Generate ToUnicodeMap bfrange in multiple ranges (#1498) (#1499)

* Generate ToUnicodeMap bfrange in multiple ranges (#1498)

This resolves #1498.

* Add unit test for bfrange lines in toUnicodeMap

* Add changelog line

CHANGELOG.md

@@ -7,6 +7,7 @@
 - Fix for soft hyphen not being replaced by visible hyphen if necessary (#457)
 - Optimize output files by ignoring identity transforms
 - Fix for Acroforms - setting an option to false will still apply the flag (#1495)
+- Fix for text extraction in PDFium-based viewers due to invalid ToUnicodeMap (#1498)
 
 ### [v0.14.0] - 2023-11-09
 

lib/font/embedded.js

@@ -252,6 +252,15 @@ class EmbeddedFont extends PDFFont {
       entries.push(`<${encoded.join(' ')}>`);
     }
 
+    const chunkSize = 256;
+    const chunks = Math.ceil(entries.length / chunkSize);
+    const ranges = [];
+    for (let i = 0; i < chunks; i++) {
+      const start = i * chunkSize;
+      const end = Math.min((i + 1) * chunkSize, entries.length);
+      ranges.push(`<${toHex(start)}> <${toHex(end - 1)}> [${entries.slice(start, end).join(' ')}]`);
+    }
+
     cmap.end(`\
 /CIDInit /ProcSet findresource begin
 12 dict begin
@@ -267,7 +276,7 @@ begincmap
 <0000><ffff>
 endcodespacerange
 1 beginbfrange
-<0000> <${toHex(entries.length - 1)}> [${entries.join(' ')}]
+${ranges.join('\n')}
 endbfrange
 endcmap
 CMapName currentdict /CMap defineresource pop

tests/unit/font.spec.js

@@ -1,5 +1,6 @@
-import PDFFontFactory from '../../lib/font_factory';
 import PDFDocument from '../../lib/document';
+import PDFFontFactory from '../../lib/font_factory';
+import { logData } from './helpers';
 
 describe('EmbeddedFont', () => {
   test('no fontLayoutCache option', () => {
@@ -52,4 +53,46 @@ describe('EmbeddedFont', () => {
       expect(dictionary.data.BaseFont).toBe('BAJJZZ+Roboto-Regular');
     });
   });
+
+  describe.only('toUnicodeMap', () => {
+    test('bfrange lines should not cross highcode boundary', () => {
+      const doc = new PDFDocument({ compress: false });
+      const font = PDFFontFactory.open(
+        doc,
+        'tests/fonts/Roboto-Regular.ttf',
+        undefined,
+        'F1099'
+      );
+
+      // 398 different glyphs
+      font.encode('ABCDEFGHIJKLMNOPQRSTUVWXYZ');
+      font.encode('abcdefghijklmnopqrstuvwxyz');
+      font.encode('ÁÀÂÄÅÃÆÇÐÉÈÊËÍÌÎÏÑÓÒÔÖÕØŒÞÚÙÛÜÝŸ');
+      font.encode('áàâäãåæçðéèêëíìîïıñóòôöõøœßþúùûüýÿ');
+      font.encode('ĀĂĄĆČĎĐĒĖĘĚĞĢĪĮİĶŁĹĻĽŃŅŇŌŐŔŖŘŠŚŞȘŢȚŤŪŮŰŲŽŹŻ');
+      font.encode('āăąćčďđēėęěğģīįķłĺļľńņňōőŕŗřšśşșţțťūůűųžźż');
+      font.encode('ΑΒΓ∆ΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩΆΈΉΊΌΎΏΪΫ');
+      font.encode('αβγδεζηθικλµνξοπρςστυφχψωάέήίόύώϊϋΐΰ');
+      font.encode('АБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ');
+      font.encode('абвгдежзийклмнопрстуфхцчшщъыьэюя');
+      font.encode('ЀЁЂЃЄЅІЇЈЉЊЋЌЍЎЏҐӁҒҖҚҢҮҰҲҶҺӘӢӨӮ');
+      font.encode('ѐёђѓєѕіїјљњћќѝўџґӂғҗқңүұҳҷһәӣөӯ');
+
+      const docData = logData(doc);
+      font.toUnicodeCmap();
+      const text = docData.map((d) => d.toString("utf8")).join("");
+
+      let glyphs = 0
+      for (const block of text.matchAll(/beginbfrange\n((?:.|\n)*?)\nendbfrange/g)) {
+        for (const line of block[1].matchAll(/^<([0-9a-f]+)>\s+<([0-9a-f]+)>\s+\[/igm)) {
+          const low = parseInt(line[1], 16);
+          const high = parseInt(line[2], 16);
+          glyphs += high - low + 1;
+          expect(high & 0xFFFFFF00).toBe(low & 0xFFFFFF00);
+        }
+      }
+
+      expect(glyphs).toBe(398 + 1);
+    });
+  });
 });