Update app.py

app.py

@@ -293,26 +293,37 @@ def initialize_chat_from_existing():
     return False
 
 def initialize_chat_from_collection():
-    """Initialize chat system from existing collection or documents."""
+    """Initialize chat system with vector store reuse."""
     try:
-        if not st.session_state.chat_ready:
-            documents = None
-            if st.session_state.get('current_collection'):
-                documents = get_collection_documents(st.session_state.db_conn, 
-                                                    st.session_state.current_collection['id'])
-            else:
-                documents = get_all_documents(st.session_state.db_conn)
+        documents = None
+        if st.session_state.get('current_collection'):
+            documents = get_collection_documents(st.session_state.db_conn, 
+                                              st.session_state.current_collection['id'])
+        else:
+            documents = get_all_documents(st.session_state.db_conn)
 
-            if documents:
+        if documents:
+            document_ids = [doc['id'] for doc in documents]
+            
+            # Check for existing vector store
+            vector_store = get_existing_vector_store(document_ids)
+            
+            if vector_store:
+                # Reuse existing vector store
+                st.session_state.vector_store = vector_store
+                st.session_state.qa_system = initialize_qa_system(vector_store)
+                return True
+            
+            # If no existing vector store, create new one
+            with st.spinner("Initializing chat system..."):
                 embeddings = get_embeddings_model()
                 text_splitter = RecursiveCharacterTextSplitter(
-                    chunk_size=700,
-                    chunk_overlap=100,
+                    chunk_size=500,
+                    chunk_overlap=50,
                     length_function=len,
                     separators=["\n\n", "\n", " ", ""]
                 )
                 
-                # Process documents into chunks
                 chunks = []
                 for doc in documents:
                     doc_chunks = text_splitter.split_text(doc['content'])
@@ -322,7 +333,6 @@ def initialize_chat_from_collection():
                             'metadata': {'source': doc['name'], 'document_id': doc['id']}
                         })
                 
-                # Initialize vector store
                 vector_store = FAISS.from_texts(
                     [chunk['content'] for chunk in chunks],
                     embeddings,
@@ -331,9 +341,10 @@ def initialize_chat_from_collection():
                 
                 st.session_state.vector_store = vector_store
                 st.session_state.qa_system = initialize_qa_system(vector_store)
-                st.session_state.chat_ready = True
                 return True
-        return st.session_state.chat_ready
+                
+        return False
+        
     except Exception as e:
         st.error(f"Error initializing chat: {e}")
         return False