All pull requests

Pull requests

• fix(hybrid): report the pages docling blamed in failed_pages

  #754·bundolee opened Sep 23, 2026Collaborator·

  ·
  4

• ci: land the version bump through a pull request

  #753·bundolee opened Sep 23, 2026Collaborator·

  ·
  8

• fix(hybrid): read picture descriptions from docling's meta field

  #750·bundolee opened Sep 23, 2026Collaborator·

  ·
  1

  4

• fix(hybrid): describe every picture, and report the ones skipped

  #747·bundolee opened Sep 23, 2026Collaborator·

  ·
  1

  4

• fix: measure tiny text by font size, not by ink height

  #733·hrosb opened Sep 16, 2026·

  ·
  1

  4

• feat(images): keep vector overlays on extracted figures; fix lone-number lists and 1x1 URL tables

  #728·hoiyada7-maker opened Sep 10, 2026·

  ·
  5

• fix(hybrid): keep the backend text when bbox-matched stream chunks do not resemble it

  #722·geunamja-gimhae opened Sep 6, 2026·

  ·
  8

• fix(markdown): keep the first character of unordered list items that carry no bullet glyph

  #721·geunamja-gimhae opened Sep 6, 2026·

  ·
  13

• feat(python): add timeout parameter to run_jar and convert

  #718·anshulbansod2 opened Sep 2, 2026·

  ·
  4

  

• feat(hybrid): let --device directml run RapidOCR on DirectML

  #716·FamousMonsterr opened Aug 31, 2026·

  ·
  4

  

• Add turnkey high-precision pipeline with MPS/CUDA acceleration

  #712·denis-samatov opened Aug 30, 2026·

  ·
  12

  

• Fix veraPDF artifact memory retention

  #705·petar030 opened Aug 28, 2026·

  ·
  4

• fix: avoid emitting empty tables from line-only content

  #704·Destined-at-Dawn opened Aug 28, 2026·

  ·
  1

  2

• Add Simplified Chinese README translation

  #703·imzihuailin opened Aug 27, 2026·

  ·
  4

  

• fix(hybrid): re-exec into UTF-8 mode to fix cp949 crash on Korean Windows

  #683·SomSamantray opened Aug 14, 2026Contributor·

  ·
  1

  20

  

• feat: add marker to indicate typing support for Python package

  #675·maxdswain opened Aug 10, 2026·

  ·
  1

  3

  

• Fix sanitize phone pattern to match unprefixed US numbers

  #671·cr-sbarbouche opened Aug 6, 2026·

  ·
  1

  4

  

• Fix: carry OCR-recovered text into hybrid-mode tagged-pdf/pdf output

  #669·ViolanteCodes opened Aug 5, 2026·

  ·
  11

• fix: improve heading detection by adding exclusion filters for equati…

  #668·MoBahgat010 opened Aug 4, 2026·

  ·
  1

  14

• fix(triage): route full-page image-only scans to the backend

  #620·ysys143 opened Jul 1, 2026·

  ·
  1

  4

  

• Fix merged-cell Markdown table rendering by filling spanned cells and add regression tests

  #613·julie743 opened Jun 25, 2026·

  ·
  4

  

• Feature/decouple picture extraction and description

  #612·julie743 opened Jun 24, 2026·

  ·
  6

  

• feat(cli): add --filter-hidden-text to enable hidden-text filtering

  #576·i-Veni-Vidi-Vici opened Jun 16, 2026Contributor·

  ·
  13

  

• fix: handle invalid input in Java API so batch processing continues (#430)

  #564·hnc-jglee opened Jun 11, 2026Contributor·

  ·
  1

  4

  

• fix(python): validate to-stdout+format compat and add batch_convert()

  #554·charlesaurav13 opened Jun 3, 2026·

  ·
  2

  4