Add real-time CoT (Chain of Thought) streaming display

app.py

@@ -145,19 +145,67 @@ def process_message(message, history, system_prompt, temperature, max_tokens, to
         # 流式输出
         full_response = ""
         chunk_count = 0
+        in_reasoning = False
+        reasoning_content = ""
+        final_content = ""
+        
         for chunk in response:
             chunk_count += 1
             if chunk.choices and len(chunk.choices) > 0:
                 delta = chunk.choices[0].delta
                 if hasattr(delta, 'content') and delta.content:
-                    full_response += delta.content
-                    history[-1][1] = full_response
-                    if chunk_count % 10 == 0:
+                    content = delta.content
+                    full_response += content
+                    
+                    # 检测 <reasoning> 标签
+                    if '<reasoning>' in content:
+                        in_reasoning = True
+                        # 分割内容
+                        parts = content.split('<reasoning>')
+                        final_content += parts[0]
+                        if len(parts) > 1:
+                            reasoning_content += parts[1]
+                    elif '</reasoning>' in content:
+                        # 结束推理部分
+                        parts = content.split('</reasoning>')
+                        if parts[0]:
+                            reasoning_content += parts[0]
+                        in_reasoning = False
+                        if len(parts) > 1:
+                            final_content += parts[1]
+                    elif in_reasoning:
+                        # 在推理标签内
+                        reasoning_content += content
+                    else:
+                        # 在推理标签外
+                        final_content += content
+                    
+                    # 实时更新显示
+                    if reasoning_content and final_content:
+                        # 有推理和最终答案
+                        display_text = f"💭 **Chain of Thought:**\n\n{reasoning_content}\n\n---\n\n📝 **Answer:**\n\n{final_content}"
+                    elif reasoning_content:
+                        # 只有推理过程
+                        display_text = f"💭 **Chain of Thought:**\n\n{reasoning_content}\n\n---\n\n📝 **Answer:**\n\n*Generating...*"
+                    else:
+                        # 只有答案或普通回复
+                        display_text = full_response
+                    
+                    history[-1][1] = display_text
+                    
+                    if chunk_count % 5 == 0:
                         print(f"[DEBUG] Received {chunk_count} chunks, {len(full_response)} chars")
                     yield history
         
         print(f"[DEBUG] Stream complete. Total chunks: {chunk_count}, Total chars: {len(full_response)}")
         
+        # 最终格式化
+        if reasoning_content:
+            # 如果有推理内容，使用格式化显示
+            final_display = f"💭 **Chain of Thought:**\n\n{reasoning_content}\n\n---\n\n📝 **Answer:**\n\n{final_content.strip()}"
+            history[-1][1] = final_display
+            yield history
+        
         if not full_response:
             print("[DEBUG] No response content received")
             history[-1][1] = "⚠️ No response received from API"