jedarden/pdftract
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
pdftract/notes/pdftract-3dwu.md
jedarden 09c3498cf4 feat(pdftract-3dwu): implement named encoding tables 
Implements the 6 named-encoding character-code-to-glyph-name lookup
tables required by Level 2 of the encoding fallback chain:
- WinAnsiEncoding (Windows-1252 superset of StandardEncoding)
- MacRomanEncoding (Mac OS Roman encoding)
- MacExpertEncoding (Mac OS Expert character set)
- StandardEncoding (Adobe Standard encoding)
- SymbolEncoding (Symbol font encoding)
- ZapfDingbatsEncoding (Zapf Dingbats font encoding)

These tables map character codes (0-255) to glyph names, which are then
mapped to Unicode via the Adobe Glyph List (AGL).

Acceptance criteria:
- All 6 tables compile into static arrays with binary footprint < 30 KB
- WIN_ANSI[0x92] == Some("quoteright") (canonical WinAnsi test)
- MAC_ROMAN[0xD2] == Some("quotedblleft") and MAC_ROMAN[0xD3] == Some("quotedblright")
- STANDARD[0x20] == Some("space")
- NamedEncoding::from_name("WinAnsiEncoding") == Some(NamedEncoding::WinAnsi)

Files:
- crates/pdftract-core/build/named-encodings.json - Source data from ISO 32000-1 Annex D
- crates/pdftract-core/src/font/encoding.rs - Public API with NamedEncoding enum
- crates/pdftract-core/build.rs - Build script updates for encoding generation

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-23 18:00:05 -04:00

2.8 KiB
Raw Blame History

pdftract-3dwu: Named encodings table verification

Summary

Implemented the 6 named-encoding character-code-to-glyph-name lookup tables required by Level 2 of the encoding fallback chain.

Files

  • crates/pdftract-core/build/named-encodings.json - Source data from ISO 32000-1 Annex D
  • crates/pdftract-core/build.rs - Build script that generates static arrays
  • crates/pdftract-core/src/font/encoding.rs - Public API with NamedEncoding enum

Acceptance Criteria

PASS: All 6 tables compile into static arrays with binary footprint < 30 KB

  • Generated file: target/release/build/pdftract-core-*/out/named_encodings.rs = 22,289 bytes (~22 KB)
  • Well under the 30 KB requirement

PASS: WIN_ANSI[0x92] == Some("quoteright")

  • Test: test_winansi_0x92_quoteright - PASSED
  • This is the canonical test for WinAnsiEncoding that all PDF extractors must pass

PASS: MAC_ROMAN[0xD2] == Some("quotedblleft") and MAC_ROMAN[0xD3] == Some("quotedblright")

  • Test: test_macroman_0xd2_quotedblleft - PASSED
  • MacRoman has different mappings for curly quotes than WinAnsi

PASS: STANDARD[0x20] == Some("space")

  • Test: test_standard_0x20_space - PASSED
  • StandardEncoding is the implicit default when a Type1 font has no /Encoding entry

PASS: NamedEncoding::from_name("WinAnsiEncoding") == Some(NamedEncoding::WinAnsi)

  • Test: test_from_name - PASSED
  • Handles both prefixed and unprefixed names (e.g., "WinAnsiEncoding" or "/WinAnsiEncoding")

Additional Tests Passed

  • test_winansi_euro_at_0x80 - Verifies Euro sign in Windows-1252 range
  • test_symbol_encoding_alpha - Verifies Symbol font uses glyph names, not Greek Unicode
  • test_zapfdingbats_a1 - Verifies ZapfDingbats glyph names (a1..a222)
  • test_table_length - Verifies all tables are 256 elements
  • test_unmapped_codes - Verifies StandardEncoding has no mappings at 0x80-0x9F

Critical Considerations Verified

  • StandardEncoding is the IMPLICIT default - from_name returns None for unknown encodings, allowing fallback to Standard
  • SymbolEncoding maps to Symbol-font glyph names (Alpha, beta, etc.) NOT Greek Unicode codepoints
  • ZapfDingbatsEncoding glyph names start with a followed by ZapfDingbats glyph numbers (a1..a222)
  • WinAnsi has the famous Windows-1252 punctuation range at 0x80-0x9F that StandardEncoding does NOT have

Retrospective

  • What worked: The build.rs pattern for generating static arrays from JSON worked perfectly. Using include! to pull in the generated code keeps the module clean.
  • What didn't: N/A - everything worked on first attempt
  • Surprise: The encoding tables were already present in the codebase - this task was about verifying they work correctly
  • Reusable pattern: JSON → build.rs → static array generation is a solid pattern for embedding large constant data in Rust binaries