LangExtract OpenAI Plugin

OpenAI and Azure OpenAI provider plugin for LangExtract that enables structured data extraction using OpenAI's language models.

Features

• OpenAI Support: Direct integration with OpenAI's API
• Azure OpenAI Support: Full support for Azure OpenAI deployments
• Structured Output: JSON and YAML format support
• Parallel Processing: Efficient batch processing with configurable concurrency
• Plugin System: Seamless integration with LangExtract's provider system

Structure

langextract-openai/
├── pyproject.toml                    # Package configuration and metadata
├── README.md                         # This file
├── langextract_openai/              # Package directory
│   ├── __init__.py                  # Package initialization and exports
│   └── openai_providers.py         # OpenAI and Azure OpenAI providers
├── examples/                        # Usage examples
│   └── usage_examples.ipynb        # Jupyter notebook with examples
└── LICENSE

Provider Implementations

• OpenAI (OpenAILanguageModel): Direct OpenAI API integration
• Azure OpenAI (AzureOpenAILanguageModel): Azure OpenAI service integration (inherits OpenAI implementation)

Package Configuration (pyproject.toml)

[project.entry-points."langextract.providers"]
openai = "langextract_openai:OpenAILanguageModel"
azure_openai = "langextract_openai:AzureOpenAILanguageModel"

This entry point allows LangExtract to automatically discover your provider.

Installation

Prerequisites

First install the latest LangExtract from source:

git clone https://github.com/google/langextract.git
cd langextract
pip install -e .

Install Plugin

# Clone this plugin
git clone <this-repo-url>
cd langextract-openai

# Install in development mode
pip install -e .

# Run the example notebook
# (ensure you have Jupyter installed: pip install jupyter)
jupyter notebook examples/usage_examples.ipynb

Quick Start

OpenAI

import langextract as lx

# Extract structured data with OpenAI
result = lx.extract(
    text_or_documents="John Smith is a software engineer at Tech Corp.",
    model_id="gpt-4o-mini",
    api_key="your-openai-api-key",
    prompt_description="Extract person's name, job title, and company",
    examples=[{
        "input": "Jane Doe works as a data scientist at DataCorp.",
        "output": {"name": "Jane Doe", "job_title": "data scientist", "company": "DataCorp"}
    }]
)

# Tip: If there are multiple providers matching your model_id in your environment,
# you can disambiguate by explicitly specifying the provider name:
# result = lx.extract(
#     text_or_documents="...",
#     model_id="gpt-4o-mini",
#     api_key="...",
#     provider="OpenAILanguageModel",
#     prompt_description="...",
# )

Azure OpenAI

import langextract as lx

# Extract with Azure OpenAI
result = lx.extract(
    text_or_documents="John Smith is a software engineer at Tech Corp.",
    model_id="azure:your-deployment-name",
    api_key="your-azure-api-key",
    azure_endpoint="https://your-resource.openai.azure.com",
    prompt_description="Extract person's name, job title, and company",
)

Environment Variables

Set these environment variables for the examples and tests:

• OPENAI_API_KEY: Your OpenAI API key
• AZURE_OPENAI_API_KEY: Your Azure OpenAI API key
• AZURE_OPENAI_ENDPOINT: Your Azure OpenAI endpoint URL
• AZURE_OPENAI_DEPLOYMENT: Your Azure deployment name (optional, defaults to 'gpt-4o-mini')

Release

1. Bump version in pyproject.toml under [project] version.

2. Build and upload to PyPI:

python -m pip install --upgrade build twine
rm -rf dist build *.egg-info
python -m build
twine upload dist/*

Optional: Upload to TestPyPI first:

twine upload --repository testpypi dist/*

Optional: Tag the release in git:

git tag vX.Y.Z
git push origin vX.Y.Z

Notes:

• Use a PyPI API token (username: __token__, password: your token), or configure ~/.pypirc.
• Ensure you have a clean tree and tests/examples pass before publishing.

License

Apache License 2.0