added text streamer

backend.py

@@ -41,15 +41,24 @@ parser = SentenceSplitter.from_defaults(
                 chunk_size=256, chunk_overlap=64, paragraph_separator="\n\n"
             )
 
+@spaces.GPU(duration=20)
+def build_index():
+    # Load documents from a file
+    documents = SimpleDirectoryReader(input_files=["data/blockchainprova.txt"]).load_data()
+    # Parse the documents into nodes
+    nodes = parser.get_nodes_from_documents(documents)
+    # Build the vector store index from the nodes
+    index = VectorStoreIndex(nodes)
+    
+    return index
+
 
+index = build_index()
 
 @spaces.GPU(duration=20)
 def handle_query(query_str, chathistory):
   
-    # build the vector
-    documents = SimpleDirectoryReader(input_files=["data/blockchainprova.txt"]).load_data()
-    nodes = parser.get_nodes_from_documents(documents)
-    index = VectorStoreIndex(nodes)
+
   
     qa_prompt_str = (
         "Context information is below.\n"
@@ -71,17 +80,19 @@ def handle_query(query_str, chathistory):
     text_qa_template = ChatPromptTemplate.from_messages(chat_text_qa_msgs)
 
     try:
-        result = index.as_query_engine(text_qa_template=text_qa_template, streaming=True).query(query_str)
+        # Create a streaming query engine
+        query_engine = index.as_query_engine(text_qa_template=text_qa_template, streaming=True)
         
-        for text in result.response_gen:
-          print(text)
-          yield text
-          
+        # Execute the query
+        streaming_response = query_engine.query(query_str)
         
-        # Remove any unwanted tokens like <end_of_turn>
-        #cleaned_result = response_text#.replace("<end_of_turn>", "").strip()
+        streaming_response.print_response_stream()
 
-        #yield cleaned_result
+        
+        # Stream the response
+        for text in streaming_response.response_gen:
+            yield text
+          
         
     except Exception as e:
         yield f"Error processing query: {str(e)}"