What's New

Features

• Prompt alignment validation for few-shot examples (#215)
  • Validates that example extractions exist in their source text
  • Three modes: OFF, WARNING (default), ERROR
  • New parameters: prompt_validation_level and prompt_validation_strict
• Vertex AI authentication support for Gemini provider (#60)
• llama-cpp-python community provider added (#202)

Improvements

• Changed debug=False as default in extract() for cleaner output
• Fixed router typings for provider plugins (#190)
• Allow T-prefixed TypeVars in pylint (#194)

Full Changelog: v1.0.8...v1.0.9

What's Changed

Features

• Ollama timeout improvements (#154)
  • Increased default timeout from 30s to 120s
  • Made timeout configurable via ModelConfig
  • Fixed kwargs not being passed through

Documentation

• Improved visualization examples for Jupyter/Colab (#153)
• Added Romeo & Juliet Colab notebook

Full Changelog: v1.0.7...v1.0.8

What's New

• Debug logging support when debug=True in lx.extract() (#142)
• GPT-5 model registration fixes (#143)
• Improved documentation for provider plugins and schema support
• Automated plugin generator script for external providers
• Base URL support for OpenAI-compatible endpoints (#138)

See the full changelog for details.

Major Features

Custom Model Provider Plugin Support

• New provider registry infrastructure for extending LangExtract with custom LLM providers
• Plugin discovery via entry points allows third-party packages to register providers
• Example implementation available at examples/custom_provider_plugin

Schema System Refactor

• Refactored schema system to support provider-specific schema implementations
• Providers can now define their own schema constraints and validation
• Better separation of concerns between core schema logic and provider implementations

Enhancements

• Ollama Provider: Added support for Hugging Face style model IDs (e.g., meta-llama/Llama-3.2-1B-Instruct)
• Extract API: Added model and config parameters to extract() for more flexible model configuration
• Examples: Updated Ollama quickstart to demonstrate ModelConfig pattern with JSON mode
• Testing: Improved test infrastructure for provider registry and plugin system

Bug Fixes

• Fixed lazy loading for provider pattern registration
• Fixed unicode escaping in example generation
• Fixed test failures related to provider registry initialization

Installation

pip install langextract==1.0.6

Full Changelog: v1.0.5...v1.0.6

What's Changed

Bug Fixes

• Fix chunking bug when newlines fall at chunk boundaries (#88) - Resolves issue where content was incorrectly chunked when newline characters appeared at chunk boundaries
• Fix IPython import warnings and improve notebook detection (#86) - Eliminates import warnings in Jupyter notebooks and improves compatibility

New Features

• Add base_url parameter to OpenAILanguageModel (#51) - Enables using custom OpenAI-compatible endpoints for alternative LLM providers

Full Changelog: v1.0.4...v1.0.5

What's Changed

• Added Ollama language model integration – Full support for local LLMs via Ollama
• Docker deployment support – Production-ready docker-compose setup with health checks
• Comprehensive examples – Quickstart script and detailed documentation in examples/ollama/
• Fixed OllamaLanguageModel parameter – Changed from model to model_id for consistency (#57)
• Enhanced CI/CD – Added Ollama integration tests that run on every PR
• Improved documentation – Consistent API examples across all language models

Technical Details

• Supports all Ollama models (gemma2:2b, llama3.2, mistral, etc.)
• Secure setup with localhost-only binding by default
• Integration tests use lightweight models for faster CI runs
• Docker setup includes automatic model pulling and health checks

Usage Example

import langextract as lx

result = lx.extract(
    text_or_documents=input_text,
    prompt_description=prompt,
    examples=examples,
    language_model_type=lx.inference.OllamaLanguageModel,
    model_id="gemma2:2b",
    model_url="http://localhost:11434",
    fence_output=False,
    use_schema_constraints=False
)

Quick setup: Install Ollama from ollama.com, run ollama pull gemma2:2b, then ollama serve.

For detailed installation, Docker setup, and more examples, see examples/ollama/.

Full Changelog: v1.0.3...v1.0.4

v1.0.3 – OpenAI language model support

What's Changed

• Added OpenAI language model integration – Support for GPT-4o, GPT-4o-mini, and other OpenAI models
• Enhanced documentation – Added OpenAI usage examples and API key setup instructions to README
• Comprehensive test coverage – Added unit tests for OpenAI backend

Technical Details

• Uses modern OpenAI v1.x client API with parallel processing support
• Note: Schema constraints for OpenAI are not yet implemented (use use_schema_constraints=False)

Full Changelog: v1.0.2...v1.0.3

v1.0.2 – Slimmer install, Windows fix, OpenAI v1.x support

What’s Changed

• Removed langfun and pylibmagic dependencies – lighter install; no libmagic needed on Windows
• Fixed Windows-installation failure [#25]
• Restored compatibility with modern OpenAI SDK v1.x [#16]
• Updated README and Dockerfile to match the new, slimmer dependency set

Note

LangFunLanguageModel has been removed.
If you still need LangFun support, please open a new issue so we can discuss re-adding it in a cross-platform way.

Full Changelog: v1.0.1...v1.0.2

What's Changed

• Fixed libmagic ImportError by adding pylibmagic dependency (#6)
• Added [full] install option for easier setup
• Added Docker support with pre-installed libmagic
• Updated troubleshooting documentation

Bug Fixes

• Resolve "failed to find libmagic" error when importing langextract (#6)

Installation

# Standard install (now includes pylibmagic)
pip install langextract

# Full install (explicit all dependencies)
pip install langextract[full]

# Docker (libmagic pre-installed)
docker run --rm -e LANGEXTRACT_API_KEY="your-key" langextract python script.py

Full Changelog: v1.0.0...v1.0.1

A Python library for extracting structured information from unstructured text using LLMs with precise source grounding and interactive visualization.

Key Features

• Extract structured data from any text using few-shot examples
• Support for Gemini and Ollama models
• Interactive HTML visualizations with source highlighting
• Optimized for long documents with parallel processing and multiple extraction passes
• Precise source grounding - every extraction maps to its location in the original text

Installation

pip install langextract

See the documentation for full usage examples.