feat: add CRON job for static schedules

config/cron.sh

@@ -1,5 +1,21 @@
-# This script adds commands to the current crontab
+#!/bin/sh
 
-# run the daily script at 1am every morning
-# TODO: make sure timezone is PST
-crontab -l | { cat; echo "0 1 * * * /home/csss-site/csss-site-backend/src/cron/daily.py"; } | crontab -
+# Install the weekly TransLink static schedule refresh without duplicating it.
+set -eu
+
+cron_file=$(mktemp)
+trap 'rm -f "$cron_file"' EXIT
+
+crontab -l 2>/dev/null | sed \
+    -e '/# BEGIN CSSS TRANSLINK STATIC/,/# END CSSS TRANSLINK STATIC/d' \
+    -e '\|scripts.refresh_translink_static|d' > "$cron_file" || true
+
+{
+    cat "$cron_file"
+    printf '%s\n' \
+        '# BEGIN CSSS TRANSLINK STATIC' \
+        'PATH=/home/csss-site/.local/bin:/usr/local/bin:/usr/bin:/bin' \
+        'CRON_TZ=America/Vancouver' \
+        '0 23 * * 5 cd /home/csss-site/csss-site-backend/src && uv run python -m scripts.refresh_translink_static' \
+        '# END CSSS TRANSLINK STATIC'
+} | crontab -

src/scripts/refresh_translink_static.py

@@ -0,0 +1,47 @@
+"""Download, preprocess, and store the TransLink static GTFS schedule."""
+
+import asyncio
+import logging
+
+import httpx
+
+import database
+from translink.crud import refresh_static_schedule
+
+_logger = logging.getLogger(__name__)
+
+
+async def refresh() -> None:
+    await database.setup_database()
+    if database.sessionmanager is None:
+        raise RuntimeError("Database has not been initialized")
+
+    manager = database.sessionmanager
+    try:
+        async with httpx.AsyncClient(timeout=60, follow_redirects=True) as client:
+            async with manager.session() as session:
+                cache = await refresh_static_schedule(session, client)
+        departure_count = sum(len(rows) for rows in cache["departures"].values())
+        _logger.info(
+            "Stored TransLink static schedule version %s with %s departures covering %s through %s",
+            cache["version"],
+            departure_count,
+            cache["coverage"]["start_date"],
+            cache["coverage"]["end_date"],
+        )
+    finally:
+        await manager.close()
+
+
+def main() -> int:
+    logging.basicConfig(level=logging.INFO)
+    try:
+        asyncio.run(refresh())
+    except Exception:
+        _logger.exception("Failed to refresh the TransLink static schedule")
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/translink/README.md

@@ -9,8 +9,30 @@ All dates are adjusted for the America/Vancouver timezone.
 3. Make sure your database has the correct migrations `alembic upgrade head`. Reload your test database as well `python src/load_test_db.py`
 4. Start (or restart) the web server to test the endpoints
 
+## Static schedule refresh
+
+The static GTFS archive is downloaded and preprocessed outside HTTP requests. Before serving the static schedule for
+the first time, populate the cache manually from the `src` directory:
+
+```bash
+# in ./src
+uv run python -m scripts.refresh_translink_static
+```
+
+If deploying on the web server run the cron job.
+```bash
+# in root
+sh config/cron.sh
+```
+Production refreshes it every Friday at 11:00 PM in the America/Vancouver timezone. Install or update that cron entry
+by running `sh config/cron.sh` from the repository root. The installer is idempotent.
+
+If a refresh fails, the prior database row is preserved and the command exits unsuccessfully. If no compatible cache
+can serve the current date, the static and combined schedule endpoints return HTTP 503; requests never download or
+parse the static GTFS archive.
+
 ## Endpoints
 You can see the exact schemas in the `/docs` page. At the time this was written there are three endpoints:
 1. `translink/realtime`: returns realtime data for buses that are at or are approaching SFU
-2. `translink/static`: returns the schedule for the current day
+2. `translink/static`: returns the preprocessed schedule for the current day
 3. `translink/schedule`: combines the realtime and static data to show if a bus is at the loop, is running late, or was cancelled