Skip to content

docling

v2.113.0 Feature

This release adds 2 notable features for engineering teams evaluating rollout.

Published 12d RAG & Retrieval
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai convert document-parser document-parsing documents docx
+9 more
html markdown pdf pdf-converter pdf-to-json pdf-to-text pptx tables xlsx

Summary

AI summary

Preserve paragraph formatting and table structure in DOCX imports.

Changes in this release

Feature Medium

Adds GCS, Azure Blob, and Google Drive source/target types in service.

Adds GCS, Azure Blob, and Google Drive source/target types in service.

Source: llm_adapter@2026-07-14

Confidence: high

Feature Medium

Parses native PowerPoint charts as classified pictures with data in pptx module.

Parses native PowerPoint charts as classified pictures with data in pptx module.

Source: llm_adapter@2026-07-14

Confidence: high

Performance Low

Adds request timeout to model download in utils to avoid indefinite hangs.

Adds request timeout to model download in utils to avoid indefinite hangs.

Source: llm_adapter@2026-07-14

Confidence: high

Bugfix Medium

Preserves paragraph formatting when a trailing run is whitespace-only in msword.

Preserves paragraph formatting when a trailing run is whitespace-only in msword.

Source: llm_adapter@2026-07-14

Confidence: high

Bugfix Medium

Drops phantom cell from trailing-pipe table rows in asciidoc.

Drops phantom cell from trailing-pipe table rows in asciidoc.

Source: llm_adapter@2026-07-14

Confidence: high

Bugfix Medium

Stops duplicating table cells in the Markdown backend.

Stops duplicating table cells in the Markdown backend.

Source: llm_adapter@2026-07-14

Confidence: high

Bugfix Medium

Preserves cells and merges in DOCX tables with gridBefore rows in msword.

Preserves cells and merges in DOCX tables with gridBefore rows in msword.

Source: llm_adapter@2026-07-14

Confidence: high

Bugfix Medium

Chains original exception in ConversionError for error classification.

Chains original exception in ConversionError for error classification.

Source: llm_adapter@2026-07-14

Confidence: high

Refactor Low

Imports pypdfium2 lazily in pdf_outline.

Imports pypdfium2 lazily in pdf_outline.

Source: llm_adapter@2026-07-14

Confidence: high

Full changelog

Feature

  • pptx: Parse native PowerPoint charts as classified pictures with data (#3794) (3b5891c)
  • service: Expose GCS, Azure Blob, and Google Drive source/target types (#3795) (e1e2fbc)

Fix

  • msword: Keep paragraph formatting when a trailing run is whitespace-only (#3799) (1db1c32)
  • asciidoc: Drop phantom cell from trailing-pipe table rows (#3787) (f976796)
  • md: Stop duplicating table cells in the Markdown backend (#3781) (6cfdc5f)
  • Import pypdfium2 lazily in pdf_outline (#3796) (06e9c89)
  • msword: Preserve cells and merges in DOCX tables with gridBefore rows (#3745) (1e197a0)
  • Chain original exception in ConversionError for error classification (#3757) (5f2a3b6)
  • utils: Add request timeout to model download to avoid indefinite hangs (#3784) (e2329d9)

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track docling

Get notified when new releases ship.

Sign up free

About docling

Get your documents ready for gen AI

All releases →

Related context

Beta — feedback welcome: [email protected]