Merge pull request #6590 from yurydelendik/combinechars

Combines standalone chars into text groups.

src/core/evaluator.js

@@ -920,9 +920,26 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         items: [],
         styles: Object.create(null)
       };
-      var bidiTexts = textContent.items;
+      var textContentItem = {
+        initialized: false,
+        str: [],
+        width: 0,
+        height: 0,
+        vertical: false,
+        lastAdvanceWidth: 0,
+        lastAdvanceHeight: 0,
+        textAdvanceScale: 0,
+        spaceWidth: 0,
+        fakeSpaceMin: Infinity,
+        fakeMultiSpaceMin: Infinity,
+        fakeMultiSpaceMax: -0,
+        textRunBreakAllowed: false,
+        transform: null,
+        fontName: null
+      };
       var SPACE_FACTOR = 0.3;
       var MULTI_SPACE_FACTOR = 1.5;
+      var MULTI_SPACE_FACTOR_MAX = 4;
 
       var self = this;
       var xref = this.xref;
@@ -937,7 +954,10 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
 
       var textState;
 
-      function newTextChunk() {
+      function ensureTextContentItem() {
+        if (textContentItem.initialized) {
+          return textContentItem;
+        }
         var font = textState.font;
         if (!(font.loadedName in textContent.styles)) {
           textContent.styles[font.loadedName] = {
@@ -947,24 +967,79 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
             vertical: font.vertical
           };
         }
-        return {
+        textContentItem.fontName = font.loadedName;
-          // |str| is initially an array which we push individual chars to, and
+
-          // then runBidi() overwrites it with the final string.
+        // 9.4.4 Text Space Details
-          str: [],
+        var tsm = [textState.fontSize * textState.textHScale, 0,
-          dir: null,
+                   0, textState.fontSize,
-          width: 0,
+                   0, textState.textRise];
-          height: 0,
+
-          transform: null,
+        if (font.isType3Font &&
-          fontName: font.loadedName
+            textState.fontMatrix !== FONT_IDENTITY_MATRIX &&
-        };
+            textState.fontSize === 1) {
+          var glyphHeight = font.bbox[3] - font.bbox[1];
+          if (glyphHeight > 0) {
+            glyphHeight = glyphHeight * textState.fontMatrix[3];
+            tsm[3] *= glyphHeight;
+          }
+        }
+
+        var trm = Util.transform(textState.ctm,
+                                 Util.transform(textState.textMatrix, tsm));
+        textContentItem.transform = trm;
+        if (!font.vertical) {
+          textContentItem.width = 0;
+          textContentItem.height = Math.sqrt(trm[2] * trm[2] + trm[3] * trm[3]);
+          textContentItem.vertical = false;
+        } else {
+          textContentItem.width = Math.sqrt(trm[0] * trm[0] + trm[1] * trm[1]);
+          textContentItem.height = 0;
+          textContentItem.vertical = true;
+        }
+
+        var a = textState.textLineMatrix[0];
+        var b = textState.textLineMatrix[1];
+        var scaleLineX = Math.sqrt(a * a + b * b);
+        a = textState.ctm[0];
+        b = textState.ctm[1];
+        var scaleCtmX = Math.sqrt(a * a + b * b);
+        textContentItem.textAdvanceScale = scaleCtmX * scaleLineX;
+        textContentItem.lastAdvanceWidth = 0;
+        textContentItem.lastAdvanceHeight = 0;
+
+        var spaceWidth = font.spaceWidth / 1000 * textState.fontSize;
+        if (spaceWidth) {
+          textContentItem.spaceWidth = spaceWidth;
+          textContentItem.fakeSpaceMin = spaceWidth * SPACE_FACTOR;
+          textContentItem.fakeMultiSpaceMin = spaceWidth * MULTI_SPACE_FACTOR;
+          textContentItem.fakeMultiSpaceMax =
+            spaceWidth * MULTI_SPACE_FACTOR_MAX;
+          // It's okay for monospace fonts to fake as much space as needed.
+          textContentItem.textRunBreakAllowed = !font.isMonospace;
+        } else {
+          textContentItem.spaceWidth = 0;
+          textContentItem.fakeSpaceMin = Infinity;
+          textContentItem.fakeMultiSpaceMin = Infinity;
+          textContentItem.fakeMultiSpaceMax = 0;
+          textContentItem.textRunBreakAllowed = false;
+        }
+
+
+        textContentItem.initialized = true;
+        return textContentItem;
       }
 
-      function runBidi(textChunk) {
+      function runBidiTransform(textChunk) {
         var str = textChunk.str.join('');
-        var bidiResult = PDFJS.bidi(str, -1, textState.font.vertical);
+        var bidiResult = PDFJS.bidi(str, -1, textChunk.vertical);
-        textChunk.str = bidiResult.str;
+        return {
-        textChunk.dir = bidiResult.dir;
+          str: bidiResult.str,
-        return textChunk;
+          dir: bidiResult.dir,
+          width: textChunk.width,
+          height: textChunk.height,
+          transform: textChunk.transform,
+          fontName: textChunk.fontName
+        };
       }
 
       function handleSetFont(fontName, fontRef) {
@@ -976,33 +1051,9 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           });
       }
 
-      function buildTextGeometry(chars, textChunk) {
+      function buildTextContentItem(chars) {
         var font = textState.font;
-        textChunk = textChunk || newTextChunk();
+        var textChunk = ensureTextContentItem();
-        if (!textChunk.transform) {
-          // 9.4.4 Text Space Details
-          var tsm = [textState.fontSize * textState.textHScale, 0,
-                     0, textState.fontSize,
-                     0, textState.textRise];
-
-          if (font.isType3Font &&
-              textState.fontMatrix !== FONT_IDENTITY_MATRIX &&
-              textState.fontSize === 1) {
-            var glyphHeight = font.bbox[3] - font.bbox[1];
-            if (glyphHeight > 0) {
-              glyphHeight = glyphHeight * textState.fontMatrix[3];
-              tsm[3] *= glyphHeight;
-            }
-          }
-
-          var trm = textChunk.transform = Util.transform(textState.ctm,
-                                    Util.transform(textState.textMatrix, tsm));
-          if (!font.vertical) {
-            textChunk.height = Math.sqrt(trm[2] * trm[2] + trm[3] * trm[3]);
-          } else {
-            textChunk.width = Math.sqrt(trm[0] * trm[0] + trm[1] * trm[1]);
-          }
-        }
         var width = 0;
         var height = 0;
         var glyphs = font.charsToGlyphs(chars);
@@ -1049,8 +1100,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
             var wordSpacing = textState.wordSpacing;
             charSpacing += wordSpacing;
             if (wordSpacing > 0) {
-              addFakeSpaces(wordSpacing * 1000 / textState.fontSize,
+              addFakeSpaces(wordSpacing, textChunk.str);
-                            textChunk.str);
             }
           }
 
@@ -1071,36 +1121,41 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           textChunk.str.push(glyphUnicode);
         }
 
-        var a = textState.textLineMatrix[0];
-        var b = textState.textLineMatrix[1];
-        var scaleLineX = Math.sqrt(a * a + b * b);
-        a = textState.ctm[0];
-        b = textState.ctm[1];
-        var scaleCtmX = Math.sqrt(a * a + b * b);
         if (!font.vertical) {
-          textChunk.width += width * scaleCtmX * scaleLineX;
+          textChunk.lastAdvanceWidth = width;
+          textChunk.width += width * textChunk.textAdvanceScale;
         } else {
-          textChunk.height += Math.abs(height * scaleCtmX * scaleLineX);
+          textChunk.lastAdvanceHeight = height;
+          textChunk.height += Math.abs(height * textChunk.textAdvanceScale);
         }
+
         return textChunk;
       }
 
       function addFakeSpaces(width, strBuf) {
-        var spaceWidth = textState.font.spaceWidth;
+        if (width < textContentItem.fakeSpaceMin) {
-        if (spaceWidth <= 0) {
           return;
         }
-        var fakeSpaces = width / spaceWidth;
+        if (width < textContentItem.fakeMultiSpaceMin) {
-        if (fakeSpaces > MULTI_SPACE_FACTOR) {
+          strBuf.push(' ');
-          fakeSpaces = Math.round(fakeSpaces);
+          return;
-          while (fakeSpaces--) {
+        }
-            strBuf.push(' ');
+        var fakeSpaces = Math.round(width / textContentItem.spaceWidth);
-          }
+        while (fakeSpaces-- > 0) {
-        } else if (fakeSpaces > SPACE_FACTOR) {
           strBuf.push(' ');
         }
       }
 
+      function flushTextContentItem() {
+        if (!textContentItem.initialized) {
+          return;
+        }
+        textContent.items.push(runBidiTransform(textContentItem));
+
+        textContentItem.initialized = false;
+        textContentItem.str.length = 0;
+      }
+
       var timeSlotManager = new TimeSlotManager();
 
       return new Promise(function next(resolve, reject) {
@@ -1119,35 +1174,62 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           textState = stateManager.state;
           var fn = operation.fn;
           args = operation.args;
+          var advance;
 
           switch (fn | 0) {
             case OPS.setFont:
+              flushTextContentItem();
               textState.fontSize = args[1];
               return handleSetFont(args[0].name).then(function() {
                 next(resolve, reject);
               }, reject);
             case OPS.setTextRise:
+              flushTextContentItem();
               textState.textRise = args[0];
               break;
             case OPS.setHScale:
+              flushTextContentItem();
               textState.textHScale = args[0] / 100;
               break;
             case OPS.setLeading:
+              flushTextContentItem();
               textState.leading = args[0];
               break;
             case OPS.moveText:
+              // Optimization to treat same line movement as advance
+              var isSameTextLine = !textState.font ? false :
+                ((textState.font.vertical ? args[0] : args[1]) === 0);
+              advance = args[0] - args[1];
+              if (isSameTextLine && textContentItem.initialized &&
+                  advance > 0 &&
+                  advance <= textContentItem.fakeMultiSpaceMax) {
+                textState.translateTextLineMatrix(args[0], args[1]);
+                textContentItem.width +=
+                  (args[0] - textContentItem.lastAdvanceWidth);
+                textContentItem.height +=
+                  (args[1] - textContentItem.lastAdvanceHeight);
+                var diff = (args[0] - textContentItem.lastAdvanceWidth) -
+                           (args[1] - textContentItem.lastAdvanceHeight);
+                addFakeSpaces(diff, textContentItem.str);
+                break;
+              }
+
+              flushTextContentItem();
               textState.translateTextLineMatrix(args[0], args[1]);
               textState.textMatrix = textState.textLineMatrix.slice();
               break;
             case OPS.setLeadingMoveText:
+              flushTextContentItem();
               textState.leading = -args[1];
               textState.translateTextLineMatrix(args[0], args[1]);
               textState.textMatrix = textState.textLineMatrix.slice();
               break;
             case OPS.nextLine:
+              flushTextContentItem();
               textState.carriageReturn();
               break;
             case OPS.setTextMatrix:
+              flushTextContentItem();
               textState.setTextMatrix(args[0], args[1], args[2], args[3],
                 args[4], args[5]);
               textState.setTextLineMatrix(args[0], args[1], args[2], args[3],
@@ -1160,17 +1242,19 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               textState.wordSpacing = args[0];
               break;
             case OPS.beginText:
+              flushTextContentItem();
               textState.textMatrix = IDENTITY_MATRIX.slice();
               textState.textLineMatrix = IDENTITY_MATRIX.slice();
               break;
             case OPS.showSpacedText:
               var items = args[0];
-              var textChunk = newTextChunk();
               var offset;
               for (var j = 0, jj = items.length; j < jj; j++) {
                 if (typeof items[j] === 'string') {
-                  buildTextGeometry(items[j], textChunk);
+                  buildTextContentItem(items[j]);
                 } else {
+                  ensureTextContentItem();
+
                   // PDF Specification 5.3.2 states:
                   // The number is expressed in thousandths of a unit of text
                   // space.
@@ -1179,45 +1263,57 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
                   // In the default coordinate system, a positive adjustment
                   // has the effect of moving the next glyph painted either to
                   // the left or down by the given amount.
-                  var advance = items[j];
+                  advance = items[j] * textState.fontSize / 1000;
-                  var val = advance * textState.fontSize / 1000;
+                  var breakTextRun = false;
                   if (textState.font.vertical) {
-                    offset = val *
+                    offset = advance *
                       (textState.textHScale * textState.textMatrix[2] +
                        textState.textMatrix[3]);
-                    textState.translateTextMatrix(0, val);
+                    textState.translateTextMatrix(0, advance);
-                    // Value needs to be added to height to paint down.
+                    breakTextRun = textContentItem.textRunBreakAllowed &&
-                    textChunk.height += offset;
+                                   advance > textContentItem.fakeMultiSpaceMax;
+                    if (!breakTextRun) {
+                      // Value needs to be added to height to paint down.
+                      textContentItem.height += offset;
+                    }
                   } else {
-                    offset = val * (
+                    advance = -advance;
+                    offset = advance * (
                       textState.textHScale * textState.textMatrix[0] +
                       textState.textMatrix[1]);
-                    textState.translateTextMatrix(-val, 0);
+                    textState.translateTextMatrix(advance, 0);
-                    // Value needs to be subtracted from width to paint left.
+                    breakTextRun = textContentItem.textRunBreakAllowed &&
-                    textChunk.width -= offset;
+                                   advance > textContentItem.fakeMultiSpaceMax;
-                    advance = -advance;
+                    if (!breakTextRun) {
+                      // Value needs to be subtracted from width to paint left.
+                      textContentItem.width += offset;
+                    }
                   }
-                  if (advance > 0) {
+                  if (breakTextRun) {
-                    addFakeSpaces(advance, textChunk.str);
+                    flushTextContentItem();
+                  } else if (advance > 0) {
+                    addFakeSpaces(advance, textContentItem.str);
                   }
                 }
               }
-              bidiTexts.push(runBidi(textChunk));
               break;
             case OPS.showText:
-              bidiTexts.push(runBidi(buildTextGeometry(args[0])));
+              buildTextContentItem(args[0]);
               break;
             case OPS.nextLineShowText:
+              flushTextContentItem();
               textState.carriageReturn();
-              bidiTexts.push(runBidi(buildTextGeometry(args[0])));
+              buildTextContentItem(args[0]);
               break;
             case OPS.nextLineSetSpacingShowText:
+              flushTextContentItem();
               textState.wordSpacing = args[0];
               textState.charSpacing = args[1];
               textState.carriageReturn();
-              bidiTexts.push(runBidi(buildTextGeometry(args[2])));
+              buildTextContentItem(args[2]);
               break;
             case OPS.paintXObject:
+              flushTextContentItem();
               if (args[0].code) {
                 break;
               }
@@ -1229,7 +1325,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               var name = args[0].name;
               if (xobjsCache.key === name) {
                 if (xobjsCache.texts) {
-                  Util.appendToArray(bidiTexts, xobjsCache.texts.items);
+                  Util.appendToArray(textContent.items, xobjsCache.texts.items);
                   Util.extendObj(textContent.styles, xobjsCache.texts.styles);
                 }
                 break;
@@ -1260,7 +1356,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               return self.getTextContent(xobj, task,
                 xobj.dict.get('Resources') || resources, stateManager).
                 then(function (formTextContent) {
-                  Util.appendToArray(bidiTexts, formTextContent.items);
+                  Util.appendToArray(textContent.items, formTextContent.items);
                   Util.extendObj(textContent.styles, formTextContent.styles);
                   stateManager.restore();
 
@@ -1270,6 +1366,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
                   next(resolve, reject);
                 }, reject);
             case OPS.setGState:
+              flushTextContentItem();
               var dictName = args[0];
               var extGState = resources.get('ExtGState');
 
@@ -1300,6 +1397,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
           }, reject);
           return;
         }
+        flushTextContentItem();
         resolve(textContent);
       });
     },

test/pdfs/.gitignore

@@ -137,6 +137,7 @@
 !vertical.pdf
 !ZapfDingbats.pdf
 !bug878026.pdf
+!issue1045.pdf
 !issue5010.pdf
 !issue4934.pdf
 !issue4650.pdf

test/test_manifest.json

@@ -607,6 +607,13 @@
        "link": false,
        "type": "load"
     },
+    {  "id": "issue1045",
+       "file": "pdfs/issue1045.pdf",
+       "md5": "61d7e9bfbc03cd457dcefeab3e78a687",
+       "rounds": 1,
+       "link": false,
+       "type": "text"
+    },
     {  "id": "issue5677",
        "file": "pdfs/issue5677.pdf",
        "md5": "c9101578fcb806269145132724d24ac1",