LangChain Service is an opinionated processing platform that allows you to develop and deploy langchain as a service with batteries included (Kong Authentication, Qdrant Vector Database, Postgresql Database, Swagger, etc.). This documentation will guide you through the installation process, as well as provide an overview of the available features and API reference.
Based on python-poetry-docker-example.
This project uses Poetry as a dependency manager. Check out Poetry's documentation on how to install it on your system before proceeding.
- create and activate a new Conda env
conda create -n langchain-service python=3.11 - Install Poetry (see introduction)
curl -sSL https://install.python-poetry.org | python3 - - Tell Poetry to use the virtualenv python environment
poetry config virtualenvs.prefer-active-python truepoetry shell - Install dependencies
poetry install -E all
See the poetry docs for information on how to add/update dependencies.
Start a development server locally:
poetry run uvicorn app.main:app --reload --host localhost --port 8000API will be available at localhost:8000/
FastAPI comes with built-in API generated documentation. Available at
- Swagger UI docs at localhost:8000/docs
- ReDoc docs at localhost:8000/redoc
Build images with:
docker build -t langchain-service .The Dockerfile uses multi-stage builds to run lint and test stages before building the production stage.
You can stop the build at specific stages with the --target option with the following values: development, lint, test, production, e.g.:
docker build -t langchain-service --target $STAGE .If you do not specify a target the resulting image will be the last image defined which in our case is the 'production' image.
Run the 'production' image:
docker run -it -p 8000:8000 langchain-service
Open your browser and go to http://localhost:8000/redoc to see the API spec in ReDoc.
This project offers a complete and convenient setup using Docker Compose, allowing you to run multiple services as part of the LangChain ecosystem. Docker Compose simplifies the process of managing and organizing the various services required for a fully functional environment. In this setup, the following services are included:
- Kong: Kong serves as the API Gateway and manages authentication for the LangChain API. It acts as a reverse proxy, handling requests from clients and forwarding them to the appropriate service, while also providing authentication and other middleware features.
- Qdrant: Qdrant is a vector database that allows efficient search and retrieval of similar items based on their vector representation. It enables storing and searching for language model embeddings.
- PostgreSQL: PostgreSQL provides a robust and reliable database solution for storing and managing data associated with the LangChain API.
To build and run the containers for these services using Docker Compose, use the following command:
docker compose upIf you prefer to run the containers in detached mode, which means they will run in the background without displaying logs in the terminal, use the -d flag:
docker compose up -d