1; - } - } - // TODO cache rendered images? + while (true) { + if (!('firstEntryNum' in tableState) || !('entryCount' in tableState)) { + if (isCmd(obj = parser.getObj(), 'trailer')) { + break; + } + tableState.firstEntryNum = obj; + tableState.entryCount = parser.getObj(); + } - this.width = dict.get('Width', 'W'); - this.height = dict.get('Height', 'H'); + var first = tableState.firstEntryNum; + var count = tableState.entryCount; + if (!isInt(first) || !isInt(count)) { + error('Invalid XRef table: wrong types in subsection header'); + } + // Inner loop is over objects themselves + for (var i = tableState.entryNum; i < count; i++) { + tableState.streamPos = stream.pos; + tableState.entryNum = i; + tableState.parserBuf1 = parser.buf1; + tableState.parserBuf2 = parser.buf2; - if (this.width < 1 || this.height < 1) { - error('Invalid image width: ' + this.width + ' or height: ' + - this.height); - } + var entry = {}; + entry.offset = parser.getObj(); + entry.gen = parser.getObj(); + var type = parser.getObj(); - this.interpolate = dict.get('Interpolate', 'I') || false; - this.imageMask = dict.get('ImageMask', 'IM') || false; - this.matte = dict.get('Matte') || false; + if (isCmd(type, 'f')) { + entry.free = true; + } else if (isCmd(type, 'n')) { + entry.uncompressed = true; + } - var bitsPerComponent = image.bitsPerComponent; - if (!bitsPerComponent) { - bitsPerComponent = dict.get('BitsPerComponent', 'BPC'); - if (!bitsPerComponent) { - if (this.imageMask) { - bitsPerComponent = 1; - } else { - error('Bits per component missing in image: ' + this.imageMask); + // Validate entry obj + if (!isInt(entry.offset) || !isInt(entry.gen) || + !(entry.free || entry.uncompressed)) { + error('Invalid entry in XRef subsection: ' + first + ', ' + count); + } + + if (!this.entries[i + first]) { + this.entries[i + first] = entry; + } } + + tableState.entryNum = 0; + tableState.streamPos = stream.pos; + tableState.parserBuf1 = parser.buf1; + tableState.parserBuf2 = parser.buf2; + delete tableState.firstEntryNum; + delete tableState.entryCount; } - } - this.bpc = bitsPerComponent; - if (!this.imageMask) { - var colorSpace = dict.get('ColorSpace', 'CS'); - if (!colorSpace) { - info('JPX images (which do not require color spaces)'); - switch (image.numComps) { - case 1: - colorSpace = Name.get('DeviceGray'); - break; - case 3: - colorSpace = Name.get('DeviceRGB'); - break; - case 4: - colorSpace = Name.get('DeviceCMYK'); - break; - default: - error('JPX images with ' + this.numComps + - ' color components not supported.'); - } + // Per issue 3248: hp scanners generate bad XRef + if (first === 1 && this.entries[1] && this.entries[1].free) { + // shifting the entries + this.entries.shift(); } - this.colorSpace = ColorSpace.parse(colorSpace, xref, res); - this.numComps = this.colorSpace.numComps; - } - this.decode = dict.get('Decode', 'D'); - this.needsDecode = false; - if (this.decode && - ((this.colorSpace && !this.colorSpace.isDefaultDecode(this.decode)) || - (isMask && !ColorSpace.isDefaultDecode(this.decode, 1)))) { - this.needsDecode = true; - // Do some preprocessing to avoid more math. - var max = (1 << bitsPerComponent) - 1; - this.decodeCoefficients = []; - this.decodeAddends = []; - for (var i = 0, j = 0; i < this.decode.length; i += 2, ++j) { - var dmin = this.decode[i]; - var dmax = this.decode[i + 1]; - this.decodeCoefficients[j] = dmax - dmin; - this.decodeAddends[j] = max * dmin; + // Sanity check: as per spec, first object must be free + if (this.entries[0] && !this.entries[0].free) { + error('Invalid XRef table: unexpected first object'); } - } + return obj; + }, - if (smask) { - this.smask = new PDFImage(xref, res, smask, false); - } else if (mask) { - if (isStream(mask)) { - var maskDict = mask.dict, imageMask = maskDict.get('ImageMask', 'IM'); - if (!imageMask) { - warn('Ignoring /Mask in image without /ImageMask.'); - } else { - this.mask = new PDFImage(xref, res, mask, false, null, null, true); + processXRefStream: function XRef_processXRefStream(stream) { + if (!('streamState' in this)) { + // Stores state of the stream as we process it so we can resume + // from middle of stream in case of missing data error + var streamParameters = stream.dict; + var byteWidths = streamParameters.get('W'); + var range = streamParameters.get('Index'); + if (!range) { + range = [0, streamParameters.get('Size')]; } - } else { - // Color key mask (just an array). - this.mask = mask; + + this.streamState = { + entryRanges: range, + byteWidths: byteWidths, + entryNum: 0, + streamPos: stream.pos + }; } - } - } - /** - * Handles processing of image data and returns the Promise that is resolved - * with a PDFImage when the image is ready to be used. - */ - PDFImage.buildImage = function PDFImage_buildImage(handler, xref, - res, image, inline) { - var imagePromise = handleImageData(handler, xref, res, image); - var smaskPromise; - var maskPromise; + this.readXRefStream(stream); + delete this.streamState; - var smask = image.dict.get('SMask'); - var mask = image.dict.get('Mask'); + return stream.dict; + }, - if (smask) { - smaskPromise = handleImageData(handler, xref, res, smask); - maskPromise = Promise.resolve(null); - } else { - smaskPromise = Promise.resolve(null); - if (mask) { - if (isStream(mask)) { - maskPromise = handleImageData(handler, xref, res, mask); - } else if (isArray(mask)) { - maskPromise = Promise.resolve(mask); - } else { - warn('Unsupported mask format.'); - maskPromise = Promise.resolve(null); + readXRefStream: function XRef_readXRefStream(stream) { + var i, j; + var streamState = this.streamState; + stream.pos = streamState.streamPos; + + var byteWidths = streamState.byteWidths; + var typeFieldWidth = byteWidths[0]; + var offsetFieldWidth = byteWidths[1]; + var generationFieldWidth = byteWidths[2]; + + var entryRanges = streamState.entryRanges; + while (entryRanges.length > 0) { + var first = entryRanges[0]; + var n = entryRanges[1]; + + if (!isInt(first) || !isInt(n)) { + error('Invalid XRef range fields: ' + first + ', ' + n); } - } else { - maskPromise = Promise.resolve(null); + if (!isInt(typeFieldWidth) || !isInt(offsetFieldWidth) || + !isInt(generationFieldWidth)) { + error('Invalid XRef entry fields length: ' + first + ', ' + n); + } + for (i = streamState.entryNum; i < n; ++i) { + streamState.entryNum = i; + streamState.streamPos = stream.pos; + + var type = 0, offset = 0, generation = 0; + for (j = 0; j < typeFieldWidth; ++j) { + type = (type << 8) | stream.getByte(); + } + // if type field is absent, its default value is 1 + if (typeFieldWidth === 0) { + type = 1; + } + for (j = 0; j < offsetFieldWidth; ++j) { + offset = (offset << 8) | stream.getByte(); + } + for (j = 0; j < generationFieldWidth; ++j) { + generation = (generation << 8) | stream.getByte(); + } + var entry = {}; + entry.offset = offset; + entry.gen = generation; + switch (type) { + case 0: + entry.free = true; + break; + case 1: + entry.uncompressed = true; + break; + case 2: + break; + default: + error('Invalid XRef entry type: ' + type); + } + if (!this.entries[first + i]) { + this.entries[first + i] = entry; + } + } + + streamState.entryNum = 0; + streamState.streamPos = stream.pos; + entryRanges.splice(0, 2); } - } - return Promise.all([imagePromise, smaskPromise, maskPromise]).then( - function(results) { - var imageData = results[0]; - var smaskData = results[1]; - var maskData = results[2]; - return new PDFImage(xref, res, imageData, inline, smaskData, maskData); - }); - }; + }, - /** - * Resize an image using the nearest neighbor algorithm. Currently only - * supports one and three component images. - * @param {TypedArray} pixels The original image with one component. - * @param {Number} bpc Number of bits per component. - * @param {Number} components Number of color components, 1 or 3 is supported. - * @param {Number} w1 Original width. - * @param {Number} h1 Original height. - * @param {Number} w2 New width. - * @param {Number} h2 New height. - * @param {TypedArray} dest (Optional) The destination buffer. - * @param {Number} alpha01 (Optional) Size reserved for the alpha channel. - * @return {TypedArray} Resized image data. - */ - PDFImage.resize = function PDFImage_resize(pixels, bpc, components, - w1, h1, w2, h2, dest, alpha01) { + indexObjects: function XRef_indexObjects() { + // Simple scan through the PDF content to find objects, + // trailers and XRef streams. + var TAB = 0x9, LF = 0xA, CR = 0xD, SPACE = 0x20; + var PERCENT = 0x25, LT = 0x3C; + + function readToken(data, offset) { + var token = '', ch = data[offset]; + while (ch !== LF && ch !== CR && ch !== LT) { + if (++offset >= data.length) { + break; + } + token += String.fromCharCode(ch); + ch = data[offset]; + } + return token; + } + function skipUntil(data, offset, what) { + var length = what.length, dataLength = data.length; + var skipped = 0; + // finding byte sequence + while (offset < dataLength) { + var i = 0; + while (i < length && data[offset + i] === what[i]) { + ++i; + } + if (i >= length) { + break; // sequence found + } + offset++; + skipped++; + } + return skipped; + } + var objRegExp = /^(\d+)\s+(\d+)\s+obj\b/; + var trailerBytes = new Uint8Array([116, 114, 97, 105, 108, 101, 114]); + var startxrefBytes = new Uint8Array([115, 116, 97, 114, 116, 120, 114, + 101, 102]); + var endobjBytes = new Uint8Array([101, 110, 100, 111, 98, 106]); + var xrefBytes = new Uint8Array([47, 88, 82, 101, 102]); - if (components !== 1 && components !== 3) { - error('Unsupported component count for resizing.'); - } + // Clear out any existing entries, since they may be bogus. + this.entries.length = 0; - var length = w2 * h2 * components; - var temp = dest ? dest : (bpc <= 8 ? new Uint8Array(length) : - (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); - var xRatio = w1 / w2; - var yRatio = h1 / h2; - var i, j, py, newIndex = 0, oldIndex; - var xScaled = new Uint16Array(w2); - var w1Scanline = w1 * components; - if (alpha01 !== 1) { - alpha01 = 0; - } + var stream = this.stream; + stream.pos = 0; + var buffer = stream.getBytes(); + var position = stream.start, length = buffer.length; + var trailers = [], xrefStms = []; + while (position < length) { + var ch = buffer[position]; + if (ch === TAB || ch === LF || ch === CR || ch === SPACE) { + ++position; + continue; + } + if (ch === PERCENT) { // %-comment + do { + ++position; + if (position >= length) { + break; + } + ch = buffer[position]; + } while (ch !== LF && ch !== CR); + continue; + } + var token = readToken(buffer, position); + var m; + if (token.indexOf('xref') === 0 && + (token.length === 4 || /\s/.test(token[4]))) { + position += skipUntil(buffer, position, trailerBytes); + trailers.push(position); + position += skipUntil(buffer, position, startxrefBytes); + } else if ((m = objRegExp.exec(token))) { + if (typeof this.entries[m[1]] === 'undefined') { + this.entries[m[1]] = { + offset: position - stream.start, + gen: m[2] | 0, + uncompressed: true + }; + } + var contentLength = skipUntil(buffer, position, endobjBytes) + 7; + var content = buffer.subarray(position, position + contentLength); - for (j = 0; j < w2; j++) { - xScaled[j] = Math.floor(j * xRatio) * components; - } + // checking XRef stream suspect + // (it shall have '/XRef' and next char is not a letter) + var xrefTagOffset = skipUntil(content, 0, xrefBytes); + if (xrefTagOffset < contentLength && + content[xrefTagOffset + 5] < 64) { + xrefStms.push(position - stream.start); + this.xrefstms[position - stream.start] = 1; // Avoid recursion + } - if (components === 1) { - for (i = 0; i < h2; i++) { - py = Math.floor(i * yRatio) * w1Scanline; - for (j = 0; j < w2; j++) { - oldIndex = py + xScaled[j]; - temp[newIndex++] = pixels[oldIndex]; + position += contentLength; + } else if (token.indexOf('trailer') === 0 && + (token.length === 7 || /\s/.test(token[7]))) { + trailers.push(position); + position += skipUntil(buffer, position, startxrefBytes); + } else { + position += token.length + 1; } } - } else if (components === 3) { - for (i = 0; i < h2; i++) { - py = Math.floor(i * yRatio) * w1Scanline; - for (j = 0; j < w2; j++) { - oldIndex = py + xScaled[j]; - temp[newIndex++] = pixels[oldIndex++]; - temp[newIndex++] = pixels[oldIndex++]; - temp[newIndex++] = pixels[oldIndex++]; - newIndex += alpha01; + // reading XRef streams + var i, ii; + for (i = 0, ii = xrefStms.length; i < ii; ++i) { + this.startXRefQueue.push(xrefStms[i]); + this.readXRef(/* recoveryMode */ true); + } + // finding main trailer + var dict; + for (i = 0, ii = trailers.length; i < ii; ++i) { + stream.pos = trailers[i]; + var parser = new Parser(new Lexer(stream), true, this); + var obj = parser.getObj(); + if (!isCmd(obj, 'trailer')) { + continue; + } + // read the trailer dictionary + if (!isDict(dict = parser.getObj())) { + continue; + } + // taking the first one with 'ID' + if (dict.has('ID')) { + return dict; } } - } - return temp; - }; - - PDFImage.createMask = - function PDFImage_createMask(imgArray, width, height, - imageIsFromDecodeStream, inverseDecode) { - - // |imgArray| might not contain full data for every pixel of the mask, so - // we need to distinguish between |computedLength| and |actualLength|. - // In particular, if inverseDecode is true, then the array we return must - // have a length of |computedLength|. + // no tailer with 'ID', taking last one (if exists) + if (dict) { + return dict; + } + // nothing helps + // calling error() would reject worker with an UnknownErrorException. + throw new InvalidPDFException('Invalid PDF structure'); + }, - var computedLength = ((width + 7) >> 3) * height; - var actualLength = imgArray.byteLength; - var haveFullData = computedLength === actualLength; - var data, i; + readXRef: function XRef_readXRef(recoveryMode) { + var stream = this.stream; - if (imageIsFromDecodeStream && (!inverseDecode || haveFullData)) { - // imgArray came from a DecodeStream and its data is in an appropriate - // form, so we can just transfer it. - data = imgArray; - } else if (!inverseDecode) { - data = new Uint8Array(actualLength); - data.set(imgArray); - } else { - data = new Uint8Array(computedLength); - data.set(imgArray); - for (i = actualLength; i < computedLength; i++) { - data[i] = 0xff; - } - } + try { + while (this.startXRefQueue.length) { + var startXRef = this.startXRefQueue[0]; - // If necessary, invert the original mask data (but not any extra we might - // have added above). It's safe to modify the array -- whether it's the - // original or a copy, we're about to transfer it anyway, so nothing else - // in this thread can be relying on its contents. - if (inverseDecode) { - for (i = 0; i < actualLength; i++) { - data[i] = ~data[i]; - } - } + stream.pos = startXRef + stream.start; - return {data: data, width: width, height: height}; - }; + var parser = new Parser(new Lexer(stream), true, this); + var obj = parser.getObj(); + var dict; - PDFImage.prototype = { - get drawWidth() { - return Math.max(this.width, - this.smask && this.smask.width || 0, - this.mask && this.mask.width || 0); - }, + // Get dictionary + if (isCmd(obj, 'xref')) { + // Parse end-of-file XRef + dict = this.processXRefTable(parser); + if (!this.topDict) { + this.topDict = dict; + } - get drawHeight() { - return Math.max(this.height, - this.smask && this.smask.height || 0, - this.mask && this.mask.height || 0); - }, + // Recursively get other XRefs 'XRefStm', if any + obj = dict.get('XRefStm'); + if (isInt(obj)) { + var pos = obj; + // ignore previously loaded xref streams + // (possible infinite recursion) + if (!(pos in this.xrefstms)) { + this.xrefstms[pos] = 1; + this.startXRefQueue.push(pos); + } + } + } else if (isInt(obj)) { + // Parse in-stream XRef + if (!isInt(parser.getObj()) || + !isCmd(parser.getObj(), 'obj') || + !isStream(obj = parser.getObj())) { + error('Invalid XRef stream'); + } + dict = this.processXRefStream(obj); + if (!this.topDict) { + this.topDict = dict; + } + if (!dict) { + error('Failed to read XRef stream'); + } + } else { + error('Invalid XRef stream header'); + } - decodeBuffer: function PDFImage_decodeBuffer(buffer) { - var bpc = this.bpc; - var numComps = this.numComps; + // Recursively get previous dictionary, if any + obj = dict.get('Prev'); + if (isInt(obj)) { + this.startXRefQueue.push(obj); + } else if (isRef(obj)) { + // The spec says Prev must not be a reference, i.e. "/Prev NNN" + // This is a fallback for non-compliant PDFs, i.e. "/Prev NNN 0 R" + this.startXRefQueue.push(obj.num); + } - var decodeAddends = this.decodeAddends; - var decodeCoefficients = this.decodeCoefficients; - var max = (1 << bpc) - 1; - var i, ii; + this.startXRefQueue.shift(); + } - if (bpc === 1) { - // If the buffer needed decode that means it just needs to be inverted. - for (i = 0, ii = buffer.length; i < ii; i++) { - buffer[i] = +!(buffer[i]); + return this.topDict; + } catch (e) { + if (e instanceof MissingDataException) { + throw e; } - return; + info('(while reading XRef): ' + e); } - var index = 0; - for (i = 0, ii = this.width * this.height; i < ii; i++) { - for (var j = 0; j < numComps; j++) { - buffer[index] = decodeAndClamp(buffer[index], decodeAddends[j], - decodeCoefficients[j], max); - index++; - } + + if (recoveryMode) { + return; } + throw new XRefParseException(); }, - getComponents: function PDFImage_getComponents(buffer) { - var bpc = this.bpc; - - // This image doesn't require any extra work. - if (bpc === 8) { - return buffer; + getEntry: function XRef_getEntry(i) { + var xrefEntry = this.entries[i]; + if (xrefEntry && !xrefEntry.free && xrefEntry.offset) { + return xrefEntry; } + return null; + }, - var width = this.width; - var height = this.height; - var numComps = this.numComps; - - var length = width * height * numComps; - var bufferPos = 0; - var output = (bpc <= 8 ? new Uint8Array(length) : - (bpc <= 16 ? new Uint16Array(length) : new Uint32Array(length))); - var rowComps = width * numComps; + fetchIfRef: function XRef_fetchIfRef(obj) { + if (!isRef(obj)) { + return obj; + } + return this.fetch(obj); + }, - var max = (1 << bpc) - 1; - var i = 0, ii, buf; + fetch: function XRef_fetch(ref, suppressEncryption) { + assert(isRef(ref), 'ref object is not a reference'); + var num = ref.num; + if (num in this.cache) { + var cacheEntry = this.cache[num]; + return cacheEntry; + } - if (bpc === 1) { - // Optimization for reading 1 bpc images. - var mask, loop1End, loop2End; - for (var j = 0; j < height; j++) { - loop1End = i + (rowComps & ~7); - loop2End = i + rowComps; + var xrefEntry = this.getEntry(num); - // unroll loop for all full bytes - while (i < loop1End) { - buf = buffer[bufferPos++]; - output[i] = (buf >> 7) & 1; - output[i + 1] = (buf >> 6) & 1; - output[i + 2] = (buf >> 5) & 1; - output[i + 3] = (buf >> 4) & 1; - output[i + 4] = (buf >> 3) & 1; - output[i + 5] = (buf >> 2) & 1; - output[i + 6] = (buf >> 1) & 1; - output[i + 7] = buf & 1; - i += 8; - } + // the referenced entry can be free + if (xrefEntry === null) { + return (this.cache[num] = null); + } - // handle remaing bits - if (i < loop2End) { - buf = buffer[bufferPos++]; - mask = 128; - while (i < loop2End) { - output[i++] = +!!(buf & mask); - mask >>= 1; - } - } - } + if (xrefEntry.uncompressed) { + xrefEntry = this.fetchUncompressed(ref, xrefEntry, suppressEncryption); } else { - // The general case that handles all other bpc values. - var bits = 0; - buf = 0; - for (i = 0, ii = length; i < ii; ++i) { - if (i % rowComps === 0) { - buf = 0; - bits = 0; - } + xrefEntry = this.fetchCompressed(xrefEntry, suppressEncryption); + } + if (isDict(xrefEntry)){ + xrefEntry.objId = ref.toString(); + } else if (isStream(xrefEntry)) { + xrefEntry.dict.objId = ref.toString(); + } + return xrefEntry; + }, - while (bits < bpc) { - buf = (buf << 8) | buffer[bufferPos++]; - bits += 8; + fetchUncompressed: function XRef_fetchUncompressed(ref, xrefEntry, + suppressEncryption) { + var gen = ref.gen; + var num = ref.num; + if (xrefEntry.gen !== gen) { + error('inconsistent generation in XRef'); + } + var stream = this.stream.makeSubStream(xrefEntry.offset + + this.stream.start); + var parser = new Parser(new Lexer(stream), true, this); + var obj1 = parser.getObj(); + var obj2 = parser.getObj(); + var obj3 = parser.getObj(); + if (!isInt(obj1) || parseInt(obj1, 10) !== num || + !isInt(obj2) || parseInt(obj2, 10) !== gen || + !isCmd(obj3)) { + error('bad XRef entry'); + } + if (!isCmd(obj3, 'obj')) { + // some bad PDFs use "obj1234" and really mean 1234 + if (obj3.cmd.indexOf('obj') === 0) { + num = parseInt(obj3.cmd.substring(3), 10); + if (!isNaN(num)) { + return num; } - - var remainingBits = bits - bpc; - var value = buf >> remainingBits; - output[i] = (value < 0 ? 0 : (value > max ? max : value)); - buf = buf & ((1 << remainingBits) - 1); - bits = remainingBits; } + error('bad XRef entry'); } - return output; + if (this.encrypt && !suppressEncryption) { + xrefEntry = parser.getObj(this.encrypt.createCipherTransform(num, gen)); + } else { + xrefEntry = parser.getObj(); + } + if (!isStream(xrefEntry)) { + this.cache[num] = xrefEntry; + } + return xrefEntry; }, - fillOpacity: function PDFImage_fillOpacity(rgbaBuf, width, height, - actualHeight, image) { - var smask = this.smask; - var mask = this.mask; - var alphaBuf, sw, sh, i, ii, j; - - if (smask) { - sw = smask.width; - sh = smask.height; - alphaBuf = new Uint8Array(sw * sh); - smask.fillGrayBuffer(alphaBuf); - if (sw !== width || sh !== height) { - alphaBuf = PDFImage.resize(alphaBuf, smask.bpc, 1, sw, sh, width, - height); + fetchCompressed: function XRef_fetchCompressed(xrefEntry, + suppressEncryption) { + var tableOffset = xrefEntry.offset; + var stream = this.fetch(new Ref(tableOffset, 0)); + if (!isStream(stream)) { + error('bad ObjStm stream'); + } + var first = stream.dict.get('First'); + var n = stream.dict.get('N'); + if (!isInt(first) || !isInt(n)) { + error('invalid first and n parameters for ObjStm stream'); + } + var parser = new Parser(new Lexer(stream), false, this); + parser.allowStreams = true; + var i, entries = [], num, nums = []; + // read the object numbers to populate cache + for (i = 0; i < n; ++i) { + num = parser.getObj(); + if (!isInt(num)) { + error('invalid object number in the ObjStm stream: ' + num); } - } else if (mask) { - if (mask instanceof PDFImage) { - sw = mask.width; - sh = mask.height; - alphaBuf = new Uint8Array(sw * sh); - mask.numComps = 1; - mask.fillGrayBuffer(alphaBuf); - - // Need to invert values in rgbaBuf - for (i = 0, ii = sw * sh; i < ii; ++i) { - alphaBuf[i] = 255 - alphaBuf[i]; - } - - if (sw !== width || sh !== height) { - alphaBuf = PDFImage.resize(alphaBuf, mask.bpc, 1, sw, sh, width, - height); - } - } else if (isArray(mask)) { - // Color key mask: if any of the compontents are outside the range - // then they should be painted. - alphaBuf = new Uint8Array(width * height); - var numComps = this.numComps; - for (i = 0, ii = width * height; i < ii; ++i) { - var opacity = 0; - var imageOffset = i * numComps; - for (j = 0; j < numComps; ++j) { - var color = image[imageOffset + j]; - var maskOffset = j * 2; - if (color < mask[maskOffset] || color > mask[maskOffset + 1]) { - opacity = 255; - break; - } - } - alphaBuf[i] = opacity; - } - } else { - error('Unknown mask format.'); + nums.push(num); + var offset = parser.getObj(); + if (!isInt(offset)) { + error('invalid object offset in the ObjStm stream: ' + offset); } } - - if (alphaBuf) { - for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { - rgbaBuf[j] = alphaBuf[i]; + // read stream objects for cache + for (i = 0; i < n; ++i) { + entries.push(parser.getObj()); + num = nums[i]; + var entry = this.entries[num]; + if (entry && entry.offset === tableOffset && entry.gen === i) { + this.cache[num] = entries[i]; } - } else { - // No mask. - for (i = 0, j = 3, ii = width * actualHeight; i < ii; ++i, j += 4) { - rgbaBuf[j] = 255; + } + xrefEntry = entries[xrefEntry.gen]; + if (xrefEntry === undefined) { + error('bad XRef entry for compressed object'); + } + return xrefEntry; + }, + + fetchIfRefAsync: function XRef_fetchIfRefAsync(obj) { + if (!isRef(obj)) { + return Promise.resolve(obj); + } + return this.fetchAsync(obj); + }, + + fetchAsync: function XRef_fetchAsync(ref, suppressEncryption) { + var streamManager = this.stream.manager; + var xref = this; + return new Promise(function tryFetch(resolve, reject) { + try { + resolve(xref.fetch(ref, suppressEncryption)); + } catch (e) { + if (e instanceof MissingDataException) { + streamManager.requestRange(e.begin, e.end).then(function () { + tryFetch(resolve, reject); + }, reject); + return; + } + reject(e); } - } + }); }, - undoPreblend: function PDFImage_undoPreblend(buffer, width, height) { - var matte = this.smask && this.smask.matte; - if (!matte) { - return; + getCatalogObj: function XRef_getCatalogObj() { + return this.root; + } + }; + + return XRef; +})(); + +/** + * A NameTree/NumberTree is like a Dict but has some advantageous properties, + * see the specification (7.9.6 and 7.9.7) for additional details. + * TODO: implement all the Dict functions and make this more efficient. + */ +var NameOrNumberTree = (function NameOrNumberTreeClosure() { + function NameOrNumberTree(root, xref) { + throw new Error('Cannot initialize NameOrNumberTree.'); + } + + NameOrNumberTree.prototype = { + getAll: function NameOrNumberTree_getAll() { + var dict = Object.create(null); + if (!this.root) { + return dict; } - var matteRgb = this.colorSpace.getRgb(matte, 0); - var matteR = matteRgb[0]; - var matteG = matteRgb[1]; - var matteB = matteRgb[2]; - var length = width * height * 4; - var r, g, b; - for (var i = 0; i < length; i += 4) { - var alpha = buffer[i + 3]; - if (alpha === 0) { - // according formula we have to get Infinity in all components - // making it white (typical paper color) should be okay - buffer[i] = 255; - buffer[i + 1] = 255; - buffer[i + 2] = 255; + var xref = this.xref; + // Reading Name/Number tree. + var processed = new RefSet(); + processed.put(this.root); + var queue = [this.root]; + while (queue.length > 0) { + var i, n; + var obj = xref.fetchIfRef(queue.shift()); + if (!isDict(obj)) { continue; } - var k = 255 / alpha; - r = (buffer[i] - matteR) * k + matteR; - g = (buffer[i + 1] - matteG) * k + matteG; - b = (buffer[i + 2] - matteB) * k + matteB; - buffer[i] = r <= 0 ? 0 : r >= 255 ? 255 : r | 0; - buffer[i + 1] = g <= 0 ? 0 : g >= 255 ? 255 : g | 0; - buffer[i + 2] = b <= 0 ? 0 : b >= 255 ? 255 : b | 0; + if (obj.has('Kids')) { + var kids = obj.get('Kids'); + for (i = 0, n = kids.length; i < n; i++) { + var kid = kids[i]; + assert(!processed.has(kid), + 'Duplicate entry in "' + this._type + '" tree.'); + queue.push(kid); + processed.put(kid); + } + continue; + } + var entries = obj.get(this._type); + if (isArray(entries)) { + for (i = 0, n = entries.length; i < n; i += 2) { + dict[xref.fetchIfRef(entries[i])] = xref.fetchIfRef(entries[i + 1]); + } + } } + return dict; }, - createImageData: function PDFImage_createImageData(forceRGBA) { - var drawWidth = this.drawWidth; - var drawHeight = this.drawHeight; - var imgData = { // other fields are filled in below - width: drawWidth, - height: drawHeight - }; + get: function NameOrNumberTree_get(key) { + if (!this.root) { + return null; + } - var numComps = this.numComps; - var originalWidth = this.width; - var originalHeight = this.height; - var bpc = this.bpc; + var xref = this.xref; + var kidsOrEntries = xref.fetchIfRef(this.root); + var loopCount = 0; + var MAX_LEVELS = 10; + var l, r, m; - // Rows start at byte boundary. - var rowBytes = (originalWidth * numComps * bpc + 7) >> 3; - var imgArray; + // Perform a binary search to quickly find the entry that + // contains the key we are looking for. + while (kidsOrEntries.has('Kids')) { + if (++loopCount > MAX_LEVELS) { + warn('Search depth limit reached for "' + this._type + '" tree.'); + return null; + } - if (!forceRGBA) { - // If it is a 1-bit-per-pixel grayscale (i.e. black-and-white) image - // without any complications, we pass a same-sized copy to the main - // thread rather than expanding by 32x to RGBA form. This saves *lots* - // of memory for many scanned documents. It's also much faster. - // - // Similarly, if it is a 24-bit-per pixel RGB image without any - // complications, we avoid expanding by 1.333x to RGBA form. - var kind; - if (this.colorSpace.name === 'DeviceGray' && bpc === 1) { - kind = ImageKind.GRAYSCALE_1BPP; - } else if (this.colorSpace.name === 'DeviceRGB' && bpc === 8 && - !this.needsDecode) { - kind = ImageKind.RGB_24BPP; + var kids = kidsOrEntries.get('Kids'); + if (!isArray(kids)) { + return null; } - if (kind && !this.smask && !this.mask && - drawWidth === originalWidth && drawHeight === originalHeight) { - imgData.kind = kind; - imgArray = this.getImageBytes(originalHeight * rowBytes); - // If imgArray came from a DecodeStream, we're safe to transfer it - // (and thus detach its underlying buffer) because it will constitute - // the entire DecodeStream's data. But if it came from a Stream, we - // need to copy it because it'll only be a portion of the Stream's - // data, and the rest will be read later on. - if (this.image instanceof DecodeStream) { - imgData.data = imgArray; + l = 0; + r = kids.length - 1; + while (l <= r) { + m = (l + r) >> 1; + var kid = xref.fetchIfRef(kids[m]); + var limits = kid.get('Limits'); + + if (key < xref.fetchIfRef(limits[0])) { + r = m - 1; + } else if (key > xref.fetchIfRef(limits[1])) { + l = m + 1; } else { - var newArray = new Uint8Array(imgArray.length); - newArray.set(imgArray); - imgData.data = newArray; - } - if (this.needsDecode) { - // Invert the buffer (which must be grayscale if we reached here). - assert(kind === ImageKind.GRAYSCALE_1BPP); - var buffer = imgData.data; - for (var i = 0, ii = buffer.length; i < ii; i++) { - buffer[i] ^= 0xff; - } + kidsOrEntries = xref.fetchIfRef(kids[m]); + break; } - return imgData; } - if (this.image instanceof JpegStream && !this.smask && !this.mask && - (this.colorSpace.name === 'DeviceGray' || - this.colorSpace.name === 'DeviceRGB' || - this.colorSpace.name === 'DeviceCMYK')) { - imgData.kind = ImageKind.RGB_24BPP; - imgData.data = this.getImageBytes(originalHeight * rowBytes, - drawWidth, drawHeight, true); - return imgData; + if (l > r) { + return null; } } - imgArray = this.getImageBytes(originalHeight * rowBytes); - // imgArray can be incomplete (e.g. after CCITT fax encoding). - var actualHeight = 0 | (imgArray.length / rowBytes * - drawHeight / originalHeight); + // If we get here, then we have found the right entry. Now go through the + // entries in the dictionary until we find the key we're looking for. + var entries = kidsOrEntries.get(this._type); + if (isArray(entries)) { + // Perform a binary search to reduce the lookup time. + l = 0; + r = entries.length - 2; + while (l <= r) { + // Check only even indices (0, 2, 4, ...) because the + // odd indices contain the actual data. + m = (l + r) & ~1; + var currentKey = xref.fetchIfRef(entries[m]); + if (key < currentKey) { + r = m - 2; + } else if (key > currentKey) { + l = m + 2; + } else { + return xref.fetchIfRef(entries[m + 1]); + } + } + } + return null; + } + }; + return NameOrNumberTree; +})(); - var comps = this.getComponents(imgArray); +var NameTree = (function NameTreeClosure() { + function NameTree(root, xref) { + this.root = root; + this.xref = xref; + this._type = 'Names'; + } - // If opacity data is present, use RGBA_32BPP form. Otherwise, use the - // more compact RGB_24BPP form if allowable. - var alpha01, maybeUndoPreblend; - if (!forceRGBA && !this.smask && !this.mask) { - imgData.kind = ImageKind.RGB_24BPP; - imgData.data = new Uint8Array(drawWidth * drawHeight * 3); - alpha01 = 0; - maybeUndoPreblend = false; - } else { - imgData.kind = ImageKind.RGBA_32BPP; - imgData.data = new Uint8Array(drawWidth * drawHeight * 4); - alpha01 = 1; - maybeUndoPreblend = true; + Util.inherit(NameTree, NameOrNumberTree, {}); - // Color key masking (opacity) must be performed before decoding. - this.fillOpacity(imgData.data, drawWidth, drawHeight, actualHeight, - comps); + return NameTree; +})(); + +var NumberTree = (function NumberTreeClosure() { + function NumberTree(root, xref) { + this.root = root; + this.xref = xref; + this._type = 'Nums'; + } + + Util.inherit(NumberTree, NameOrNumberTree, {}); + + return NumberTree; +})(); + +/** + * "A PDF file can refer to the contents of another file by using a File + * Specification (PDF 1.1)", see the spec (7.11) for more details. + * NOTE: Only embedded files are supported (as part of the attachments support) + * TODO: support the 'URL' file system (with caching if !/V), portable + * collections attributes and related files (/RF) + */ +var FileSpec = (function FileSpecClosure() { + function FileSpec(root, xref) { + if (!root || !isDict(root)) { + return; + } + this.xref = xref; + this.root = root; + if (root.has('FS')) { + this.fs = root.get('FS'); + } + this.description = root.has('Desc') ? + stringToPDFString(root.get('Desc')) : + ''; + if (root.has('RF')) { + warn('Related file specifications are not supported'); + } + this.contentAvailable = true; + if (!root.has('EF')) { + this.contentAvailable = false; + warn('Non-embedded file specifications are not supported'); + } + } + + function pickPlatformItem(dict) { + // Look for the filename in this order: + // UF, F, Unix, Mac, DOS + if (dict.has('UF')) { + return dict.get('UF'); + } else if (dict.has('F')) { + return dict.get('F'); + } else if (dict.has('Unix')) { + return dict.get('Unix'); + } else if (dict.has('Mac')) { + return dict.get('Mac'); + } else if (dict.has('DOS')) { + return dict.get('DOS'); + } else { + return null; + } + } + + FileSpec.prototype = { + get filename() { + if (!this._filename && this.root) { + var filename = pickPlatformItem(this.root) || 'unnamed'; + this._filename = stringToPDFString(filename). + replace(/\\\\/g, '\\'). + replace(/\\\//g, '/'). + replace(/\\/g, '/'); + } + return this._filename; + }, + get content() { + if (!this.contentAvailable) { + return null; + } + if (!this.contentRef && this.root) { + this.contentRef = pickPlatformItem(this.root.get('EF')); + } + var content = null; + if (this.contentRef) { + var xref = this.xref; + var fileObj = xref.fetchIfRef(this.contentRef); + if (fileObj && isStream(fileObj)) { + content = fileObj.getBytes(); + } else { + warn('Embedded file specification points to non-existing/invalid ' + + 'content'); + } + } else { + warn('Embedded file specification does not have a content'); } + return content; + }, + get serializable() { + return { + filename: this.filename, + content: this.content + }; + } + }; + return FileSpec; +})(); - if (this.needsDecode) { - this.decodeBuffer(comps); +/** + * A helper for loading missing data in object graphs. It traverses the graph + * depth first and queues up any objects that have missing data. Once it has + * has traversed as many objects that are available it attempts to bundle the + * missing data requests and then resume from the nodes that weren't ready. + * + * NOTE: It provides protection from circular references by keeping track of + * of loaded references. However, you must be careful not to load any graphs + * that have references to the catalog or other pages since that will cause the + * entire PDF document object graph to be traversed. + */ +var ObjectLoader = (function() { + function mayHaveChildren(value) { + return isRef(value) || isDict(value) || isArray(value) || isStream(value); + } + + function addChildren(node, nodesToVisit) { + var value; + if (isDict(node) || isStream(node)) { + var map; + if (isDict(node)) { + map = node.map; + } else { + map = node.dict.map; } - this.colorSpace.fillRgb(imgData.data, originalWidth, originalHeight, - drawWidth, drawHeight, actualHeight, bpc, comps, - alpha01); - if (maybeUndoPreblend) { - this.undoPreblend(imgData.data, drawWidth, actualHeight); + for (var key in map) { + value = map[key]; + if (mayHaveChildren(value)) { + nodesToVisit.push(value); + } + } + } else if (isArray(node)) { + for (var i = 0, ii = node.length; i < ii; i++) { + value = node[i]; + if (mayHaveChildren(value)) { + nodesToVisit.push(value); + } } + } + } - return imgData; - }, + function ObjectLoader(obj, keys, xref) { + this.obj = obj; + this.keys = keys; + this.xref = xref; + this.refSet = null; + this.capability = null; + } - fillGrayBuffer: function PDFImage_fillGrayBuffer(buffer) { - var numComps = this.numComps; - if (numComps !== 1) { - error('Reading gray scale from a color image: ' + numComps); + ObjectLoader.prototype = { + load: function ObjectLoader_load() { + var keys = this.keys; + this.capability = createPromiseCapability(); + // Don't walk the graph if all the data is already loaded. + if (!(this.xref.stream instanceof ChunkedStream) || + this.xref.stream.getMissingChunks().length === 0) { + this.capability.resolve(); + return this.capability.promise; } - var width = this.width; - var height = this.height; - var bpc = this.bpc; + this.refSet = new RefSet(); + // Setup the initial nodes to visit. + var nodesToVisit = []; + for (var i = 0; i < keys.length; i++) { + nodesToVisit.push(this.obj[keys[i]]); + } - // rows start at byte boundary - var rowBytes = (width * numComps * bpc + 7) >> 3; - var imgArray = this.getImageBytes(height * rowBytes); + this._walk(nodesToVisit); + return this.capability.promise; + }, - var comps = this.getComponents(imgArray); - var i, length; + _walk: function ObjectLoader_walk(nodesToVisit) { + var nodesToRevisit = []; + var pendingRequests = []; + // DFS walk of the object graph. + while (nodesToVisit.length) { + var currentNode = nodesToVisit.pop(); - if (bpc === 1) { - // inline decoding (= inversion) for 1 bpc images - length = width * height; - if (this.needsDecode) { - // invert and scale to {0, 255} - for (i = 0; i < length; ++i) { - buffer[i] = (comps[i] - 1) & 255; + // Only references or chunked streams can cause missing data exceptions. + if (isRef(currentNode)) { + // Skip nodes that have already been visited. + if (this.refSet.has(currentNode)) { + continue; } - } else { - // scale to {0, 255} - for (i = 0; i < length; ++i) { - buffer[i] = (-comps[i]) & 255; + try { + var ref = currentNode; + this.refSet.put(ref); + currentNode = this.xref.fetch(currentNode); + } catch (e) { + if (!(e instanceof MissingDataException)) { + throw e; + } + nodesToRevisit.push(currentNode); + pendingRequests.push({ begin: e.begin, end: e.end }); + } + } + if (currentNode && currentNode.getBaseStreams) { + var baseStreams = currentNode.getBaseStreams(); + var foundMissingData = false; + for (var i = 0; i < baseStreams.length; i++) { + var stream = baseStreams[i]; + if (stream.getMissingChunks && stream.getMissingChunks().length) { + foundMissingData = true; + pendingRequests.push({ + begin: stream.start, + end: stream.end + }); + } + } + if (foundMissingData) { + nodesToRevisit.push(currentNode); } } - return; - } - if (this.needsDecode) { - this.decodeBuffer(comps); - } - length = width * height; - // we aren't using a colorspace so we need to scale the value - var scale = 255 / ((1 << bpc) - 1); - for (i = 0; i < length; ++i) { - buffer[i] = (scale * comps[i]) | 0; + addChildren(currentNode, nodesToVisit); } - }, - getImageBytes: function PDFImage_getImageBytes(length, - drawWidth, drawHeight, - forceRGB) { - this.image.reset(); - this.image.drawWidth = drawWidth || this.width; - this.image.drawHeight = drawHeight || this.height; - this.image.forceRGB = !!forceRGB; - return this.image.getBytes(length); + if (pendingRequests.length) { + this.xref.stream.manager.requestRanges(pendingRequests).then( + function pendingRequestCallback() { + nodesToVisit = nodesToRevisit; + for (var i = 0; i < nodesToRevisit.length; i++) { + var node = nodesToRevisit[i]; + // Remove any reference nodes from the currrent refset so they + // aren't skipped when we revist them. + if (isRef(node)) { + this.refSet.remove(node); + } + } + this._walk(nodesToVisit); + }.bind(this), this.capability.reject); + return; + } + // Everything is loaded. + this.refSet = null; + this.capability.resolve(); } }; - return PDFImage; -})(); -exports.PDFImage = PDFImage; + return ObjectLoader; +})(); -// TODO refactor to remove dependency on colorspace.js -coreColorSpace._setCoreImage(exports); +exports.Catalog = Catalog; +exports.ObjectLoader = ObjectLoader; +exports.XRef = XRef; })); diff --git a/package.json b/package.json index 5b06e8a0d..69e1bb319 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "pdfjs-dist", - "version": "1.4.76", + "version": "1.4.79", "main": "build/pdf.js", "description": "Generic build of Mozilla's PDF.js library.", "keywords": [