Merge upstream and change to error parameter for callback.

README.md

@@ -1,6 +1,6 @@
 # pdf.js
 
-
+ 
 
 ## Overview
 

extensions/firefox/install.rdf

@@ -6,13 +6,13 @@
   <Description about="urn:mozilla:install-manifest">
     <em:id>uriloader@pdf.js</em:id>
     <em:name>pdf.js</em:name>
-    <em:version>0.1</em:version>
+    <em:version>0.1.0</em:version>
 	  <em:iconURL>chrome://pdf.js/skin/logo.png</em:iconURL>
     <em:targetApplication>
       <Description>
        <em:id>{ec8030f7-c20a-464f-9b0e-13a3a9e97384}</em:id>
        <em:minVersion>6.0</em:minVersion>
-       <em:maxVersion>11.0.*</em:maxVersion>
+       <em:maxVersion>11.0a1</em:maxVersion>
      </Description>
     </em:targetApplication>
     <em:bootstrap>true</em:bootstrap>
@@ -20,5 +20,6 @@
     <em:creator>Vivien Nicolas</em:creator>
     <em:description>pdf.js uri loader</em:description>
     <em:homepageURL>https://github.com/mozilla/pdf.js/</em:homepageURL>
+    <em:type>2</em:type>
   </Description>
 </RDF>

src/canvas.js

@@ -174,7 +174,7 @@ var CanvasGraphics = (function canvasGraphics() {
   // before it stops and shedules a continue of execution.
   var kExecutionTime = 50;
 
-  function constructor(canvasCtx, objs) {
+  function constructor(canvasCtx, objs, textLayer) {
     this.ctx = canvasCtx;
     this.current = new CanvasExtraState();
     this.stateStack = [];
@@ -183,7 +183,7 @@ var CanvasGraphics = (function canvasGraphics() {
     this.xobjs = null;
     this.ScratchCanvas = ScratchCanvas;
     this.objs = objs;
-
+    this.textLayer = textLayer;
     if (canvasCtx) {
       addContextCurrentTransform(canvasCtx);
     }
@@ -212,7 +212,13 @@ var CanvasGraphics = (function canvasGraphics() {
           this.ctx.transform(0, -1, -1, 0, cw, ch);
           break;
       }
+      // Scale so that canvas units are the same as PDF user space units
       this.ctx.scale(cw / mediaBox.width, ch / mediaBox.height);
+      this.textDivs = [];
+      this.textLayerQueue = [];
+      // Prevent textLayerQueue from being rendered while rendering a new page
+      if (this.textLayerTimer)
+        clearTimeout(this.textLayerTimer);
     },
 
     executeIRQueue: function canvasGraphicsExecuteIRQueue(codeIR,
@@ -270,6 +276,37 @@ var CanvasGraphics = (function canvasGraphics() {
 
     endDrawing: function canvasGraphicsEndDrawing() {
       this.ctx.restore();
+
+      var textLayer = this.textLayer;
+      if (!textLayer)
+        return;
+
+      var self = this;
+      var renderTextLayer = function canvasRenderTextLayer() {
+        var textDivs = self.textDivs;
+        for (var i = 0, length = textDivs.length; i < length; ++i) {
+          if (textDivs[i].dataset.textLength > 1) { // avoid div by zero
+            textLayer.appendChild(textDivs[i]);
+            // Adjust div width (via letterSpacing) to match canvas text
+            // Due to the .offsetWidth calls, this is slow
+            textDivs[i].style.letterSpacing =
+              ((textDivs[i].dataset.canvasWidth - textDivs[i].offsetWidth) /
+               (textDivs[i].dataset.textLength - 1)) + 'px';
+          }
+        }
+      }
+      var textLayerQueue = this.textLayerQueue;
+      textLayerQueue.push(renderTextLayer);
+
+      // Lazy textLayer rendering (to prevent UI hangs)
+      // Only render queue if activity has stopped, where "no activity" ==
+      // "no beginDrawing() calls in the last N ms"
+      this.textLayerTimer = setTimeout(function renderTextLayerQueue() {
+        // Render most recent (==most relevant) layers first
+        for (var i = textLayerQueue.length - 1; i >= 0; i--) {
+          textLayerQueue.pop().call();
+        }
+      }, 500);
     },
 
     // Graphics state
@@ -528,23 +565,93 @@ var CanvasGraphics = (function canvasGraphics() {
     nextLine: function canvasGraphicsNextLine() {
       this.moveText(0, this.current.leading);
     },
-    showText: function canvasGraphicsShowText(text) {
+    applyTextTransforms: function canvasApplyTransforms() {
+      var ctx = this.ctx;
+      var current = this.current;
+      var textHScale = current.textHScale;
+      var fontMatrix = current.font.fontMatrix || IDENTITY_MATRIX;
+
+      ctx.transform.apply(ctx, current.textMatrix);
+      ctx.scale(1, -1);
+      ctx.translate(current.x, -1 * current.y);
+      ctx.transform.apply(ctx, fontMatrix);
+      ctx.scale(textHScale, 1);
+    },
+    getTextGeometry: function canvasGetTextGeometry() {
+      var geometry = {};
+      var ctx = this.ctx;
+      var font = this.current.font;
+      var ctxMatrix = ctx.mozCurrentTransform;
+      if (ctxMatrix) {
+        var bl = Util.applyTransform([0, 0], ctxMatrix);
+        var tr = Util.applyTransform([1, 1], ctxMatrix);
+        geometry.x = bl[0];
+        geometry.y = bl[1];
+        geometry.hScale = tr[0] - bl[0];
+        geometry.vScale = tr[1] - bl[1];
+      }
+      var spaceGlyph = font.charsToGlyphs(' ');
+
+      // Hack (sometimes space is not encoded)
+      if (spaceGlyph.length === 0 || spaceGlyph[0].width === 0)
+        spaceGlyph = font.charsToGlyphs('i');
+
+      // Fallback
+      if (spaceGlyph.length === 0 || spaceGlyph[0].width === 0)
+        spaceGlyph = [{width: 0}];
+
+      geometry.spaceWidth = spaceGlyph[0].width;
+      return geometry;
+    },
+
+    pushTextDivs: function canvasGraphicsPushTextDivs(text) {
+      var div = document.createElement('div');
+      var fontSize = this.current.fontSize;
+
+      // vScale and hScale already contain the scaling to pixel units
+      // as mozCurrentTransform reflects ctx.scale() changes
+      // (see beginDrawing())
+      var fontHeight = fontSize * text.geom.vScale;
+      div.dataset.canvasWidth = text.canvasWidth * text.geom.hScale;
+
+      div.style.fontSize = fontHeight + 'px';
+      div.style.fontFamily = this.current.font.loadedName || 'sans-serif';
+      div.style.left = text.geom.x + 'px';
+      div.style.top = (text.geom.y - fontHeight) + 'px';
+      div.innerHTML = text.str;
+      div.dataset.textLength = text.length;
+      this.textDivs.push(div);
+    },
+    showText: function canvasGraphicsShowText(str, skipTextSelection) {
       var ctx = this.ctx;
       var current = this.current;
       var font = current.font;
-      var glyphs = font.charsToGlyphs(text);
+      var glyphs = font.charsToGlyphs(str);
       var fontSize = current.fontSize;
       var charSpacing = current.charSpacing;
       var wordSpacing = current.wordSpacing;
       var textHScale = current.textHScale;
+      var fontMatrix = font.fontMatrix || IDENTITY_MATRIX;
+      var textHScale2 = textHScale * fontMatrix[0];
       var glyphsLength = glyphs.length;
+      var textLayer = this.textLayer;
+      var text = {str: '', length: 0, canvasWidth: 0, geom: {}};
+      var textSelection = textLayer && !skipTextSelection ? true : false;
+
+      if (textSelection) {
+        ctx.save();
+        this.applyTextTransforms();
+        text.geom = this.getTextGeometry();
+        ctx.restore();
+      }
+
+      // Type3 fonts - each glyph is a "mini-PDF"
       if (font.coded) {
         ctx.save();
         ctx.transform.apply(ctx, current.textMatrix);
         ctx.translate(current.x, current.y);
 
-        var fontMatrix = font.fontMatrix || IDENTITY_MATRIX;
+        ctx.scale(textHScale, 1);
-        ctx.scale(1 / textHScale, 1);
         for (var i = 0; i < glyphsLength; ++i) {
 
           var glyph = glyphs[i];
@@ -564,18 +671,16 @@ var CanvasGraphics = (function canvasGraphics() {
           var width = transformed[0] * fontSize + charSpacing;
 
           ctx.translate(width, 0);
-          current.x += width;
+          current.x += width * textHScale2;
 
+          text.str += glyph.unicode;
+          text.length++;
+          text.canvasWidth += width;
         }
         ctx.restore();
       } else {
         ctx.save();
-        ctx.transform.apply(ctx, current.textMatrix);
+        this.applyTextTransforms();
-        ctx.scale(1, -1);
-        ctx.translate(current.x, -1 * current.y);
-        ctx.transform.apply(ctx, font.fontMatrix || IDENTITY_MATRIX);
-
-        ctx.scale(1 / textHScale, 1);
 
         var width = 0;
         for (var i = 0; i < glyphsLength; ++i) {
@@ -586,36 +691,78 @@ var CanvasGraphics = (function canvasGraphics() {
             continue;
           }
 
-          var unicode = glyph.unicode;
+          var char = glyph.fontChar;
-          var char = (unicode >= 0x10000) ?
+          var charWidth = glyph.width * fontSize * 0.001 + charSpacing;
-            String.fromCharCode(0xD800 | ((unicode - 0x10000) >> 10),
-            0xDC00 | (unicode & 0x3FF)) : String.fromCharCode(unicode);
-
           ctx.fillText(char, width, 0);
-          width += glyph.width * fontSize * 0.001 + charSpacing;
+          width += charWidth;
-        }
-        current.x += width;
 
+          text.str += glyph.unicode === ' ' ? '&nbsp;' : glyph.unicode;
+          text.length++;
+          text.canvasWidth += charWidth;
+        }
+        current.x += width * textHScale2;
         ctx.restore();
       }
-    },
 
+      if (textSelection)
+        this.pushTextDivs(text);
+
+      return text;
+    },
     showSpacedText: function canvasGraphicsShowSpacedText(arr) {
       var ctx = this.ctx;
       var current = this.current;
       var fontSize = current.fontSize;
-      var textHScale = current.textHScale;
+      var textHScale2 = current.textHScale *
+        (current.font.fontMatrix || IDENTITY_MATRIX)[0];
       var arrLength = arr.length;
+      var textLayer = this.textLayer;
+      var font = current.font;
+      var text = {str: '', length: 0, canvasWidth: 0, geom: {}};
+      var textSelection = textLayer ? true : false;
+
+      if (textSelection) {
+        ctx.save();
+        this.applyTextTransforms();
+        text.geom = this.getTextGeometry();
+        ctx.restore();
+      }
+
       for (var i = 0; i < arrLength; ++i) {
         var e = arr[i];
         if (isNum(e)) {
-          current.x -= e * 0.001 * fontSize * textHScale;
+          var spacingLength = -e * 0.001 * fontSize * textHScale2;
+          current.x += spacingLength;
+
+          if (textSelection) {
+            // Emulate precise spacing via HTML spaces
+            text.canvasWidth += spacingLength;
+            if (e < 0 && text.geom.spaceWidth > 0) { // avoid div by zero
+              var numFakeSpaces = Math.round(-e / text.geom.spaceWidth);
+              for (var j = 0; j < numFakeSpaces; ++j)
+                text.str += '&nbsp;';
+              text.length += numFakeSpaces > 0 ? 1 : 0;
+            }
+          }
         } else if (isString(e)) {
-          this.showText(e);
+          var shownText = this.showText(e, true);
+
+          if (textSelection) {
+            if (shownText.str === ' ') {
+              text.str += '&nbsp;';
+            } else {
+              text.str += shownText.str;
+            }
+            text.canvasWidth += shownText.canvasWidth;
+            text.length += e.length;
+          }
         } else {
           malformed('TJ array element ' + e + ' is not string or num');
         }
       }
+
+      if (textSelection)
+        this.pushTextDivs(text);
     },
     nextLineShowText: function canvasGraphicsNextLineShowText(text) {
       this.nextLine();

src/core.js

@@ -70,7 +70,6 @@ var Page = (function pagePage() {
 
     this.ctx = null;
     this.callback = null;
-    this.errorback = null;
   }
 
   constructor.prototype = {
@@ -164,7 +163,7 @@ var Page = (function pagePage() {
                                                 IRQueue, fonts) {
       var self = this;
       this.IRQueue = IRQueue;
-      var gfx = new CanvasGraphics(this.ctx, this.objs);
+      var gfx = new CanvasGraphics(this.ctx, this.objs, this.textLayer);
 
       var displayContinuation = function pageDisplayContinuation() {
         // Always defer call to display() to work around bug in
@@ -173,8 +172,8 @@ var Page = (function pagePage() {
           try {
             self.display(gfx, self.callback);
           } catch (e) {
-            if (self.errorback)
+            if (self.callback)
-              self.errorback(e);
+              self.callback(e);
             else
               throw e;
           }
@@ -251,6 +250,7 @@ var Page = (function pagePage() {
         startIdx = gfx.executeIRQueue(IRQueue, startIdx, next);
         if (startIdx == length) {
           self.stats.render = Date.now();
+          gfx.endDrawing();
           if (callback) callback();
         }
       }
@@ -313,10 +313,10 @@ var Page = (function pagePage() {
       }
       return links;
     },
-    startRendering: function pageStartRendering(ctx, callback, errorback)  {
+    startRendering: function pageStartRendering(ctx, callback, textLayer)  {
       this.ctx = ctx;
       this.callback = callback;
-      this.errorback = errorback;
+      this.textLayer = textLayer;
 
       this.startRenderingTime = Date.now();
       this.pdf.startRendering(this);
@@ -569,20 +569,9 @@ var PDFDoc = (function pdfDoc() {
             var properties = data[4];
 
             if (file) {
+              // Rewrap the ArrayBuffer in a stream.
               var fontFileDict = new Dict();
-              fontFileDict.map = file.dict.map;
+              file = new Stream(file, 0, file.length, fontFileDict);
-
-              var fontFile = new Stream(file.bytes, file.start,
-                                        file.end - file.start, fontFileDict);
-
-              // Check if this is a FlateStream. Otherwise just use the created
-              // Stream one. This makes complex_ttf_font.pdf work.
-              var cmf = file.bytes[0];
-              if ((cmf & 0x0f) == 0x08) {
-                file = new FlateStream(fontFile);
-              } else {
-                file = fontFile;
-              }
             }
 
             // For now, resolve the font object here direclty. The real font
@@ -612,8 +601,8 @@ var PDFDoc = (function pdfDoc() {
 
       messageHandler.on('page_error', function pdfDocError(data) {
         var page = this.pageCache[data.pageNum];
-        if (page.errorback)
+        if (page.callback)
-          page.errorback(data.error);
+          page.callback(data.error);
         else
           throw data.error;
       }, this);

src/evaluator.js

@@ -155,6 +155,11 @@ var PartialEvaluator = (function partialEvaluator() {
             font.loadedName = loadedName;
 
             var translated = font.translated;
+            // Convert the file to an ArrayBuffer which will be turned back into
+            // a Stream in the main thread.
+            if (translated.file)
+              translated.file = translated.file.getBytes();
+
             handler.send('obj', [
                 loadedName,
                 'Font',
@@ -493,6 +498,8 @@ var PartialEvaluator = (function partialEvaluator() {
           var baseName = encoding.get('BaseEncoding');
           if (baseName)
             baseEncoding = Encodings[baseName.name];
+          else
+            hasEncoding = false; // base encoding was not provided
 
           // Load the differences between the base and original
           if (encoding.has('Differences')) {
@@ -512,6 +519,7 @@ var PartialEvaluator = (function partialEvaluator() {
           error('Encoding is not a Name nor a Dict');
         }
       }
+
       properties.differences = differences;
       properties.baseEncoding = baseEncoding;
       properties.hasEncoding = hasEncoding;
@@ -554,9 +562,21 @@ var PartialEvaluator = (function partialEvaluator() {
                   var startRange = tokens[j];
                   var endRange = tokens[j + 1];
                   var code = tokens[j + 2];
-                  while (startRange <= endRange) {
+                  if (code == 0xFFFF) {
-                    charToUnicode[startRange] = code++;
+                    // CMap is broken, assuming code == startRange
-                    ++startRange;
+                    code = startRange;
+                  }
+                  if (isArray(code)) {
+                    var codeindex = 0;
+                    while (startRange <= endRange) {
+                      charToUnicode[startRange] = code[codeindex++];
+                      ++startRange;
+                    }
+                  } else {
+                    while (startRange <= endRange) {
+                      charToUnicode[startRange] = code++;
+                      ++startRange;
+                    }
                   }
                 }
                 break;
@@ -595,9 +615,18 @@ var PartialEvaluator = (function partialEvaluator() {
             }
           } else if (byte == 0x3E) {
             if (token.length) {
-              // parsing hex number
+              if (token.length <= 4) {
-              tokens.push(parseInt(token, 16));
+                // parsing hex number
-              token = '';
+                tokens.push(parseInt(token, 16));
+                token = '';
+              } else {
+                // parsing hex UTF-16BE numbers
+                var str = [];
+                for (var i = 0, ii = token.length; i < ii; i += 4)
+                  str.push(parseInt(token.substr(i, 4), 16));
+                tokens.push(String.fromCharCode.apply(String, str));
+                token = '';
+              }
             }
           } else {
             token += String.fromCharCode(byte);

src/fonts.js

@@ -719,20 +719,10 @@ function getUnicodeRangeFor(value) {
   return -1;
 }
 
-function adaptUnicode(unicode) {
-  return (unicode <= 0x1F || (unicode >= 127 && unicode < kSizeOfGlyphArea)) ?
-    unicode + kCmapGlyphOffset : unicode;
-}
-
-function isAdaptedUnicode(unicode) {
-  return unicode >= kCmapGlyphOffset &&
-    unicode < kCmapGlyphOffset + kSizeOfGlyphArea;
-}
-
 function isSpecialUnicode(unicode) {
   return (unicode <= 0x1F || (unicode >= 127 && unicode < kSizeOfGlyphArea)) ||
-    unicode >= kCmapGlyphOffset &&
+    (unicode >= kCmapGlyphOffset &&
-    unicode < kCmapGlyphOffset + kSizeOfGlyphArea;
+    unicode < kCmapGlyphOffset + kSizeOfGlyphArea);
 }
 
 /**
@@ -771,16 +761,21 @@ var Font = (function Font() {
     this.widths = properties.widths;
     this.defaultWidth = properties.defaultWidth;
     this.composite = properties.composite;
-    this.toUnicode = properties.toUnicode;
     this.hasEncoding = properties.hasEncoding;
 
     this.fontMatrix = properties.fontMatrix;
+    this.widthMultiplier = 1.0;
     if (properties.type == 'Type3')
       return;
 
     // Trying to fix encoding using glyph CIDSystemInfo.
     this.loadCidToUnicode(properties);
 
+    if (properties.toUnicode)
+      this.toUnicode = properties.toUnicode;
+    else
+      this.rebuildToUnicode(properties);
+
     if (!file) {
       // The file data is not specified. Trying to fix the font name
       // to be used with the canvas.font.
@@ -832,6 +827,8 @@ var Font = (function Font() {
 
     this.data = data;
     this.fontMatrix = properties.fontMatrix;
+    this.widthMultiplier = !properties.fontMatrix ? 1.0 :
+      1.0 / properties.fontMatrix[0];
     this.encoding = properties.baseEncoding;
     this.hasShortCmap = properties.hasShortCmap;
     this.loadedName = getUniqueName();
@@ -961,15 +958,15 @@ var Font = (function Font() {
     var ranges = [];
     for (var n = 0; n < length; ) {
       var start = codes[n].unicode;
-      var startCode = codes[n].code;
+      var codeIndices = [codes[n].code];
       ++n;
       var end = start;
       while (n < length && end + 1 == codes[n].unicode) {
+        codeIndices.push(codes[n].code);
         ++end;
         ++n;
       }
-      var endCode = codes[n - 1].code;
+      ranges.push([start, end, codeIndices]);
-      ranges.push([start, end, startCode, endCode]);
     }
 
     return ranges;
@@ -1012,17 +1009,16 @@ var Font = (function Font() {
         idDeltas += string16(0);
         idRangeOffsets += string16(offset);
 
-        var startCode = range[2];
+        var codes = range[2];
-        var endCode = range[3];
+        for (var j = 0, jj = codes.length; j < jj; ++j)
-        for (var j = startCode; j <= endCode; ++j)
+          glyphsIds += string16(deltas[codes[j]]);
-          glyphsIds += string16(deltas[j]);
       }
     } else {
       for (var i = 0; i < segCount - 1; i++) {
         var range = ranges[i];
         var start = range[0];
         var end = range[1];
-        var startCode = range[2];
+        var startCode = range[2][0];
 
         startCount += string16(start);
         endCount += string16(end);
@@ -1299,7 +1295,7 @@ var Font = (function Font() {
           properties.baseEncoding = encoding;
       }
 
-      function replaceCMapTable(cmap, font, properties) {
+      function readCMapTable(cmap, font) {
         var start = (font.start ? font.start : 0) + cmap.offset;
         font.pos = start;
 
@@ -1316,7 +1312,7 @@ var Font = (function Font() {
         }
 
         // Check that table are sorted by platformID then encodingID,
-        records.sort(function fontReplaceCMapTableSort(a, b) {
+        records.sort(function fontReadCMapTableSort(a, b) {
           return ((a.platformID << 16) + a.encodingID) -
                  ((b.platformID << 16) + b.encodingID);
         });
@@ -1371,16 +1367,15 @@ var Font = (function Font() {
             for (var j = 0; j < 256; j++) {
               var index = font.getByte();
               if (index) {
-                var unicode = adaptUnicode(j);
+                glyphs.push({ unicode: j, code: j });
-                glyphs.push({ unicode: unicode, code: j });
                 ids.push(index);
               }
             }
-
+            return {
-            properties.hasShortCmap = true;
+              glyphs: glyphs,
-
+              ids: ids,
-            createGlyphNameMap(glyphs, ids, properties);
+              hasShortCmap: true
-            return cmap.data = createCMapTable(glyphs, ids);
+            };
           } else if (format == 4) {
             // re-creating the table in format 4 since the encoding
             // might be changed
@@ -1432,17 +1427,18 @@ var Font = (function Font() {
                 var glyphCode = offsetIndex < 0 ? j :
                   offsets[offsetIndex + j - start];
                 glyphCode = (glyphCode + delta) & 0xFFFF;
-                if (glyphCode == 0 || isAdaptedUnicode(j))
+                if (glyphCode == 0)
                   continue;
 
-                var unicode = adaptUnicode(j);
+                glyphs.push({ unicode: j, code: j });
-                glyphs.push({ unicode: unicode, code: j });
                 ids.push(glyphCode);
               }
             }
 
-            createGlyphNameMap(glyphs, ids, properties);
+            return {
-            return cmap.data = createCMapTable(glyphs, ids);
+              glyphs: glyphs,
+              ids: ids
+            };
           } 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
@@ -1457,19 +1453,18 @@ var Font = (function Font() {
             for (var j = 0; j < entryCount; j++) {
               var glyphCode = int16(font.getBytes(2));
               var code = firstCode + j;
-              if (isAdaptedUnicode(glyphCode))
-                continue;
 
-              var unicode = adaptUnicode(code);
+              glyphs.push({ unicode: code, code: code });
-              glyphs.push({ unicode: unicode, code: code });
               ids.push(glyphCode);
             }
 
-            createGlyphNameMap(glyphs, ids, properties);
+            return {
-            return cmap.data = createCMapTable(glyphs, ids);
+              glyphs: glyphs,
+              ids: ids
+            };
           }
         }
-        return cmap.data;
+        error('Unsupported cmap table format');
       };
 
       function sanitizeMetrics(font, header, metrics, numGlyphs) {
@@ -1708,17 +1703,85 @@ var Font = (function Font() {
           tables.push(cmap);
         }
 
-        var glyphs = [];
+        var cidToGidMap = properties.cidToGidMap || [];
-        for (i = 1; i < numGlyphs; i++) {
+        var gidToCidMap = [0];
-          if (isAdaptedUnicode(i))
+        if (cidToGidMap.length > 0) {
-            continue;
+          for (var j = cidToGidMap.length - 1; j >= 0; j--) {
-
+            var gid = cidToGidMap[j];
-          glyphs.push({ unicode: adaptUnicode(i) });
+            if (gid)
+              gidToCidMap[gid] = j;
+          }
+          // filling the gaps using CID above the CIDs currently used in font
+          var nextCid = cidToGidMap.length;
+          for (var i = 1; i < numGlyphs; i++) {
+            if (!gidToCidMap[i])
+              gidToCidMap[i] = nextCid++;
+          }
         }
-        cmap.data = createCMapTable(glyphs);
+
+        var glyphs = [], ids = [];
+        var usedUnicodes = [];
+        var unassignedUnicodeItems = [];
+        for (var i = 1; i < numGlyphs; i++) {
+          var cid = gidToCidMap[i] || i;
+          var unicode = this.toUnicode[cid];
+          if (!unicode || isSpecialUnicode(unicode) ||
+              unicode in usedUnicodes) {
+            unassignedUnicodeItems.push(i);
+            continue;
+          }
+          usedUnicodes[unicode] = true;
+          glyphs.push({ unicode: unicode, code: cid });
+          ids.push(i);
+        }
+        // trying to fit as many unassigned symbols as we can
+        // in the range allocated for the user defined symbols
+        var unusedUnicode = kCmapGlyphOffset;
+        for (var j = 0, jj = unassignedUnicodeItems.length; j < jj; j++) {
+          var i = unassignedUnicodeItems[j];
+          var cid = gidToCidMap[i] || i;
+          while (unusedUnicode in usedUnicodes)
+            unusedUnicode++;
+          if (unusedUnicode >= kCmapGlyphOffset + kSizeOfGlyphArea)
+            break;
+          var unicode = unusedUnicode++;
+          this.toUnicode[cid] = unicode;
+          usedUnicodes[unicode] = true;
+          glyphs.push({ unicode: unicode, code: cid });
+          ids.push(i);
+        }
+        cmap.data = createCMapTable(glyphs, ids);
       } else {
-        replaceCMapTable(cmap, font, properties);
+        var cmapTable = readCMapTable(cmap, font);
+        var glyphs = cmapTable.glyphs;
+        var ids = cmapTable.ids;
+        var hasShortCmap = !!cmapTable.hasShortCmap;
+        var toUnicode = this.toUnicode;
+
+        if (hasShortCmap && toUnicode) {
+          // checking if cmap is just identity map
+          var isIdentity = true;
+          for (var i = 0, ii = glyphs.length; i < ii; i++) {
+            if (glyphs[i].unicode != i + 1) {
+              isIdentity = false;
+              break;
+            }
+          }
+          // if it is, replacing with meaningful toUnicode values
+          if (isIdentity) {
+            for (var i = 0, ii = glyphs.length; i < ii; i++) {
+              var unicode = toUnicode[i + 1] || i + 1;
+              glyphs[i].unicode = unicode;
+            }
+            this.useToUnicode = true;
+          }
+        }
+        properties.hasShortCmap = hasShortCmap;
+
+        createGlyphNameMap(glyphs, ids, properties);
         this.glyphNameMap = properties.glyphNameMap;
+
+        cmap.data = createCMapTable(glyphs, ids);
       }
 
       // Rewrite the 'post' table if needed
@@ -1808,6 +1871,14 @@ var Font = (function Font() {
         }
         properties.baseEncoding = encoding;
       }
+      if (properties.subtype == 'CIDFontType0C') {
+        var toUnicode = [];
+        for (var i = 0; i < charstrings.length; ++i) {
+          var charstring = charstrings[i];
+          toUnicode[charstring.code] = charstring.unicode;
+        }
+        this.toUnicode = toUnicode;
+      }
 
       var fields = {
         // PostScript Font Program
@@ -1868,8 +1939,11 @@ var Font = (function Font() {
         // Horizontal metrics
         'hmtx': (function fontFieldsHmtx() {
           var hmtx = '\x00\x00\x00\x00'; // Fake .notdef
-          for (var i = 0, ii = charstrings.length; i < ii; i++)
+          for (var i = 0, ii = charstrings.length; i < ii; i++) {
-            hmtx += string16(charstrings[i].width) + string16(0);
+            var charstring = charstrings[i];
+            var width = 'width' in charstring ? charstring.width : 0;
+            hmtx += string16(width) + string16(0);
+          }
           return stringToArray(hmtx);
         })(),
 
@@ -1898,17 +1972,35 @@ var Font = (function Font() {
       return stringToArray(otf.file);
     },
 
-    loadCidToUnicode: function font_loadCidToUnicode(properties) {
+    rebuildToUnicode: function font_rebuildToUnicode(properties) {
-      if (properties.cidToGidMap) {
+      var firstChar = properties.firstChar, lastChar = properties.lastChar;
-        this.cidToUnicode = properties.cidToGidMap;
+      var map = [];
-        return;
+      if (properties.composite) {
+        var isIdentityMap = this.cidToUnicode.length == 0;
+        for (var i = firstChar, ii = lastChar; i <= ii; i++) {
+          // TODO missing map the character according font's CMap
+          var cid = i;
+          map[i] = isIdentityMap ? cid : this.cidToUnicode[cid];
+        }
+      } else {
+        for (var i = firstChar, ii = lastChar; i <= ii; i++) {
+          var glyph = properties.differences[i];
+          if (!glyph)
+            glyph = properties.baseEncoding[i];
+          if (!!glyph && (glyph in GlyphsUnicode))
+            map[i] = GlyphsUnicode[glyph];
+        }
       }
+      this.toUnicode = map;
+    },
 
+    loadCidToUnicode: function font_loadCidToUnicode(properties) {
       if (!properties.cidSystemInfo)
         return;
 
-      var cidToUnicodeMap = [];
+      var cidToUnicodeMap = [], unicodeToCIDMap = [];
       this.cidToUnicode = cidToUnicodeMap;
+      this.unicodeToCID = unicodeToCIDMap;
 
       var cidSystemInfo = properties.cidSystemInfo;
       var cidToUnicode;
@@ -1920,28 +2012,34 @@ var Font = (function Font() {
       if (!cidToUnicode)
         return; // identity encoding
 
-      var glyph = 1, i, j, k, ii;
+      var cid = 1, i, j, k, ii;
       for (i = 0, ii = cidToUnicode.length; i < ii; ++i) {
         var unicode = cidToUnicode[i];
         if (isArray(unicode)) {
           var length = unicode.length;
-          for (j = 0; j < length; j++)
+          for (j = 0; j < length; j++) {
-            cidToUnicodeMap[unicode[j]] = glyph;
+            cidToUnicodeMap[cid] = unicode[j];
-          glyph++;
+            unicodeToCIDMap[unicode[j]] = cid;
+          }
+          cid++;
         } else if (typeof unicode === 'object') {
           var fillLength = unicode.f;
           if (fillLength) {
             k = unicode.c;
             for (j = 0; j < fillLength; ++j) {
-              cidToUnicodeMap[k] = glyph++;
+              cidToUnicodeMap[cid] = k;
+              unicodeToCIDMap[k] = cid;
+              cid++;
               k++;
             }
           } else
-            glyph += unicode.s;
+            cid += unicode.s;
         } else if (unicode) {
-          cidToUnicodeMap[unicode] = glyph++;
+          cidToUnicodeMap[cid] = unicode;
+          unicodeToCIDMap[unicode] = cid;
+          cid++;
         } else
-          glyph++;
+          cid++;
       }
     },
 
@@ -1981,19 +2079,19 @@ var Font = (function Font() {
       switch (this.type) {
         case 'CIDFontType0':
           if (this.noUnicodeAdaptation) {
-            width = this.widths[this.cidToUnicode[charcode]];
+            width = this.widths[this.unicodeToCID[charcode] || charcode];
             unicode = charcode;
             break;
           }
-          unicode = adaptUnicode(this.cidToUnicode[charcode] || charcode);
+          unicode = this.toUnicode[charcode] || charcode;
           break;
         case 'CIDFontType2':
           if (this.noUnicodeAdaptation) {
-            width = this.widths[this.cidToUnicode[charcode]];
+            width = this.widths[this.unicodeToCID[charcode] || charcode];
             unicode = charcode;
             break;
           }
-          unicode = adaptUnicode(this.cidToUnicode[charcode] || charcode);
+          unicode = this.toUnicode[charcode] || charcode;
           break;
         case 'Type1':
           var glyphName = this.differences[charcode] || this.encoding[charcode];
@@ -2004,7 +2102,7 @@ var Font = (function Font() {
             break;
           }
           unicode = this.glyphNameMap[glyphName] ||
-            adaptUnicode(GlyphsUnicode[glyphName] || charcode);
+            GlyphsUnicode[glyphName] || charcode;
           break;
         case 'Type3':
           var glyphName = this.differences[charcode] || this.encoding[charcode];
@@ -2022,16 +2120,16 @@ var Font = (function Font() {
             break;
           }
           if (!this.hasEncoding) {
-            unicode = adaptUnicode(charcode);
+            unicode = this.useToUnicode ? this.toUnicode[charcode] : charcode;
             break;
           }
-          if (this.hasShortCmap) {
+          if (this.hasShortCmap && false) {
             var j = Encodings.MacRomanEncoding.indexOf(glyphName);
-            unicode = j >= 0 && !isSpecialUnicode(j) ? j :
+            unicode = j >= 0 ? j :
               this.glyphNameMap[glyphName];
           } else {
             unicode = glyphName in GlyphsUnicode ?
-              adaptUnicode(GlyphsUnicode[glyphName]) :
+              GlyphsUnicode[glyphName] :
               this.glyphNameMap[glyphName];
           }
           break;
@@ -2039,9 +2137,17 @@ var Font = (function Font() {
           warn('Unsupported font type: ' + this.type);
           break;
       }
+
+      var unicodeChars = this.toUnicode ? this.toUnicode[charcode] : charcode;
+      if (typeof unicodeChars === 'number')
+        unicodeChars = String.fromCharCode(unicodeChars);
+
+      width = (isNum(width) ? width : this.defaultWidth) * this.widthMultiplier;
+
       return {
-        unicode: unicode,
+        fontChar: String.fromCharCode(unicode),
-        width: isNum(width) ? width : this.defaultWidth,
+        unicode: unicodeChars,
+        width: width,
         codeIRQueue: codeIRQueue
       };
     },
@@ -2753,22 +2859,13 @@ CFF.prototype = {
   getOrderedCharStrings: function cff_getOrderedCharStrings(glyphs,
                                                             properties) {
     var charstrings = [];
-    var reverseMapping = {};
-    var encoding = properties.baseEncoding;
     var i, length, glyphName;
-    for (i = 0, length = encoding.length; i < length; ++i) {
-      glyphName = encoding[i];
-      if (!glyphName || isSpecialUnicode(i))
-        continue;
-      reverseMapping[glyphName] = i;
-    }
-    reverseMapping['.notdef'] = 0;
     var unusedUnicode = kCmapGlyphOffset;
     for (i = 0, length = glyphs.length; i < length; i++) {
       var item = glyphs[i];
       var glyphName = item.glyph;
-      var unicode = glyphName in reverseMapping ?
+      var unicode = glyphName in GlyphsUnicode ?
-        reverseMapping[glyphName] : unusedUnicode++;
+        GlyphsUnicode[glyphName] : unusedUnicode++;
       charstrings.push({
         glyph: glyphName,
         unicode: unicode,
@@ -3055,16 +3152,14 @@ var Type2CFF = (function type2CFF() {
       }
 
       var charStrings = this.parseIndex(topDict.CharStrings);
-      var charset = this.parseCharsets(topDict.charset,
-                                       charStrings.length, strings);
-      var encoding = this.parseEncoding(topDict.Encoding, properties,
-                                             strings, charset);
 
       var charset, encoding;
       var isCIDFont = properties.subtype == 'CIDFontType0C';
       if (isCIDFont) {
-        charset = [];
+        charset = ['.notdef'];
-        charset.length = charStrings.length;
+        for (var i = 1, ii = charStrings.length; i < ii; ++i)
+          charset.push('glyph' + i);
+
         encoding = this.parseCidMap(topDict.charset,
                                     charStrings.length);
       } else {
@@ -3133,38 +3228,44 @@ var Type2CFF = (function type2CFF() {
       var charstrings = [];
       var unicodeUsed = [];
       var unassignedUnicodeItems = [];
+      var inverseEncoding = [];
+      for (var charcode in encoding)
+        inverseEncoding[encoding[charcode]] = charcode | 0;
       for (var i = 0, ii = charsets.length; i < ii; i++) {
         var glyph = charsets[i];
-        var encodingFound = false;
+        if (glyph == '.notdef') {
-        for (var charcode in encoding) {
+          charstrings.push({
-          if (encoding[charcode] == i) {
+            unicode: 0,
-            var code = charcode | 0;
+            code: 0,
-            charstrings.push({
+            gid: i,
-              unicode: adaptUnicode(code),
+            glyph: glyph
-              code: code,
+          });
-              gid: i,
+          continue;
-              glyph: glyph
-            });
-            unicodeUsed[code] = true;
-            encodingFound = true;
-            break;
-          }
         }
-        if (!encodingFound) {
+        var code = inverseEncoding[i];
+        if (!code || isSpecialUnicode(code)) {
           unassignedUnicodeItems.push(i);
+          continue;
         }
+        charstrings.push({
+          unicode: code,
+          code: code,
+          gid: i,
+          glyph: glyph
+        });
+        unicodeUsed[code] = true;
       }
 
-      var nextUnusedUnicode = 0x21;
+      var nextUnusedUnicode = kCmapGlyphOffset;
       for (var j = 0, jj = unassignedUnicodeItems.length; j < jj; ++j) {
         var i = unassignedUnicodeItems[j];
         // giving unicode value anyway
-        while (unicodeUsed[nextUnusedUnicode])
+        while (nextUnusedUnicode in unicodeUsed)
           nextUnusedUnicode++;
-        var code = nextUnusedUnicode++;
+        var unicode = nextUnusedUnicode++;
         charstrings.push({
-          unicode: adaptUnicode(code),
+          unicode: unicode,
-          code: code,
+          code: inverseEncoding[i] || 0,
           gid: i,
           glyph: charsets[i]
         });

src/glyphlist.js

@@ -4287,6 +4287,7 @@ var GlyphsUnicode = {
   zretroflexhook: 0x0290,
   zstroke: 0x01B6,
   zuhiragana: 0x305A,
-  zukatakana: 0x30BA
+  zukatakana: 0x30BA,
+  '.notdef': 0x0000
 };
 

src/metrics.js

@@ -3,6 +3,9 @@
 
 'use strict';
 
+// The Metrics object contains glyph widths (in glyph space units).
+// As per PDF spec, for most fonts (Type 3 being an exception) a glyph
+// space unit corresponds to 1/1000th of text space unit.
 var Metrics = {
   'Courier': 600,
   'Courier-Bold': 600,

test/driver.js

@@ -162,11 +162,11 @@ function nextPage(task, loadError) {
 
       page.startRendering(
         ctx,
-        function nextPageStartRendering() {
+        function nextPageStartRendering(error) {
-          snapshotCurrentPage(task, false);
+          var failureMessage = false;
-        },
+          if (error)
-        function errorNextPageStartRendering(e) {
+            failureMessage = 'render : ' + error.message;
-          snapshotCurrentPage(task, 'render : ' + e.message);
+          snapshotCurrentPage(task, failureMessage);
         }
       );
     } catch (e) {

test/pdfs/.gitignore

@@ -16,3 +16,4 @@
 !alphatrans.pdf
 !devicen.pdf
 !cmykjpeg.pdf
+!issue840.pdf

test/pdfs/piperine.pdf.link

@@ -0,0 +1 @@
+http://www.erowid.org/archive/rhodium/chemistry/3base/piperonal.pepper/piperine.pepper/465e03piperine.pdf

test/pdfs/protectip.pdf.link

@@ -0,0 +1 @@
+http://leahy.senate.gov/imo/media/doc/BillText-PROTECTIPAct.pdf

test/test_manifest.json

@@ -276,5 +276,25 @@
        "link": false,
        "rounds": 1,
        "type": "eq"
+    },
+    {  "id": "protectip",
+       "file": "pdfs/protectip.pdf",
+       "md5": "676e7a7b8f96d04825361832b1838a93",
+       "link": true,
+       "rounds": 1,
+       "type": "eq"
+    },
+    {  "id": "piperine",
+       "file": "pdfs/piperine.pdf",
+       "md5": "603ca43dc5732dbba1579f122958c0c2",
+       "link": true,
+       "rounds": 1,
+       "type": "eq"
+    },
+    {  "id": "issue840",
+       "file": "pdfs/issue840.pdf",
+       "md5": "20d88011dd7e3c4fb5274979094dab93",
+       "rounds": 1,
+       "type": "eq"
     }
 ]

web/viewer.css

@@ -232,6 +232,27 @@ canvas {
   -webkit-box-shadow: 0px 2px 10px #ff0;
 }
 
+.textLayer {
+  position: absolute;
+  left: 0;
+  top: 0;
+  right: 0;
+  bottom: 0;
+  color: #000;
+}
+
+.textLayer > div {
+  color: transparent;
+  position: absolute;
+  line-height:1.3;
+}
+
+/* TODO: file FF bug to support ::-moz-selection:window-inactive
+   so we can override the opaque grey background when the window is inactive;
+   see https://bugzilla.mozilla.org/show_bug.cgi?id=706209 */
+::selection { background:rgba(0,0,255,0.3); }
+::-moz-selection { background:rgba(0,0,255,0.3); }
+
 #viewer {
   margin: 44px 0px 0px;
   padding: 8px 0px;

web/viewer.js

@@ -263,7 +263,7 @@ var PDFView = {
     var container = document.getElementById('viewer');
     while (container.hasChildNodes())
       container.removeChild(container.lastChild);
-    
+
     var pdf;
     try {
       pdf = new PDFJS.PDFDoc(data);
@@ -291,10 +291,10 @@ var PDFView = {
       pagesRefMap[pageRef.num + ' ' + pageRef.gen + ' R'] = i;
     }
 
-    this.setScale(scale || kDefaultScale, true);
-
     this.pagesRefMap = pagesRefMap;
     this.destinations = pdf.catalog.destinations;
+    this.setScale(scale || kDefaultScale, true);
+
     if (pdf.catalog.documentOutline) {
       this.outline = new DocumentOutlineView(pdf.catalog.documentOutline);
       var outlineSwitchButton = document.getElementById('outlineSwitch');
@@ -542,6 +542,10 @@ var PageView = function pageView(container, content, id, pageWidth, pageHeight,
     div.appendChild(canvas);
     this.canvas = canvas;
 
+    var textLayer = document.createElement('div');
+    textLayer.className = 'textLayer';
+    div.appendChild(textLayer);
+
     var scale = this.scale;
     canvas.width = pageWidth * scale;
     canvas.height = pageHeight * scale;
@@ -555,14 +559,13 @@ var PageView = function pageView(container, content, id, pageWidth, pageHeight,
 
     stats.begin = Date.now();
     this.content.startRendering(ctx,
-      (function pageViewDrawCallback() {
+      (function pageViewDrawCallback(error) {
+        if (error)
+          PDFView.error('An error occurred while rendering the page.', error);
         this.updateStats();
         if (this.onAfterDraw)
           this.onAfterDraw();
-      }).bind(this),
+      }).bind(this), textLayer
-      function pageViewErrorback(e) {
-        PDFView.error('An error occurred while rendering the page.', e);
-      }
     );
 
     setupLinks(this.content, this.scale);