This project provides an example of how to create a Rust websocket endpoint which streams responses from Python workers through Redis.

System setup requires

1. A local Redis server running on the default port
2. websocat for a terminal based websocket client

Setup & start a python worker

1. Setup OPENAI_API_KEY environment variable. If you don't have/want one, use slow_verbose_echo_worker.py instead of chatgpt_worker.py.
2. cd workers/
3. Create & activate a venv
4. pip install -r requirements.txt
5. python3 chatgpt_worker.py

Setup & start the Rust API

1. cd api
2. cargo run

Interact via websockets

1. websocat --no-line ws://127.0.0.1:8080/ws
2. type a prompt, press enter

Expected result

1. The Rust server logging that it writes to the task queue
2. The Python worker picking up a task
3. Websocat writing the ChatGPT response as it comes back

demo.webm.mov