Merge pull request #556 from notmasteryet/cid0

Initial CID-0 font encodings

fonts.js

@@ -429,6 +429,9 @@ var Font = (function Font() {
       return;
     }
 
+    // Trying to fix encoding using glyph widths and CIDSystemInfo.
+    this.fixWidths(properties);
+
     if (!file) {
       // The file data is not specified. Trying to fix the font name
       // to be used with the canvas.font.
@@ -443,6 +446,7 @@ var Font = (function Font() {
 
       this.defaultWidth = properties.defaultWidth;
       this.loadedName = fontName.split('-')[0];
+      this.composite = properties.composite;
       this.loading = false;
       return;
     }
@@ -1208,25 +1212,11 @@ var Font = (function Font() {
         }
 
         var encoding = properties.encoding, i;
-        if (!encoding[0]) {
+        for (i in encoding) {
-          // the font is directly characters to glyphs with no encoding
+          if (encoding.hasOwnProperty(i)) {
-          // so create an identity encoding
+            var unicode = encoding[i].unicode;
-          var widths = properties.widths;
+            if (unicode <= 0x1f || (unicode >= 127 && unicode <= 255))
-          for (i = 0; i < numGlyphs; i++) {
+              encoding[i].unicode = unicode += kCmapGlyphOffset;
-            var width = widths[i];
-            encoding[i] = {
-              unicode: i <= 0x1f || (i >= 127 && i <= 255) ?
-                i + kCmapGlyphOffset : i,
-              width: isNum(width) ? width : properties.defaultWidth
-            };
-          }
-        } else {
-          for (i in encoding) {
-            if (encoding.hasOwnProperty(i)) {
-              var unicode = encoding[i].unicode;
-              if (unicode <= 0x1f || (unicode >= 127 && unicode <= 255))
-                encoding[i].unicode = unicode += kCmapGlyphOffset;
-            }
           }
         }
 
@@ -1402,6 +1392,87 @@ var Font = (function Font() {
       return stringToArray(otf.file);
     },
 
+    fixWidths: function font_fixWidths(properties) {
+      if (properties.type !== 'CIDFontType0' &&
+          properties.type !== 'CIDFontType2')
+          return;
+
+      var encoding = properties.encoding;
+      if (encoding[0])
+        return;
+      var glyphsWidths = properties.widths;
+      if (!glyphsWidths)
+        return;
+
+      var defaultWidth = properties.defaultWidth;
+      var cidSystemInfo = properties.cidSystemInfo;
+      var cidToUnicode;
+      if (cidSystemInfo) {
+        cidToUnicode = CIDToUnicodeMaps[
+          cidSystemInfo.registry + '-' + cidSystemInfo.ordering];
+      }
+      if (!cidToUnicode) {
+        // the font is directly characters to glyphs with no encoding
+        // so create an identity encoding
+        for (i = 0; i < 0xD800; i++) {
+          var width = glyphsWidths[i];
+          encoding[i] = {
+            unicode: i,
+            width: isNum(width) ? width : defaultWidth
+          };
+        }
+        // skipping surrogates + 256-user defined
+        for (i = 0xE100; i <= 0xFFFF; i++) {
+          var width = glyphsWidths[i];
+          encoding[i] = {
+            unicode: i,
+            width: isNum(width) ? width : defaultWidth
+          };
+        }
+        return;
+      }
+
+      encoding[0] = { unicode: 0, width: 0 };
+      var glyph = 1, i, j, k;
+      for (i = 0; i < cidToUnicode.length; ++i) {
+        var unicode = cidToUnicode[i];
+        var width;
+        if (isArray(unicode)) {
+          var length = unicode.length;
+          width = glyphsWidths[glyph];
+          for (j = 0; j < length; j++) {
+            k = unicode[j];
+            encoding[k] = {
+              unicode: k,
+              width: isNum(width) ? width : defaultWidth
+            };
+          }
+          glyph++;
+        } else if (typeof unicode === 'object') {
+          var fillLength = unicode.f;
+          if (fillLength) {
+            k = unicode.c;
+            for (j = 0; j < fillLength; ++j) {
+              width = glyphsWidths[glyph++];
+              encoding[k] = {
+                unicode: k,
+                width: isNum(width) ? width : defaultWidth
+              };
+              k++;
+            }
+          } else
+            glyph += unicode.s;
+        } else if (unicode) {
+          width = glyphsWidths[glyph++];
+          encoding[unicode] = {
+            unicode: unicode,
+            width: isNum(width) ? width : defaultWidth
+          };
+        } else
+          glyph++;
+      }
+    },
+
     bindWorker: function font_bindWorker(data) {
       postMessage({
         action: 'font',

pdf.js

@@ -4455,72 +4455,70 @@ var PartialEvaluator = (function partialEvaluator() {
                                                               properties) {
       var type = properties.type, encoding;
       if (properties.composite) {
-        if (type == 'CIDFontType2') {
+        var defaultWidth = xref.fetchIfRef(dict.get('DW')) || 1000;
-          var defaultWidth = xref.fetchIfRef(dict.get('DW')) || 1000;
+        properties.defaultWidth = defaultWidth;
-          properties.defaultWidth = defaultWidth;
 
-          var glyphsWidths = {};
+        var glyphsWidths = {};
-          var widths = xref.fetchIfRef(dict.get('W'));
+        var widths = xref.fetchIfRef(dict.get('W'));
-          if (widths) {
+        if (widths) {
-            var start = 0, end = 0;
+          var start = 0, end = 0;
-            for (var i = 0; i < widths.length; i++) {
+          for (var i = 0; i < widths.length; i++) {
-              var code = widths[i];
+            var code = widths[i];
-              if (isArray(code)) {
+            if (isArray(code)) {
-                for (var j = 0; j < code.length; j++)
+              for (var j = 0; j < code.length; j++)
-                  glyphsWidths[start++] = code[j];
+                glyphsWidths[start++] = code[j];
-                start = 0;
+              start = 0;
-              } else if (start) {
+            } else if (start) {
-                var width = widths[++i];
+              var width = widths[++i];
-                for (var j = start; j <= code; j++)
+              for (var j = start; j <= code; j++)
-                  glyphsWidths[j] = width;
+                glyphsWidths[j] = width;
-                start = 0;
+              start = 0;
-              } else {
-                start = code;
-              }
-            }
-          }
-          properties.widths = glyphsWidths;
-
-          var cidToGidMap = dict.get('CIDToGIDMap');
-          if (!cidToGidMap || !isRef(cidToGidMap)) {
-            return Object.create(GlyphsUnicode);
-          }
-
-          // Extract the encoding from the CIDToGIDMap
-          var glyphsStream = xref.fetchIfRef(cidToGidMap);
-          var glyphsData = glyphsStream.getBytes(0);
-
-          // Glyph ids are big-endian 2-byte values
-          encoding = properties.encoding;
-
-          // Set encoding 0 to later verify the font has an encoding
-          encoding[0] = { unicode: 0, width: 0 };
-          for (var j = 0; j < glyphsData.length; j++) {
-            var glyphID = (glyphsData[j++] << 8) | glyphsData[j];
-            if (glyphID == 0)
-              continue;
-
-            var code = j >> 1;
-            var width = glyphsWidths[code];
-            encoding[code] = {
-              unicode: glyphID,
-              width: isNum(width) ? width : defaultWidth
-            };
-          }
-        } else if (type == 'CIDFontType0') {
-          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 ' +
+              start = code;
-                   '9.7.5.2 Predefined CMaps');
             }
-          } else {
-            TODO('Need to support encoding streams see PDF 32000-1:2008 ' +
-                 '9.7.5.3');
           }
         }
+        properties.widths = glyphsWidths;
+
+        // Glyph ids are big-endian 2-byte values
+        encoding = properties.encoding;
+
+        // CIDSystemInfo might help to match width and glyphs
+        var cidSystemInfo = dict.get('CIDSystemInfo');
+        if (isDict(cidSystemInfo)) {
+          properties.cidSystemInfo = {
+            registry: cidSystemInfo.get('Registry'),
+            ordering: cidSystemInfo.get('Ordering'),
+            supplement: cidSystemInfo.get('Supplement')
+          };
+        }
+
+        var cidToGidMap = dict.get('CIDToGIDMap');
+        if (!cidToGidMap || !isRef(cidToGidMap)) {
+
+
+          return Object.create(GlyphsUnicode);
+        }
+
+        // Extract the encoding from the CIDToGIDMap
+        var glyphsStream = xref.fetchIfRef(cidToGidMap);
+        var glyphsData = glyphsStream.getBytes(0);
+
+        // Set encoding 0 to later verify the font has an encoding
+        encoding[0] = { unicode: 0, width: 0 };
+        for (var j = 0; j < glyphsData.length; j++) {
+          var glyphID = (glyphsData[j++] << 8) | glyphsData[j];
+          if (glyphID == 0)
+            continue;
+
+          var code = j >> 1;
+          var width = glyphsWidths[code];
+          encoding[code] = {
+            unicode: glyphID,
+            width: isNum(width) ? width : defaultWidth
+          };
+        }
+
         return Object.create(GlyphsUnicode);
       }
 

test/pdfs/tcpdf_033.pdf.link

@@ -0,0 +1 @@
+http://www.tcpdf.org/examples/example_033.pdf

test/test_manifest.json

@@ -170,5 +170,11 @@
        "rounds": 1,
        "skipPages": [ 16 ],
        "type": "load"
+    },
+    {  "id": "tcpdf_033",
+       "file": "pdfs/tcpdf_033.pdf",
+       "link": true,
+       "rounds": 1,
+       "type": "eq"
     }
 ]

test/test_slave.html

@@ -8,6 +8,7 @@
   <script type="text/javascript" src="/glyphlist.js"></script>
   <script type="text/javascript" src="/metrics.js"></script>
   <script type="text/javascript" src="/charsets.js"></script>
+  <script type="text/javascript" src="/cidmaps.js"></script>
   <script type="text/javascript" src="driver.js"></script>
 </head>
 

web/viewer.html

@@ -12,6 +12,7 @@
         <script type="text/javascript" src="../glyphlist.js"></script>
         <script type="text/javascript" src="../metrics.js"></script>
         <script type="text/javascript" src="../charsets.js"></script>
+        <script type="text/javascript" src="../cidmaps.js"></script>
   </head>
 
   <body>