kreuzberg-dev/kreuzberg v4.0.3

on GitHub

Added

HTML Configuration Support

• Full html_options configuration: The html_options field in ExtractionConfig is now fully configurable from config files (TOML/YAML/JSON) and all language bindings (#282)
  • Upgraded html-to-markdown-rs to v2.21.1 with serde support
  • Configure heading styles, code block styles, list formatting, text wrapping, and more
  • Replaces v3's HTMLToMarkdownConfig with more comprehensive options
  • See migration guide for available options and examples

Fixed

Go Module

• Fixed header include path for external users: plugins_test_helpers.go now uses the bundled header at internal/ffi/kreuzberg.h instead of a relative path to the monorepo (#280)

C# SDK

• Keyword extraction deserialization: Fixed JsonException when using keyword extraction - keywords are now properly deserialized as ExtractedKeyword objects (#285)

Documentation

• Rust OCR code examples: Fixed incorrect Some(...) wrapper in OcrConfig examples (#284)

Tests

• Flaky concurrent interning test: Marked as #[ignore] to prevent intermittent CI failures

Distribution

• Homebrew tap visibility: Made kreuzberg-dev/homebrew-tap repository public (#283)

Full Changelog: v4.0.2...v4.0.3