Pull request #37: RED-5277 fix heartbeat issue

Merge in RR/cv-analysis from RED-5277-fix-heartbeat-issue to master

* commit '5bb9282da6aa1d75182c2172c601bed534099b0f':
  use python 3.8 in build
  update serve.py to work with new pyinfra version
  update reference to pyinfra

Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.10
+FROM python:3.8
 
 RUN python -m venv /app/venv
 ENV PATH="/app/venv/bin:$PATH"

incl/pyinfra

@@ -1 +1 @@
-Subproject commit 64d6a8cec62eeddf26bd71a9aabc28b40dcec901
+Subproject commit c97ae3d2c242dfc88a342955311dd488cb9a5f60

src/serve.py

@@ -22,27 +22,29 @@ def analysis_callback(queue_message: dict):
         "dossierId", "fileId", "targetFileExtension", "responseFileExtension", "operation"
     )(queue_message)
     bucket = PYINFRA_CONFIG.storage_bucket
-    logging.info(f"Processing {dossier_id=}/{file_id=}, {operation=}.")
+    logging.info("running operation %s file_id=%s and dossier_id=%s", operation, file_id, dossier_id)
 
     storage = get_storage(PYINFRA_CONFIG)
     object_name = f"{dossier_id}/{file_id}.{target_file_ext}"
 
     if storage.exists(bucket, object_name):
-        should_publish_result = True
-
         object_bytes = gzip.decompress(storage.get_object(bucket, object_name))
         analysis_fn = get_analysis_pipeline(operation, CV_CONFIG.table_parsing_skip_pages_without_images)
 
         results = analysis_fn(object_bytes)
+        logging.info("predictions ready for file_id=%s and dossier_id=%s", file_id, dossier_id)
+
         response = {**queue_message, "data": list(results)}
         response = gzip.compress(json.dumps(response).encode())
         response_name = f"{dossier_id}/{file_id}.{response_file_ext}"
 
+        logging.info("storing predictions for file_id=%s and dossier_id=%s", file_id, dossier_id)
         storage.put_object(bucket, response_name, response)
-    else:
-        should_publish_result = False
+        
+        return {"dossierId": dossier_id, "fileId": file_id}
 
-    return should_publish_result, {"dossierId": dossier_id, "fileId": file_id}
+    else:
+        return None
 
 
 if __name__ == "__main__":