Make getOperatorList() calls independent and merge queues at end

src/core.js

@@ -18,7 +18,7 @@
            isArrayBuffer, isDict, isName, isStream, isString, Lexer,
            Linearization, NullStream, PartialEvaluator, shadow, Stream,
            StreamsSequenceStream, stringToPDFString, TODO, Util, warn, XRef,
-           MissingDataException */
+           MissingDataException, PDFJS */
 
 'use strict';
 
@@ -60,7 +60,8 @@ var Page = (function PageClosure() {
     return appearance;
   }
 
-  function Page(xref, pageIndex, pageDict, ref) {
+  function Page(pdfManager, xref, pageIndex, pageDict, ref) {
+    this.pdfManager = pdfManager;
     this.pageIndex = pageIndex;
     this.pageDict = pageDict;
     this.xref = xref;
@@ -146,28 +147,71 @@ var Page = (function PageClosure() {
       }
       return content;
     },
-    getOperatorList: function Page_getOperatorList(handler, dependency) {
+    getOperatorList: function Page_getOperatorList(handler) {
-      var xref = this.xref;
+      var self = this;
-      var contentStream = this.getContentStream();
+      var promise = new PDFJS.Promise();
-      var resources = this.resources;
+
-      var pe = this.pe = new PartialEvaluator(
+      var pageListPromise = new PDFJS.Promise();
-                                xref, handler, this.pageIndex,
+      var annotationListPromise = new PDFJS.Promise();
-                                'p' + this.pageIndex + '_');
+
-
+      var pdfManager = this.pdfManager;
-      var list = pe.getOperatorList(contentStream, resources, dependency);
+      var contentStreamPromise = pdfManager.ensure(this, 'getContentStream',
+                                                   []);
+      var resourcesPromise = pdfManager.ensure(this, 'resources');
+      var dataPromises = PDFJS.Promise.all(
+          [contentStreamPromise, resourcesPromise]);
+      dataPromises.then(function(data) {
+        var contentStream = data[0];
+        var resources = data[1];
+        var pe = self.pe = new PartialEvaluator(
+                                  self.xref, handler, self.pageIndex,
+                                  'p' + self.pageIndex + '_');
+
+        pdfManager.ensure(pe, 'getOperatorList',
+                          [contentStream, resources]).then(
+          function(opListPromise) {
+            opListPromise.then(function(data) {
+              pageListPromise.resolve(data);
+            });
+          }
+        );
+      });
 
-      var annotations = this.getAnnotationsForDraw();
+      pdfManager.ensure(this, 'getAnnotationsForDraw', []).then(
+        function(annotations) {
           var annotationEvaluator = new PartialEvaluator(
-        xref, handler, this.pageIndex,
+            self.xref, handler, self.pageIndex,
-        'p' + this.pageIndex + '_annotation');
+            'p' + self.pageIndex + '_annotation');
-      var annotationsList = annotationEvaluator.getAnnotationsOperatorList(
+
-          annotations, dependency);
+          pdfManager.ensure(annotationEvaluator, 'getAnnotationsOperatorList',
+                            [annotations]).then(
+            function(opListPromise) {
+              opListPromise.then(function(data) {
+                annotationListPromise.resolve(data);
+              });
+            }
+          );
+        }
+      );
+
+      PDFJS.Promise.all([pageListPromise, annotationListPromise]).then(
+        function(datas) {
+          var pageData = datas[0];
+          var pageQueue = pageData.queue;
+          var annotationData = datas[1];
+          var annotationQueue = annotationData.queue;
+          Util.concatenateToArray(pageQueue.fnArray, annotationQueue.fnArray);
+          Util.concatenateToArray(pageQueue.argsArray,
+                                  annotationQueue.argsArray);
+          PartialEvaluator.optimizeQueue(pageQueue);
+          Util.extendObj(pageData.dependencies, annotationData.dependencies);
+
+          promise.resolve(pageData);
+        }
+      );
 
-      Util.concatenateToArray(list.fnArray, annotationsList.fnArray);
+      return promise;
-      Util.concatenateToArray(list.argsArray, annotationsList.argsArray);
 
-      pe.optimizeQueue(list);
-      return list;
     },
     extractTextContent: function Page_extractTextContent() {
       var handler = {
@@ -175,14 +219,39 @@ var Page = (function PageClosure() {
         send: function nullHandlerSend() {}
       };
 
-      var xref = this.xref;
+      var self = this;
-      var contentStream = this.getContentStream();
+
-      var resources = xref.fetchIfRef(this.resources);
+      var textContentPromise = new PDFJS.Promise();
+
+      var pdfManager = this.pdfManager;
+      var contentStreamPromise = pdfManager.ensure(this, 'getContentStream',
+                                                   []);
+      var resourcesPromise = new PDFJS.Promise();
+      pdfManager.ensure(this, 'resources').then(function(resources) {
+        pdfManager.ensure(self.xref, 'fetchIfRef', [resources]).then(
+          function(resources) {
+            resourcesPromise.resolve(resources);
+          }
+        );
+      });
 
+      var dataPromises = PDFJS.Promise.all([contentStreamPromise,
+                                            resourcesPromise]);
+      dataPromises.then(function(data) {
+        var contentStream = data[0];
+        var resources = data[1];
         var pe = new PartialEvaluator(
-                     xref, handler, this.pageIndex,
+                       self.xref, handler, self.pageIndex,
-                     'p' + this.pageIndex + '_');
+                       'p' + self.pageIndex + '_');
-      return pe.getTextContent(contentStream, resources);
+
+        pe.getTextContent(contentStream, resources).then(function(bidiTexts) {
+          textContentPromise.resolve({
+            bidiTexts: bidiTexts
+          });
+        });
+      });
+
+      return textContentPromise;
     },
     getLinks: function Page_getLinks() {
       var links = [];
@@ -422,17 +491,18 @@ var Page = (function PageClosure() {
  * `PDFDocument` objects on the main thread created.
  */
 var PDFDocument = (function PDFDocumentClosure() {
-  function PDFDocument(arg, password) {
+  function PDFDocument(pdfManager, arg, password) {
     if (isStream(arg))
-      init.call(this, arg, password);
+      init.call(this, pdfManager, arg, password);
     else if (isArrayBuffer(arg))
-      init.call(this, new Stream(arg), password);
+      init.call(this, pdfManager, new Stream(arg), password);
     else
       error('PDFDocument: Unknown argument type');
   }
 
-  function init(stream, password) {
+  function init(pdfManager, stream, password) {
     assertWellFormed(stream.length > 0, 'stream must have data');
+    this.pdfManager = pdfManager;
     this.stream = stream;
     var xref = new XRef(this.stream, password);
     this.xref = xref;
@@ -576,7 +646,7 @@ var PDFDocument = (function PDFDocumentClosure() {
     },
     setup: function PDFDocument_setup(recoveryMode) {
       this.xref.parse(recoveryMode);
-      this.catalog = new Catalog(this.xref);
+      this.catalog = new Catalog(this.pdfManager, this.xref);
     },
     get numPages() {
       var linearization = this.linearization;

src/obj.js

@@ -18,7 +18,7 @@
            InvalidPDFException, isArray, isCmd, isDict, isInt, isName, isRef,
            isStream, JpegStream, Lexer, log, Page, Parser, Promise, shadow,
            stringToPDFString, stringToUTF8String, warn, isString, assert, PDFJS,
-           MissingDataException, XRefParseException */
+           MissingDataException, XRefParseException, Stream */
 
 'use strict';
 
@@ -151,7 +151,8 @@ var RefSet = (function RefSetClosure() {
 })();
 
 var Catalog = (function CatalogClosure() {
-  function Catalog(xref) {
+  function Catalog(pdfManager, xref) {
+    this.pdfManager = pdfManager;
     this.xref = xref;
     this.catDict = xref.getCatalogObj();
     assertWellFormed(isDict(this.catDict),
@@ -363,7 +364,8 @@ var Catalog = (function CatalogClosure() {
         var kid = this.xref.fetch(kidRef);
         if (isDict(kid, 'Page') || (isDict(kid) && !kid.has('Kids'))) {
           var pageIndex = this.currPageIndex++;
-          var page = new Page(this.xref, pageIndex, kid, kidRef);
+          var page = new Page(this.pdfManager, this.xref, pageIndex, kid,
+                              kidRef);
           if (!(pageIndex in this.pagePromises)) {
             this.pagePromises[pageIndex] = new PDFJS.Promise();
           }
@@ -832,10 +834,16 @@ var XRef = (function XRefClosure() {
     fetch: function XRef_fetch(ref, suppressEncryption) {
       assertWellFormed(isRef(ref), 'ref object is not a reference');
       var num = ref.num;
-      if (num in this.cache)
+      var e;
-        return this.cache[num];
+      if (num in this.cache) {
+        e = this.cache[num];
+        if (e instanceof Stream) {
+          return e.makeSubStream(e.start, e.length, e.dict);
+        }
+        return e;
+      }
 
-      var e = this.getEntry(num);
+      e = this.getEntry(num);
 
       // the referenced entry can be free
       if (e === null)
@@ -877,9 +885,16 @@ var XRef = (function XRefClosure() {
         } else {
           e = parser.getObj();
         }
-        // Don't cache streams since they are mutable (except images).
+        if (!isStream(e) || e instanceof JpegStream) {
-        if (!isStream(e) || e instanceof JpegStream)
           this.cache[num] = e;
+        } else if (e instanceof Stream) {
+          e = e.makeSubStream(e.start, e.length, e.dict);
+          this.cache[num] = e;
+        } else if ('readBlock' in e) {
+          e.getBytes();
+          e = e.makeSubStream(0, e.bufferLength, e.dict);
+          this.cache[num] = e;
+        }
         return e;
       }
 

src/parser.js

@@ -36,6 +36,21 @@ var Parser = (function ParserClosure() {
   }
 
   Parser.prototype = {
+    saveState: function Parser_saveState() {
+      this.state = {
+        buf1: this.buf1,
+        buf2: this.buf2,
+        streamPos: this.lexer.stream.pos
+      };
+    },
+
+    restoreState: function Parser_restoreState() {
+      var state = this.state;
+      this.buf1 = state.buf1;
+      this.buf2 = state.buf2;
+      this.lexer.stream.pos = state.streamPos;
+    },
+
     refill: function Parser_refill() {
       this.buf1 = this.lexer.getObj();
       this.buf2 = this.lexer.getObj();

src/pdf_manager.js

@@ -67,7 +67,7 @@ var BasePdfManager = (function BasePdfManagerClosure() {
 var LocalPdfManager = (function LocalPdfManagerClosure() {
   function LocalPdfManager(data, password) {
     var stream = new Stream(data);
-    this.pdfModel = new PDFDocument(stream, password);
+    this.pdfModel = new PDFDocument(this, stream, password);
     this.loadedStream = new PDFJS.Promise();
     this.loadedStream.resolve(stream);
   }
@@ -124,13 +124,14 @@ var NetworkPdfManager = (function NetworkPdfManagerClosure() {
     this.streamManager = new ChunkedStreamManager(args.length, CHUNK_SIZE,
                                                   args.url, params);
 
-    this.pdfModel = new PDFDocument(this.streamManager.getStream(),
+    this.pdfModel = new PDFDocument(this, this.streamManager.getStream(),
                                     args.password);
   }
 
   NetworkPdfManager.prototype = Object.create(BasePdfManager.prototype);
   NetworkPdfManager.prototype.constructor = NetworkPdfManager;
 
+  // FIXME(mack): Make ensure() use array for all arguments
   NetworkPdfManager.prototype.ensure =
       function NetworkPdfManager_ensure(obj, prop) {
     var promise = new PDFJS.Promise();

src/stream.js

@@ -26,7 +26,7 @@ var Stream = (function StreamClosure() {
     this.start = start || 0;
     this.pos = this.start;
     this.end = (start + length) || this.bytes.length;
-    this.dict = dict;
+    this.parameters = this.dict = dict;
   }
 
   // required methods for a stream. if a particular stream does not
@@ -645,6 +645,10 @@ var PredictorStream = (function PredictorStreamClosure() {
     var colors = this.colors;
 
     var rawBytes = this.stream.getBytes(rowBytes);
+    this.eof = !rawBytes.length;
+    if (this.eof) {
+      return;
+    }
 
     var inbuf = 0, outbuf = 0;
     var inbits = 0, outbits = 0;
@@ -705,6 +709,10 @@ var PredictorStream = (function PredictorStreamClosure() {
 
     var predictor = this.stream.getByte();
     var rawBytes = this.stream.getBytes(rowBytes);
+    this.eof = !rawBytes.length;
+    if (this.eof) {
+      return;
+    }
 
     var bufferLength = this.bufferLength;
     var buffer = this.ensureBuffer(bufferLength + rowBytes);
@@ -853,6 +861,7 @@ var JpegStream = (function JpegStreamClosure() {
       var data = jpegImage.getData(width, height);
       this.buffer = data;
       this.bufferLength = data.length;
+      this.eof = true;
     } catch (e) {
       error('JPEG error: ' + e);
     }
@@ -988,6 +997,7 @@ var JpxStream = (function JpxStreamClosure() {
 
     this.buffer = data;
     this.bufferLength = data.length;
+    this.eof = true;
   };
   JpxStream.prototype.getChar = function JpxStream_getChar() {
     error('internal error: getChar is not valid on JpxStream');
@@ -1032,6 +1042,7 @@ var Jbig2Stream = (function Jbig2StreamClosure() {
 
     this.buffer = data;
     this.bufferLength = dataLength;
+    this.eof = true;
   };
   Jbig2Stream.prototype.getChar = function Jbig2Stream_getChar() {
     error('internal error: getChar is not valid on Jbig2Stream');

src/util.js

@@ -397,8 +397,19 @@ var Util = PDFJS.Util = (function UtilClosure() {
     return num < 0 ? -1 : 1;
   };
 
+  // TODO(mack): Rename appendToArray
   Util.concatenateToArray = function concatenateToArray(arr1, arr2) {
-    return Array.prototype.push.apply(arr1, arr2);
+    Array.prototype.push.apply(arr1, arr2);
+  };
+
+  Util.prependToArray = function concatenateToArray(arr1, arr2) {
+    Array.prototype.unshift.apply(arr1, arr2);
+  };
+
+  Util.extendObj = function extendObj(obj1, obj2) {
+    for (var key in obj2) {
+      obj1[key] = obj2[key];
+    }
   };
 
   return Util;

src/worker.js

@@ -285,6 +285,7 @@ var WorkerMessageHandler = {
       };
 
       getPdfManager(data).then(function() {
+        globalScope.pdfManager = pdfManager;
         loadDocument(false).then(onSuccess, function(ex) {
           // Try again with recoveryMode == true
           if (!(ex instanceof XRefParseException)) {
@@ -358,10 +359,11 @@ var WorkerMessageHandler = {
 
         var pageNum = data.pageIndex + 1;
         var start = Date.now();
-        var dependency = [];
         // Pre compile the pdf page and fetch the fonts/images.
-        pdfManager.ensure(page, 'getOperatorList', handler,
+        page.getOperatorList(handler).then(function(opListData) {
-            dependency).then(function(operatorList) {
+
+          var operatorList = opListData.queue;
+          var dependency = Object.keys(opListData.dependencies);
 
           // The following code does quite the same as
           // Page.prototype.startRendering, but stops at one point and sends the
@@ -420,8 +422,7 @@ var WorkerMessageHandler = {
       pdfManager.getPage(data.pageIndex).then(function(page) {
         var pageNum = data.pageIndex + 1;
         var start = Date.now();
-        pdfManager.ensure(page,
+        page.extractTextContent().then(function(textContent) {
-            'extractTextContent').then(function(textContent) {
           promise.resolve(textContent);
           log('text indexing: page=%d - time=%dms', pageNum,
               Date.now() - start);

test/test.py

@@ -158,7 +158,6 @@ class TestHandlerBase(BaseHTTPRequestHandler):
             elif v[0] == errno.EPIPE:
                 print 'Detected remote peer disconnected'
             elif v[0] == 10053:
-                # FIXME(mack): Address this issue
                 print 'An established connection was aborted by the' \
                     ' software in your host machine'
             else:

test/unit/evaluator_spec.js

@@ -33,24 +33,28 @@ describe('evaluator', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('qTT');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(1);
         expect(result.fnArray[0]).toEqual('save');
         expect(result.argsArray[0].length).toEqual(0);
       });
+    });
 
     it('should handle one operations', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('Q');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(1);
         expect(result.fnArray[0]).toEqual('restore');
       });
+    });
 
     it('should handle two glued operations', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
@@ -58,26 +62,30 @@ describe('evaluator', function() {
       var resources = new ResourcesMock();
       resources.Res1 = {};
       var stream = new StringStream('/Res1 DoQ');
-      var result = evaluator.getOperatorList(stream, resources, []);
+      var promise = evaluator.getOperatorList(stream, resources);
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(2);
         expect(result.fnArray[0]).toEqual('paintXObject');
         expect(result.fnArray[1]).toEqual('restore');
       });
+    });
 
     it('should handle tree glued operations', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('qqq');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(3);
         expect(result.fnArray[0]).toEqual('save');
         expect(result.fnArray[1]).toEqual('save');
         expect(result.fnArray[2]).toEqual('save');
       });
+    });
 
     it('should handle three glued operations #2', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
@@ -85,21 +93,24 @@ describe('evaluator', function() {
       var resources = new ResourcesMock();
       resources.Res1 = {};
       var stream = new StringStream('B*Bf*');
-      var result = evaluator.getOperatorList(stream, resources, []);
+      var promise = evaluator.getOperatorList(stream, resources);
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(3);
         expect(result.fnArray[0]).toEqual('eoFillStroke');
         expect(result.fnArray[1]).toEqual('fillStroke');
         expect(result.fnArray[2]).toEqual('eoFill');
       });
+    });
 
     it('should handle glued operations and operands', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('q5 Ts');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise  = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(2);
         expect(result.fnArray[0]).toEqual('save');
@@ -108,13 +119,15 @@ describe('evaluator', function() {
         expect(result.argsArray[1].length).toEqual(1);
         expect(result.argsArray[1][0]).toEqual(5);
       });
+    });
 
     it('should handle glued operations and literals', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('trueifalserinullq');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(!!result.fnArray && !!result.argsArray).toEqual(true);
         expect(result.fnArray.length).toEqual(3);
         expect(result.fnArray[0]).toEqual('setFlatness');
@@ -128,38 +141,46 @@ describe('evaluator', function() {
         expect(result.argsArray[2].length).toEqual(0);
       });
     });
+  });
 
   describe('validateNumberOfArgs', function() {
     it('should execute if correct number of arguments', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('5 1 d0');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      console.log('here!');
-
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
+      promise.then(function(data) {
+        var result = data.queue;
         expect(result.argsArray[0][0]).toEqual(5);
         expect(result.argsArray[0][1]).toEqual(1);
         expect(result.fnArray[0]).toEqual('setCharWidth');
       });
+    });
     it('should execute if too many arguments', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('5 1 4 d0');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(result.argsArray[0][0]).toEqual(5);
         expect(result.argsArray[0][1]).toEqual(1);
         expect(result.argsArray[0][2]).toEqual(4);
         expect(result.fnArray[0]).toEqual('setCharWidth');
       });
+    });
     it('should skip if too few arguments', function() {
       var evaluator = new PartialEvaluator(new XrefMock(), new HandlerMock(),
                                            'prefix');
       var stream = new StringStream('5 d0');
-      var result = evaluator.getOperatorList(stream, new ResourcesMock(), []);
+      var promise = evaluator.getOperatorList(stream, new ResourcesMock());
-
+      promise.then(function(data) {
+        var result = data.queue;
         expect(result.argsArray).toEqual([]);
         expect(result.fnArray).toEqual([]);
       });
     });
+  });
 });