This document describes how to run DataGenFlow using Docker.

1. Build and start the application:

   docker-compose up -d

2. Access the application:

   • Frontend: http://localhost:8000
   • API: http://localhost:8000/api

3. Stop the application:

   docker-compose down

Custom blocks can be added to lib/blocks/custom/ on your host system. They will be automatically available after restarting the backend container:

docker-compose restart backend

The lib/blocks/custom/ directory is mounted as a volume, so you can add new block files directly from your host system without rebuilding the image.

You can configure the application using environment variables. Create a .env file in the project root:

LLM_ENDPOINT=http://localhost:11434/api/generate
LLM_API_KEY=
LLM_MODEL=llama3
DEBUG=false

These variables are automatically passed to the container via docker-compose.yml.

The data/ directory is mounted as a volume, so your database and other data will persist between container restarts.

To rebuild the images:

docker-compose build

Or rebuild without cache:

docker-compose build --no-cache

For development, you may want to mount additional directories or use volume mounts for live code reloading. Modify docker-compose.yml as needed.

• Backend: Python 3.11 with uv, serves both API and frontend
• Frontend: Built with yarn/vite, served as static files by the backend
• Port: 8000 (both API and frontend)

The backend Dockerfile:

• Uses multi-stage builds for optimization
• Compiles Python bytecode for faster startup
• Builds the frontend and includes it in the final image
• Serves the frontend at the root path via FastAPI