Abimael MartellandClaude Opus 4.6 04aa6d4ae5 feat(text): width-based joining for Canva single-char items ...

Use character-width ratios instead of font_size for Canva-style PDFs:
- Single-char prev: gap/prev.width < 1.25
- Multi→single: gap/avg_prev_char_width < 1.25
- Multi→multi: page-level threshold (gap/font_size)

Also adds second Canva detection path for per-character rendering
without embedded spaces (>50% single-char items).

Improves ebgt7isj04ophcq word accuracy from ~3% to 64%.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-03-12 23:23:02 -07:00

..

bin

refactor: redesign public API for better usability

2026-02-25 13:00:21 -08:00

extractor

feat(text): adaptive join threshold for Canva-style letter-spaced PDFs

2026-03-12 17:54:49 -07:00

markdown

feat(text): adaptive join threshold for Canva-style letter-spaced PDFs

2026-03-12 17:54:49 -07:00

tables

perf(tables): remove cosmetic padding from markdown tables

2026-03-12 11:19:30 -07:00

adobe_korea1.rs

fix(fonts): Fix decoding for custom encodings and CID Korean fonts

2026-02-18 15:32:11 -08:00

detector.rs

feat(detector): recurse into Form XObjects for image detection and classify vector-text pages

2026-03-09 21:12:25 -07:00

glyph_names.rs

Fix PUA glyph decoding, Symbol font cmap mapping, and UTF-8 detection

2026-02-21 13:54:12 -08:00

lib.rs

feat(text): adaptive join threshold for Canva-style letter-spaced PDFs

2026-03-12 17:54:49 -07:00

process_mode.rs

feat: Add configurable ProcessMode (detect-only / analyze / full)

2026-02-19 10:06:30 -08:00

text_utils.rs

feat(text): width-based joining for Canva single-char items

2026-03-12 23:23:02 -07:00

tounicode.rs

fix(fonts): prefer TrueType cmap over sequential remap for CJK subset fonts

2026-03-11 17:37:42 -07:00

types.rs

feat(text): adaptive join threshold for Canva-style letter-spaced PDFs

2026-03-12 17:54:49 -07:00