All pull requests

Pull requests

• fix(cleaners): anchor every alternative in clean_prefix and clean_postfix

  #4546·chrikrah opened Oct 11, 2026Contributor·

  ·

• fix(xlsx): read .xls files whose Workbook stream ends in padding

  #4545·usmanmateen opened Oct 10, 2026·

  ·
  1

• fix(email): keep non-ASCII sender and recipient names readable

  #4543·usmanmateen opened Oct 10, 2026·

  ·
  1

• fix(pdf): extract vector-heavy PDFs with the fast strategy instead of returning nothing

  #4540·CyMule opened Oct 9, 2026Contributor·

  ·
  1

  7

• fix(docx): classify paragraphs using document languages

  #4539·cpruijsen opened Oct 9, 2026·

  ·
  1

• fix: label numbered headings without making them list items, and skip section-break paragraphs

  #4538·qued opened Oct 8, 2026Contributor·

  ·

• perf: render the PDF once in hi_res instead of once per pipeline stage

  #4530·CyMule opened Oct 6, 2026Contributor·

  ·
  16

• fix: keep overlap category names from element 100 on

  #4529·jipeng6036-del opened Oct 6, 2026·

  ·
  1

• fix(chunking): measure combine thresholds in tokens when max_tokens is set

  #4528·joaquinhuigomez opened Oct 4, 2026·

  ·

• fix(staging): preserve caller metadata in Prodigy export

  #4527·fallintoplace opened Oct 4, 2026·

  ·

• fix(image): charge GIF frames at the canvas size Pillow decodes them at

  #4526·badGarnet opened Oct 2, 2026Collaborator·

  ·
  6

• fix(xlsx): collapse empty-row runs instead of rejecting far-away cells

  #4525·palaniappa opened Oct 2, 2026·

  ·

• perf: Free decode buffers sooner when restoring original elements

  #4524·CyMule opened Oct 2, 2026Contributor·

  ·
  1

• perf: Free image buffers sooner during figure extraction

  #4523·CyMule opened Oct 2, 2026Contributor·

  ·
  3

• perf(ci): Publish multiarch tags without pulling images onto the runner

  #4522·CyMule opened Oct 2, 2026Contributor·

  ·
  2

• perf(chunking): bound image payload and original-element serialization memory (0.27.14)

  #4517·CyMule opened Oct 2, 2026Contributor·

  ·
  11

• perf(pdf): bound OCR upload and rendered-page memory (0.27.13)

  #4516·CyMule opened Oct 2, 2026Contributor·

  ·
  9

• fix(tsv): honor infer_table_structure when partitioning TSV files

  #4513·feiiiiii5 opened Oct 1, 2026Contributor·

  ·
  1

• docs: fix GitHub Docs contributing guidelines link

  #4511·pratikgx opened Sep 30, 2026·

  ·

• fix(metrics): let _mean ignore unmeasured scores like its siblings

  #4510·sclfcz opened Sep 30, 2026·

  ·
  1

• fix(utils): recognize a box nested when the tolerance is zero

  #4507·SashaMIT opened Sep 29, 2026·

  ·

• fix: write text output from an elements iterator in create_file_from_elements

  #4506·MohammadHijjawi97 opened Sep 28, 2026·

  ·

• fix(xlsx): keep cell text like N/A and None instead of emptying it

  #4505·wenqiw777 opened Sep 28, 2026·

  ·
  1

• fix(json): preserve rehydrated element identity

  #4504·rohan-patnaik opened Sep 27, 2026·

  ·
  1

  9

• fix(html): extract <figure> content instead of discarding it

  #4503·r0h1tb opened Sep 26, 2026Contributor·

  ·
  1

  1