feat: 1325 add global and per feed configurationintegrate feeds download

api/src/shared/common/config_reader.py

@@ -0,0 +1,48 @@
+from typing import Optional, Any
+from sqlalchemy.orm import Session
+from shared.database.database import with_db_session
+from shared.database_gen.sqlacodegen_models import ConfigKey, ConfigValueFeed
+
+
+@with_db_session
+def get_config_value(
+    namespace: str,
+    key: str,
+    feed_id: Optional[str] = None,
+    db_session: Session = None,
+) -> Optional[Any]:
+    """
+    Retrieves a configuration value from the database using the provided session.
+
+    It first looks for a feed-specific override. If a feed_stable_id is provided and a
+    value is found, it returns that value.
+
+    If no feed-specific value is found or no feed_stable_id is provided, it looks for
+    the global default value in the `config_key` table.
+
+    :param namespace: The namespace of the configuration key.
+    :param key: The configuration key.
+    :param feed_stable_id: The optional feed_stable_id for a specific override.
+    :param db_session: The SQLAlchemy session, injected by the `with_db_session` decorator.
+    :return: The configuration value, or None if not found.
+    """
+    # 1. Try to get feed-specific value if feed_id is provided
+    if feed_id:
+        feed_config = (
+            db_session.query(ConfigValueFeed.value)
+            .filter(
+                ConfigValueFeed.feed_id == feed_id,
+                ConfigValueFeed.namespace == namespace,
+                ConfigValueFeed.key == key,
+            )
+            .first()
+        )
+        if feed_config:
+            return feed_config.value
+
+    # 2. If not found or no feed_id, get the default value
+    default_config = (
+        db_session.query(ConfigKey.default_value).filter(ConfigKey.namespace == namespace, ConfigKey.key == key).first()
+    )
+
+    return default_config.default_value if default_config else None

functions-python/batch_process_dataset/src/main.py

@@ -122,18 +122,20 @@ def create_dataset_stable_id(feed_stable_id, timestamp):
         """
         return f"{feed_stable_id}-{timestamp}"
 
-    def download_content(self, temporary_file_path):
+    def download_content(self, temporary_file_path, feed_id):
         """
         Downloads the content of a URL and return the hash of the file
         """
         file_hash = download_and_get_hash(
             self.producer_url,
-            temporary_file_path,
+            file_path=temporary_file_path,
+            feed_id=feed_id,
             authentication_type=self.authentication_type,
             api_key_parameter_name=self.api_key_parameter_name,
             credentials=self.feed_credentials,
             logger=self.logger,
         )
+        self.logger.info(f"hash is: {file_hash}")
         is_zip = zipfile.is_zipfile(temporary_file_path)
         return file_hash, is_zip
 
@@ -193,7 +195,7 @@ def upload_files_to_storage(
                 )
         return blob, extracted_files
 
-    def upload_dataset(self, public=True) -> DatasetFile or None:
+    def upload_dataset(self, feed_id, public=True) -> DatasetFile or None:
         """
         Uploads a dataset to a GCP bucket as <feed_stable_id>/latest.zip and
         <feed_stable_id>/<feed_stable_id>-<upload_datetime>.zip
@@ -204,7 +206,7 @@ def upload_dataset(self, public=True) -> DatasetFile or None:
         try:
             self.logger.info("Accessing URL %s", self.producer_url)
             temp_file_path = self.generate_temp_filename()
-            file_sha256_hash, is_zip = self.download_content(temp_file_path)
+            file_sha256_hash, is_zip = self.download_content(temp_file_path, feed_id)
             if not is_zip:
                 self.logger.error(
                     f"[{self.feed_stable_id}] The downloaded file from {self.producer_url} is not a valid ZIP file."
@@ -417,12 +419,14 @@ def _get_unzipped_size(dataset_file):
         )
 
     @with_db_session
-    def process_from_producer_url(self, db_session) -> Optional[DatasetFile]:
+    def process_from_producer_url(
+        self, feed_id, db_session: Session
+    ) -> Optional[DatasetFile]:
         """
         Process the dataset and store new version in GCP bucket if any changes are detected
         :return: the DatasetFile object created
         """
-        dataset_file = self.upload_dataset()
+        dataset_file = self.upload_dataset(feed_id)
 
         if dataset_file is None:
             self.logger.info(f"[{self.feed_stable_id}] No database update required.")
@@ -543,7 +547,7 @@ def process_dataset(cloud_event: CloudEvent):
         if json_payload.get("use_bucket_latest", False):
             dataset_file = processor.process_from_bucket()
         else:
-            dataset_file = processor.process_from_producer_url()
+            dataset_file = processor.process_from_producer_url(json_payload["feed_id"])
     except Exception as e:
         # This makes sure the logger is initialized
         logger = get_logger("process_dataset", stable_id if stable_id else "UNKNOWN")

functions-python/batch_process_dataset/src/scripts/download_verifier.py

@@ -33,7 +33,8 @@ def verify_download_content(producer_url: str):
     )
     tempfile = processor.generate_temp_filename()
     logging.info(f"Temp filename: {tempfile}")
-    file_hash, is_zip = processor.download_content(tempfile)
+    file_hash, is_zip = processor.download_content(tempfile, "feed_id")
+    logging.info(f"Downloaded file from {producer_url} is a valid ZIP file: {is_zip}")
     logging.info(f"File hash: {file_hash}")
 
 
@@ -48,7 +49,7 @@ def verify_upload_dataset(producer_url: str):
     """
     processor = DatasetProcessor(
         producer_url=producer_url,
-        feed_id="feed_id",
+        feed_id="feed_id_2126",
         feed_stable_id="feed_stable_id",
         execution_id=None,
         latest_hash="123",
@@ -59,7 +60,7 @@ def verify_upload_dataset(producer_url: str):
     )
     tempfile = processor.generate_temp_filename()
     logging.info(f"Temp filename: {tempfile}")
-    dataset_file = processor.upload_dataset(public=False)
+    dataset_file = processor.upload_dataset("feed_id_2126", False)
     logging.info(f"Dataset File: {dataset_file}")
 
 
@@ -72,6 +73,7 @@ def verify_upload_dataset(producer_url: str):
         # create working dir if not exists
         if not os.path.exists(os.environ["WORKING_DIR"]):
             os.makedirs(os.environ["WORKING_DIR"])
+
         server = create_server(
             host=HOST, port=PORT, in_memory=False, default_bucket=BUCKET_NAME
         )
@@ -80,7 +82,6 @@ def verify_upload_dataset(producer_url: str):
         verify_download_content(producer_url=PRODUCER_URL)
         logging.info("Download content verification completed successfully.")
         verify_upload_dataset(producer_url=PRODUCER_URL)
-        verify_upload_dataset(producer_url=PRODUCER_URL)
     except Exception as e:
         logging.error(f"Error verifying download content: {e}")
     finally:

functions-python/batch_process_dataset/tests/test_batch_process_dataset_main.py

@@ -73,7 +73,7 @@ def test_upload_dataset_diff_hash(
             test_hosted_public_url,
         )
         with patch.object(processor, "date", "mocked_timestamp"):
-            result = processor.upload_dataset()
+            result = processor.upload_dataset("feed_id")
 
         self.assertIsNotNone(result)
         mock_download_url_content.assert_called_once()
@@ -111,7 +111,7 @@ def test_upload_dataset_same_hash(
             test_hosted_public_url,
         )
 
-        result = processor.upload_dataset()
+        result = processor.upload_dataset("feed_id")
 
         self.assertIsNone(result)
         upload_files_to_storage.blob.assert_not_called()
@@ -143,7 +143,7 @@ def test_upload_dataset_not_zip(
             test_hosted_public_url,
         )
 
-        result = processor.upload_dataset()
+        result = processor.upload_dataset("feed_id")
 
         self.assertIsNone(result)
         upload_files_to_storage.blob.assert_not_called()
@@ -176,7 +176,7 @@ def test_upload_dataset_download_exception(
         )
 
         with self.assertRaises(Exception):
-            processor.upload_dataset()
+            processor.upload_dataset("feed_id")
 
     def test_upload_files_to_storage(self):
         bucket_name = "test-bucket"
@@ -256,7 +256,7 @@ def test_process(self, db_session):
         )
         db_url = os.getenv("TEST_FEEDS_DATABASE_URL", default=default_db_url)
         os.environ["FEEDS_DATABASE_URL"] = db_url
-        result = processor.process_from_producer_url()
+        result = processor.process_from_producer_url(feed_id)
 
         self.assertIsNotNone(result)
         self.assertEqual(result.file_sha256_hash, new_hash)
@@ -357,7 +357,7 @@ def test_process_no_change(self):
 
         processor.upload_dataset = MagicMock(return_value=None)
         processor.create_dataset_entities = MagicMock()
-        result = processor.process_from_producer_url()
+        result = processor.process_from_producer_url(feed_id)
 
         self.assertIsNone(result)
         processor.create_dataset_entities.assert_not_called()

functions-python/helpers/tests/test_helpers.py

@@ -45,7 +45,8 @@ def test_create_bucket_already_exists(self, mock_storage_client):
         )
         mock_storage_client.return_value.create_bucket.assert_not_called()
 
-    def test_download_and_get_hash(self):
+    @patch("shared.common.config_reader.get_config_value", return_value=None)
+    def test_download_and_get_hash(self, mock_get_config):
         mock_binary_data = b"file content data"
         expected_hash = hashlib.sha256(mock_binary_data).hexdigest()
         file_path = "file_path"
@@ -65,7 +66,8 @@ def test_download_and_get_hash(self):
             if os.path.exists(file_path):
                 os.remove(file_path)
 
-    def test_download_and_get_hash_auth_type_header(self):
+    @patch("shared.common.config_reader.get_config_value", return_value=None)
+    def test_download_and_get_hash_auth_type_header(self, mock_get_config):
         """
         Test the download_and_get_hash function for authentication type 2 (headers).
         This test verifies that the download_and_get_hash function correctly handles authentication type 2,
@@ -88,7 +90,11 @@ def test_download_and_get_hash_auth_type_header(self):
             "urllib3.PoolManager.request", return_value=mock_response
         ) as mock_request:
             result_hash = download_and_get_hash(
-                url, file_path, "sha256", 8192, 2, api_key_parameter_name, credentials
+                url=url,
+                file_path=file_path,
+                authentication_type=2,
+                api_key_parameter_name=api_key_parameter_name,
+                credentials=credentials,
             )
 
             self.assertEqual(
@@ -113,7 +119,8 @@ def test_download_and_get_hash_auth_type_header(self):
             if os.path.exists(file_path):
                 os.remove(file_path)
 
-    def test_download_and_get_hash_auth_type_api_key(self):
+    @patch("shared.common.config_reader.get_config_value", return_value=None)
+    def test_download_and_get_hash_auth_type_api_key(self, mock_get_config):
         """
         Test the download_and_get_hash function for authentication type 1 (API key).
         """
@@ -137,13 +144,11 @@ def test_download_and_get_hash_auth_type_api_key(self):
 
         with patch("urllib3.PoolManager", return_value=mock_http):
             result_hash = download_and_get_hash(
-                base_url,
-                file_path,
-                "sha256",
-                8192,
-                1,
-                api_key_parameter_name,
-                credentials,
+                url=base_url,
+                file_path=file_path,
+                authentication_type=1,
+                api_key_parameter_name=api_key_parameter_name,
+                credentials=credentials,
             )
 
             self.assertEqual(
@@ -163,7 +168,8 @@ def test_download_and_get_hash_auth_type_api_key(self):
         if os.path.exists(file_path):
             os.remove(file_path)
 
-    def test_download_and_get_hash_exception(self):
+    @patch("shared.common.config_reader.get_config_value", return_value=None)
+    def test_download_and_get_hash_exception(self, mock_get_config):
         file_path = "test_file.txt"
         url = "https://test.com/"
 

functions-python/helpers/utils.py

@@ -126,12 +126,15 @@ def download_and_get_hash(
     file_path,
     hash_algorithm="sha256",
     chunk_size=8192,
+    feed_id=None,
     authentication_type=0,
     api_key_parameter_name=None,
     credentials=None,
     logger=None,
     trusted_certs=False,  # If True, disables SSL verification
 ):
+    from shared.common.config_reader import get_config_value
+
     """
     Downloads the content of a URL and stores it in a file and returns the hash of the file
     """
@@ -145,13 +148,18 @@ def download_and_get_hash(
         ctx.load_default_certs()
         ctx.options |= 0x4  # ssl.OP_LEGACY_SERVER_CONNECT
 
+        headers = get_config_value(
+            namespace="feed_download", key="http_headers", feed_id=feed_id
+        )
+        if headers is None:
+            headers = {
+                "User-Agent": "Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) "
+                "AppleWebKit/537.36 (KHTML, like Gecko) "
+                "Chrome/126.0.0.0 Mobile Safari/537.36",
+                "Referer": url,
+            }
+
         # authentication_type == 1 -> the credentials are passed in the url
-        headers = {
-            "User-Agent": "Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) "
-            "AppleWebKit/537.36 (KHTML, like Gecko) "
-            "Chrome/126.0.0.0 Mobile Safari/537.36",
-            "Referer": url,
-        }
         # Careful, some URLs may already contain a query string
         # (e.g. http://api.511.org/transit/datafeeds?operator_id=CE)
         if authentication_type == 1 and api_key_parameter_name and credentials:

liquibase/changelog.xml

@@ -69,4 +69,5 @@
     <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"/>
+    <include file="changes/feat_1325.sql" relativeToChangelogFile="true"/>
 </databaseChangeLog>

liquibase/changes/feat_1325.sql

@@ -0,0 +1,25 @@
+ALTER TABLE feed ADD CONSTRAINT feed_stable_id_unique UNIQUE (stable_id);
+-- 1. Catalog of allowed keys + optional global values
+CREATE TABLE config_key (
+  namespace     text        NOT NULL,              -- e.g. 'reverse_geolocation'
+  key           text        NOT NULL,              -- e.g. 'country_fallback'
+  description   text,
+  default_value jsonb,                             -- fallback if nothing else
+  updated_at    timestamptz NOT NULL DEFAULT now(),
+  PRIMARY KEY (namespace, key)
+);
+
+-- 2. Per-feed overrides
+CREATE TABLE config_value_feed (
+  feed_id       varchar(255) NOT NULL,
+  feed_stable_id       varchar(255) NOT NULL,
+  namespace     text         NOT NULL,
+  key           text         NOT NULL,
+  value         jsonb        NOT NULL,
+  updated_at    timestamptz  NOT NULL DEFAULT now(),
+  PRIMARY KEY (feed_id, namespace, key),
+  FOREIGN KEY (namespace, key) REFERENCES config_key(namespace, key) ON DELETE CASCADE
+);
+
+-- Helpful index
+CREATE INDEX config_value_feed_gin ON config_value_feed USING GIN (value);