Minimal images support (updated)

.gitignore

@@ -5,4 +5,7 @@
 *.swp
 *.out
 node_modules/
+target/
+.idea
+.npmrc
 

.npmignore

@@ -0,0 +1,11 @@
+## generic files to ignore
+*~
+*.lock
+*.DS_Store
+*.swp
+*.out
+node_modules/
+target/
+
+.gitignore
+test

.travis.yml

@@ -0,0 +1,3 @@
+language: node_js
+node_js:
+  - "14.18.0"

base/core/core.js

@@ -170,13 +170,17 @@ var Page = (function PageClosure() {
 
 
         var opList = new OperatorList(handler, self.pageIndex);
-
-        handler.send('StartRenderPage', {
-          transparency: partialEvaluator.hasBlendModes(self.resources),
-          pageIndex: self.pageIndex
-        });
-        partialEvaluator.getOperatorList(contentStream, self.resources, opList);
-        pageListPromise.resolve(opList);
+        try {
+            handler.send('StartRenderPage', {
+            transparency: partialEvaluator.hasBlendModes(self.resources),
+            pageIndex: self.pageIndex
+            });
+            partialEvaluator.getOperatorList(contentStream, self.resources, opList);
+            pageListPromise.resolve(opList);
+        }
+        catch(e) {
+            pageListPromise.reject(e);
+        }
       });
 
       var annotationsPromise = pdfManager.ensure(this, 'annotations');

base/core/crypto.js

@@ -553,17 +553,17 @@ var CipherTransformFactory = (function CipherTransformFactoryClosure() {
   function CipherTransformFactory(dict, fileId, password) {
     var filter = dict.get('Filter');
     if (!isName(filter) || filter.name != 'Standard')
-      error('unknown encryption method');
+      error('Error: unknown encryption method');
     this.dict = dict;
     var algorithm = dict.get('V');
     if (!isInt(algorithm) ||
       (algorithm != 1 && algorithm != 2 && algorithm != 4))
-      error('unsupported encryption algorithm');
+      error('Error: unsupported encryption algorithm');
     this.algorithm = algorithm;
     var keyLength = dict.get('Length') || 40;
     if (!isInt(keyLength) ||
       keyLength < 40 || (keyLength % 8) !== 0)
-      error('invalid key length');
+      error('Error: invalid key length');
     // prepare keys
     var ownerPassword = stringToBytes(dict.get('O')).subarray(0, 32);
     var userPassword = stringToBytes(dict.get('U')).subarray(0, 32);

base/core/fonts.js

@@ -2220,8 +2220,9 @@ var Font = (function FontClosure() {
       // name ArialBlack for example will be replaced by Helvetica.
       this.black = (name.search(/Black/g) != -1);
 
+      //MQZ Dec.03.2013 Disable font.remeasure
       // if at least one width is present, remeasure all chars when exists
-      this.remeasure = Object.keys(this.widths).length > 0;
+      //this.remeasure = Object.keys(this.widths).length > 0;
 
       this.encoding = properties.baseEncoding;
       this.noUnicodeAdaptation = true;

base/core/obj.js

@@ -198,7 +198,8 @@ var RefSetCache = (function RefSetCacheClosure() {
     },
 
     has: function RefSetCache_has(ref) {
-      return ('R' + ref.num + '.' + ref.gen) in this.dict;
+      //MQZ. 03/08/2016 fix https://github.com/modesty/pdf2json/issues/26
+      return !!ref ? ('R' + ref.num + '.' + ref.gen) in this.dict : false;
     },
 
     put: function RefSetCache_put(ref, obj) {
@@ -1000,6 +1001,7 @@ var XRef = (function XRefClosure() {
           throw e;
         }
         log('(while reading XRef): ' + e);
+        error(e);
       }
 
       if (recoveryMode)

base/core/parser.js

@@ -534,9 +534,8 @@ var Lexer = (function LexerClosure() {
           str += String.fromCharCode(ch);
         }
       }
-      if (str.length > 128) {
-        error('Warning: name token is longer than allowed by the spec: ' +
-              str.length);
+      if (str.length > 127) {
+        warn('Name token is longer than allowed by the spec: ' + str.length);
       }
       return new Name(str);
     },

base/core/worker.js

@@ -375,8 +375,9 @@ var WorkerMessageHandler = PDFJS.WorkerMessageHandler = {
         var start = Date.now();
         page.extractTextContent().then(function(textContent) {
           promise.resolve(textContent);
-          log('text indexing: page=%d - time=%dms', pageNum,
-              Date.now() - start);
+          //MQZ 03/17/2016 comment out log
+          //log('text indexing: page=%d - time=%dms', pageNum,
+          //    Date.now() - start);
         }, function (e) {
           // Skip errored pages
           promise.reject(e);

base/display/api.js

@@ -412,8 +412,13 @@ var PDFPageProxy = (function PDFPageProxyClosure() {
             return;
           }
           stats.time('Rendering');
-          internalRenderTask.initalizeGraphics(transparency);
-          internalRenderTask.operatorListChanged();
+          try {//MQZ. catch canvas drawing exceptions
+            internalRenderTask.initalizeGraphics(transparency);
+            internalRenderTask.operatorListChanged();
+          }
+          catch(err) {
+            complete(err); 
+          }
         },
         function pageDisplayReadPromiseError(reason) {
           complete(reason);
@@ -862,8 +867,8 @@ var WorkerTransport = (function WorkerTransportClosure() {
           promise.resolve({ data: buf, width: width, height: height});
         }).bind(this);
 //MQZ. Oct.17.2012 - disable image drawing
-//          img.src = imageUrl;
-          img.src = 'data:image/jpeg;base64,' + img.btoa(imageUrl);
+          img.src = imageUrl;
+//          img.src = 'data:image/jpeg;base64,' + img.btoa(imageUrl);
       });
     },
 

base/display/canvas.js

@@ -918,12 +918,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
       var rule = italic + ' ' + bold + ' ' + browserFontSize + 'px ' + typeface;
       this.ctx.font = rule;
 
-        //MQZ.Oct.23.2012. enable font detection
-        if (!fontObj.spaceWidth) {
-            var spaceId = isArray(fontObj.toFontChar) ? fontObj.toFontChar.indexOf(32) : -1;
-            fontObj.spaceWidth = (spaceId >= 0 && isArray(fontObj.widths)) ? fontObj.widths[spaceId] : 250;
-        }
-        this.ctx.setFont(fontObj);
+      this.ctx.setFont(fontObj);
     },
     setTextRenderingMode: function CanvasGraphics_setTextRenderingMode(mode) {
       this.current.textRenderingMode = mode;
@@ -1041,7 +1036,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
     showText: function CanvasGraphics_showText(glyphs, skipTextSelection) {
       var ctx = this.ctx;
       var current = this.current;
-      var font = current.font;
+      var font = current.font || {};
       var fontSize = current.fontSize;
       var fontSizeScale = current.fontSizeScale;
       var charSpacing = current.charSpacing;
@@ -1100,19 +1095,20 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
         this.processingType3 = null;
       } else {
         ctx.save();
-          var tx = 0;
 
 //MQZ Dec.04.2013 handles leading word spacing
+          var tx = 0;
           if (wordSpacing !== 0) {
-              var firstGlyph = _.find(glyphs, function(g) { return _.isObject(g);});
+              var firstGlyph = glyphs.filter(g => g && ('fontChar' in g || 'unicode' in g))[0];
               if (firstGlyph && (firstGlyph.fontChar === ' ' || firstGlyph.unicode === ' ')) {
-                  if (_.find(glyphs, function(g) { return _.isObject(g) && g.unicode !== ' ';})) {
-                    current.x += wordSpacing * fontSize * textHScale;
-                  }
+                tx = wordSpacing * fontSize * textHScale;
               }
           }
 
+        current.x += tx
         this.applyTextTransforms();
+        current.x -= tx
+        // MQZ-GYJ Apr.20.2017 handles leading word spacing over
 
         var lineWidth = current.lineWidth;
         var a1 = current.textMatrix[0], b1 = current.textMatrix[1];
@@ -1167,9 +1163,6 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
               scaledY = 0;
             }
 
-            //MQZ Dec.03.2013 Disable font.remeasure
-              font.remeasure = false;
-
             if (font.remeasure && width > 0) {
               // some standard fonts may not have the exact width, trying to
               // rescale per character
@@ -1274,7 +1267,12 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
       }
 
 //MQZ Nov.28.2012 Adjust Text Positions, and also make it a string
-        var stGlyphs = [];
+      var stGlyphs = [];
+      var spaceWidth = font.spaceWidth;
+      if (!font.spaceWidth) {
+          var spaceId = isArray(font.toFontChar) ? font.toFontChar.indexOf(32) : -1;
+          spaceWidth = (spaceId >= 0 && isArray(font.widths)) ? font.widths[spaceId] : 250;
+      }
 
       for (var i = 0; i < arrLength; ++i) {
         var e = arr[i];
@@ -1289,7 +1287,8 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
               }
           }
           else {
-              if (-e >= font.spaceWidth) {
+            //MQZ-GYJ. Apr.20.2017 split word when spacing is a positive number but very big
+              if (Math.abs(e) >= spaceWidth) {
                   if (vertical) {
                       current.y += spacingLength;
                   } else {
@@ -1536,6 +1535,7 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
       var depth = this.current.paintFormXObjectDepth;
       do {
         this.restore();
+        this.current.paintFormXObjectDepth--;
         // some pdf don't close all restores inside object
         // closing those for them
       } while (this.current.paintFormXObjectDepth >= depth);
@@ -1619,6 +1619,10 @@ var CanvasGraphics = (function CanvasGraphicsClosure() {
     },
 
     endGroup: function CanvasGraphics_endGroup(group) {
+        //MQZ. make sure endGroup is always invoked after beginGroup
+        if (this.groupLevel == 0)
+            this.beginGroup(group);
+
       this.groupLevel--;
       var groupCtx = this.ctx;
       this.ctx = this.groupStack.pop();

base/display/metadata.js

@@ -57,7 +57,7 @@ var Metadata = PDFJS.Metadata = (function MetadataClosure() {
       var doc = this.metaDocument;
       var rdf = doc.documentElement;
 
-      if (rdf.nodeName.toLowerCase() !== 'rdf:rdf') { // Wrapped in <xmpmeta>
+      if (rdf && rdf.nodeName.toLowerCase() !== 'rdf:rdf') { // Wrapped in <xmpmeta>
         rdf = rdf.firstChild;
         while (rdf && rdf.nodeName.toLowerCase() !== 'rdf:rdf')
           rdf = rdf.nextSibling;

base/shared/annotation.js

@@ -667,7 +667,7 @@ var LinkAnnotation = (function LinkAnnotationClosure() {
 
   // Lets URLs beginning with 'www.' default to using the 'http://' protocol.
   function addDefaultProtocolToUrl(url) {
-    if (url.indexOf('www.') === 0) {
+    if (url && url.indexOf('www.') === 0) {
       return ('http://' + url);
     }
     return url;

base/shared/util.js

@@ -145,8 +145,11 @@ var OPS = PDFJS.OPS = {
 
 //MQZ.Mar.22 Disabled Operators (to prevent image painting & annotation default appearance)
 //paintJpegXObject, paintImageMaskXObject, paintImageMaskXObjectGroup, paintImageXObject, paintInlineImageXObject, paintInlineImageXObjectGroup
-var NO_OPS = PDFJS.NO_OPS = [82, 83, 84, 85, 86, 87];
+var NO_OPS = PDFJS.NO_OPS = [83, 84, 86, 87];
 var NO_OPS_RANGE = PDFJS.NO_OPS_RANGE = [78, 79, 80, 81]; //range pairs, all ops with each pair will be skipped. !important!
+var HTMLElement = typeof HTMLElement === "undefined" ?
+  function() { } : HTMLElement;
+
 
 // Use only for debugging purposes. This should not be used in any code that is
 // in mozilla master.
@@ -188,10 +191,10 @@ function error(msg) {
     // Join the arguments into a single string for the lines below.
     msg = [].join.call(arguments, ' ');
   } else {
-    log('Error: ' + msg);
+    //log('Error: ' + msg);
   }
-  log(backtrace());
-  PDFJS.LogManager.notify('error', msg);
+  //log(backtrace());
+  //PDFJS.LogManager.notify('error', msg);
   throw new Error(msg);
 }
 
@@ -1229,9 +1232,9 @@ function loadJpegStream(id, imageUrl, objs) {
   img.onload = (function loadJpegStream_onloadClosure() {
     objs.resolve(id, img);
   });
-//  img.src = imageUrl;
+  img.src = imageUrl;
     //MQZ. Apr.09.2013 calls windows.btoa safely
-    img.src = 'data:image/jpeg;base64,' + img.btoa(imageUrl);
+//    img.src = 'data:image/jpeg;base64,' + img.btoa(imageUrl);
 }
 
 //MQZ Oct.18.2013 expose util methods
@@ -1241,18 +1244,18 @@ nodeUtil.p2jinfo = info;
 nodeUtil.p2jwarn = warn;
 nodeUtil.p2jerror = error;
 nodeUtil.verbosity = function(verbo) {
-    if (!isNaN(verbo)) {
+    if (isNaN(verbo)) {
+        verbosity = WARNINGS;
+    }
+    else {
         if (verbo <= ERRORS) {
-            verbo = ERRORS;
+            verbosity = ERRORS;
         }
         else if (verbo >= INFOS) {
-            verbo = INFOS;
+            verbosity = INFOS;
         }
-
-        verbosity = verbo;
-    }
-    else {
-        verbosity = ERRORS;
+        else
+            verbosity = verbo;
     }
 };
 nodeUtil.verbosity();

bin/pdf2json

@@ -1,6 +1,4 @@
 #!/usr/bin/env node
 
-'use strict';
-
-var P2JCMD = require('../lib/p2jcmd');
+const P2JCMD = require('../lib/p2jcmd');
 new P2JCMD().start();

lib/imagedata.js

@@ -0,0 +1,13 @@
+function ImageData(width, height) {
+  if (width && width instanceof Object) {
+    this.width = width.width;
+    this.height = width.height;
+    this.data = width.data;
+  } else {
+    this.width = width;
+    this.height = height;
+    this.data = new Uint8Array(width * height * 4);
+  }
+};
+
+module.exports = ImageData;