Update app.py

app.py

@@ -583,7 +583,6 @@ def handsome_chat_completions():
     
     if model_name in image_models:
         # Handle image generation
-        # Map OpenAI-style parameters to SiliconFlow's parameters
         siliconflow_data = {
             "model": model_name,
             "prompt": data.get("messages", [{}])[0].get("content") if isinstance(data.get("messages"), list) else "",
@@ -644,22 +643,30 @@ def handsome_chat_completions():
                   image_url = images[0]
                   logging.info(f"Extracted image URL: {image_url}")
                 
-                # Construct the expected JSON output - Mimicking OpenAI
+                #  Construct the expected JSON output - Mimicking OpenAI with system_fingerprint and usage
+
+                #  对于图像模型，usage 可以设置为固定的值， 因为实际的token消耗难以计算
                 response_data = {
-                   "id": f"chatcmpl-{uuid.uuid4()}",
-                   "object": "chat.completion",
-                   "created": int(time.time()),
-                   "model": model_name,
-                   "choices": [
-                      {
-                        "index": 0,
-                        "message": {
-                            "role": "assistant",
-                            "content": image_url if image_url else "Failed to generate image", # Directly return the URL in content
-                        },
-                        "finish_reason": "stop",
-                     }
-                   ],
+                    "id": f"chatcmpl-{uuid.uuid4()}",
+                    "object": "chat.completion",
+                    "created": int(time.time()),
+                    "model": model_name,
+                    "system_fingerprint": system_fingerprint,
+                    "choices": [
+                        {
+                            "index": 0,
+                            "message": {
+                                "role": "assistant",
+                                "content": image_url if image_url else "Failed to generate image",
+                            },
+                            "finish_reason": "stop",
+                        }
+                    ],
+                    "usage": {
+                        "completion_tokens": 0,
+                        "prompt_tokens": 0,
+                        "total_tokens": 0,
+                    },
                 }
 
             except (KeyError, ValueError, IndexError) as e:
@@ -668,20 +675,26 @@ def handsome_chat_completions():
                     f"完整内容: {response_json}"
                 )
                 response_data = {
-                   "id": f"chatcmpl-{uuid.uuid4()}",
-                   "object": "chat.completion",
-                   "created": int(time.time()),
-                   "model": model_name,
-                   "choices": [
-                      {
-                        "index": 0,
-                        "message": {
-                            "role": "assistant",
-                            "content": "Failed to process image data",
-                        },
-                        "finish_reason": "stop",
-                      }
-                   ],
+                    "id": f"chatcmpl-{uuid.uuid4()}",
+                    "object": "chat.completion",
+                    "created": int(time.time()),
+                    "model": model_name,
+                    "system_fingerprint": system_fingerprint,
+                    "choices": [
+                        {
+                            "index": 0,
+                            "message": {
+                                "role": "assistant",
+                                "content": "Failed to process image data",
+                            },
+                            "finish_reason": "stop",
+                        }
+                    ],
+                    "usage": {
+                        "completion_tokens": 0,
+                        "prompt_tokens": 0,
+                        "total_tokens": 0,
+                    },
                 }
 
             logging.info(
@@ -820,10 +833,36 @@ 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']
-                )
+                        
+                    # 构建符合 OpenAI 的响应格式，包括 system_fingerprint 和 usage
+                    response_data = {
+                        "id": f"chatcmpl-{uuid.uuid4()}",
+                        "object": "chat.completion",
+                        "created": int(time.time()),
+                        "model": model_name,
+                        "system_fingerprint": system_fingerprint,
+                        "choices": [
+                             {
+                               "index": 0,
+                                "message": {
+                                    "role": "assistant",
+                                    "content": response_content,
+                                },
+                                "finish_reason": "stop",
+                             }
+                        ],
+                        "usage": {
+                            "completion_tokens": completion_tokens,
+                            "prompt_tokens": prompt_tokens,
+                            "total_tokens": prompt_tokens + completion_tokens,
+                        },
+                    }
+
+                    return Response(
+                        json.dumps(response_data),
+                        content_type="application/json"
+                    )
+                return generate()
             else:
                 response.raise_for_status()
                 end_time = time.time()
@@ -846,6 +885,30 @@ def handsome_chat_completions():
                     prompt_tokens = 0
                     completion_tokens = 0
                     response_content = ""
+                # 构建符合 OpenAI 的响应格式，包括 system_fingerprint 和 usage
+                response_data = {
+                    "id": f"chatcmpl-{uuid.uuid4()}",
+                    "object": "chat.completion",
+                    "created": int(time.time()),
+                    "model": model_name,
+                    "system_fingerprint": system_fingerprint,
+                    "choices": [
+                        {
+                           "index": 0,
+                            "message": {
+                                "role": "assistant",
+                                "content": response_content,
+                            },
+                            "finish_reason": "stop",
+                        }
+                    ],
+                    "usage": {
+                        "completion_tokens": completion_tokens,
+                        "prompt_tokens": prompt_tokens,
+                        "total_tokens": prompt_tokens + completion_tokens,
+                    },
+                }
+
 
                 user_content = ""
                 messages = data.get("messages", [])
@@ -890,7 +953,7 @@ def handsome_chat_completions():
                     else:
                         token_counts.append(0)
 
-                return jsonify(response_json)
+                return jsonify(response_data)
 
         except requests.exceptions.RequestException as e:
             logging.error(f"请求转发异常: {e}")