fix(chroma): use get_or_create_collection to prevent race conditions

src/config.py

@@ -14,7 +14,7 @@ SUPPORTED_FORMATS = {".txt", ".md", ".markdown"}
 CORPUS_DIRECTORY = "synthetic_policies"
 
 # Vector Database Settings
-VECTOR_STORAGE_TYPE = os.getenv("VECTOR_STORAGE_TYPE", "postgres")  # "chroma" or "postgres"
+VECTOR_STORAGE_TYPE = os.getenv("VECTOR_STORAGE_TYPE", "chroma")  # "chroma" or "postgres"
 VECTOR_DB_PERSIST_PATH = "data/chroma_db"  # Used for ChromaDB
 DATABASE_URL = os.getenv("DATABASE_URL")  # Used for PostgreSQL
 COLLECTION_NAME = "policy_documents"

src/embedding/embedding_service.py

@@ -151,9 +151,11 @@ class EmbeddingService:
                             file_name,
                             e,
                         )
+                        # The key change: we now pass the file_name to the fallback as well
                         model = ORTModelForFeatureExtraction.from_pretrained(
                             self.model_name,
                             provider=provider,
+                            file_name=file_name,  # Added this line
                             session_options=so,
                         )
                         logging.info(

src/vector_store/vector_db.py

@@ -72,11 +72,7 @@ class VectorDatabase:
         log_memory_checkpoint("vector_db_after_client_init")
 
         # Get or create collection
-        try:
-            self.collection = self.client.get_collection(name=collection_name)
-        except ValueError:
-            # Collection doesn't exist, create it
-            self.collection = self.client.create_collection(name=collection_name)
+        self.collection = self.client.get_or_create_collection(name=collection_name)
 
         logging.info(f"Initialized VectorDatabase with collection " f"'{collection_name}' at '{persist_path}'")