CineRaaga

Curated Indian cinema discovery and reviews across Hindi, Tamil, Telugu, Malayalam and Punjabi. Editorial authority, not a crowd platform.

Navras is the internal method layer: nine-rasa mood taxonomy and the hand-assigned Navras Score (0–100). A score only exists when it has been personally assigned — nothing is ever derived from crowd ratings.

Architecture (launch mode: static export)

backend/    FastAPI + SQLAlchemy. Source of truth (Postgres via docker-compose,
            SQLite for quick local work). Seed → DB → static JSON export.
frontend/   Static HTML/CSS/JS served by Vercel. Reads frontend/data/*.json.

At launch nothing runs live — the export output is committed and deployed. The FastAPI routers become the live layer in month 2.

Workflow

cd backend
pip install -r requirements.txt
export DATABASE_URL=sqlite:///./cineraaga.db   # or Postgres from docker-compose
python -m app.services.seed      # idempotent
python -m app.services.export    # writes ../frontend/data/
pytest -q

Data rules

• Film.navras_score is nullable. score_status is placeholder for numbers imported from the old prototype and official only when hand-scored. The frontend renders a score badge only for official.
• RasaTag.source is human or ai — the basis for the month-3 classification eval set.
• Reviews (body_md) are written by a human. Always.