Upload 4 files

app.py

@@ -177,10 +177,10 @@ def increment_request_count(api_key):
         request_counts[api_key] = deque()
     request_counts[api_key].append(now)
 
-def handle_api_error(error, attempt, stream=False):
+def handle_api_error(error, attempt):
     if attempt > MAX_RETRIES:
         logger.error(f"{MAX_RETRIES} 次尝试后仍然失败，请修改预设或输入")
-        return False, jsonify({
+        return 0, jsonify({
                 'error': {
                     'message': f"{MAX_RETRIES} 次尝试后仍然失败，请修改预设或输入",
                     'type': 'max_retries_exceeded'
@@ -191,7 +191,7 @@ def handle_api_error(error, attempt, stream=False):
         logger.error(f"{current_api_key[:11]} → 无效，可能已过期或被删除")
         key_manager.blacklist_key(current_api_key)
         switch_api_key()
-        return False, None
+        return 0, None
 
     elif isinstance(error, (ResourceExhausted, Aborted, InternalServerError, ServiceUnavailable)):
         delay = min(RETRY_DELAY * (2 ** attempt), MAX_RETRY_DELAY)
@@ -202,16 +202,16 @@ def handle_api_error(error, attempt, stream=False):
         key_manager.blacklist_key(current_api_key)
         switch_api_key()
         time.sleep(delay)
-        return False, None
+        return 0, None
 
     elif isinstance(error, generation_types.StopCandidateException):
         logger.warning(f"AI输出内容被Gemini官方阻挡，代理没有得到有效回复")
         switch_api_key()
-        return False, None
+        return 0, None
 
     else:
-        logger.error(f"未知错误↙\n {error}")
-        return False, None
+        logger.error(f"该模型暂时不可用↙\n {error}")
+        return 2, None
 
 @app.route('/hf/v1/chat/completions', methods=['POST'])
 def chat_completions():
@@ -239,7 +239,7 @@ def chat_completions():
         if not isok:
             logger.warning(f"{current_api_key[:11]} → 暂时超过限额，该API key将在 {time} 秒后启用...")
             switch_api_key()
-            return False, None
+            return 0, None
 
         increment_request_count(current_api_key)
 
@@ -262,7 +262,7 @@ def chat_completions():
                 response = chat_session.send_message(user_message, stream=stream)
             else:
                 response = gen_model.generate_content(user_message, stream=stream)
-            return True, response
+            return 1, response
         except Exception as e:
             return handle_api_error(e, attempt, stream)
 
@@ -321,15 +321,27 @@ def chat_completions():
             yield f"data: {json.dumps(data)}\n\n"
 
     attempt = 0
-    success = False
+    success = 0
     response = None
 
-    while attempt < MAX_RETRIES and not success:
-        attempt += 1
+    for attempt in range(1, MAX_RETRIES + 1):
         logger.info(f"第 {attempt}/{MAX_RETRIES} 次尝试 ...")
         success, response = do_request(current_api_key, attempt)
 
-    if not success:
+        if success == 1:
+            break
+        elif success == 2:
+
+            logger.error("该模型暂时不可用，请更换模型或稍后重试")
+            response = {
+                'error': {
+                    'message': '该模型暂时不可用，请更换模型或稍后重试',
+                    'type': 'internal_server_error'
+                }
+            }
+            return jsonify(response), 503
+
+    else:
         logger.error(f"{MAX_RETRIES} 次尝试均失败，请调整配置或向Moonfanz反馈")
         response = {
             'error': {