github docling-project/docling-core v2.100.0

4 hours ago

Feature

  • Lenient escape repair on doclang deserialization (#803) (367ed14)
  • doclang: Add include_namespace & image_mode options to export & archive saving (#809) (eac575a)

Fix

  • vtt: Carry millisecond rounding across second/minute/hour boundaries (#829) (c44fcf8)
  • serializer: Render relative hyperlinks with portable separators (#780) (cff563f)
  • Support "application/rtf" MIME type in document validation (#827) (3e42dba)
  • doclang: Keep the nested content of float footnotes and captions (#824) (d7cef4c)
  • doclang: Fix deserialization of without element body (#811) (01c2443)
  • doclang: Don't replay parent children for empty list item bodies (#820) (bb89daa)
  • doclang: Keep label, provenance, layer and leading text of mixed-content elements (#804) (39aec16)
  • Renumber all items on deletion; add opt-in validation rules and repairs (#810) (61ab797)
  • markdown: Keep non-ASCII list markers in AUTO mode (#799) (25ade75)

Performance

  • chunking: Make _split_by_doc_items linear in the number of doc items (#808) (e551f14)
  • Stop building a throwaway serializer for every rich table cell (#771) (ed78fd3)

Don't miss a new docling-core release

NewReleases is sending notifications on new releases.