Update agent.py

agent.py

@@ -429,31 +429,42 @@ embedding_model = HuggingFaceEmbeddings(
 # -----------------------------
 # Create FAISS index and save it
 # -----------------------------
-try:
-    # Make sure to pass the directory, not the file path
-    index_dir = "/home/wendy/my_hf_agent_course_projects/faiss_index"
-    vector_store = FAISS.load_local(index_dir, embedding_model, allow_dangerous_deserialization=True)
-except Exception as e:
-    print(f"Error loading index: {str(e)}")
-    # Fallback to rebuilding the index if loading fails
-    vector_store = FAISS.from_documents(docs, embedding_model)
+class ChatState(TypedDict):
+    messages: Annotated[
+        List[str], 
+        gr.State(render=False),
+        "Stores chat history"
+    ]
 
-# -----------------------------
-# Load FAISS index properly
-# -----------------------------
-# Ensure the directory path is correct and points to the index file directly
-index_path = "/home/wendy/my_hf_agent_course_projects/faiss_index"
-try:
-    loaded_store = FAISS.load_local(index_path, embedding_model, allow_dangerous_deserialization=True)
-    print("FAISS index loaded successfully")
-except Exception as e:
-    print(f"Error loading FAISS index: {str(e)}")
+def initialize_vector_store():
+    embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+    index_path = "/home/wendy/my_hf_agent_course_projects/faiss_index"
+    
+    if os.path.exists(os.path.join(index_path, "index.faiss")):
+        try:
+            return FAISS.load_local(
+                index_path,
+                embedding_model,
+                allow_dangerous_deserialization=True
+            )
+        except Exception as e:
+            print(f"Error loading index: {e}")
+    
+    # Fallback: Create new index
+    print("Building new vector store...")
+    docs = [...]  # Your document loading logic here
+    vector_store = FAISS.from_documents(docs, embedding_model)
+    vector_store.save_local(index_path)
+    return vector_store
 
+# Initialize at module level
+loaded_store = initialize_vector_store()
+retriever = loaded_store.as_retriever()
 
 # -----------------------------
 # Create LangChain Retriever Tool
 # -----------------------------
-retriever = loaded_store.as_retriever()
+#retriever = loaded_store.as_retriever()
 
 question_retriever_tool = create_retriever_tool(
     retriever=retriever,