Merge pull request #262 from ironymark/CIDFont

First Pass CIDFont support

fonts.js

@@ -7,7 +7,7 @@ var isWorker = (typeof window == 'undefined');
 /**
  * Maximum file size of the font.
  */
-var kMaxFontFileSize = 40000;
+var kMaxFontFileSize = 200000;
 
 /**
  * Maximum time to wait for a font to be loaded by font-face rules.
@@ -389,6 +389,7 @@ var Font = (function() {
     var data;
     switch (properties.type) {
       case 'Type1':
+      case 'CIDFontType0':
         var cff = new CFF(name, file, properties);
         this.mimetype = 'font/opentype';
 
@@ -397,6 +398,7 @@ var Font = (function() {
         break;
 
       case 'TrueType':
+      case 'CIDFontType2':
         this.mimetype = 'font/opentype';
 
         // Repair the TrueType file if it is can be damaged in the point of
@@ -409,9 +411,10 @@ var Font = (function() {
         break;
     }
     this.data = data;
-
+    this.type = properties.type;
     this.id = Fonts.registerFont(name, data, properties);
     this.loadedName = 'pdfFont' + this.id;
+    this.compositeFont = properties.compositeFont;
   };
 
   function stringToArray(str) {
@@ -679,6 +682,65 @@ var Font = (function() {
            '\x00\x00\x00\x00';  // maxMemType1
   };
 
+  function createNameTable(name) {
+    var strings = [
+      'Original licence',  // 0.Copyright
+      name,                // 1.Font family
+      'Unknown',           // 2.Font subfamily (font weight)
+      'uniqueID',          // 3.Unique ID
+      name,                // 4.Full font name
+      'Version 0.11',      // 5.Version
+      '',                  // 6.Postscript name
+      'Unknown',           // 7.Trademark
+      'Unknown',           // 8.Manufacturer
+      'Unknown'            // 9.Designer
+    ];
+  
+    // Mac want 1-byte per character strings while Windows want
+    // 2-bytes per character, so duplicate the names table
+    var stringsUnicode = [];
+    for (var i = 0; i < strings.length; i++) {
+      var str = strings[i];
+  
+      var strUnicode = '';
+      for (var j = 0; j < str.length; j++)
+        strUnicode += string16(str.charCodeAt(j));
+      stringsUnicode.push(strUnicode);
+    }
+  
+    var names = [strings, stringsUnicode];
+    var platforms = ['\x00\x01', '\x00\x03'];
+    var encodings = ['\x00\x00', '\x00\x01'];
+    var languages = ['\x00\x00', '\x04\x09'];
+  
+    var namesRecordCount = strings.length * platforms.length;
+    var nameTable =
+      '\x00\x00' +                           // format
+      string16(namesRecordCount) +           // Number of names Record
+      string16(namesRecordCount * 12 + 6);   // Storage
+  
+    // Build the name records field
+    var strOffset = 0;
+    for (var i = 0; i < platforms.length; i++) {
+      var strs = names[i];
+      for (var j = 0; j < strs.length; j++) {
+        var str = strs[j];
+        var nameRecord =
+          platforms[i] + // platform ID
+          encodings[i] + // encoding ID
+          languages[i] + // language ID
+          string16(j) + // name ID
+          string16(str.length) +
+          string16(strOffset);
+        nameTable += nameRecord;
+        strOffset += str.length;
+      }
+    }
+  
+    nameTable += strings.join('') + stringsUnicode.join('');
+    return nameTable;
+  }
+
   constructor.prototype = {
     name: null,
     font: null,
@@ -811,7 +873,7 @@ var Font = (function() {
 
       // This keep a reference to the CMap and the post tables since they can
       // be rewritted
-      var cmap, post;
+      var cmap, post, nameTable, maxp;
 
       var tables = [];
       for (var i = 0; i < numTables; i++) {
@@ -822,6 +884,10 @@ var Font = (function() {
             cmap = table;
           else if (table.tag == 'post')
             post = table;
+          else if (table.tag == 'name')
+            nameTable = table;
+          else if (table.tag == 'maxp')
+            maxp = table;
 
           requiredTables.splice(index, 1);
         }
@@ -857,7 +923,51 @@ var Font = (function() {
         });
 
         // Replace the old CMAP table with a shiny new one
+        if (properties.type == 'CIDFontType2') {
+          // Type2 composite fonts map characters directly to glyphs so the cmap
+          // table must be replaced.
+          
+          var glyphs = [];
+          var charset = properties.charset;
+          if (!charset.length) {
+            // PDF did not contain a GIDMap for the font so create an identity cmap
+            
+            // First get the number of glyphs from the maxp table
+            font.pos = (font.start ? font.start : 0) + maxp.length;
+            var version = int16(font.getBytes(4));
+            var numGlyphs = int16(font.getBytes(2));
+            
+            // Now create an identity mapping
+            for (var i = 1; i < numGlyphs; i++) {
+              glyphs.push({
+                unicode: i
+              });
+            }
+          } else {
+            for (var i = 1; i < charset.length; i++) {
+              var index = charset.indexOf(i);
+              if (index != -1) {
+                glyphs.push({
+                  unicode: index
+                });
+              } else {
+                break;
+              }
+            }
+          }
+          
+          if (!cmap) {
+            // Font did not contain a cmap
+            tables.push({
+              tag: 'cmap',
+              data: createCMapTable(glyphs)
+            })
+          } else {
+            cmap.data = createCMapTable(glyphs);
+          }
+        } else {
           replaceCMapTable(cmap, font, properties);          
+        }
 
         // Rewrite the 'post' table if needed
         if (!post) {
@@ -867,6 +977,14 @@ var Font = (function() {
           });
         }
 
+        // Rewrite the 'name' table if needed
+        if (!nameTable) {
+          tables.push({
+            tag: 'name',
+            data: stringToArray(createNameTable(this.name))
+          });
+        }
+
         // Tables needs to be written by ascendant alphabetic order
         tables.sort(function tables_sort(a, b) {
           return (a.tag > b.tag) - (a.tag < b.tag);
@@ -909,65 +1027,6 @@ var Font = (function() {
     },
 
     convert: function font_convert(fontName, font, properties) {
-      function createNameTable(name) {
-        var strings = [
-          'Original licence',  // 0.Copyright
-          name,                // 1.Font family
-          'Unknown',           // 2.Font subfamily (font weight)
-          'uniqueID',          // 3.Unique ID
-          name,                // 4.Full font name
-          'Version 0.11',      // 5.Version
-          '',                  // 6.Postscript name
-          'Unknown',           // 7.Trademark
-          'Unknown',           // 8.Manufacturer
-          'Unknown'            // 9.Designer
-        ];
-
-        // Mac want 1-byte per character strings while Windows want
-        // 2-bytes per character, so duplicate the names table
-        var stringsUnicode = [];
-        for (var i = 0; i < strings.length; i++) {
-          var str = strings[i];
-
-          var strUnicode = '';
-          for (var j = 0; j < str.length; j++)
-            strUnicode += string16(str.charCodeAt(j));
-          stringsUnicode.push(strUnicode);
-        }
-
-        var names = [strings, stringsUnicode];
-        var platforms = ['\x00\x01', '\x00\x03'];
-        var encodings = ['\x00\x00', '\x00\x01'];
-        var languages = ['\x00\x00', '\x04\x09'];
-
-        var namesRecordCount = strings.length * platforms.length;
-        var nameTable =
-          '\x00\x00' +                           // format
-          string16(namesRecordCount) +           // Number of names Record
-          string16(namesRecordCount * 12 + 6);   // Storage
-
-        // Build the name records field
-        var strOffset = 0;
-        for (var i = 0; i < platforms.length; i++) {
-          var strs = names[i];
-          for (var j = 0; j < strs.length; j++) {
-            var str = strs[j];
-            var nameRecord =
-              platforms[i] + // platform ID
-              encodings[i] + // encoding ID
-              languages[i] + // language ID
-              string16(j) + // name ID
-              string16(str.length) +
-              string16(strOffset);
-            nameTable += nameRecord;
-            strOffset += str.length;
-          }
-        }
-
-        nameTable += strings.join('') + stringsUnicode.join('');
-        return nameTable;
-      }
-
       function isFixedPitch(glyphs) {
         for (var i = 0; i < glyphs.length - 1; i++) {
           if (glyphs[i] != glyphs[i + 1])
@@ -1110,23 +1169,44 @@ var Font = (function() {
 
     charsToUnicode: function fonts_chars2Unicode(chars) {
       var charsCache = this.charsCache;
+      var str;
       
       // if we translated this string before, just grab it from the cache
       if (charsCache) {
-        var str = charsCache[chars];
+        str = charsCache[chars];
         if (str)
           return str;
       }
       
+      // lazily create the translation cache
+      if (!charsCache)
+        charsCache = this.charsCache = Object.create(null);
+      
+      if (this.compositeFont) {
+        // composite fonts have multi-byte strings
+        // convert the string from single-byte to multi-byte
+        // XXX assuming CIDFonts are two-byte - later need to extract the correct byte encoding
+        //     according to the PDF spec
+        str = '';
+        var multiByteStr = "";
+        var length = chars.length;
+        for (var i = 0; i < length; i++) {
+          var byte1 = chars.charCodeAt(i++) & 0xFF;
+          var byte2;
+          if (i == length)
+            byte2 = 0;
+          else
+            byte2 = chars.charCodeAt(i) & 0xFF;
+          multiByteStr += String.fromCharCode((byte1 << 8) | byte2);
+        }
+        str = multiByteStr;
+      }
+      else {
         // translate the string using the font's encoding
         var encoding = this.encoding;
         if (!encoding)
           return chars;
   
-      // lazily create the translation cache
-      if (!charsCache)
-        charsCache = this.charsCache = Object.create(null);
-
         str = '';
         for (var i = 0; i < chars.length; ++i) {
           var charcode = chars.charCodeAt(i);
@@ -1149,6 +1229,7 @@ var Font = (function() {
           }
           str += String.fromCharCode(unicode);
         }
+      }
 
       // Enter the translated string into the cache
       return charsCache[chars] = str;

pdf.js

@@ -3624,11 +3624,31 @@ var PartialEvaluator = (function() {
     },
 
     translateFont: function(fontDict, xref, resources) {
-      var fd = fontDict.get('FontDescriptor');
+      var fd;
-      if (!fd)
+      var descendant = [];
-        // XXX deprecated "special treatment" for standard
+      var subType = fontDict.get('Subtype');
-        // fonts?  What do we need to do here?
+      var compositeFont = false;
+      assertWellFormed(IsName(subType), 'invalid font Subtype');
+      
+      // If font is a composite 
+      //  - get the descendant font
+      //  - set the type according to the descendant font
+      //  - get the FontDescriptor from the descendant font
+      if (subType.name == 'Type0') {
+        var df = fontDict.get('DescendantFonts');
+        if (!df)
           return null;
+        compositeFont = true;
+        descendant = xref.fetch(df[0]);
+        subType = descendant.get('Subtype');
+        fd = descendant.get('FontDescriptor');
+      } else {
+        fd = fontDict.get('FontDescriptor');
+      }
+      
+      if (!fd)
+        return null;
+      
       var descriptor = xref.fetch(fd);
 
       var fontName = descriptor.get('FontName');
@@ -3642,7 +3662,38 @@ var PartialEvaluator = (function() {
 
       var encodingMap = {};
       var charset = [];
-      if (fontDict.has('Encoding')) {
+      if (compositeFont) {
+        // Special CIDFont support
+        // XXX only CIDFontType2 supported for now
+        if (subType.name == 'CIDFontType2') {
+          var cidToGidMap = descendant.get('CIDToGIDMap');
+          if (cidToGidMap && IsRef(cidToGidMap)) {
+            // Extract the charset from the CIDToGIDMap
+            var glyphsStream = xref.fetchIfRef(cidToGidMap);
+            var glyphsData = glyphsStream.getBytes(0);
+            var i = 0;
+            // Glyph ids are big-endian 2-byte values
+            for (var j=0; j<glyphsData.length; j++) {
+              var glyphID = (glyphsData[j++] << 8) | glyphsData[j];
+              charset.push(glyphID);
+            }
+          }
+        }
+        else {
+          // XXX This is a placeholder for handling of the encoding of CIDFontType0 fonts
+          var encoding = xref.fetchIfRef(fontDict.get('Encoding'));
+          if (IsName(encoding)) {
+            // Encoding is a predefined CMap
+            if (encoding.name == 'Identity-H') {
+              TODO ('Need to create an identity cmap')
+            } else {
+              TODO ('Need to support predefined CMaps see PDF 32000-1:2008 9.7.5.2 Predefined CMaps')
+            }
+          } else {
+            TODO ('Need to support encoding streams see PDF 32000-1:2008  9.7.5.3'); 
+          }
+        }
+      } else if (fontDict.has('Encoding')) {
         var encoding = xref.fetchIfRef(fontDict.get('Encoding'));
         if (IsDict(encoding)) {
           // Build a map of between codes and glyphs
@@ -3758,9 +3809,6 @@ var PartialEvaluator = (function() {
         }
       }
 
-      var subType = fontDict.get('Subtype');
-      assertWellFormed(IsName(subType), 'invalid font Subtype');
-
       var properties = {
         type: subType.name,
         encoding: encodingMap,
@@ -3775,7 +3823,8 @@ var PartialEvaluator = (function() {
         flags: descriptor.get('Flags'),
         italicAngle: descriptor.get('ItalicAngle'),
         fixedPitch: false,
-        textMatrix: IDENTITY_MATRIX
+        textMatrix: IDENTITY_MATRIX,
+        compositeFont: compositeFont
       };
 
       return {

test/test_manifest.json

@@ -48,5 +48,20 @@
        "link": true,
        "rounds": 1,
        "type": "load"
+    },
+    {  "id": "openofficecidtruetype-pdf",
+       "file": "pdfs/arial_unicode_en_cidfont.pdf",
+       "rounds": 1,
+       "type": "load"
+    },
+    {  "id": "openofficearabiccidtruetype-pdf",
+       "file": "pdfs/arial_unicode_ab_cidfont.pdf",
+       "rounds": 1,
+       "type": "load"
+    },
+    {  "id": "arabiccidtruetype-pdf",
+       "file": "pdfs/ArabicCIDTrueType.pdf",
+       "rounds": 1,
+       "type": "load"
     }
 ]