Fix accented character rendering (á, é, ñ, €, etc.) with Standard 14 fonts

[Mythie]

Summary

Fixes corrupted rendering of accented/Latin-1 characters (á, é, ñ, ö, €, curly quotes, em dashes, etc.) when using Standard 14 fonts like Helvetica, Times-Roman, and Courier.

Problem

Three compounding bugs caused non-ASCII characters to render as mojibake:

1. Wrong text encoding — encodeTextForFont() used PDFDocEncoding (a metadata encoding) instead of WinAnsiEncoding, producing incorrect bytes in the 0x80–0x9F range (€, curly quotes, em dash, etc.)
2. UTF-8 round-trip corruption — The pipeline shuttled content stream bytes through Operator.toString() → TextDecoder (UTF-8) → TextEncoder (UTF-8), destroying any non-ASCII byte (e.g., 0xE9 for é became U+FFFD)
3. Missing /Encoding in font dict — Without an explicit /Encoding WinAnsiEncoding entry, PDF viewers fell back to the font's built-in StandardEncoding, mapping bytes to wrong glyphs
4. Wrong width measurement — getGlyphName() only mapped ASCII, returning "space" for all accented characters, breaking text layout and measurement

Solution

• Use proper font encoding: encodeTextForFont() now uses WinAnsiEncoding for Helvetica/Times/Courier, SymbolEncoding for Symbol, and ZapfDingbatsEncoding for ZapfDingbats. Unencodable characters (CJK, emoji) substitute with .notdef (byte 0x00).
• Hex-format PdfString: Standard 14 text is encoded as hex strings (<636166E9>) — pure ASCII that's immune to any encoding transformation. Matches pdf-lib's approach.
• Bytes-first pipeline: appendOperators() uses Operator.toBytes() directly. createContentStream/appendContent/prependContent accept string | Uint8Array, eliminating the UTF-8 round-trip.
• Font dict /Encoding: Standard 14 font dicts now include /Encoding /WinAnsiEncoding (omitted for Symbol/ZapfDingbats per PDF spec Table 5.15).
• Extended glyph map: CHAR_TO_GLYPH now covers all WinAnsi non-ASCII characters (~95 entries), fixing width measurement for accented text.

Changes

File	Change
src/fonts/standard-14.ts	Encoding helpers, extended CHAR_TO_GLYPH map
src/api/pdf-page.ts	Encoding fix, bytes pipeline, font dict /Encoding
src/api/drawing/path-builder.ts	ContentAppender type accepts string | Uint8Array
src/api/drawing/latin1-encoding.test.ts	29 new tests

Test coverage

• Font encoding selection (WinAnsi vs Symbol vs ZapfDingbats)
• Glyph name mapping and width measurement for accented characters
• Font dict /Encoding presence/absence verification
• Hex string encoding in content streams (é → <E9>, not UTF-8 C3A1)
• Unencodable character .notdef substitution
• Round-trip PDF generation with all Standard 14 font families
• Backward compatibility (shapes, paths, images still work)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Review	Updated (UTC)
core	Ready	Preview, Comment	Jan 28, 2026 4:54am

[Copilot]

Pull request overview

This PR fixes corrupted rendering of accented and Latin-1 characters (á, é, ñ, €, curly quotes, em dashes, etc.) when using Standard 14 fonts like Helvetica, Times-Roman, and Courier. The fix addresses four compounding bugs: wrong text encoding (PDFDocEncoding instead of WinAnsiEncoding), UTF-8 round-trip corruption through the content stream pipeline, missing /Encoding entries in font dictionaries, and incorrect width measurements for accented characters.

Changes:

• Implements proper font encoding selection (WinAnsi for Helvetica/Times/Courier, Symbol/ZapfDingbats for those respective fonts)
• Refactors content stream pipeline to work with Uint8Array throughout, eliminating UTF-8 corruption
• Adds /Encoding /WinAnsiEncoding to Standard 14 font dictionaries (except Symbol/ZapfDingbats)
• Extends CHAR_TO_GLYPH map to cover all WinAnsi non-ASCII characters for correct width measurement
• Uses hex-format PdfStrings for Standard 14 text as defense-in-depth against encoding transformations

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

File	Description
src/fonts/standard-14.ts	Adds encoding helper functions (getEncodingForStandard14, isWinAnsiStandard14) and extends CHAR_TO_GLYPH map with ~95 WinAnsi non-ASCII entries
src/api/pdf-page.ts	Refactors encodeTextForFont() to use proper font encodings, updates appendContent/prependContent/createContentStream to accept bytes, implements bytes-first appendOperators(), adds /Encoding to font dicts
src/api/drawing/path-builder.ts	Updates ContentAppender type to accept string | Uint8Array for backward-compatible bytes support
src/api/drawing/latin1-encoding.test.ts	Adds comprehensive test coverage with 29 tests covering encoding selection, glyph mapping, font dictionary structure, content stream encoding, and round-trip rendering

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.