v0.6.0
v0.6.0 (2026-07-18)
Feat
- show OCR text snippets in HTML report
- wire the OCR demoter into the scan CLI
- demote screenshot matches whose OCR text disagrees
- add macOS Vision OCR backend with text normalization and similarity
- add metadata-only screenshot predicate
- persist OCR and camera-EXIF fields in the scan cache (schema v3)
- detect camera EXIF in the perceptual pass
- add OCR fields to FileRecord
Fix
- default has_camera_exif to None so cached EXIF survives the exact-pass cache merge
Refactor
- apply CodeRabbit review nits (redundant set, zero-width escapes, test annotations)
[main fb2584d] bump: version 0.5.0 → 0.6.0
2 files changed, 22 insertions(+), 1 deletion(-)