Chess Calendar (India) — static / GitHub-hosted build

A free, serverless deployment of the chess-calendar app. There is no live backend: a scheduled GitHub Actions job runs the scrapers and commits the result as a static JSON file, and GitHub Pages serves the frontend.

This repo was derived from the FastAPI app; the Python backend/ is copied verbatim and reused by scripts/export_json.py (which calls the same refresh_all() / query_tournaments()), so scraping/geocoding/dedup behaviour is identical — only the delivery changed.

How it works

GitHub Actions (cron, 2×/day)          GitHub Pages
  refresh.yml                            pages.yml
    └─ python scripts/export_json.py       └─ serves site/
         ├─ refresh_all()   → backend/data/chess.sqlite   (committed)
         └─ query_tournaments() → site/tournaments.json   (committed)

• site/ — the static frontend (index.html, app.js, styles.css) plus the generated tournaments.json. The frontend fetches ./tournaments.json instead of a live API; the old "Refresh sources" button is removed.
• backend/data/chess.sqlite is committed on purpose. It holds the per-tournament detail cache. Persisting it between runs keeps each scheduled refresh to a handful of requests instead of a ~1500-request cold crawl — essential for staying under chess-results.com's 2000 requests/IP/day cap across two runs a day.

One-time setup on GitHub

1. Create a repo and push this directory.
2. Settings → Pages → Build and deployment → Source: GitHub Actions.
3. refresh.yml needs write access to commit data: Settings → Actions → General → Workflow permissions → Read and write permissions.
4. (Optional) trigger the first run manually: Actions → "Refresh tournament data" → Run workflow.

Gotchas

• Freshness is bounded by the cron (2×/day). No on-demand refresh from the page.
• Shared runner IPs: Actions runners egress from shared, rotating IPs. The small per-run request count (thanks to the committed cache) keeps this well within chess-results' cap, but a cold crawl on a fresh DB is ~1500 requests — don't run several cold refreshes in one day.
• Cron auto-disable: GitHub disables scheduled workflows after 60 days of repository inactivity. The 2×/day data commits count as activity, so this won't trigger in normal operation.

Local preview

python scripts/export_json.py          # refresh + regenerate site/tournaments.json
python -m http.server -d site 8000     # then open http://localhost:8000