How filigott/dat640-project works
Project type and purpose
• This is a full-stack, conversational music recommendation application.
• Frontend: Vite + React (TypeScript) SPA with chat widget, playlist and song list components.
• Backend: FastAPI service exposing REST and WebSocket endpoints, using SQLAlchemy/Alembic for PostgreSQL persistence, plus a chat-agent layer.
• NLP/NLU: Rasa agent running as a separate service for intent parsing and dialogue management.
• Simulation: A built-in user simulation module for testing chat flows.
File structure
• Top-level folders:
• backend_fastAPI/ – FastAPI code, Alembic migrations, Docker Compose for Postgres
• frontend/ – Vite + React application
• rasa_agent/ – Rasa NLU project (actions, data, models, endpoints)
• backend_fastAPI/app/ packages:
• routers/ – HTTP & WebSocket route definitions (song, playlist, seed, client-only, chat)
• repository/ – data access layer (Repository pattern over SQLAlchemy models)
• chat_agent/ & chat_agent_service/ – orchestration and external service calls for chat
• simulation/ – user‐simulator code driving WebSocket client for testing
• database.py, init_db.py, models.py, schemas.py – DB setup and ORM mapping
• Configuration/build files:
• backend_fastAPI/alembic/ – DB migration scripts
• backend_fastAPI/postgres/docker-compose.yml – Postgres container definition
• frontend/vite.config.ts, tsconfig*.json, package.json – frontend build/dev config
• rasa_agent/*.yml, models/ – Rasa config, training data, and pre-built model
• run.sh, requirements.txt – top-level orchestration and Python deps
Key points from README
• Services to run: Docker Compose for Postgres, FastAPI (uvicorn), React (npm run dev), Rasa (rasa run).
• Endpoints: REST (8000), WebSocket (8000), frontend (8001), Rasa API (5005), Adminer (8080).