PDF.js version 1.3.34 - See mozilla/pdf.js@0819d71a027020d44b37a31808bf778fc2c0ea14

bower.json

@@ -1,6 +1,6 @@
 {
   "name": "pdfjs-dist",
-  "version": "1.3.32",
+  "version": "1.3.34",
   "main": [
     "build/pdf.js",
     "build/pdf.worker.js"

build/pdf.combined.js

@@ -20,8 +20,8 @@ if (typeof PDFJS === 'undefined') {
   (typeof window !== 'undefined' ? window : this).PDFJS = {};
 }
 
-PDFJS.version = '1.3.32';
-PDFJS.build = 'c2dfe9e';
+PDFJS.version = '1.3.34';
+PDFJS.build = '0819d71';
 
 (function pdfjsWrapper() {
   // Use strict in our context only - users might not want it
@@ -2354,6 +2354,14 @@ var PDFDocumentProxy = (function PDFDocumentProxyClosure() {
   return PDFDocumentProxy;
 })();
 
+/**
+ * Page getTextContent parameters.
+ *
+ * @typedef {Object} getTextContentParameters
+ * @param {boolean} normalizeWhitespace - replaces all occurrences of
+ *   whitespace with standard spaces (0x20). The default value is `false`.
+ */
+
 /**
  * Page text content.
  *
@@ -2632,12 +2640,16 @@ var PDFPageProxy = (function PDFPageProxyClosure() {
     },
 
     /**
+     * @param {getTextContentParameters} params - getTextContent parameters.
      * @return {Promise} That is resolved a {@link TextContent}
      * object that represent the page text content.
      */
-    getTextContent: function PDFPageProxy_getTextContent() {
+    getTextContent: function PDFPageProxy_getTextContent(params) {
+      var normalizeWhitespace = (params && params.normalizeWhitespace) || false;
+
       return this.transport.messageHandler.sendWithPromise('GetTextContent', {
-        pageIndex: this.pageNumber - 1
+        pageIndex: this.pageNumber - 1,
+        normalizeWhitespace: normalizeWhitespace,
       });
     },
 
@@ -9909,7 +9921,8 @@ var Page = (function PageClosure() {
       });
     },
 
-    extractTextContent: function Page_extractTextContent(task) {
+    extractTextContent: function Page_extractTextContent(task,
+                                                         normalizeWhitespace) {
       var handler = {
         on: function nullHandlerOn() {},
         send: function nullHandlerSend() {}
@@ -9939,7 +9952,9 @@ var Page = (function PageClosure() {
 
         return partialEvaluator.getTextContent(contentStream,
                                                task,
-                                               self.resources);
+                                               self.resources,
+                                               /* stateManager = */ null,
+                                               normalizeWhitespace);
       });
     },
 
@@ -18842,12 +18857,15 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
       });
     },
 
-    getTextContent: function PartialEvaluator_getTextContent(stream, task,
-                                                             resources,
-                                                             stateManager) {
+    getTextContent:
+        function PartialEvaluator_getTextContent(stream, task, resources,
+                                                 stateManager,
+                                                 normalizeWhitespace) {
 
       stateManager = (stateManager || new StateManager(new TextState()));
 
+      var WhitespaceRegexp = /\s/g;
+
       var textContent = {
         items: [],
         styles: Object.create(null)
@@ -18961,11 +18979,23 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         return textContentItem;
       }
 
+      function replaceWhitespace(str) {
+        // Replaces all whitespaces with standard spaces (0x20), to avoid
+        // alignment issues between the textLayer and the canvas if the text
+        // contains e.g. tabs (fixes issue6612.pdf).
+        var i = 0, ii = str.length, code;
+        while (i < ii && (code = str.charCodeAt(i)) >= 0x20 && code <= 0x7F) {
+          i++;
+        }
+        return (i < ii ? str.replace(WhitespaceRegexp, ' ') : str);
+      }
+
       function runBidiTransform(textChunk) {
         var str = textChunk.str.join('');
         var bidiResult = PDFJS.bidi(str, -1, textChunk.vertical);
         return {
-          str: bidiResult.str,
+          str: (normalizeWhitespace ? replaceWhitespace(bidiResult.str) :
+                                      bidiResult.str),
           dir: bidiResult.dir,
           width: textChunk.width,
           height: textChunk.height,
@@ -19286,8 +19316,8 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               }
 
               return self.getTextContent(xobj, task,
-                xobj.dict.get('Resources') || resources, stateManager).
-                then(function (formTextContent) {
+                xobj.dict.get('Resources') || resources, stateManager,
+                normalizeWhitespace).then(function (formTextContent) {
                   Util.appendToArray(textContent.items, formTextContent.items);
                   Util.extendObj(textContent.styles, formTextContent.styles);
                   stateManager.restore();
@@ -41996,12 +42026,14 @@ var WorkerMessageHandler = PDFJS.WorkerMessageHandler = {
 
     handler.on('GetTextContent', function wphExtractText(data) {
       var pageIndex = data.pageIndex;
+      var normalizeWhitespace = data.normalizeWhitespace;
       return pdfManager.getPage(pageIndex).then(function(page) {
         var task = new WorkerTask('GetTextContent: page ' + pageIndex);
         startWorkerTask(task);
         var pageNum = pageIndex + 1;
         var start = Date.now();
-        return page.extractTextContent(task).then(function(textContent) {
+        return page.extractTextContent(task, normalizeWhitespace).then(
+            function(textContent) {
           finishWorkerTask(task);
           info('text indexing: page=' + pageNum + ' - time=' +
                (Date.now() - start) + 'ms');

build/pdf.js

@@ -20,8 +20,8 @@ if (typeof PDFJS === 'undefined') {
   (typeof window !== 'undefined' ? window : this).PDFJS = {};
 }
 
-PDFJS.version = '1.3.32';
-PDFJS.build = 'c2dfe9e';
+PDFJS.version = '1.3.34';
+PDFJS.build = '0819d71';
 
 (function pdfjsWrapper() {
   // Use strict in our context only - users might not want it
@@ -2354,6 +2354,14 @@ var PDFDocumentProxy = (function PDFDocumentProxyClosure() {
   return PDFDocumentProxy;
 })();
 
+/**
+ * Page getTextContent parameters.
+ *
+ * @typedef {Object} getTextContentParameters
+ * @param {boolean} normalizeWhitespace - replaces all occurrences of
+ *   whitespace with standard spaces (0x20). The default value is `false`.
+ */
+
 /**
  * Page text content.
  *
@@ -2632,12 +2640,16 @@ var PDFPageProxy = (function PDFPageProxyClosure() {
     },
 
     /**
+     * @param {getTextContentParameters} params - getTextContent parameters.
      * @return {Promise} That is resolved a {@link TextContent}
      * object that represent the page text content.
      */
-    getTextContent: function PDFPageProxy_getTextContent() {
+    getTextContent: function PDFPageProxy_getTextContent(params) {
+      var normalizeWhitespace = (params && params.normalizeWhitespace) || false;
+
       return this.transport.messageHandler.sendWithPromise('GetTextContent', {
-        pageIndex: this.pageNumber - 1
+        pageIndex: this.pageNumber - 1,
+        normalizeWhitespace: normalizeWhitespace,
       });
     },
 

build/pdf.worker.js

@@ -20,8 +20,8 @@ if (typeof PDFJS === 'undefined') {
   (typeof window !== 'undefined' ? window : this).PDFJS = {};
 }
 
-PDFJS.version = '1.3.32';
-PDFJS.build = 'c2dfe9e';
+PDFJS.version = '1.3.34';
+PDFJS.build = '0819d71';
 
 (function pdfjsWrapper() {
   // Use strict in our context only - users might not want it
@@ -2849,7 +2849,8 @@ var Page = (function PageClosure() {
       });
     },
 
-    extractTextContent: function Page_extractTextContent(task) {
+    extractTextContent: function Page_extractTextContent(task,
+                                                         normalizeWhitespace) {
       var handler = {
         on: function nullHandlerOn() {},
         send: function nullHandlerSend() {}
@@ -2879,7 +2880,9 @@ var Page = (function PageClosure() {
 
         return partialEvaluator.getTextContent(contentStream,
                                                task,
-                                               self.resources);
+                                               self.resources,
+                                               /* stateManager = */ null,
+                                               normalizeWhitespace);
       });
     },
 
@@ -11782,12 +11785,15 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
       });
     },
 
-    getTextContent: function PartialEvaluator_getTextContent(stream, task,
-                                                             resources,
-                                                             stateManager) {
+    getTextContent:
+        function PartialEvaluator_getTextContent(stream, task, resources,
+                                                 stateManager,
+                                                 normalizeWhitespace) {
 
       stateManager = (stateManager || new StateManager(new TextState()));
 
+      var WhitespaceRegexp = /\s/g;
+
       var textContent = {
         items: [],
         styles: Object.create(null)
@@ -11901,11 +11907,23 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
         return textContentItem;
       }
 
+      function replaceWhitespace(str) {
+        // Replaces all whitespaces with standard spaces (0x20), to avoid
+        // alignment issues between the textLayer and the canvas if the text
+        // contains e.g. tabs (fixes issue6612.pdf).
+        var i = 0, ii = str.length, code;
+        while (i < ii && (code = str.charCodeAt(i)) >= 0x20 && code <= 0x7F) {
+          i++;
+        }
+        return (i < ii ? str.replace(WhitespaceRegexp, ' ') : str);
+      }
+
       function runBidiTransform(textChunk) {
         var str = textChunk.str.join('');
         var bidiResult = PDFJS.bidi(str, -1, textChunk.vertical);
         return {
-          str: bidiResult.str,
+          str: (normalizeWhitespace ? replaceWhitespace(bidiResult.str) :
+                                      bidiResult.str),
           dir: bidiResult.dir,
           width: textChunk.width,
           height: textChunk.height,
@@ -12226,8 +12244,8 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
               }
 
               return self.getTextContent(xobj, task,
-                xobj.dict.get('Resources') || resources, stateManager).
-                then(function (formTextContent) {
+                xobj.dict.get('Resources') || resources, stateManager,
+                normalizeWhitespace).then(function (formTextContent) {
                   Util.appendToArray(textContent.items, formTextContent.items);
                   Util.extendObj(textContent.styles, formTextContent.styles);
                   stateManager.restore();
@@ -34936,12 +34954,14 @@ var WorkerMessageHandler = PDFJS.WorkerMessageHandler = {
 
     handler.on('GetTextContent', function wphExtractText(data) {
       var pageIndex = data.pageIndex;
+      var normalizeWhitespace = data.normalizeWhitespace;
       return pdfManager.getPage(pageIndex).then(function(page) {
         var task = new WorkerTask('GetTextContent: page ' + pageIndex);
         startWorkerTask(task);
         var pageNum = pageIndex + 1;
         var start = Date.now();
-        return page.extractTextContent(task).then(function(textContent) {
+        return page.extractTextContent(task, normalizeWhitespace).then(
+            function(textContent) {
           finishWorkerTask(task);
           info('text indexing: page=' + pageNum + ' - time=' +
                (Date.now() - start) + 'ms');

package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdfjs-dist",
-  "version": "1.3.32",
+  "version": "1.3.34",
   "description": "Generic build of Mozilla's PDF.js library.",
   "keywords": [
     "Mozilla",

web/pdf_viewer.js

@@ -1345,7 +1345,7 @@ var PDFPageView = (function PDFPageViewClosure() {
         function pdfPageRenderCallback() {
           pageViewDrawCallback(null);
           if (textLayer) {
-            self.pdfPage.getTextContent().then(
+            self.pdfPage.getTextContent({ normalizeWhitespace: true }).then(
               function textContentResolved(textContent) {
                 textLayer.setTextContent(textContent);
                 textLayer.render(TEXT_LAYER_RENDER_DELAY);
@@ -2368,7 +2368,7 @@ var PDFViewer = (function pdfViewer() {
       if (!this.pdfDocument) {
         return;
       }
-      
+
       var pageView = this._pages[pageNumber - 1];
 
       if (this.isInPresentationMode) {
@@ -2626,7 +2626,7 @@ var PDFViewer = (function pdfViewer() {
 
     getPageTextContent: function (pageIndex) {
       return this.pdfDocument.getPage(pageIndex + 1).then(function (page) {
-        return page.getTextContent();
+        return page.getTextContent({ normalizeWhitespace: true });
       });
     },