Ensure GTFS ingest (#100)

* GTFS ingest should have a week lookback

* Run at 8am UTC instead of 7

.gitignore

@@ -1,4 +1,5 @@
 .env.local
 **/__pycache__
 .python-version
-.temp
+.temp
+feeds

ingestor/app.py

@@ -104,13 +104,6 @@ def update_delivered_trip_metrics_yesterday(event):
     daily_speeds.update_daily_table(two_days_ago)
 
 
-# 7am UTC -> 2/3am ET
-@app.schedule(Cron(0, 7, "*", "*", "?", "*"))
-def update_gtfs(event):
-    today = date.today()
-    gtfs.ingest_gtfs_feeds_to_dynamo_and_s3(date_range=(today, today))
-
-
 # 7:10am UTC -> 2:10/3:10am ET every day
 @app.schedule(Cron(10, 7, "*", "*", "?", "*"))
 def update_ridership(event):
@@ -129,6 +122,14 @@ def update_time_predictions(event):
     predictions.update_predictions()
 
 
+# 8:00am UTC -> 3:00/4:00am ET every day
+@app.schedule(Cron(0, 8, "*", "*", "?", "*"))
+def update_gtfs(event):
+    today = datetime.now()
+    last_week = (today - timedelta(days=7)).date()
+    gtfs.ingest_gtfs_feeds_to_dynamo_and_s3(date_range=(last_week, today.date()))
+
+
 # 4:40am UTC -> 2:40/3:40am ET every day
 @app.schedule(Cron(40, 7, "*", "*", "?", "*"))
 def update_trip_metrics(event):

ingestor/chalicelib/s3.py

@@ -55,7 +55,7 @@ def ls(bucket, prefix):
     return all_keys
 
 
-def clear_cf_cache(distribution, keys):
+def clear_cf_cache(distribution: str, keys: list[str]):
     cloudfront.create_invalidation(
         DistributionId=distribution,
         InvalidationBatch={

ingestor/chalicelib/trip_metrics/backfill.py

@@ -3,8 +3,8 @@
 
 from .ingest import ingest_trip_metrics, get_date_ranges
 
-START_DATE = date(2023, 8, 29)
-END_DATE = date(2023, 8, 30)
+START_DATE = date(2024, 3, 1)
+END_DATE = date(2024, 5, 14)
 MAX_RANGE_SIZE = 90
 
 if __name__ == "__main__":