Update app.py

app.py

@@ -1,25 +1,26 @@
 #!/usr/bin/env python3
-# gradio_timer_streaming.py ── Gradio 4.x 相容，真正逐字即時顯示
+# gradio_timer_buffer.py ── 定時刷新目前 buffer 的所有 token
+
 import gradio as gr
 import requests, json, re, sys, time
 from threading import Thread
-from queue import Queue
 
 URL = "http://localhost:8000/v1/chat/completions"
 HEADERS = {"Content-Type": "application/json"}
 
 def sanitize(text: str):
-    return re.sub(r"[\ud800-\udfff]", "", text)  # ✅ 修正：補上右括號
+    return re.sub(r"[\ud800-\udfff]", "", text)
 
 history_openai = []
-output_queue = Queue()
 current_chatbot = []
+assistant_buffer = ""  # 累積 token
 
-def stream_to_queue(user_input: str):
+def stream_to_buffer(user_input: str):
+    global assistant_buffer
     user_input = sanitize(user_input)
     history_openai.append({"role": "user", "content": user_input})
     payload = {"messages": history_openai, "stream": True, "temperature": 0.7}
-    assistant = ""
+    assistant_buffer = ""
     try:
         with requests.post(URL, headers=HEADERS, json=payload, stream=True, timeout=60) as r:
             r.raise_for_status()
@@ -41,54 +42,36 @@ def stream_to_queue(user_input: str):
                     except json.JSONDecodeError:
                         byte_buf = line_bytes + b"\n" + byte_buf
                         continue
-                    if tok is None:
-                        continue
-                    assistant += tok
-                    output_queue.put(assistant)
-                    time.sleep(0.02)
-            history_openai.append({"role": "assistant", "content": sanitize(assistant)})
+                    if tok:
+                        assistant_buffer += tok
+            # 完整回答放入 history
+            history_openai.append({"role": "assistant", "content": sanitize(assistant_buffer)})
     except Exception as e:
-        output_queue.put(f"请求失败: {e}")
+        assistant_buffer = f"请求失败: {e}"
 
 def user_submit(user_msg, chatbot):
-    global current_chatbot
+    global current_chatbot, assistant_buffer
     current_chatbot = chatbot + [[user_msg, ""]]
-    Thread(target=stream_to_queue, args=(user_msg,), daemon=True).start()
+    assistant_buffer = ""
+    Thread(target=stream_to_buffer, args=(user_msg,), daemon=True).start()
     return "", current_chatbot
 
-def poll_queue():
-    global current_chatbot
-    try:
-        latest = output_queue.get_nowait()
-        current_chatbot[-1][1] = latest
-    except:
-        pass
+def flush_buffer():
+    """每 0.3 秒刷新 Chatbot，顯示目前 buffer 的所有 token"""
+    global current_chatbot, assistant_buffer
+    if not current_chatbot:
+        return current_chatbot
+    current_chatbot[-1][1] = assistant_buffer
     return current_chatbot
 
 with gr.Blocks() as demo:
-    gr.Markdown("## Chat8000 Lab2 - 真正即時逐字顯示（Gradio 4.x 相容）")
+    gr.Markdown("## Chat8000 Lab2 - 0.3 秒刷新 buffer 顯示")
     chatbot = gr.Chatbot()
     msg = gr.Textbox(show_label=False, placeholder="輸入訊息後按 Enter")
     msg.submit(user_submit, [msg, chatbot], [msg, chatbot], queue=False)
 
-    timer = gr.Timer(value=0.03, active=True)
-    timer.tick(poll_queue, outputs=chatbot)
+    timer = gr.Timer(value=0.3, active=True)  # 每 0.3 秒刷新一次
+    timer.tick(flush_buffer, outputs=chatbot)
 
 if __name__ == "__main__":
-    if len(sys.argv) > 1 and sys.argv[1] == "--cli":
-        print("=== CLI 模式（支援 Streaming）===")
-        while True:
-            try:
-                prompt = input(">>> ").strip()
-            except (KeyboardInterrupt, EOFError):
-                break
-            if prompt.lower() in {"exit", "quit"}:
-                break
-            if not prompt:
-                continue
-            for token in stream_chat(prompt):
-                print(token, end="", flush=True)
-            print()
-        print("\n再见！")
-    else:
-        demo.launch(server_name="0.0.0.0", server_port=7860)
+    demo.launch(server_name="0.0.0.0", server_port=7860)