Merge pull request #1148 from MobilityData/1105-implement-or-filter-for-feature-search

feat: implemented OR filter for feature search

Author: qcdyx

.github/workflows/build-test.yml

@@ -119,7 +119,6 @@ jobs:
 
       - name: Upload DB models
         uses: actions/upload-artifact@v4
-        if: ${{ steps.set-should-run-tests.outputs.result == 'true' }}
         with:
           name: database_gen
           path: api/src/shared/database_gen/

api/src/feeds/impl/models/search_feed_item_result_impl.py

@@ -33,37 +33,41 @@ def from_orm_gtfs(cls, feed_search_row: t_feedsearch):
             feed_contact_email=feed_search_row.feed_contact_email,
             source_info=SourceInfo(
                 producer_url=feed_search_row.producer_url,
-                authentication_type=int(feed_search_row.authentication_type)
-                if feed_search_row.authentication_type
-                else None,
+                authentication_type=(
+                    int(feed_search_row.authentication_type) if feed_search_row.authentication_type else None
+                ),
                 authentication_info_url=feed_search_row.authentication_info_url,
                 api_key_parameter_name=feed_search_row.api_key_parameter_name,
                 license_url=feed_search_row.license_url,
             ),
             redirects=feed_search_row.redirect_ids,
             locations=cls.resolve_locations(feed_search_row.locations),
-            latest_dataset=LatestDataset(
-                id=feed_search_row.latest_dataset_id,
-                hosted_url=feed_search_row.latest_dataset_hosted_url,
-                downloaded_at=feed_search_row.latest_dataset_downloaded_at,
-                hash=feed_search_row.latest_dataset_hash,
-                service_date_range_start=feed_search_row.latest_dataset_service_date_range_start,
-                service_date_range_end=feed_search_row.latest_dataset_service_date_range_end,
-                agency_timezone=feed_search_row.latest_dataset_agency_timezone,
-                validation_report=LatestDatasetValidationReport(
-                    total_error=feed_search_row.latest_total_error,
-                    total_warning=feed_search_row.latest_total_warning,
-                    total_info=feed_search_row.latest_total_info,
-                    unique_error_count=feed_search_row.latest_unique_error_count,
-                    unique_warning_count=feed_search_row.latest_unique_warning_count,
-                    unique_info_count=feed_search_row.latest_unique_info_count,
-                    features=sorted([feature for feature in feed_search_row.latest_dataset_features])
-                    if feed_search_row.latest_dataset_features
-                    else [],
-                ),
-            )
-            if feed_search_row.latest_dataset_id
-            else None,
+            latest_dataset=(
+                LatestDataset(
+                    id=feed_search_row.latest_dataset_id,
+                    hosted_url=feed_search_row.latest_dataset_hosted_url,
+                    downloaded_at=feed_search_row.latest_dataset_downloaded_at,
+                    hash=feed_search_row.latest_dataset_hash,
+                    service_date_range_start=feed_search_row.latest_dataset_service_date_range_start,
+                    service_date_range_end=feed_search_row.latest_dataset_service_date_range_end,
+                    agency_timezone=feed_search_row.latest_dataset_agency_timezone,
+                    validation_report=LatestDatasetValidationReport(
+                        total_error=feed_search_row.latest_total_error,
+                        total_warning=feed_search_row.latest_total_warning,
+                        total_info=feed_search_row.latest_total_info,
+                        unique_error_count=feed_search_row.latest_unique_error_count,
+                        unique_warning_count=feed_search_row.latest_unique_warning_count,
+                        unique_info_count=feed_search_row.latest_unique_info_count,
+                        features=(
+                            sorted([feature for feature in feed_search_row.latest_dataset_features])
+                            if feed_search_row.latest_dataset_features
+                            else []
+                        ),
+                    ),
+                )
+                if feed_search_row.latest_dataset_id
+                else None
+            ),
         )
 
     @classmethod
@@ -78,9 +82,9 @@ def from_orm_gbfs(cls, feed_search_row):
             feed_contact_email=feed_search_row.feed_contact_email,
             source_info=SourceInfo(
                 producer_url=feed_search_row.producer_url,
-                authentication_type=int(feed_search_row.authentication_type)
-                if feed_search_row.authentication_type
-                else None,
+                authentication_type=(
+                    int(feed_search_row.authentication_type) if feed_search_row.authentication_type else None
+                ),
                 authentication_info_url=feed_search_row.authentication_info_url,
                 api_key_parameter_name=feed_search_row.api_key_parameter_name,
                 license_url=feed_search_row.license_url,
@@ -104,9 +108,9 @@ def from_orm_gtfs_rt(cls, feed_search_row):
             feed_contact_email=feed_search_row.feed_contact_email,
             source_info=SourceInfo(
                 producer_url=feed_search_row.producer_url,
-                authentication_type=int(feed_search_row.authentication_type)
-                if feed_search_row.authentication_type
-                else None,
+                authentication_type=(
+                    int(feed_search_row.authentication_type) if feed_search_row.authentication_type else None
+                ),
                 authentication_info_url=feed_search_row.authentication_info_url,
                 api_key_parameter_name=feed_search_row.api_key_parameter_name,
                 license_url=feed_search_row.license_url,
@@ -125,9 +129,9 @@ def resolve_locations(cls, locations):
         return [
             {
                 **location,
-                "country": location.get("country")
-                if location.get("country")
-                else cls.resolve_country_by_code(location),
+                "country": (
+                    location.get("country") if location.get("country") else cls.resolve_country_by_code(location)
+                ),
             }
             for location in locations
         ]

api/src/feeds/impl/search_api_impl.py

@@ -31,7 +31,7 @@ def get_parsed_search_tsquery(search_query: str) -> str:
         return func.plainto_tsquery("english", unaccent(parsed_query))
 
     @staticmethod
-    def add_search_query_filters(query, search_query, data_type, feed_id, status, is_official) -> Query:
+    def add_search_query_filters(query, search_query, data_type, feed_id, status, is_official, features) -> Query:
         """
         Add filters to the search query.
         Filter values are trimmed and converted to lowercase.
@@ -62,6 +62,13 @@ def add_search_query_filters(query, search_query, data_type, feed_id, status, is
             query = query.filter(
                 t_feedsearch.c.document.op("@@")(SearchApiImpl.get_parsed_search_tsquery(search_query))
             )
+        # Add feature filter with OR logic
+        if features:
+            features_list = [s.strip() for s in features[0].split(",") if s]
+            if features_list:
+                query = query.filter(
+                    t_feedsearch.c.latest_dataset_features.op("&&")(features_list)
+                )  # overlap: Test if elements are a superset of the elements of the argument array expression.
         return query
 
     @staticmethod
@@ -71,16 +78,24 @@ def create_count_search_query(
         data_type: str,
         is_official: bool,
         search_query: str,
+        features,
     ) -> Query:
         """
         Create a search query for the database.
         """
         query = select(func.count(t_feedsearch.c.feed_id))
-        return SearchApiImpl.add_search_query_filters(query, search_query, data_type, feed_id, status, is_official)
+        return SearchApiImpl.add_search_query_filters(
+            query, search_query, data_type, feed_id, status, is_official, features
+        )
 
     @staticmethod
     def create_search_query(
-        status: List[str], feed_id: str, data_type: str, is_official: bool, search_query: str
+        status: List[str],
+        feed_id: str,
+        data_type: str,
+        is_official: bool,
+        search_query: str,
+        features: List[str],
     ) -> Query:
         """
         Create a search query for the database.
@@ -93,7 +108,9 @@ def create_search_query(
             rank_expression,
             *feed_search_columns,
         )
-        query = SearchApiImpl.add_search_query_filters(query, search_query, data_type, feed_id, status, is_official)
+        query = SearchApiImpl.add_search_query_filters(
+            query, search_query, data_type, feed_id, status, is_official, features
+        )
         return query.order_by(rank_expression.desc())
 
     @with_db_session
@@ -106,10 +123,11 @@ def search_feeds(
         data_type: str,
         is_official: bool,
         search_query: str,
+        features: List[str],
         db_session: "Session",
     ) -> SearchFeeds200Response:
         """Search feeds using full-text search on feed, location and provider's information."""
-        query = self.create_search_query(status, feed_id, data_type, is_official, search_query)
+        query = self.create_search_query(status, feed_id, data_type, is_official, search_query, features)
         feed_rows = Database().select(
             session=db_session,
             query=query,
@@ -118,7 +136,7 @@ def search_feeds(
         )
         feed_total_count = Database().select(
             session=db_session,
-            query=self.create_count_search_query(status, feed_id, data_type, is_official, search_query),
+            query=self.create_count_search_query(status, feed_id, data_type, is_official, search_query, features),
         )
         if feed_rows is None or feed_total_count is None:
             return SearchFeeds200Response(

api/tests/integration/test_search_api.py

@@ -424,3 +424,56 @@ def test_search_filter_by_official_status(client: TestClient, values: dict):
     assert (
         response_body.total == expected_count
     ), f"There should be {expected_count} feeds for official={values['official']}"
+
+
+@pytest.mark.parametrize(
+    "values",
+    [
+        {"feature": "", "expected_count": 16},
+        {"feature": "Bike Allowed", "expected_count": 2},
+        {"feature": "Stops Wheelchair Accessibility", "expected_count": 0},
+    ],
+    ids=[
+        "All",
+        "Bike Allowed",
+        "Stops Wheelchair Accessibility",
+    ],
+)
+def test_search_filter_by_feature(client: TestClient, values: dict):
+    """
+    Retrieve feeds that contain specific features.
+    """
+    params = None
+    if values["feature"] is not None:
+        params = [
+            ("feature", values["feature"]),
+        ]
+
+    headers = {
+        "Authentication": "special-key",
+    }
+    response = client.request(
+        "GET",
+        "/v1/search",
+        headers=headers,
+        params=params,
+    )
+    # Assert the status code of the HTTP response
+    assert response.status_code == 200
+
+    # Parse the response body into a Python object
+    response_body = SearchFeeds200Response.model_validate(response.json())
+    expected_count = values["expected_count"]
+    assert (
+        response_body.total == expected_count
+    ), f"There should be {expected_count} feeds with feature={values['feature']}"
+
+    # Verify all returned feeds have at least one of the requested features
+    if values["feature"] and expected_count > 0:
+        requested_features = set(values["feature"].split(","))
+        for result in response_body.results:
+            features = result.latest_dataset.validation_report.features
+            # Check that at least one of the feed's features is in the requested features
+            assert requested_features.intersection(features), (
+                f"Feed {result.id} with features {features} does not match " f"requested features {requested_features}"
+            )

docs/DatabaseCatalogAPI.yaml

@@ -335,6 +335,7 @@ paths:
         - $ref: "#/components/parameters/data_type_query_param"
         - $ref: "#/components/parameters/is_official_query_param"
         - $ref: "#/components/parameters/search_text_query_param"
+        - $ref: "#/components/parameters/feature"
       security:
         - Authentication: []
       responses:
@@ -1120,6 +1121,17 @@ components:
             - inactive
             - development
             - future
+    feature:
+      name: feature
+      in: query
+      description: Filter feeds by their GTFS features. [GTFS features definitions defined here](https://gtfs.org/getting-started/features/overview)
+      required: false
+      style: form
+      explode: false
+      schema:
+        type: array
+        items:
+          type: string
     provider:
       name: provider
       in: query

web-app/src/app/screens/Feeds/index.tsx

@@ -121,10 +121,13 @@ export default function Feed(): React.ReactElement {
               selectedFeedTypes,
               config.enableGbfsInSearchPage,
             ),
-            is_official: isOfficialFeedSearch || undefined,
+            is_official: isOfficialTagFilterEnabled
+              ? isOfficialFeedSearch || undefined
+              : undefined,
             // Fixed status values for now, until a status filter is implemented
             // Filtering out deprecated feeds
             status: ['active', 'inactive', 'development', 'future'],
+            feature: areFeatureFiltersEnabled ? selectedFeatures : undefined,
           },
         },
       }),

web-app/src/app/services/feeds/types.ts

@@ -643,6 +643,8 @@ export interface components {
     statuses?: Array<
       'active' | 'deprecated' | 'inactive' | 'development' | 'future'
     >;
+    /** @description Filter feeds by their GTFS features. [GTFS features definitions defined here](https://gtfs.org/getting-started/features/overview) */
+    feature?: string[];
     /** @description List only feeds with the specified value. Can be a partial match. Case insensitive. */
     provider?: string;
     /** @description List only feeds with the specified value. Can be a partial match. Case insensitive. */
@@ -991,6 +993,7 @@ export interface operations {
         data_type?: components['parameters']['data_type_query_param'];
         is_official?: components['parameters']['is_official_query_param'];
         search_query?: components['parameters']['search_text_query_param'];
+        feature?: components['parameters']['feature'];
       };
     };
     responses: {