Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Andr21Da16/UNITRU-ACADEMIC/llms.txt

Use this file to discover all available pages before exploring further.

Both the backend and frontend services are configured entirely through environment variables. No source files or config files need to be edited manually — set the variables in your shell, Docker Compose file, or Railway dashboard before starting the services.

​
Backend environment variables

The backend is a FastAPI application served by Uvicorn. It reads its configuration at startup from the process environment.
VariableDefaultDescriptionRequired
ALLOWED_ORIGINShttp://localhost:3000Comma-separated list of CORS-allowed origins. In production, set this to the Railway-generated frontend domain (e.g. https://unitru-academic.up.railway.app). Required for cross-origin WebSocket connections to succeed.Yes
PORT8000Server listen port. The backend CMD expands ${PORT:-8000}, so it works with or without the variable.No
Railway injects PORT automatically into every service at runtime. Do not set it manually in the Railway dashboard — the backend’s Dockerfile CMD already handles it with ${PORT:-8000}.

​
Frontend environment variables

The frontend is a Next.js application compiled to a standalone Node.js server. Because it uses a NEXT_PUBLIC_* variable, the value is embedded into the JavaScript bundle at build time, not at runtime.
VariableDefaultDescriptionRequired
NEXT_PUBLIC_BACKEND_WS_URLws://localhost:8000/wsWebSocket URL the student’s browser connects to. In production this must use the wss:// scheme with the Railway backend domain: wss://unitru-backend.up.railway.app/ws.Yes (production)
NEXT_PUBLIC_BACKEND_WS_URL is a build-time variable. It must be present as a Docker ARG before npm run build runs. Changing it in the Railway dashboard after the image has already been built has no effect until you trigger a full redeploy so the image is rebuilt with the new value.

​
Docker Compose configuration

For local development, both variables are set directly in docker-compose.yml. The frontend receives NEXT_PUBLIC_BACKEND_WS_URL as a Docker build argument (args:), not as a runtime environment: entry, because Next.js embeds it at build time.
docker-compose.yml
services:
  backend:
    build: ./backend
    image: unitru-academic-backend
    ports:
      - "8000:8000"
    environment:
      # Comma-separated CORS-allowed origins — see src/main.py
      ALLOWED_ORIGINS: "http://localhost:3000"
    restart: unless-stopped

  frontend:
    build:
      context: ./frontend
      args:
        # Embedded in build; the student's browser connects to this URL.
        NEXT_PUBLIC_BACKEND_WS_URL: "ws://localhost:8000/ws"
    image: unitru-academic-frontend
    ports:
      - "3000:3000"
    depends_on:
      - backend
    restart: unless-stopped

​
Railway cross-service references

Railway can resolve references between services in the same project using the ${{service.VARIABLE}} syntax. This means you never have to paste a domain manually — Railway fills it in when the referenced service generates its public domain. Backend service — Variables tab: 
ALLOWED_ORIGINS = https://${{frontend.RAILWAY_PUBLIC_DOMAIN}}
Frontend service — Variables tab (used as a build arg): 
NEXT_PUBLIC_BACKEND_WS_URL = wss://${{backend.RAILWAY_PUBLIC_DOMAIN}}/ws
Note the wss:// scheme (TLS) and the /ws path suffix. Railway terminates TLS at its proxy layer and forwards traffic to the backend Uvicorn process, which must be started with --proxy-headers --forwarded-allow-ips="*" — the backend Dockerfile already does this.

​
Schedule catalog

The schedule catalog (data/horarios_catalogo.json) is a JSON file that maps course names to their official section timetables for the current semester. It is baked into the backend Docker image at build time via the COPY . . instruction in the backend Dockerfile — it is not fetched at runtime. To update the catalog when the university publishes a new semester schedule:
  1. Re-run scripts/download_horarios.py with the new Google Sheets URL.
  2. Re-run scripts/parse_horarios.py to regenerate data/horarios_catalogo.json.
  3. Commit the updated JSON file to the repository.
  4. Redeploy the backend service so Docker rebuilds the image with the new catalog baked in.
See the Scripts page for detailed usage of each catalog script.

Build docs developers (and LLMs) love

Get started for freeTalk to us