fix circular FK between feed and gtfsdataset

Author: davidgamez

functions-python/reverse_geolocation/src/reverse_geolocation_processor.py

@@ -33,6 +33,7 @@
     Osmlocationgroup,
     Gtfsdataset,
     Gtfsfeed,
+    Gbfsfeed,
 )
 from shared.dataset_service.dataset_service_commons import Status
 
@@ -141,7 +142,7 @@ def create_geojson_aggregate(
     bounding_box: shapely.Polygon,
     data_type: str,
     logger,
-    feed: Feed,
+    feed: Gtfsfeed | Gbfsfeed,
     gtfs_dataset: Gtfsdataset = None,
     extraction_urls: List[str] = None,
     public: bool = True,
@@ -345,30 +346,15 @@ def reverse_geolocation_process(
         # Update the bounding box of the dataset
         if dataset_id:
             gtfs_dataset = load_dataset(dataset_id, db_session)
-            feed = gtfs_dataset.feed
-        if not feed:
-            feed = (
-                db_session.query(Feed).filter(Feed.stable_id == stable_id).one_or_none()
-            )
-            if not feed:
-                no_feed_message = f"No feed found for stable ID {stable_id}."
-                logger.warning(no_feed_message)
-                record_execution_trace(
-                    execution_id=execution_id,
-                    stable_id=stable_id,
-                    status=Status.FAILED,
-                    logger=logger,
-                    dataset_file=None,
-                    error_message=no_feed_message,
-                )
-                return no_feed_message, ERROR_STATUS_CODE
+        feed = load_feed(stable_id, data_type, logger, db_session)
 
         bounding_box = update_dataset_bounding_box(gtfs_dataset, stops_df, db_session)
 
         location_groups = reverse_geolocation(
             strategy=strategy,
             stable_id=stable_id,
             stops_df=stops_df,
+            data_type=data_type,
             logger=logger,
             use_cache=use_cache,
             db_session=db_session,
@@ -447,6 +433,7 @@ def reverse_geolocation(
     strategy,
     stable_id,
     stops_df,
+    data_type,
     logger,
     use_cache,
     db_session: Session = None,
@@ -456,7 +443,7 @@ def reverse_geolocation(
     """
     logger.info("Processing geopolygons with strategy: %s.", strategy)
 
-    feed = load_feed(stable_id, logger, db_session)
+    feed = load_feed(stable_id, data_type, logger, db_session)
 
     # Get Geopolygons with Geometry and cached location groups
     cache_location_groups, unmatched_stops_df = get_geopolygons_with_geometry(
@@ -495,9 +482,10 @@ def reverse_geolocation(
     return cache_location_groups
 
 
-def load_feed(stable_id, logger, db_session):
+def load_feed(stable_id, data_type, logger, db_session) -> Gtfsfeed | Gbfsfeed:
+    """Load feed from the database using the stable ID and data type."""
     feed = (
-        db_session.query(Feed)
+        db_session.query(Gbfsfeed if data_type == "gbfs" else Gtfsfeed)
         .options(joinedload(Feed.feedlocationgrouppoints))
         .filter(Feed.stable_id == stable_id)
         .one_or_none()

functions-python/reverse_geolocation/tests/test_reverse_geolocation_processor.py

@@ -451,10 +451,12 @@ def test_update_dataset_bounding_box_exception(self, db_session):
     @patch("reverse_geolocation_processor.check_maximum_executions")
     @patch("reverse_geolocation_processor.get_execution_id")
     @patch("reverse_geolocation_processor.load_dataset")
+    @patch("reverse_geolocation_processor.load_feed")
     @patch("reverse_geolocation_processor.record_execution_trace")
     def test_valid_request(
         self,
         _,
+        mock_load_feed,
         mock_load_dataset,
         mock_get_execution_id,
         mock_check_maximum_executions,
@@ -482,9 +484,10 @@ def test_valid_request(
         mock_update_bounding_box.return_value = MagicMock()
         mock_reverse_geolocation.return_value = {"group_id": MagicMock()}
         mock_create_geojson_aggregate.return_value = MagicMock()
+        mock_load_feed.return_value = MagicMock()
+        mock_load_dataset.return_value = MagicMock()
         # Mocking a Flask request
         request = MagicMock(spec=Request)
-
         # Call the function
         response, status_code = reverse_geolocation_process(request)
 
@@ -520,10 +523,12 @@ def test_invalid_request(self, mock_parse_request_parameters):
     @patch("reverse_geolocation_processor.check_maximum_executions")
     @patch("reverse_geolocation_processor.get_execution_id")
     @patch("reverse_geolocation_processor.load_dataset")
+    @patch("reverse_geolocation_processor.load_feed")
     @patch("reverse_geolocation_processor.record_execution_trace")
     def test_exception_handling(
         self,
         _,
+        mock_load_feed,
         mock_load_dataset,
         mock_check_get_execution_id,
         mock_check_maximum_executions,
@@ -549,6 +554,7 @@ def test_exception_handling(
             1,
         )
         mock_load_dataset.return_value = MagicMock()
+        mock_load_feed.return_value = MagicMock()
         mock_update_bounding_box.side_effect = Exception("Unexpected error")
 
         # Mocking a Flask request
@@ -632,6 +638,7 @@ def test_valid_per_point_strategy(
             stable_id="test_stable_id",
             stops_df=pd.DataFrame({"stop_lat": [1.0], "stop_lon": [1.0]}),
             logger=MagicMock(),
+            data_type="gtfs",
             use_cache=True,
             db_session=MagicMock(spec=Session),
         )
@@ -670,6 +677,7 @@ def test_valid_per_polygon_strategy(
             strategy=ReverseGeocodingStrategy.PER_POLYGON,
             stable_id="test_stable_id",
             stops_df=pd.DataFrame({"stop_lat": [1.0], "stop_lon": [1.0]}),
+            data_type="gtfs",
             logger=MagicMock(),
             use_cache=True,
             db_session=MagicMock(spec=Session),
@@ -689,7 +697,7 @@ def test_invalid_strategy(
         from reverse_geolocation_processor import reverse_geolocation
 
         id = str(uuid.uuid4())
-        feed = Feed(
+        feed = Gtfsfeed(
             id=id,
             stable_id=f"test_feed{id}",
             status="active",
@@ -705,6 +713,7 @@ def test_invalid_strategy(
             stable_id=f"test_feed{id}",
             stops_df=pd.DataFrame({"stop_lat": [1.0], "stop_lon": [1.0]}),
             logger=MagicMock(),
+            data_type=feed.data_type,
             use_cache=True,
             db_session=db_session,
         )
@@ -724,6 +733,7 @@ def test_load_feed_missing_feed(self, db_session):
                 strategy=ReverseGeocodingStrategy.PER_POINT,
                 stable_id="missing_stable_id",
                 stops_df=pd.DataFrame({"stop_lat": [1.0], "stop_lon": [1.0]}),
+                data_type="gtfs",
                 logger=MagicMock(),
                 use_cache=True,
                 db_session=db_session,

functions-python/tasks_executor/tests/tasks/geojson/test_update_geojson_files_precision.py

@@ -282,7 +282,6 @@ def test_handler_uploads_and_updates_gbfs_feed_info(self, db_session: Session):
             .limit(1)
             .first()
         )
-        self.assertIsNone(reloaded_testing_feed.geolocation_file_dataset_id)
         self.assertIsNotNone(reloaded_testing_feed.geolocation_file_created_date)
 
     @with_db_session(db_url=default_db_url)

liquibase/changelog.xml

@@ -68,4 +68,5 @@
     <include file="changes/feat_1260.sql" relativeToChangelogFile="true"/>
     <include file="changes/feat_1333.sql" relativeToChangelogFile="true"/>
     <include file="changes/feat_pt_152.sql" relativeToChangelogFile="true"/>
+    <include file="changes/feat_fix_geolocation_circular_dep.sql" relativeToChangelogFile="true"/>
 </databaseChangeLog>