handle CMAP format 0 for platform 1, encoding 0 (legacy Macintosh encoding)

Author: Connum

src/tables/cmap.js

@@ -4,6 +4,25 @@
 import check from '../check.js';
 import parse from '../parse.js';
 import table from '../table.js';
+import { eightBitMacEncodings } from '../types.js';
+import { getEncoding } from '../tables/name.js';
+
+function parseCmapTableFormat0(cmap, p, platformID, encodingID) {
+    // Length in bytes of the index map
+    cmap.length = p.parseUShort();
+    // see https://developer.apple.com/fonts/TrueType-Reference-Manual/RM06/Chap6name.html
+    // section "Macintosh Language Codes"
+    cmap.language = p.parseUShort() - 1;
+
+    const indexMap = p.parseByteList(cmap.length);
+    const glyphIndexMap = Object.assign({}, indexMap);
+    const encoding = getEncoding(platformID, encodingID, cmap.language);
+    const decodingTable = eightBitMacEncodings[encoding];
+    for (let i = 0; i < decodingTable.length; i++) {
+        glyphIndexMap[decodingTable.charCodeAt(i)] = indexMap[0x80 + i];
+    }
+    cmap.glyphIndexMap = glyphIndexMap;
+}
 
 function parseCmapTableFormat12(cmap, p) {
     //Skip reserved.
@@ -150,11 +169,15 @@ function parseCmapTable(data, start) {
     let format14Parser = null;
     let format14offset = -1;
     let offset = -1;
+    let platformId = null;
+    let encodingId = null;
     for (let i = cmap.numTables - 1; i >= 0; i -= 1) {
-        const platformId = parse.getUShort(data, start + 4 + (i * 8));
-        const encodingId = parse.getUShort(data, start + 4 + (i * 8) + 2);
+        platformId = parse.getUShort(data, start + 4 + (i * 8));
+        encodingId = parse.getUShort(data, start + 4 + (i * 8) + 2);
         if ((platformId === 3 && (encodingId === 0 || encodingId === 1 || encodingId === 10)) ||
-            (platformId === 0 && (encodingId === 0 || encodingId === 1 || encodingId === 2 || encodingId === 3 || encodingId === 4))) {
+            (platformId === 0 && (encodingId === 0 || encodingId === 1 || encodingId === 2 || encodingId === 3 || encodingId === 4)) ||
+            (platformId === 1 && encodingId === 0) // MacOS <= 9
+        ) {
             offset = parse.getULong(data, start + 4 + (i * 8) + 4);
             // allow for early break
             if (format14Parser) {
@@ -178,12 +201,17 @@ function parseCmapTable(data, start) {
     const p = new parse.Parser(data, start + offset);
     cmap.format = p.parseUShort();
 
-    if (cmap.format === 12) {
+    if (cmap.format === 0) {
+        parseCmapTableFormat0(cmap, p, platformId, encodingId);
+    } else if (cmap.format === 12) {
         parseCmapTableFormat12(cmap, p);
     } else if (cmap.format === 4) {
         parseCmapTableFormat4(cmap, p, data, start, offset);
     } else {
-        throw new Error('Only format 4, 12 and 14 cmap tables are supported (found format ' + cmap.format + ').');
+        throw new Error(
+            'Only format 0 (platformId 1, encodingId 0), 4, 12 and 14 cmap tables are supported ' +
+            '(found format ' + cmap.format + ', platformId ' + platformId + ', encodingId ' + encodingId + ').'
+        );
     }
 
     // format 14 is the only one that's not exclusive but can be used as a supplement.
@@ -361,4 +389,4 @@ function makeCmapTable(glyphs) {
 
 export default { parse: parseCmapTable, make: makeCmapTable };
 
-export { parseCmapTableFormat14 };
+export { parseCmapTableFormat0, parseCmapTableFormat14 };

src/tables/name.js

@@ -604,7 +604,7 @@ const macLanguageEncodings = {
     146: 'x-mac-gaelic'       // langIrishGaelicScript
 };
 
-function getEncoding(platformID, encodingID, languageID) {
+export function getEncoding(platformID, encodingID, languageID) {
     switch (platformID) {
         case 0:  // Unicode
             return utf16;

src/types.js

@@ -437,7 +437,7 @@ sizeOf.UTF16 = function(v) {
 /**
  * @private
  */
-const eightBitMacEncodings = {
+export const eightBitMacEncodings = {
     'x-mac-croatian':  // Python: 'mac_croatian'
     'ÄÅÇÉÑÖÜáàâäãåçéèêëíìîïñóòôöõúùûü†°¢£§•¶ß®Š™´¨≠ŽØ∞±≤≥∆µ∂∑∏š∫ªºΩžø' +
     '¿¡¬√ƒ≈Ć«Č… ÀÃÕŒœĐ—“”‘’÷◊©⁄€‹›Æ»–·‚„‰ÂćÁčÈÍÎÏÌÓÔđÒÚÛÙıˆ˜¯πË˚¸Êæˇ',

test/fonts/LICENSE

@@ -42,6 +42,11 @@ TestCMAP14.otf
     http://www.apache.org/licenses/LICENSE-2.0
     https://github.com/unicode-org/text-rendering-tests/blob/main/LICENSE.md
 
+TestCMAPMacTurkish.ttf
+    Copyright © 2016 by Unicode Inc.
+    SIL Open Font License, Version 1.1
+    https://opensource.org/licenses/OFL-1.1
+
 Vibur.woff
     Copyright (c) 2010, Johan Kallas ([email protected]).
     SIL Open Font License, Version 1.1

test/fonts/TestCMAPMacTurkish.ttf

@@ -1,7 +1,10 @@
 import assert from 'assert';
 import { unhex } from '../testutil';
 import { Parser } from '../../src/parse';
-import { parseCmapTableFormat14 } from '../../src/tables/cmap';
+import { parseCmapTableFormat14, parseCmapTableFormat0 } from '../../src/tables/cmap';
+import { parse } from '../../src/opentype.js';
+import { readFileSync } from 'fs';
+const loadSync = (url, opt) => parse(readFileSync(url), opt);
 
 describe('tables/cmap.js', function() {
 
@@ -55,4 +58,17 @@ describe('tables/cmap.js', function() {
         assert.deepEqual(cmap.varSelectorList, expectedData);
     });
 
+    it('can parse CMAP format 0 legacy Mac encoding', function() {
+        let font;
+        assert.doesNotThrow(function() {
+            font = loadSync('./test/fonts/TestCMAPMacTurkish.ttf');
+        });
+        const testString = '“ABÇĞIİÖŞÜ”abçğıiöşüă';
+        const glyphIds = [];
+        const expectedGlyphIds = [200,34,35,126,176,42,178,140,181,145,201,66,67,154,177,222,74,168,182,174,123,184];
+        for (let i = 0; i < testString.length; i++) {
+            glyphIds.push(font.charToGlyphIndex(testString.charAt(i)));
+        }
+        assert.deepEqual(glyphIds, expectedGlyphIds);
+    });
 });