Try to skip mapping of missing TrueType and CIDFontType2 glyphs

src/core/fonts.js

@@ -2203,6 +2203,10 @@ var ToUnicodeMap = (function ToUnicodeMapClosure() {
       }
     },
 
+    has: function(i) {
+      return this._map[i] !== undefined;
+    },
+
     get: function(i) {
       return this._map[i];
     },
@@ -2232,6 +2236,10 @@ var IdentityToUnicodeMap = (function IdentityToUnicodeMapClosure() {
       }
     },
 
+    has: function (i) {
+      return this.firstChar <= i && i <= this.lastChar;
+    },
+
     get: function (i) {
       if (this.firstChar <= i && i <= this.lastChar) {
         return String.fromCharCode(i);
@@ -2664,7 +2672,6 @@ var Font = (function FontClosure() {
     var isSymbolic = !!(properties.flags & FontFlags.Symbolic);
     var isIdentityUnicode =
       properties.toUnicode instanceof IdentityToUnicodeMap;
-    var isCidFontType2 = (properties.type === 'CIDFontType2');
     var newMap = Object.create(null);
     var toFontChar = [];
     var usedFontCharCodes = [];
@@ -2675,17 +2682,11 @@ var Font = (function FontClosure() {
       var fontCharCode = originalCharCode;
       // First try to map the value to a unicode position if a non identity map
       // was created.
-      if (!isIdentityUnicode) {
-        if (toUnicode.get(originalCharCode) !== undefined) {
-          var unicode = toUnicode.get(fontCharCode);
-          // TODO: Try to map ligatures to the correct spot.
-          if (unicode.length === 1) {
-            fontCharCode = unicode.charCodeAt(0);
-          }
-        } else if (isCidFontType2) {
-          // For CIDFontType2, move characters not present in toUnicode
-          // to the private use area (fixes bug 1028735 and issue 4881).
-          fontCharCode = nextAvailableFontCharCode;
+      if (!isIdentityUnicode && toUnicode.has(originalCharCode)) {
+        var unicode = toUnicode.get(fontCharCode);
+        // TODO: Try to map ligatures to the correct spot.
+        if (unicode.length === 1) {
+          fontCharCode = unicode.charCodeAt(0);
         }
       }
       // Try to move control characters, special characters and already mapped
@@ -3526,6 +3527,7 @@ var Font = (function FontClosure() {
         var newGlyfData = new Uint8Array(oldGlyfDataLength);
         var startOffset = itemDecode(locaData, 0);
         var writeOffset = 0;
+        var missingGlyphData = {};
         itemEncode(locaData, 0, writeOffset);
         var i, j;
         for (i = 0, j = itemSize; i < numGlyphs; i++, j += itemSize) {
@@ -3543,6 +3545,10 @@ var Font = (function FontClosure() {
             continue;
           }
 
+          if (startOffset === endOffset) {
+            missingGlyphData[i] = true;
+          }
+
           var newLength = sanitizeGlyph(oldGlyfData, startOffset, endOffset,
                                         newGlyfData, writeOffset, hintsValid);
           writeOffset += newLength;
@@ -3559,7 +3565,7 @@ var Font = (function FontClosure() {
             itemEncode(locaData, j, simpleGlyph.length);
           }
           glyf.data = simpleGlyph;
-          return;
+          return missingGlyphData;
         }
 
         if (dupFirstEntry) {
@@ -3576,6 +3582,7 @@ var Font = (function FontClosure() {
         } else {
           glyf.data = newGlyfData.subarray(0, writeOffset);
         }
+        return missingGlyphData;
       }
 
       function readPostScriptTable(post, properties, maxpNumGlyphs) {
@@ -4035,11 +4042,13 @@ var Font = (function FontClosure() {
 
       sanitizeHead(tables.head, numGlyphs, isTrueType ? tables.loca.length : 0);
 
+      var missingGlyphs = {};
       if (isTrueType) {
         var isGlyphLocationsLong = int16(tables.head.data[50],
                                          tables.head.data[51]);
-        sanitizeGlyphLocations(tables.loca, tables.glyf, numGlyphs,
-                               isGlyphLocationsLong, hintsValid, dupFirstEntry);
+        missingGlyphs = sanitizeGlyphLocations(tables.loca, tables.glyf,
+                                               numGlyphs, isGlyphLocationsLong,
+                                               hintsValid, dupFirstEntry);
       }
 
       if (!tables.hhea) {
@@ -4061,19 +4070,33 @@ var Font = (function FontClosure() {
         }
       }
 
-      var charCodeToGlyphId = [], charCode;
+      var charCodeToGlyphId = [], charCode, toUnicode = properties.toUnicode;
+
+      function hasGlyph(glyphId, charCode) {
+        if (!missingGlyphs[glyphId]) {
+          return true;
+        }
+        if (charCode >= 0 && toUnicode.has(charCode)) {
+          return true;
+        }
+        return false;
+      }
+
       if (properties.type === 'CIDFontType2') {
         var cidToGidMap = properties.cidToGidMap || [];
-        var cidToGidMapLength = cidToGidMap.length;
+        var isCidToGidMapEmpty = cidToGidMap.length === 0;
+
         properties.cMap.forEach(function(charCode, cid) {
           assert(cid <= 0xffff, 'Max size of CID is 65,535');
           var glyphId = -1;
-          if (cidToGidMapLength === 0) {
+          if (isCidToGidMapEmpty) {
             glyphId = charCode;
           } else if (cidToGidMap[cid] !== undefined) {
             glyphId = cidToGidMap[cid];
           }
-          if (glyphId >= 0 && glyphId < numGlyphs) {
+
+          if (glyphId >= 0 && glyphId < numGlyphs &&
+              hasGlyph(glyphId, charCode)) {
             charCodeToGlyphId[charCode] = glyphId;
           }
         });
@@ -4133,7 +4156,8 @@ var Font = (function FontClosure() {
 
             var found = false;
             for (i = 0; i < cmapMappingsLength; ++i) {
-              if (cmapMappings[i].charCode === unicodeOrCharCode) {
+              if (cmapMappings[i].charCode === unicodeOrCharCode &&
+                  hasGlyph(cmapMappings[i].glyphId, unicodeOrCharCode)) {
                 charCodeToGlyphId[charCode] = cmapMappings[i].glyphId;
                 found = true;
                 break;
@@ -4143,7 +4167,7 @@ var Font = (function FontClosure() {
               // Try to map using the post table. There are currently no known
               // pdfs that this fixes.
               var glyphId = properties.glyphNames.indexOf(glyphName);
-              if (glyphId > 0) {
+              if (glyphId > 0 && hasGlyph(glyphId, -1)) {
                 charCodeToGlyphId[charCode] = glyphId;
               }
             }

test/pdfs/.gitignore

@@ -18,6 +18,7 @@
 !sizes.pdf
 !close-path-bug.pdf
 !issue4630.pdf
+!issue5202.pdf
 !issue5280.pdf
 !alphatrans.pdf
 !devicen.pdf
@@ -58,10 +59,13 @@
 !zerowidthline.pdf
 !bug868745.pdf
 !mmtype1.pdf
+!issue5704.pdf
+!bug893730.pdf
 !bug864847.pdf
 !issue1002.pdf
 !issue925.pdf
 !issue2840.pdf
+!issue4061.pdf
 !issue4668.pdf
 !issue5039.pdf
 !issue5070.pdf

test/test_manifest.json

@@ -485,6 +485,20 @@
        "lastPage": 1,
        "type": "eq"
     },
+    {  "id": "issue4061",
+       "file": "pdfs/issue4061.pdf",
+       "md5": "236aaa8840a47c3c061f8e3034549764",
+       "rounds": 1,
+       "link": false,
+       "type": "eq"
+    },
+    {  "id": "issue5202",
+       "file": "pdfs/issue5202.pdf",
+       "md5": "bb9cc69211112e66aab40828086a4e5a",
+       "rounds": 1,
+       "link": false,
+       "type": "eq"
+    },
     {  "id": "issue5238",
        "file": "pdfs/issue5238.pdf",
        "md5": "6ddecda00893be1793de20a70c83a3c2",
@@ -1008,6 +1022,20 @@
       "link": true,
       "type": "eq"
     },
+    {  "id": "issue5704",
+       "file": "pdfs/issue5704.pdf",
+       "md5": "6e0b62585feef24dff2d7e7687cd8128",
+       "rounds": 1,
+       "link": false,
+       "type": "eq"
+    },
+    {  "id": "bug893730",
+       "file": "pdfs/bug893730.pdf",
+       "md5": "2587379fb1b3bbff89c14f0863e78383",
+       "rounds": 1,
+       "link": false,
+       "type": "eq"
+    },
     {  "id": "pdfkit_compressed",
       "file": "pdfs/pdfkit_compressed.pdf",
       "md5": "ffe9c571d0a1572e234253e6c7cdee6c",