Example FastAPI server using Langroid

Setup

For env setup, follow the instructions in the langroid-examples repo, which means roughly the following. First ensure you have the latest poetry and you are using python 3.11+.

python3 -m venv .venv
. .venv/bin/activate
poetry env use 3.11
poetry install

Copy the .env-template to .env and fill in the values of the OPENAI_API_KEY, QDRANT_API_KEY and QDRANT_API_URL variables. (The latter two are needed to use the Qdrant Cloud vector-store; consult their docs to see how to get these values.)

Directory structure

Various files, dirs:

• fastapi_server/: FastAPI server code, defines the routes/endpoints. Contains:
  • Dockerfile: to build the docker container
  • app.py: defines the endpoints
  • requirements.txt: python dependencies
• tests/: tests for the FastAPI server
• Makefile: to build and run the docker container
• server/: auxiliary code modules that can be imported into app.py

Building, running the docker container

make build

Note that this will run on localhost:90 -- caution it's 90 not 80

make up

See docker container logs in tail -f mode:

make tail

Test the FastAPI server

This tests the server using both the test client provided by FastAPI (which lets you step into the code while it's running), and also the endpoint served via uvicorn from the docker container.

pytest -xs tests/test_fastapi.py