v1.25.0 v1.25.1
low risk
feyninc/pulpieSep 28, 2026 → Sep 28, 20260 days between releases
What's New
Fixes critical font decoding issues for ligatures and math symbols.
Improvements
- Correctly reads simple fonts when ToUnicode declares a two-byte codespace but text uses one-byte strings.
- Fallback to built-in encoding for Type1 fonts with non-standard base encodings to recover ligatures.
- Improved repair logic for short ToUnicode CMaps to prevent misreading common words.
Bug Fixes
- Fixes kerned runs of short strings losing characters (e.g., 'Income Statement').
- Fixes Type1 fonts reading ligatures, curly quotes, dashes, and math symbols.
- Fixes short common words being misread (e.g., 'engagement').
Upgrade Notes
No API changes are required. However, text extraction results for complex fonts (Type1, kerned strings) may differ from v1.25.0.
The Two Releases
v1.25.0
1.25.0 — Async positions and underline detection speed
Sep 28, 2026· 24 days ago
Introduces an asynchronous text extraction API and significantly boosts the performance of underline and strikeout detection.
Full release notesv1.25.1
1.25.1 — Font decoding fixes
Sep 28, 2026· 24 days ago
Addresses critical font decoding issues for simple fonts and Type1 encodings, ensuring ligatures and math symbols are read correctly.
Full release notes