ng2-pdfjs-viewer 26.4.3

Release v26.4.3

on Node.js NPM

Fixed

• The bundled PDF.js now polyfills Math.sumPrecise (Chrome 147). PDF.js 6 calls it
  when it rebuilds fonts, lays out XFA tables and saves a document, so on older engines
  those paths failed with Math.sumPrecise is not a function. Thanks to @delagen for
  tracking it down. (#422)
• Text search threw RegExp.escape is not a function on Chrome 125 to 135. The find
  controller in viewer.mjs needs it; it now picks it up from the polyfills that
  pdf.mjs installs, since pdf.mjs runs first.
• Uint8Array.prototype.toBase64 and Uint8Array.fromBase64 (Chrome 140) are
  polyfilled too. Saved signatures, base64 images in XFA forms and one font-loading
  fallback use them.
• bytes() on Response, Blob and Request (Chrome 132) is polyfilled by hand,
  since core-js has no module for it. PDF.js uses it to fetch standard font data
  (Symbol, ZapfDingbats) and built-in CJK CMaps, so on older engines those fonts fell
  back to substitutes and CJK text using a predefined CMap didn't render at all. It
  also backs inserting pages from another PDF and compressing streams on save.
  Thanks to @delagen for spotting it and suggesting the fix. (#445)

With these, the practical floor for the modern build is about Chrome 125, where CSS
round() sets the limit and no polyfill helps.