Fix ChatQnA streaming response issue (#88)

Signed-off-by: lvliang-intel <[email protected]>

ChatQnA/langchain/docker/qna-app/app/server.py

@@ -219,7 +219,7 @@ def stream_generator():
         for text in router.llm_chain.stream({"question": query, "chat_history": router.chat_history}):
             chat_response += text
             processed_text = post_process_text(text)
-            if text is not None:
+            if text and processed_text:
                 yield processed_text
         chat_response = chat_response.split("</s>")[0]
         print(f"[rag - chat_stream] stream response: {chat_response}")

ChatQnA/langchain/docker/qna-app/app/utils.py

@@ -344,9 +344,9 @@ def reload_retriever(embeddings, index_name):
 def post_process_text(text: str):
     if text == " ":
         return "data: @#$\n\n"
-    if text.isspace():
-        return None
     if text == "\n":
         return "data: <br/>\n\n"
+    if text.isspace():
+        return None
     new_text = text.replace(" ", "@#$")
     return f"data: {new_text}\n\n"