All pull requests

Pull requests

• fix: tolerate a dangling DOCX header/footer relationship instead of crashing

  #4624·aballman opened Oct 28, 2026Contributor·

  ·
  4

• fix(metrics): read evaluation files as UTF-8 on every platform

  #4623·tanishqbaweja opened Oct 28, 2026·2 of 3 tasks·

  ·
  2

• fix(ndjson): end an NDJSON record only at a newline

  #4622·chrikrah opened Oct 28, 2026Contributor·1 of 3 tasks·

  ·

• fix(documents): count the hash-id sequence number per page, not per contiguous run

  #4621·Eapp1e opened Oct 28, 2026Contributor·

  ·
  1

  1

• fix(pptx): keep visible text in translated, scaled and nested groups

  #4619·feiiiiii5 opened Oct 28, 2026Contributor·2 of 3 tasks·

  ·
  2

• fix(chunking): don't extend a pre-chunk that has reached new_after_n_chars

  #4618·simpleqt opened Oct 27, 2026Contributor·

  ·
  1

• fix(coordinates): default CoordinateSystem.orientation to CARTESIAN

  #4615·Eapp1e opened Oct 27, 2026Contributor·

  ·
  1

  1

• fix(html): complete a partial rel on blank target links

  #4613·Eapp1e opened Oct 27, 2026Contributor·

  ·
  4

• fix(lang): keep the languages a caller passes on short ASCII text

  #4601·chrikrah opened Oct 25, 2026Contributor·1 of 3 tasks·

  ·
  1

• fix(chunking): don't emit a chunk that holds only the overlap when overlap_all splits an element

  #4598·sinnercode228 opened Oct 25, 2026Contributor·1 of 3 tasks·

  ·

• fix(csv): read a field a row does not have as an empty cell, not "None"

  #4596·simpleqt opened Oct 24, 2026Contributor·2 of 3 tasks·

  ·
  1

• fix: evaluate XML paths against the complete document

  #4593·Kayvan-Zahiri opened Oct 23, 2026·

  ·

• fix(pdf): hi_res table text ran column by column while its HTML ran row by row

  #4586·CyMule opened Oct 22, 2026Contributor·

  ·
  8

• fix(pdf): text positioned entirely off the page no longer becomes elements

  #4585·CyMule opened Oct 22, 2026Contributor·

  ·
  2

• fix(pdf): ligatures mapped to U+0000 by the font's ToUnicode CMap came out as NUL characters

  #4584·CyMule opened Oct 22, 2026Contributor·

  ·

• fix(pdf): one complex page no longer disables the text layer for the whole document

  #4583·CyMule opened Oct 22, 2026Contributor·

  ·
  3

• fix(staging): handle CheckBox and TableChunk in convert_to_coco

  #4567·kerthans opened Oct 21, 2026·

  ·

• fix(partition): keep section parent_id after images and other non-text elements

  #4566·kerthans opened Oct 21, 2026·

  ·
  2

• fix(xml): keep the text that follows an inline node in XML partitions

  #4560·simpleqt opened Oct 20, 2026Contributor·

  ·
  1

• fix(cleaners): translate sequential chunks in translate_text (#4558)

  #4559·gunjan-l opened Oct 19, 2026·

  ·
  1

• fix(xlsx): keep whole numbers and booleans in a column with an empty cell

  #4556·L4XB opened Oct 17, 2026Contributor·

  ·
  1

• fix(csv): read a single-column CSV file as one column

  #4555·L4XB opened Oct 17, 2026Contributor·

  ·

• fix: stop dropping nested metadata values in convert_to_csv

  #4549·linhongyu510 opened Oct 13, 2026Contributor·

  ·
  1

  14

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

  #4546·chrikrah opened Oct 11, 2026Contributor·

  ·
  4

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

  #4540·CyMule opened Oct 9, 2026Contributor·

  ·
  1

  7