Update app.py

app.py

@@ -653,11 +653,21 @@ def handsome_chat_completions():
                         logging.error(f"无效的图片数据: {item}")
                         openai_images.append({"url": item})
 
-
+                # Construct the expected JSON output
                 response_data = {
+                    "images": openai_images, # use this key so the frontend can parse
                     "created": int(time.time()),
-                    "data": openai_images
+                    "timings": {
+                        "inference": total_time  # add inference time as shown in example
+                    }
                 }
+                if "seed" in response_json:
+                   response_data["seed"] = response_json["seed"]
+                if "shared_id" in response_json:
+                   response_data["shared_id"] = response_json["shared_id"]
+                if "data" in response_json:
+                   response_data["data"] = response_json["data"]
+                   
             except (KeyError, ValueError, IndexError) as e:
                 logging.error(
                     f"解析响应 JSON 失败: {e}, "
@@ -683,6 +693,7 @@ def handsome_chat_completions():
             logging.error(f"请求转发异常: {e}")
             return jsonify({"error": str(e)}), 500
     else:
+        # Existing text-based model handling logic
         try:
             start_time = time.time()
             response = requests.post(
@@ -696,7 +707,8 @@ def handsome_chat_completions():
                 return jsonify(response.json()), 429
 
             if data.get("stream", False):
-                def generate():
+              # ... (Existing stream response handling)
+              def generate():
                     first_chunk_time = None
                     full_response_content = ""
                     for chunk in response.iter_content(chunk_size=1024):
@@ -803,10 +815,14 @@ def handsome_chat_completions():
                         request_timestamps.append(time.time())
                         token_counts.append(prompt_tokens+completion_tokens)
 
-                return Response(
-                    stream_with_context(generate()),
-                    content_type=response.headers['Content-Type']
-                )
+                    
+
+                    
+
+                    return Response(
+                       stream_with_context(generate()),
+                       content_type=response.headers['Content-Type']
+                   )
             else:
                 response.raise_for_status()
                 end_time = time.time()