kreuzberg-dev/kreuzberg v3.11.2

on GitHub

What's Changed

Bug Fixes

• Fixed GMFT table extraction failing with pandas.errors.EmptyDataError when processing PDFs without tables (#104)
  • Modified isolated process serialization to handle empty DataFrames specially
  • Empty DataFrames now preserve column information during serialization/deserialization
  • Added comprehensive test coverage for PDFs without tables

Dependencies

• Updated mcp to >=1.13.0
• Updated paddleocr to >=3.1.1
• Updated ruff to >=0.12.9
• Updated pydoclint to 0.6.10
• Re-enabled ai-rulez hooks (v1.5.1)

Full Changelog: v3.11.1...v3.11.2