implement streaming to the chat

app.py

@@ -1,6 +1,7 @@
 import os
 import gradio as gr
 from dotenv import load_dotenv
+import time
 
 from implementation.chat import answer_question
 
@@ -46,13 +47,29 @@ def chat(history):
             })
 
         answer, context = answer_question(last_message, clean_history)
-        history.append({"role": "assistant", "content": answer})
-        return history, format_context(context)
+
+        # Stream the final answer into the UI (typewriter-style).
+        # This is robust even when the backend does routing/tool-calls.
+        history.append({"role": "assistant", "content": ""})
+        context_md = ""
+        yield history, context_md
+
+        chunk_size = int(os.getenv("STREAM_CHUNK_SIZE", "24"))
+        delay_s = float(os.getenv("STREAM_DELAY_S", "0.01"))
+        built = ""
+        for i in range(0, len(answer), max(1, chunk_size)):
+            built = answer[: i + chunk_size]
+            history[-1]["content"] = built
+            yield history, context_md
+            if delay_s > 0:
+                time.sleep(delay_s)
+
+        yield history, format_context(context)
     except Exception as e:
         import traceback
         traceback.print_exc()
         history.append({"role": "assistant", "content": f"Error: {str(e)}"})
-        return history, "Error occurred."
+        yield history, "Error occurred."