Fix
- table: Keep OCR word cells in table tokens (
affa54c) - table: Add OCR cells to table tokens next to native words (
ce6690b) - table: Read the page's parsed_page for table tokens in full-page OCR (
a65b29e) - pptx: Count the header row against the chart table cell cap (#5030) (
7331067) - ocr: Pass render resolution to Tesseract CLI (#5031) (
4818833) - pptx: Skip shapes whose text is only whitespace (#5028) (
af66cbf) - Widen dependency ranges to allow latest versions (
1bab069)
Performance
- table: Skip spatial indexes for empty token inputs (
103edef)