pypi docling 2.148.0

3 hours ago

Feature

  • Add surya v2.5 (#5141) (9b25840)
  • pdf: Mark superscript and subscript text from glyph size and baseline (#4815) (3bbd3fa)
  • calendar: Add an iCalendar (.ics) backend (#4880) (e461fa3)

Fix

  • backend: Keep inline formatting in JATS and ODF table cells (#4809) (6cd7a35)
  • mets: Keep pages loadable when a page div lacks a usable ORDER (#5149) (f7a535b)
  • html: Use 72 DPI fallback for HTML and MHTML images with 0 DPI metadata (#5165) (e03c1aa)
  • boxnote: Keep a later paragraph out of the list item text (#5160) (0ab06cb)
  • pdfium: Join a punctuation run to its row (#5161) (9046b8e)
  • epub: Reduce a single-quoted cross-document link to its anchor (#5170) (f18cc1b)
  • docx: Preserve table cells with incomplete grids (#5179) (e973a49)
  • epub: Decode the percent-escapes of the package document path (#5168) (6d67609)
  • asr: Pass temperature to Transformers Whisper generation (#5166) (7f6b3d4)
  • jats: Keep the text of inline elements (#5177) (fc7df25)
  • asciidoc: Number ordered lists from their start attribute (#5180) (a230f3e)
  • pdfium: Show the hyphen of a line-final glyph (#5159) (94ce021)
  • tables: Keep table text when TableFormer predicts a runaway structure (#4592) (eed760d)
  • asciidoc: Turn a trailing " +" into a hard line break (#5175) (5f6a7fc)
  • heading-hierarchy: Ignore part/chapter/article keywords used as plain words (#5154) (b927c57)

Don't miss a new docling release

NewReleases is sending notifications on new releases.