Merge pull request #466 from notmasteryet/charstoglyphs

Refactoring charsToUnicode into charsToGlyphs

fonts.js

@@ -440,6 +440,7 @@ var Font = (function Font() {
       // name ArialBlack for example will be replaced by Helvetica.
       this.black = (name.search(/Black/g) != -1);
 
+      this.defaultWidth = properties.defaultWidth;
       this.loadedName = fontName.split('-')[0];
       this.loading = false;
       return;
@@ -476,6 +477,7 @@ var Font = (function Font() {
     this.data = data;
     this.type = properties.type;
     this.textMatrix = properties.textMatrix;
+    this.defaultWidth = properties.defaultWidth;
     this.loadedName = getUniqueName();
     this.composite = properties.composite;
     this.loading = true;
@@ -1424,15 +1426,15 @@ var Font = (function Font() {
       return rule;
     },
 
-    charsToUnicode: function fonts_chars2Unicode(chars) {
+    charsToGlyphs: function fonts_chars2Glyphs(chars) {
       var charsCache = this.charsCache;
-      var str;
+      var glyphs;
 
       // if we translated this string before, just grab it from the cache
       if (charsCache) {
-        str = charsCache[chars];
+        glyphs = charsCache[chars];
-        if (str)
+        if (glyphs)
-          return str;
+          return glyphs;
       }
 
       // lazily create the translation cache
@@ -1443,7 +1445,8 @@ var Font = (function Font() {
       var encoding = this.encoding;
       if (!encoding)
         return chars;
-      str = '';
+
+      glyphs = [];
 
       if (this.composite) {
         // composite fonts have multi-byte strings convert the string from
@@ -1454,38 +1457,39 @@ var Font = (function Font() {
                                        // loop should never end on the last byte
         for (var i = 0; i < length; i++) {
           var charcode = int16([chars.charCodeAt(i++), chars.charCodeAt(i)]);
-          var unicode = encoding[charcode];
+          var glyph = encoding[charcode];
-          if ('undefined' == typeof(unicode)) {
+          if ('undefined' == typeof(glyph)) {
             warn('Unencoded charcode ' + charcode);
-            unicode = charcode;
+            glyph = {
-          } else {
+              unicode: charcode,
-            unicode = unicode.unicode;
+              width: this.defaultWidth
+            };
           }
-          str += String.fromCharCode(unicode);
+          glyphs.push(glyph);
+          // placing null after each word break charcode (ASCII SPACE)
+          if (charcode == 0x20)
+            glyphs.push(null);
         }
       }
       else {
         for (var i = 0; i < chars.length; ++i) {
           var charcode = chars.charCodeAt(i);
-          var unicode = encoding[charcode];
+          var glyph = encoding[charcode];
-          if ('undefined' == typeof(unicode)) {
+          if ('undefined' == typeof(glyph)) {
             warn('Unencoded charcode ' + charcode);
-            unicode = charcode;
+            glyph = {
-          } else {
+              unicode: charcode,
-            unicode = unicode.unicode;
+              width: this.defaultWidth
-          }
+            };
-
-          // Handle surrogate pairs
-          if (unicode > 0xFFFF) {
-            str += String.fromCharCode(unicode & 0xFFFF);
-            unicode >>= 16;
           }
-          str += String.fromCharCode(unicode);
+          glyphs.push(glyph);
+          if (charcode == 0x20)
+            glyphs.push(null);
         }
       }
 
       // Enter the translated string into the cache
-      return (charsCache[chars] = str);
+      return (charsCache[chars] = glyphs);
     }
   };
 

pdf.js

@@ -4301,7 +4301,6 @@ var PartialEvaluator = (function() {
             };
           }
         } else if (type == 'CIDFontType0') {
-          encoding = xref.fetchIfRef(dict.get('Encoding'));
           if (IsName(encoding)) {
             // Encoding is a predefined CMap
             if (encoding.name == 'Identity-H') {
@@ -4362,6 +4361,7 @@ var PartialEvaluator = (function() {
       // merge in the differences
       var firstChar = properties.firstChar;
       var lastChar = properties.lastChar;
+      var widths = properties.widths || [];
       var glyphs = {};
       for (var i = firstChar; i <= lastChar; i++) {
         var glyph = differences[i];
@@ -4372,7 +4372,7 @@ var PartialEvaluator = (function() {
             continue;
         }
         var index = GlyphsUnicode[glyph] || i;
-        var width = properties.widths[i] || properties.widths[glyph];
+        var width = widths[i] || widths[glyph];
         map[i] = {
           unicode: index,
           width: IsNum(width) ? width : properties.defaultWidth
@@ -4543,7 +4543,7 @@ var PartialEvaluator = (function() {
           type: type.name,
           encoding: map,
           differences: [],
-          widths: widths,
+          widths: widths || {},
           defaultWidth: defaultWidth,
           firstChar: 0,
           lastChar: 256
@@ -4934,7 +4934,7 @@ var CanvasGraphics = (function() {
       font = font.get(fontRef.name);
       font = this.xref.fetchIfRef(font);
       if (!font)
-        return;
+        error('Referenced font is not found');
 
       var fontObj = font.fontObj;
       this.current.font = fontObj;
@@ -4986,22 +4986,15 @@ var CanvasGraphics = (function() {
     showText: function(text) {
       var ctx = this.ctx;
       var current = this.current;
-      var originalText = text;
+      var font = current.font;
 
       ctx.save();
       ctx.transform.apply(ctx, current.textMatrix);
       ctx.scale(1, -1);
-
       ctx.translate(current.x, -1 * current.y);
+      ctx.transform.apply(ctx, font.textMatrix || IDENTITY_MATRIX);
 
-      var font = current.font;
+      var glyphs = font.charsToGlyphs(text);
-      if (font) {
-        ctx.transform.apply(ctx, font.textMatrix || IDENTITY_MATRIX);
-        text = font.charsToUnicode(text);
-      }
-
-      var composite = font.composite;
-      var encoding = font.encoding;
       var fontSize = current.fontSize;
       var charSpacing = current.charSpacing;
       var wordSpacing = current.wordSpacing;
@@ -5009,22 +5002,23 @@ var CanvasGraphics = (function() {
       ctx.scale(1 / textHScale, 1);
 
       var width = 0;
-      for (var i = 0; i < text.length; i++) {
+      for (var i = 0; i < glyphs.length; i++) {
-        if (composite) {
+        var glyph = glyphs[i];
-          var position = i * 2 + 1;
+        if (glyph === null) {
-          var charcode = (originalText.charCodeAt(position - 1) << 8) +
+          // word break
-                          originalText.charCodeAt(position);
+          width += wordSpacing;
-        } else {
+          continue;
-          var charcode = originalText.charCodeAt(i);
         }
 
-        var charWidth = font.encoding[charcode].width * fontSize * 0.001;
+        var unicode = glyph.unicode;
+        var char = unicode >= 0x10000 ?
+          String.fromCharCode(0xD800 | ((unicode - 0x10000) >> 10),
+          0xDC00 | (unicode & 0x3FF)) : String.fromCharCode(unicode);
+
+        var charWidth = glyph.width * fontSize * 0.001;
         charWidth += charSpacing;
-        if (charcode == 32)
-          charWidth += wordSpacing;
 
-        ctx.fillText(text.charAt(i), 0, 0);
+        ctx.fillText(char, width, 0);
-        ctx.translate(charWidth, 0);
         width += charWidth;
       }
       current.x += width;

test/pdfs/fips197.pdf.link

@@ -0,0 +1 @@
+http://csrc.nist.gov/publications/fips/fips197/fips-197.pdf

test/test_manifest.json

@@ -127,5 +127,11 @@
        "link": true,
        "rounds": 1,
        "type": "eq"
+    },
+    {  "id": "fips197",
+       "file": "pdfs/fips197.pdf",
+       "link": true,
+       "rounds": 1,
+       "type": "load"
     }
 ]