ADD: Initialization check

Author: lukovicaleksa

main.py

@@ -5,7 +5,7 @@
 
 from database.collections import mongodb_connection
 from routes.movies import movies_router
-from utils import initialize_db_movies_collection_from_dataset
+from utils import is_db_movies_collection_initialized, initialize_db_movies_collection_from_dataset
 
 
 logger = logging.getLogger("uvicorn")
@@ -17,11 +17,14 @@ def app_startup_event_handler():
     """
     logger.info('Initializing Movies collection from dataset ...')
 
-    start_time = time.time()
-    movies_inserted = initialize_db_movies_collection_from_dataset()
+    if is_db_movies_collection_initialized():
+        logger.info('Collection already initialized')
+    else:
+        start_time = time.time()
+        movies_inserted = initialize_db_movies_collection_from_dataset()
 
-    logger.info(f'Successfully initialized the collection with {movies_inserted} movies, '
-                f'elapsed time: {time.time() - start_time:.2f} seconds')
+        logger.info(f'Successfully initialized the collection with {movies_inserted} movies, '
+                    f'elapsed time: {time.time() - start_time:.2f} seconds')
 
 
 def app_shutdown_event_handler():

utils.py

@@ -8,6 +8,16 @@
 from language_model import embedding_model
 
 
+def is_db_movies_collection_initialized() -> bool:
+    """
+    Checks if the Movies collection is initialized.\n
+    For the sake of simplicity, collection is considered as initialized if it contains documents.
+
+    :return: True/False
+    """
+    return db_movies_collection.count_documents({}) > 0
+
+
 def initialize_db_movies_collection_from_dataset() -> int:
     """
     Initialize Movies collection using TMDB 5000 Movie Dataset from Kaggle (https://www.kaggle.com/datasets/tmdb/tmdb-movie-metadata)\n