load faiss from azure blb storage

agent/test.ipynb

@@ -191,6 +191,36 @@
     "vector_info = str(chunk_size) + \"-\" + tokenizer_name.replace(\"/\",\"-\") + \"-\" + datetime.now().date().isoformat()\n",
     "KNOWLEDGE_VECTOR_DATABASE.save_local(\"../data/FAISS/{vector_info}\")\n"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "406da007",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "from azure.storage.blob import BlobServiceClient\n",
+    "from langchain_community.vectorstores import FAISS\n",
+    "\n",
+    "\n",
+    "def load_from_azure(container_name, local_dir=\"./index\"):\n",
+    "    connection_string = os.environ[\"AZURE_CONN_STR\"]\n",
+    "    blob_service_client = BlobServiceClient.from_connection_string(connection_string)\n",
+    "    container_client = blob_service_client.get_container_client(container_name)\n",
+    "\n",
+    "    os.makedirs(local_dir, exist_ok=True)\n",
+    "    \n",
+    "    # Download all files in the container (index.faiss and index.pkl)\n",
+    "    blobs = container_client.list_blobs()\n",
+    "    for blob in blobs:\n",
+    "        download_file_path = os.path.join(local_dir, blob.name)\n",
+    "        with open(download_file_path, \"wb\") as file:\n",
+    "            file.write(container_client.download_blob(blob).readall())\n",
+    "\n",
+    "# Download files from Azure\n",
+    "load_from_azure(\"blobcontaineravatarbot\")"
+   ]
   }
  ],
  "metadata": {

app.py

@@ -32,10 +32,10 @@ def load_from_azure(container_name, local_dir="./index"):
         with open(download_file_path, "wb") as file:
             file.write(container_client.download_blob(blob).readall())
 
-# 1. Download files from Azure
+# Download files from Azure
 load_from_azure("blobcontaineravatarbot")
 
-# 2. Load into FAISS
+# Load into FAISS
 embeddings="intfloat/e5-base-v2"
 vectorstore = FAISS.load_local("./index", embeddings, allow_dangerous_deserialization=True)
 
@@ -48,7 +48,7 @@ from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
 from langchain.chat_models import init_chat_model
 
 
-def predict(message, history,request: gr.Request):
+def predict(message, history, request: gr.Request):
 
     # Get client IP and check rate limit
     client_ip = request.client.host