Django OpenAI-Powered Chatbot with HTMX Streaming

While searching the web for an example of a Django-based LLM chatbot featuring asynchronous streaming, I found no suitable examples that incorporated HTMX and async. This prompted me to put together this repository.

Key Features

• Asynchronous Streaming: This project leverages the HTMX websockets extension and Django's AsyncWebsocketConsumer to achieve real-time, streaming interactions with the chatbot.
• OpenAI Integration: Utilizes AsyncOpenAI client for seamless communication with OpenAI's API.
• Conversation History: Enhances user experience by maintaining a history of interactions, allowing the chatbot to provide contextually relevant responses.

Can be adapted to work with open-source Large Language Models like Llama-3, provided that the HTTP client handling communications with the LLM supports asynchronous operations.

Notes about deploying to production

Because websockets requires ASGI instead of WSGI server, you should use something like Uvicorn: https://www.uvicorn.org/ to serve your application. I provided an example how to set this up in the docker-compose.yml file. Check out the additional environment variables in there if you plan on deploying this. Inside the django_chatbot_app/chatbot/settings.py file we check ENVIRONMENT variable to determine if we are in production or development mode. In development mode we use daphne ASGI server that is not suitable for production. Also take a look at scripts/gunicorn/gunicorn_config.py this is where we configure the gunicorn server to use Uvicorn as the worker in production.

Running the application locally:

• Create OpenAI api key at: https://platform.openai.com/

• Create .env file to the root directory. Example: .env_example

• Create your virtual environment and install dependencies:

pip install -r requirements.txt

Note that the requirements_prod.txt file should only be used in "production" setup (docker-compose.yaml). See the Dockerfile for more information.

• Run migrations and runserver

python manage.py migrate
python manage.py runserver

• Create a user for the application and start chatting.

Running the application in docker-compose (production-like setup):

• Create OpenAI api key at: https://platform.openai.com/
• Create .env file to the root directory. Example: .env_example.

docker-compose build
docker-compose up

The application is available at http://127.0.0.1:8000