chore: improve logs on reverse geolocation functions (#1217)

Author: davidgamez

functions-python/reverse_geolocation/src/main.py

@@ -1,6 +1,12 @@
+import logging
+
 import flask
 import functions_framework
 from cloudevents.http import CloudEvent
+from shared.helpers.logger import init_logger
+
+
+init_logger()
 
 
 @functions_framework.http
@@ -11,7 +17,9 @@ def reverse_geolocation_processor(request: flask.Request):
     """
     from reverse_geolocation_processor import reverse_geolocation_process
 
-    return reverse_geolocation_process(request)
+    result = reverse_geolocation_process(request)
+    logging.info(result)
+    return result
 
 
 @functions_framework.http
@@ -22,7 +30,9 @@ def reverse_geolocation_batch(request: flask.Request):
     """
     from reverse_geolocation_batch import reverse_geolocation_batch
 
-    return reverse_geolocation_batch(request)
+    result = reverse_geolocation_batch(request)
+    logging.info(result)
+    return result
 
 
 @functions_framework.cloud_event
@@ -33,7 +43,9 @@ def reverse_geolocation(request: CloudEvent):
     """
     from reverse_geolocation import reverse_geolocation_storage_trigger
 
-    return reverse_geolocation_storage_trigger(request)
+    result = reverse_geolocation_storage_trigger(request)
+    logging.info(result)
+    return result
 
 
 @functions_framework.cloud_event
@@ -44,4 +56,6 @@ def reverse_geolocation_pubsub(request: CloudEvent):
     """
     from reverse_geolocation import reverse_geolocation_pubsub
 
-    return reverse_geolocation_pubsub(request)
+    result = reverse_geolocation_pubsub(request)
+    logging.info(result)
+    return result

functions-python/reverse_geolocation/src/parse_request.py

@@ -17,7 +17,7 @@ def parse_request_parameters(
     """
     logging.info("Parsing request parameters.")
     request_json = request.get_json(silent=True)
-    logging.info(f"Request JSON: {request_json}")
+    logging.info("Request JSON: %s", request_json)
 
     if (
         not request_json
@@ -34,7 +34,7 @@ def parse_request_parameters(
         )
 
     data_type = request_json.get("data_type", "gtfs")
-    logging.info(f"Data type: {data_type}")
+    logging.info("Data type: %s", data_type)
     if data_type == "gtfs":
         df, stable_id, dataset_id, url = parse_request_parameters_gtfs(request_json)
     elif data_type == "gbfs":

functions-python/reverse_geolocation/src/reverse_geolocation.py

@@ -8,16 +8,18 @@
 from google.cloud import storage
 from google.cloud import tasks_v2
 
-from shared.helpers.logger import Logger
+from shared.helpers.logger import init_logger
 from shared.helpers.parser import jsonify_pubsub
 from shared.helpers.utils import create_http_task
 
 
+init_logger()
+
+
 def init(request: CloudEvent) -> None:
     """
     Initializer function.
     """
-    Logger.init_logger()
     logging.info("Processing reverse geolocation request.")
     logging.info("Request: %s", request)
 
@@ -58,7 +60,7 @@ def reverse_geolocation_pubsub(request: CloudEvent) -> None:
         url = message_json["url"]
         reverse_geolocation(stable_id, dataset_id, url)
     except Exception as e:
-        logging.error(f"Execution error: {e}")
+        logging.error("Execution error: %s", e)
 
 
 def reverse_geolocation_storage_trigger(request: CloudEvent) -> None:
@@ -70,7 +72,7 @@ def reverse_geolocation_storage_trigger(request: CloudEvent) -> None:
         stable_id, dataset_id, url = parse_resource_data(request.data)
         reverse_geolocation(stable_id, dataset_id, url)
     except Exception as e:
-        logging.error(f"Execution error: {e}")
+        logging.error("Execution error: %s", e)
 
 
 def reverse_geolocation(stable_id: str, dataset_id: str, url: str) -> None:
@@ -79,7 +81,10 @@ def reverse_geolocation(stable_id: str, dataset_id: str, url: str) -> None:
     """
     try:
         logging.info(
-            f"Stable ID: {stable_id} - Dataset Stable ID: {dataset_id} - URL: {url}"
+            "Stable ID: %s - Dataset Stable ID: %s - URL: %s",
+            stable_id,
+            dataset_id,
+            url,
         )
 
         # TODO: This logic should be moved to a separate service
@@ -90,12 +95,12 @@ def reverse_geolocation(stable_id: str, dataset_id: str, url: str) -> None:
         blob = bucket.blob(f"{stable_id}/{dataset_id}/stops.txt")
         blob.upload_from_filename("stops.txt")
         blob.make_public()
-        logging.info(f"Uploaded stops.txt to {blob.public_url}")
+        logging.info("Uploaded stops.txt to %s", blob.public_url)
 
         create_http_processor_task(stable_id, dataset_id, blob.public_url)
-        logging.info(f"Reverse geolocation task created for feed {stable_id}.")
+        logging.info("Reverse geolocation task created for feed %s.", stable_id)
     except Exception as e:
-        logging.error(f"Error creating task: {e}")
+        logging.error("Error creating task: %s", e)
 
 
 def create_http_processor_task(

functions-python/reverse_geolocation/src/reverse_geolocation_batch.py

@@ -9,10 +9,11 @@
 
 from shared.database_gen.sqlacodegen_models import Gtfsfeed, Gtfsdataset, Location
 from shared.database.database import with_db_session
-from shared.helpers.logger import Logger
+from shared.helpers.logger import init_logger
 from shared.helpers.pub_sub import publish_messages
 
-logging.basicConfig(level=logging.INFO)
+
+init_logger()
 
 
 @with_db_session
@@ -30,7 +31,7 @@ def get_feeds_data(
         .filter(Gtfsdataset.latest)
     )
     if country_codes:
-        logging.info(f"Getting feeds for country codes: {country_codes}")
+        logging.info("Getting feeds for country codes: %s", country_codes)
         query = query.filter(
             Gtfsfeed.locations.any(Location.country_code.in_(country_codes))
         )
@@ -42,7 +43,7 @@ def get_feeds_data(
         query = query.filter(~Gtfsfeed.feedlocationgrouppoints.any())
 
     results = query.populate_existing().all()
-    logging.info(f"Found {len(results)} feeds.")
+    logging.info("Found %s feeds.", len(results))
 
     data = [
         {
@@ -74,18 +75,17 @@ def parse_request_parameters(request: flask.Request) -> Tuple[List[str], bool]:
 def reverse_geolocation_batch(request: flask.Request) -> Tuple[str, int]:
     """Batch function to trigger reverse geolocation for feeds."""
     try:
-        Logger.init_logger()
         country_codes, include_only_unprocessed = parse_request_parameters(request)
         feeds_data = get_feeds_data(country_codes, include_only_unprocessed)
-        logging.info(f"Valid feeds with latest dataset: {len(feeds_data)}")
+        logging.info("Valid feeds with latest dataset: %s", len(feeds_data))
 
         pubsub_topic_name = os.getenv("PUBSUB_TOPIC_NAME", None)
         project_id = os.getenv("PROJECT_ID")
 
-        logging.info(f"Publishing to topic: {pubsub_topic_name}")
+        logging.info("Publishing to topic: %s", pubsub_topic_name)
         publish_messages(feeds_data, project_id, pubsub_topic_name)
 
         return f"Batch function triggered for {len(feeds_data)} feeds.", 200
     except Exception as e:
-        logging.error(f"Execution error: {e}")
+        logging.error("Execution error: %s", e)
         return "Error while fetching feeds.", 500