Merge pull request #366 from sbarman/encoding2

Encoding2

pdf.js

@@ -4176,6 +4176,7 @@ var PartialEvaluator = (function() {
       }
 
       var encodingMap = {};
+      var glyphMap = {};
       var charset = [];
       if (compositeFont) {
         // Special CIDFont support
@@ -4213,18 +4214,16 @@ var PartialEvaluator = (function() {
                  '9.7.5.3');
           }
         }
-      } else if (fontDict.has('Encoding')) {
+      } else {
+        var baseEncoding = null, diffEncoding = [];
+        if (fontDict.has('Encoding')) {
           var encoding = xref.fetchIfRef(fontDict.get('Encoding'));
           if (IsDict(encoding)) {
             // Build a map of between codes and glyphs
             // Load the base encoding
             var baseName = encoding.get('BaseEncoding');
             if (baseName) {
-            var base = Encodings[baseName.name];
+              baseEncoding = Encodings[baseName.name].slice();
-            for (var j = 0, end = base.length; j < end; j++)
-              encodingMap[j] = GlyphsUnicode[base[j]] || 0;
-          } else {
-            TODO('need to load default encoding');
             }
 
             // Load the differences between the base and original
@@ -4235,35 +4234,45 @@ var PartialEvaluator = (function() {
               if (IsNum(data)) {
                 index = data;
               } else {
-              encodingMap[index++] = (subType.name == 'TrueType') ? j :
+                diffEncoding[index++] = data.name;
-                                     GlyphsUnicode[data.name];
               }
             }
           } else if (IsName(encoding)) {
-          var encoding = Encodings[encoding.name];
+            baseEncoding = Encodings[encoding.name].slice();
-          if (!encoding)
-            error('Unknown font encoding');
-
-          var index = 0;
-          for (var j = 0; j < encoding.length; j++)
-            encodingMap[index++] = GlyphsUnicode[encoding[j]];
-
-          // firstChar and width are required
-          // (except for 14 standard fonts)
-          var firstChar = xref.fetchIfRef(fontDict.get('FirstChar'));
-          var widths = xref.fetchIfRef(fontDict.get('Widths')) || [];
-          for (var j = 0; j < widths.length; j++) {
-            if (widths[j])
-              charset.push(encoding[j + firstChar]);
           }
         }
-      } else if (fontDict.has('ToUnicode')) {
+
-        encodingMap = {empty: true};
+        if (!baseEncoding) {
+          var type = subType.name;
+          if (type == 'TrueType') {
+            baseEncoding = Encodings.WinAnsiEncoding.slice(0);
+          } else if (type == 'Type1') {
+            baseEncoding = Encodings.StandardEncoding.slice(0);
+          } else {
+            error('Unknown type of font');
+          }
+        }
+
+        // merge in the differences
+        var length = baseEncoding.length > diffEncoding.length ? 
+            baseEncoding.length : diffEncoding.length;
+        for (var i = 0, ii = length; i < ii; ++i) {
+          var diffGlyph = diffEncoding[i];
+          var baseGlyph = baseEncoding[i];
+          if (diffGlyph) {
+            glyphMap[i] = diffGlyph;
+            encodingMap[i] = GlyphsUnicode[diffGlyph];
+          } else if (baseGlyph) {
+            glyphMap[i] = baseGlyph;
+            encodingMap[i] = GlyphsUnicode[baseGlyph];
+          }
+        }
+
+        if (fontDict.has('ToUnicode')) {
           var cmapObj = xref.fetchIfRef(fontDict.get('ToUnicode'));
           if (IsName(cmapObj)) {
             error('ToUnicode file cmap translation not implemented');
           } else if (IsStream(cmapObj)) {
-          var encoding = Encodings['WinAnsiEncoding'];
             var firstChar = xref.fetchIfRef(fontDict.get('FirstChar'));
 
             var tokens = [];
@@ -4272,7 +4281,8 @@ var PartialEvaluator = (function() {
             var cmap = cmapObj.getBytes(cmapObj.length);
             for (var i = 0; i < cmap.length; i++) {
               var byte = cmap[i];
-            if (byte == 0x20 || byte == 0x0A || byte == 0x3C || byte == 0x3E) {
+              if (byte == 0x20 || byte == 0x0A || byte == 0x3C || 
+                  byte == 0x3E) {
                 switch (token) {
                   case 'useCMap':
                     error('useCMap is not implemented');
@@ -4294,10 +4304,6 @@ var PartialEvaluator = (function() {
                       var startRange = parseInt('0x' + tokens[j]);
                       var endRange = parseInt('0x' + tokens[j + 1]);
                       var code = parseInt('0x' + tokens[j + 2]);
-
-                  for (var k = startRange; k <= endRange; k++) {
-                    charset.push(encoding[code++] || '.notdef');
-                  }
                     }
                     break;
 
@@ -4305,8 +4311,7 @@ var PartialEvaluator = (function() {
                     for (var j = 0; j < tokens.length; j += 2) {
                       var index = parseInt('0x' + tokens[j]);
                       var code = parseInt('0x' + tokens[j + 1]);
-                  encodingMap[index] = GlyphsUnicode[encoding[code]];
+                      encodingMap[index] = code;
-                  charset.push(encoding[code] || '.notdef');
                     }
                     break;
 
@@ -4326,6 +4331,16 @@ var PartialEvaluator = (function() {
           }
         }
 
+        // firstChar and width are required
+        // (except for 14 standard fonts)
+        var firstChar = xref.fetchIfRef(fontDict.get('FirstChar'));
+        var widths = xref.fetchIfRef(fontDict.get('Widths')) || [];
+        for (var j = 0; j < widths.length; j++) {
+          if (widths[j])
+            charset.push(glyphMap[j + firstChar]);
+        }
+      }
+
       if (!fd) {
         var baseFontName = fontDict.get('BaseFont');
         if (!IsName(baseFontName))