Support font with characters below 0x20 declared in format 1 cmap and be more general about the TrueType rewritting cases

fonts.js

@@ -372,7 +372,7 @@ function getUnicodeRangeFor(value) {
  *   var type1Font = new Font("MyFontName", binaryFile, propertiesObject);
  *   type1Font.bind();
  */
-var Font = (function() {
+var Font = (function Font() {
   var constructor = function font_constructor(name, file, properties) {
     this.name = name;
     this.textMatrix = properties.textMatrix || IDENTITY_MATRIX;
@@ -531,6 +531,19 @@ var Font = (function() {
       }
       ranges.push([start, end]);
     }
+    
+    // Removes duplicate ranges
+    for (var i = ranges.length - 1; i > 0; i--) {
+      var range = [];
+      var range1 = ranges[i];
+      var range2 = ranges[i - 1];
+      if (range1[0] == range2[1]) {
+        range2[0] = range2[0] - 1;
+        range2[1] = range1[1];
+        ranges.splice(i, 1);
+      }
+    }
+    
     return ranges;
   };
 
@@ -790,21 +803,57 @@ var Font = (function() {
       };
 
       function replaceCMapTable(cmap, font, properties) {
-        font.pos = (font.start ? font.start : 0) + cmap.length;
+        var start = (font.start ? font.start : 0) + cmap.length;
+        font.pos = start;
 
         var version = int16(font.getBytes(2));
-        var numTables = int16(font.getBytes(2));
+        var numRecords = int16(font.getBytes(2));
+
+        var records = [];
+        for (var i = 0; i < numRecords; i++) {
+          records.push({
+            platformID: int16(font.getBytes(2)),
+            encodingID: int16(font.getBytes(2)),
+            offset: int32(font.getBytes(4))
+          });
+        };
+
+        
+        for (var i = 0; i < numRecords; i++) {
+          var table = records[i];
+          font.pos = start + table.offset;
           
-        for (var i = 0; i < numTables; i++) {
-          var platformID = int16(font.getBytes(2));
-          var encodingID = int16(font.getBytes(2));
-          var offset = int32(font.getBytes(4));
           var format = int16(font.getBytes(2));
           var length = int16(font.getBytes(2));
           var language = int16(font.getBytes(2));
           
-          if ((format == 0 && numTables == 1) ||
+          if (format == 0 && numRecords > 1) {
-              (format == 6 && numTables == 1 && !properties.encoding.empty)) {
+            // Characters below 0x20 are controls characters that are hardcoded
+            // into the platform so if some characters in the font are assigned
+            // under this limit they will not be displayed so let's rewrite the
+            // CMap.
+            var map = [];
+            var rewrite = false;
+            for (var j = 0; j < 256; j++) {
+              var index = font.getByte();
+              if (index != 0) {
+                map.push(index);
+                if (j < 0x20)
+                  rewrite = true;
+              }
+            }
+            
+            if (rewrite) {
+              var glyphs = [];
+              for (var j = 0x20; j < 256; j++) {
+                // TODO do not hardcode WinAnsiEncoding
+                var unicode = GlyphsUnicode[Encodings["WinAnsiEncoding"][j]];
+                glyphs.push({ unicode: unicode });
+              }
+              cmap.data = createCMapTable(glyphs, true);
+            }
+          } else if ((format == 0 && numRecords == 1) ||
+                     (format == 6 && numRecords == 1 && !properties.encoding.empty)) {
             // Format 0 alone is not allowed by the sanitizer so let's rewrite
             // that to a 3-1-4 Unicode BMP table
             TODO('Use an other source of informations than ' +
@@ -818,7 +867,7 @@ var Font = (function() {
             }
 
             cmap.data = createCMapTable(glyphs);
-          } else if (format == 6 && numTables == 1) {
+          } else if (format == 6 && numRecords == 1) {
             // Format 6 is a 2-bytes dense mapping, which means the font data
             // lives glue together even if they are pretty far in the unicode
             // table. (This looks weird, so I can have missed something), this
@@ -871,10 +920,7 @@ var Font = (function() {
       var header = readOpenTypeHeader(font);
       var numTables = header.numTables;
 
-      // This keep a reference to the CMap and the post tables since they can
+      var cmap, maxp;
-      // be rewritted
-      var cmap, post, nameTable, maxp;
-
       var tables = [];
       for (var i = 0; i < numTables; i++) {
         var table = readTableEntry(font);
@@ -882,10 +928,6 @@ var Font = (function() {
         if (index != -1) {
           if (table.tag == 'cmap')
             cmap = table;
-          else if (table.tag == 'post')
-            post = table;
-          else if (table.tag == 'name')
-            nameTable = table;
           else if (table.tag == 'maxp')
             maxp = table;
 
@@ -894,10 +936,6 @@ var Font = (function() {
         tables.push(table);
       }
 
-      // If any tables are still in the array this means some required
-      // tables are missing, which means that we need to rebuild the
-      // font in order to pass the sanitizer.
-      if (requiredTables.length && requiredTables[0] == 'OS/2') {
       // Create a new file to hold the new version of our truetype with a new
       // header and new offsets
       var ttf = new Uint8Array(kMaxFontFileSize);
@@ -916,11 +954,12 @@ var Font = (function() {
       // of missing tables
       createOpenTypeHeader('\x00\x01\x00\x00', ttf, offsets, numTables);
 
-        // Insert the missing table
+			if (requiredTables.indexOf('OS/2') != -1) {
         tables.push({
           tag: 'OS/2',
           data: stringToArray(createOS2Table(properties))
         });
+			}
 
       // Replace the old CMAP table with a shiny new one
       if (properties.type == 'CIDFontType2') {
@@ -970,7 +1009,7 @@ var Font = (function() {
       }
 
       // Rewrite the 'post' table if needed
-        if (!post) {
+      if (requiredTables.indexOf('post') != -1) {
         tables.push({
           tag: 'post',
           data: stringToArray(createPostTable(properties))
@@ -978,7 +1017,7 @@ var Font = (function() {
       }
 
       // Rewrite the 'name' table if needed
-        if (!nameTable) {
+      if (requiredTables.indexOf('name') != -1) {
         tables.push({
           tag: 'name',
           data: stringToArray(createNameTable(this.name))
@@ -1018,12 +1057,6 @@ var Font = (function() {
         fontData.push(ttf[i]);
 
       return fontData;
-      } else if (requiredTables.length) {
-        error('Table ' + requiredTables[0] +
-              ' is missing from the TrueType font');
-      }
-
-      return font.getBytes();
     },
 
     convert: function font_convert(fontName, font, properties) {

web/viewer.html

@@ -8,6 +8,7 @@
         <script type="text/javascript" src="viewer.js"></script>
         <script type="text/javascript" src="../pdf.js"></script>
         <script type="text/javascript" src="../fonts.js"></script>
+        <script type="text/javascript" src="../utils/fonts_utils.js"></script>
         <script type="text/javascript" src="../crypto.js"></script>
         <script type="text/javascript" src="../glyphlist.js"></script>
   </head>