Merge pull request #2127 from jviereck/text-algo-3

Use the text extracted in the getTextContent function for the divs of the textLayer.

src/bidi.js

@@ -138,11 +138,16 @@ var bidi = PDFJS.bidi = (function bidiClosure() {
     }
   }
 
-  function bidi(text, startLevel) {
+  function BidiResult(str, isLTR) {
-    var str = text.str;
+    this.str = str;
+    this.ltr = isLTR;
+  }
+
+  function bidi(str, startLevel) {
+    var isLTR = true;
     var strLength = str.length;
     if (strLength == 0)
-      return str;
+      return new BidiResult(str, ltr);
 
     // get types, fill arrays
 
@@ -176,16 +181,16 @@ var bidi = PDFJS.bidi = (function bidiClosure() {
     //  if less than 30% chars are rtl then string is primarily ltr
     //  if more than 30% chars are rtl then string is primarily rtl
     if (numBidi == 0) {
-      text.direction = 'ltr';
+      isLTR = true;
-      return str;
+      return new BidiResult(str, isLTR);
     }
 
     if (startLevel == -1) {
       if ((strLength / numBidi) < 0.3) {
-        text.direction = 'ltr';
+        isLTR = true;
         startLevel = 0;
       } else {
-        text.direction = 'rtl';
+        isLTR = false;
         startLevel = 1;
       }
     }
@@ -438,7 +443,8 @@ var bidi = PDFJS.bidi = (function bidiClosure() {
       if (ch != '<' && ch != '>')
         result += ch;
     }
-    return result;
+
+    return new BidiResult(result, isLTR);
   }
 
   return bidi;

src/canvas.js

@@ -677,9 +677,10 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
       var textHScale2 = textHScale * fontMatrix[0];
       var glyphsLength = glyphs.length;
       var textLayer = this.textLayer;
-      var text = {str: '', length: 0, canvasWidth: 0, geom: {}};
+      var geom;
       var textSelection = textLayer && !skipTextSelection ? true : false;
       var textRenderingMode = current.textRenderingMode;
+      var canvasWidth = 0.0;
 
       // Type3 fonts - each glyph is a "mini-PDF"
       if (font.coded) {
@@ -692,7 +693,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
         if (textSelection) {
           this.save();
           ctx.scale(1, -1);
-          text.geom = this.getTextGeometry();
+          geom = this.getTextGeometry();
           this.restore();
         }
         for (var i = 0; i < glyphsLength; ++i) {
@@ -718,9 +719,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
           ctx.translate(width, 0);
           current.x += width * textHScale;
 
-          text.str += glyph.unicode;
+          canvasWidth += width;
-          text.length++;
-          text.canvasWidth += width;
         }
         ctx.restore();
       } else {
@@ -735,7 +734,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
           lineWidth /= scale;
 
         if (textSelection)
-          text.geom = this.getTextGeometry();
+          geom = this.getTextGeometry();
 
         if (fontSizeScale != 1.0) {
           ctx.scale(fontSizeScale, fontSizeScale);
@@ -784,17 +783,19 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
           var glyphUnicode = glyph.unicode === ' ' ? '\u00A0' : glyph.unicode;
           if (glyphUnicode in NormalizedUnicodes)
             glyphUnicode = NormalizedUnicodes[glyphUnicode];
-          text.str += reverseIfRtl(glyphUnicode);
+
-          text.canvasWidth += charWidth;
+          canvasWidth += charWidth;
         }
         current.x += x * textHScale2;
         ctx.restore();
       }
 
-      if (textSelection)
+      if (textSelection) {
-        this.textLayer.appendText(text, font.fallbackName, fontSize);
+        geom.canvasWidth = canvasWidth;
+        this.textLayer.appendText(font.fallbackName, fontSize, geom);
+      }
 
-      return text;
+      return canvasWidth;
     },
     showSpacedText: function CanvasGraphics_showSpacedText(arr) {
       var ctx = this.ctx;
@@ -806,7 +807,8 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
         textHScale *= (current.fontMatrix || IDENTITY_MATRIX)[0];
       var arrLength = arr.length;
       var textLayer = this.textLayer;
-      var text = {str: '', length: 0, canvasWidth: 0, geom: {}};
+      var geom;
+      var canvasWidth = 0.0;
       var textSelection = textLayer ? true : false;
 
       if (textSelection) {
@@ -819,7 +821,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
           ctx.scale(textHScale, 1);
         } else
           this.applyTextTransforms();
-        text.geom = this.getTextGeometry();
+        geom = this.getTextGeometry();
         ctx.restore();
       }
 
@@ -829,34 +831,22 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
           var spacingLength = -e * 0.001 * fontSize * textHScale;
           current.x += spacingLength;
 
-          if (textSelection) {
+          if (textSelection)
-            // Emulate precise spacing via HTML spaces
+            canvasWidth += spacingLength;
-            text.canvasWidth += spacingLength;
-            if (e < 0 && text.geom.spaceWidth > 0) { // avoid div by zero
-              var numFakeSpaces = Math.round(-e / text.geom.spaceWidth);
-              if (numFakeSpaces > 0) {
-                text.str += '\u00A0';
-              }
-            }
-          }
         } else if (isString(e)) {
-          var shownText = this.showText(e, true);
+          var shownCanvasWidth = this.showText(e, true);
 
-          if (textSelection) {
+          if (textSelection)
-            if (shownText.str === ' ') {
+            canvasWidth += shownCanvasWidth;
-              text.str += '\u00A0';
-            } else {
-              text.str += shownText.str;
-            }
-            text.canvasWidth += shownText.canvasWidth;
-          }
         } else {
           error('TJ array element ' + e + ' is not string or num');
         }
       }
 
-      if (textSelection)
+      if (textSelection) {
-        this.textLayer.appendText(text, font.fallbackName, fontSize);
+        geom.canvasWidth = canvasWidth;
+        this.textLayer.appendText(font.fallbackName, fontSize, geom);
+      }
     },
     nextLineShowText: function CanvasGraphics_nextLineShowText(text) {
       this.nextLine();

src/evaluator.js

@@ -164,6 +164,21 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           translated = { error: e };
         }
         font.translated = translated;
+
+        var data = translated;
+        if (data.loadCharProcs) {
+          delete data.loadCharProcs;
+
+          var charProcs = font.get('CharProcs').getAll();
+          var fontResources = font.get('Resources') || resources;
+          var charProcOperatorList = {};
+          for (var key in charProcs) {
+            var glyphStream = charProcs[key];
+            charProcOperatorList[key] =
+              this.getOperatorList(glyphStream, fontResources, dependency);
+          }
+          data.charProcOperatorList = charProcOperatorList;
+        }
       }
       return font;
     },
@@ -195,19 +210,6 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         var loadedName = font.loadedName;
         if (!font.sent) {
           var data = font.translated;
-          if (data.loadCharProcs) {
-            delete data.loadCharProcs;
-
-            var charProcs = font.get('CharProcs').getAll();
-            var fontResources = font.get('Resources') || resources;
-            var charProcOperatorList = {};
-            for (var key in charProcs) {
-              var glyphStream = charProcs[key];
-              charProcOperatorList[key] =
-                self.getOperatorList(glyphStream, fontResources, dependency);
-            }
-            data.charProcOperatorList = charProcOperatorList;
-          }
 
           if (data instanceof Font)
             data = data.exportData();
@@ -505,7 +507,18 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
       return queue;
     },
 
-    getTextContent: function partialEvaluatorGetIRQueue(stream, resources) {
+    getTextContent: function partialEvaluatorGetIRQueue(
+                                                    stream, resources, state) {
+      var bidiTexts;
+
+      if (!state) {
+        bidiTexts = [];
+        state = {
+          bidiTexts: bidiTexts
+        };
+      } else {
+        bidiTexts = state.bidiTexts;
+      }
 
       var self = this;
       var xref = this.xref;
@@ -515,18 +528,20 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
       }
 
       resources = xref.fetchIfRef(resources) || new Dict();
+      // The xobj is parsed iff it's needed, e.g. if there is a `DO` cmd.
+      var xobjs = null;
 
       var parser = new Parser(new Lexer(stream), false);
       var res = resources;
       var args = [], obj;
 
-      var text = '';
       var chunk = '';
       var font = null;
       while (!isEOF(obj = parser.getObj())) {
         if (isCmd(obj)) {
           var cmd = obj.cmd;
           switch (cmd) {
+            // TODO: Add support for SAVE/RESTORE and XFORM here.
             case 'Tf':
               font = handleSetFont(args[0].name).translated;
               break;
@@ -535,10 +550,11 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               for (var j = 0, jj = items.length; j < jj; j++) {
                 if (typeof items[j] === 'string') {
                   chunk += fontCharsToUnicode(items[j], font);
-                } else if (items[j] < 0) {
+                } else if (items[j] < 0 && font.spaceWidth > 0) {
-                  // making all negative offsets a space - better to have
+                  var numFakeSpaces = Math.round(-items[j] / font.spaceWidth);
-                  // a space in incorrect place than not have them at all
+                  if (numFakeSpaces > 0) {
-                  chunk += ' ';
+                    chunk += ' ';
+                  }
                 }
               }
               break;
@@ -546,14 +562,69 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               chunk += fontCharsToUnicode(args[0], font);
               break;
             case "'":
-              chunk += fontCharsToUnicode(args[0], font) + ' ';
+              // For search, adding a extra white space for line breaks would be
+              // better here, but that causes too much spaces in the
+              // text-selection divs.
+              chunk += fontCharsToUnicode(args[0], font);
               break;
             case '"':
-              chunk += fontCharsToUnicode(args[2], font) + ' ';
+              // Note comment in "'"
+              chunk += fontCharsToUnicode(args[2], font);
+              break;
+            case 'Do':
+              // Set the chunk such that the following if won't add something
+              // to the state.
+              chunk = '';
+
+              if (args[0].code) {
+                break;
+              }
+
+              if (!xobjs) {
+                xobjs = resources.get('XObject') || new Dict();
+              }
+
+              var name = args[0].name;
+              var xobj = xobjs.get(name);
+              if (!xobj)
+                break;
+              assertWellFormed(isStream(xobj), 'XObject should be a stream');
+
+              var type = xobj.dict.get('Subtype');
+              assertWellFormed(
+                isName(type),
+                'XObject should have a Name subtype'
+              );
+
+              if ('Form' !== type.name)
+                break;
+
+              state = this.getTextContent(
+                xobj,
+                xobj.dict.get('Resources') || resources,
+                state
+              );
+              break;
+            case 'gs':
+              var dictName = args[0];
+              var extGState = resources.get('ExtGState');
+
+              if (!isDict(extGState) || !extGState.has(dictName.name))
+                break;
+
+              var gsState = extGState.get(dictName.name);
+
+              for (var i = 0; i < gsState.length; i++) {
+                if (gsState[i] === 'Font') {
+                  font = handleSetFont(args[0].name).translated;
+                }
+              }
               break;
           } // switch
+
           if (chunk !== '') {
-            text += chunk;
+            bidiTexts.push(PDFJS.bidi(chunk, -1));
+
             chunk = '';
           }
 
@@ -562,9 +633,9 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           assertWellFormed(args.length <= 33, 'Too many arguments');
           args.push(obj);
         }
-      }
+      } // while
 
-      return text;
+      return state;
     },
 
     extractDataStructures: function

src/fonts.js

@@ -3886,6 +3886,10 @@ var Font = (function FontClosure() {
     },
 
     get spaceWidth() {
+      if ('_shadowWidth' in this) {
+        return this._shadowWidth;
+      }
+
       // trying to estimate space character width
       var possibleSpaceReplacements = ['space', 'minus', 'one', 'i'];
       var width;
@@ -3913,7 +3917,10 @@ var Font = (function FontClosure() {
           break; // the non-zero width found
       }
       width = (width || this.defaultWidth) * this.widthMultiplier;
-      return shadow(this, 'spaceWidth', width);
+      // 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) {

test/driver.js

@@ -159,6 +159,7 @@ NullTextLayerBuilder.prototype = {
 function SimpleTextLayerBuilder(ctx, viewport) {
   this.ctx = ctx;
   this.viewport = viewport;
+  this.textCounter = 0;
 }
 SimpleTextLayerBuilder.prototype = {
   beginLayout: function SimpleTextLayerBuilder_BeginLayout() {
@@ -167,27 +168,31 @@ SimpleTextLayerBuilder.prototype = {
   endLayout: function SimpleTextLayerBuilder_EndLayout() {
     this.ctx.restore();
   },
-  appendText: function SimpleTextLayerBuilder_AppendText(text, fontName,
+  appendText: function SimpleTextLayerBuilder_AppendText(fontName, fontSize,
-                                                         fontSize) {
+                                                          geom) {
     var ctx = this.ctx, viewport = this.viewport;
     // vScale and hScale already contain the scaling to pixel units
-    var fontHeight = fontSize * text.geom.vScale;
+    var fontHeight = fontSize * geom.vScale;
     ctx.beginPath();
     ctx.strokeStyle = 'red';
     ctx.fillStyle = 'yellow';
-    ctx.rect(text.geom.x, text.geom.y - fontHeight,
+    ctx.rect(geom.x, geom.y - fontHeight,
-             text.canvasWidth * text.geom.hScale, fontHeight);
+             geom.canvasWidth * geom.hScale, fontHeight);
     ctx.stroke();
     ctx.fill();
 
-    var textContent = bidi(text, -1);
+    var textContent = this.textContent.bidiTexts[this.textCounter].str;
     ctx.font = fontHeight + 'px ' + fontName;
     ctx.fillStyle = 'black';
-    ctx.fillText(textContent, text.geom.x, text.geom.y);
+    ctx.fillText(textContent, geom.x, geom.y);
+
+    this.textCounter++;
+  },
+  setTextContent: function SimpleTextLayerBuilder_SetTextContent(textContent) {
+    this.textContent = textContent;
   }
 };
 
-
 function nextPage(task, loadError) {
   var failure = loadError || '';
 
@@ -245,6 +250,10 @@ function nextPage(task, loadError) {
           drawContext = dummyCanvas.getContext('2d');
           // ... text builder will draw its content on the test canvas
           textLayerBuilder = new SimpleTextLayerBuilder(ctx, viewport);
+
+          page.getTextContent().then(function(textContent) {
+            textLayerBuilder.setTextContent(textContent);
+          });
         } else {
           drawContext = ctx;
           textLayerBuilder = new NullTextLayerBuilder();

web/viewer.css

@@ -1066,6 +1066,7 @@ canvas {
   color: transparent;
   position: absolute;
   line-height:1.3;
+  white-space:pre;
 }
 
 /* TODO: file FF bug to support ::-moz-selection:window-inactive

web/viewer.js

@@ -1040,7 +1040,7 @@ var PDFView = {
     function extractPageText(pageIndex) {
       self.pages[pageIndex].pdfPage.getTextContent().then(
         function textContentResolved(textContent) {
-          self.pageText[pageIndex] = textContent;
+          self.pageText[pageIndex] = textContent.join('');
           self.search();
           if ((pageIndex + 1) < self.pages.length)
             extractPageText(pageIndex + 1);
@@ -1228,6 +1228,8 @@ var PageView = function pageView(container, pdfPage, id, scale,
   this.renderingState = RenderingStates.INITIAL;
   this.resume = null;
 
+  this.textContent = null;
+
   var anchor = document.createElement('a');
   anchor.name = '' + this.id;
 
@@ -1448,6 +1450,13 @@ var PageView = function pageView(container, pdfPage, id, scale,
       }, 0);
   };
 
+  this.getTextContent = function pageviewGetTextContent() {
+    if (!this.textContent) {
+      this.textContent = this.pdfPage.getTextContent();
+    }
+    return this.textContent;
+  };
+
   this.draw = function pageviewDraw(callback) {
     if (this.renderingState !== RenderingStates.INITIAL)
       error('Must be in new state before drawing');
@@ -1528,6 +1537,14 @@ var PageView = function pageView(container, pdfPage, id, scale,
       }
     );
 
+    if (textLayer) {
+      this.getTextContent().then(
+        function textContentResolved(textContent) {
+          textLayer.setTextContent(textContent);
+        }
+      );
+    }
+
     setupAnnotations(this.pdfPage, this.viewport);
     div.setAttribute('data-loaded', true);
   };
@@ -1820,12 +1837,19 @@ var CustomStyle = (function CustomStyleClosure() {
 var TextLayerBuilder = function textLayerBuilder(textLayerDiv) {
   var textLayerFrag = document.createDocumentFragment();
   this.textLayerDiv = textLayerDiv;
+  this.layoutDone = false;
+  this.divContentDone = false;
 
   this.beginLayout = function textLayerBuilderBeginLayout() {
     this.textDivs = [];
     this.textLayerQueue = [];
   };
 
+  this.endLayout = function textLayerBuilderEndLayout() {
+    this.layoutDone = true;
+    this.insertDivContent();
+  },
+
   this.renderLayer = function textLayerBuilderRenderLayer() {
     var self = this;
     var textDivs = this.textDivs;
@@ -1857,7 +1881,7 @@ var TextLayerBuilder = function textLayerBuilder(textLayerDiv) {
     textLayerDiv.appendChild(textLayerFrag);
   };
 
-  this.endLayout = function textLayerBuilderEndLayout() {
+  this.setupRenderLayoutTimer = function textLayerSetupRenderLayoutTimer() {
     // Schedule renderLayout() if user has been scrolling, otherwise
     // run it right away
     var kRenderDelay = 200; // in ms
@@ -1870,27 +1894,56 @@ var TextLayerBuilder = function textLayerBuilder(textLayerDiv) {
       if (this.renderTimer)
         clearTimeout(this.renderTimer);
       this.renderTimer = setTimeout(function() {
-        self.endLayout();
+        self.setupRenderLayoutTimer();
       }, kRenderDelay);
     }
-  }; // endLayout
+  };
 
-  this.appendText = function textLayerBuilderAppendText(text,
+  this.appendText = function textLayerBuilderAppendText(fontName, fontSize,
-                                                        fontName, fontSize) {
+                                                        geom) {
     var textDiv = document.createElement('div');
 
     // vScale and hScale already contain the scaling to pixel units
-    var fontHeight = fontSize * text.geom.vScale;
+    var fontHeight = fontSize * geom.vScale;
-    textDiv.dataset.canvasWidth = text.canvasWidth * text.geom.hScale;
+    textDiv.dataset.canvasWidth = geom.canvasWidth * geom.hScale;
+    textDiv.dataset.fontName = fontName;
 
     textDiv.style.fontSize = fontHeight + 'px';
     textDiv.style.fontFamily = fontName;
-    textDiv.style.left = text.geom.x + 'px';
+    textDiv.style.left = geom.x + 'px';
-    textDiv.style.top = (text.geom.y - fontHeight) + 'px';
+    textDiv.style.top = (geom.y - fontHeight) + 'px';
-    textDiv.textContent = PDFJS.bidi(text, -1);
+
-    textDiv.dir = text.direction;
+    // The content of the div is set in the `setTextContent` function.
+
     this.textDivs.push(textDiv);
   };
+
+  this.insertDivContent = function textLayerUpdateTextContent() {
+    // Only set the content of the divs once layout has finished, the content
+    // for the divs is available and content is not yet set on the divs.
+    if (!this.layoutDone || this.divContentDone || !this.textContent)
+      return;
+
+    this.divContentDone = true;
+
+    var textDivs = this.textDivs;
+    var bidiTexts = this.textContent.bidiTexts;
+
+    for (var i = 0; i < bidiTexts.length; i++) {
+      var bidiText = bidiTexts[i];
+      var textDiv = textDivs[i];
+
+      textDiv.textContent = bidiText.str;
+      textDiv.dir = bidiText.ltr ? 'ltr' : 'rtl';
+    }
+
+    this.setupRenderLayoutTimer();
+  };
+
+  this.setTextContent = function textLayerBuilderSetTextContent(textContent) {
+    this.textContent = textContent;
+    this.insertDivContent();
+  };
 };
 
 document.addEventListener('DOMContentLoaded', function webViewerLoad(evt) {