global client

load_card_data.py

@@ -10,7 +10,7 @@ from chromadb.utils import embedding_functions
 from dotenv import load_dotenv
 from huggingface_hub import InferenceClient
 from tqdm.contrib.concurrent import thread_map
-from utils import get_chroma_client, get_collection
+from utils import get_collection, get_chroma_client
 
 os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
 # Set up logging
@@ -163,8 +163,8 @@ def get_inference_client():
 
 def refresh_card_data(min_len: int = 250, min_likes: Optional[int] = None):
     logger.info(f"Starting data refresh with min_len={min_len}, min_likes={min_likes}")
-    chroma_client = get_chroma_client()
     embedding_function = card_embedding_function()
+    chroma_client = get_chroma_client()
     collection = get_collection(chroma_client, embedding_function, COLLECTION_NAME)
     most_recent = get_last_modified_in_collection(collection)
 

main.py

@@ -18,7 +18,7 @@ from starlette.status import (
 
 from load_card_data import card_embedding_function, refresh_card_data
 from load_viewer_data import refresh_viewer_data
-from utils import get_save_path, get_collection
+from utils import get_save_path, get_collection, get_chroma_client
 
 # Set up logging
 logging.basicConfig(
@@ -30,9 +30,7 @@ logger = logging.getLogger(__name__)
 cache.setup("mem://?check_interval=10&size=1000")
 
 # Initialize Chroma client
-SAVE_PATH = get_save_path()
-client = chromadb.PersistentClient(path=SAVE_PATH)
-
+client = get_chroma_client()
 
 async_client = AsyncClient(
     follow_redirects=True,

utils.py

@@ -26,6 +26,8 @@ logger = logging.getLogger(__name__)
 
 load_dotenv()
 
+CHROMA_CLIENT = None
+
 
 def get_save_path() -> Literal["chroma/"] | Literal["/data/chroma/"]:
     path = "chroma/" if platform.system() == "Darwin" else "/data/chroma/"
@@ -34,13 +36,14 @@ def get_save_path() -> Literal["chroma/"] | Literal["/data/chroma/"]:
 
 
 def get_chroma_client():
-    logger.info("Initializing Chroma client")
-    SAVE_PATH = get_save_path()
-
-    return chromadb.PersistentClient(
-        path=SAVE_PATH,
-        settings=Settings(anonymized_telemetry=False, is_persistent=True),
-    )
+    global CHROMA_CLIENT
+    if CHROMA_CLIENT is None:
+        SAVE_PATH = get_save_path()
+        CHROMA_CLIENT = chromadb.PersistentClient(
+            path=SAVE_PATH,
+            settings=Settings(anonymized_telemetry=False, is_persistent=True),
+        )
+    return CHROMA_CLIENT
 
 
 def get_collection(chroma_client, embedding_function, collection_name):