Spaces:
Paused
Paused
Upload 4 files
Browse files
app.py
CHANGED
|
@@ -38,6 +38,26 @@ request_counts = {}
|
|
| 38 |
api_key_blacklist = set()
|
| 39 |
api_key_blacklist_duration = 60
|
| 40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
class APIKeyManager:
|
| 42 |
def __init__(self):
|
| 43 |
self.api_keys = os.environ.get('KeyArray').split(',')
|
|
@@ -179,10 +199,9 @@ def handle_api_error(error, attempt, stream=False):
|
|
| 179 |
logger.warning(f"{current_api_key[:11]} → 429 官方资源耗尽 → {delay} 秒后重试...")
|
| 180 |
else:
|
| 181 |
logger.warning(f"{current_api_key[:11]} → 未知错误↙ {delay} 秒后重试...\n{type(error).__name__}\n")
|
|
|
|
|
|
|
| 182 |
time.sleep(delay)
|
| 183 |
-
if isinstance(error, (ResourceExhausted)):
|
| 184 |
-
key_manager.blacklist_key(current_api_key)
|
| 185 |
-
switch_api_key()
|
| 186 |
return False, None
|
| 187 |
|
| 188 |
elif isinstance(error, generation_types.StopCandidateException):
|
|
@@ -212,8 +231,8 @@ def chat_completions():
|
|
| 212 |
gemini_history, user_message, error_response = func.process_messages_for_gemini(messages)
|
| 213 |
|
| 214 |
if error_response:
|
| 215 |
-
|
| 216 |
-
|
| 217 |
|
| 218 |
def do_request(current_api_key, attempt):
|
| 219 |
isok, time = is_within_rate_limit(current_api_key)
|
|
@@ -224,7 +243,18 @@ def chat_completions():
|
|
| 224 |
|
| 225 |
increment_request_count(current_api_key)
|
| 226 |
|
| 227 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 228 |
|
| 229 |
try:
|
| 230 |
if gemini_history:
|
|
@@ -268,7 +298,7 @@ def chat_completions():
|
|
| 268 |
yield f"data: {json.dumps(data)}\n\n"
|
| 269 |
logger.info(f"200!")
|
| 270 |
|
| 271 |
-
except Exception
|
| 272 |
logger.error(f"流式输出时截断,请关闭流式输出或修改你的输入")
|
| 273 |
error_data = {
|
| 274 |
'error': {
|
|
@@ -300,10 +330,10 @@ def chat_completions():
|
|
| 300 |
success, response = do_request(current_api_key, attempt)
|
| 301 |
|
| 302 |
if not success:
|
| 303 |
-
logger.error(f"
|
| 304 |
response = {
|
| 305 |
'error': {
|
| 306 |
-
'message': f'
|
| 307 |
'type': 'internal_server_error'
|
| 308 |
}
|
| 309 |
}
|
|
@@ -337,7 +367,7 @@ def chat_completions():
|
|
| 337 |
'finish_reason': 'stop'
|
| 338 |
}],
|
| 339 |
'usage': {
|
| 340 |
-
'prompt_tokens': 0,
|
| 341 |
'completion_tokens': 0,
|
| 342 |
'total_tokens': 0
|
| 343 |
}
|
|
|
|
| 38 |
api_key_blacklist = set()
|
| 39 |
api_key_blacklist_duration = 60
|
| 40 |
|
| 41 |
+
# 核心优势
|
| 42 |
+
safety_settings = [
|
| 43 |
+
{
|
| 44 |
+
"category": "HARM_CATEGORY_HARASSMENT",
|
| 45 |
+
"threshold": "BLOCK_NONE"
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"category": "HARM_CATEGORY_HATE_SPEECH",
|
| 49 |
+
"threshold": "BLOCK_NONE"
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
|
| 53 |
+
"threshold": "BLOCK_NONE"
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"category": "HARM_CATEGORY_DANGEROUS_CONTENT",
|
| 57 |
+
"threshold": "BLOCK_NONE"
|
| 58 |
+
},
|
| 59 |
+
]
|
| 60 |
+
|
| 61 |
class APIKeyManager:
|
| 62 |
def __init__(self):
|
| 63 |
self.api_keys = os.environ.get('KeyArray').split(',')
|
|
|
|
| 199 |
logger.warning(f"{current_api_key[:11]} → 429 官方资源耗尽 → {delay} 秒后重试...")
|
| 200 |
else:
|
| 201 |
logger.warning(f"{current_api_key[:11]} → 未知错误↙ {delay} 秒后重试...\n{type(error).__name__}\n")
|
| 202 |
+
key_manager.blacklist_key(current_api_key)
|
| 203 |
+
switch_api_key()
|
| 204 |
time.sleep(delay)
|
|
|
|
|
|
|
|
|
|
| 205 |
return False, None
|
| 206 |
|
| 207 |
elif isinstance(error, generation_types.StopCandidateException):
|
|
|
|
| 231 |
gemini_history, user_message, error_response = func.process_messages_for_gemini(messages)
|
| 232 |
|
| 233 |
if error_response:
|
| 234 |
+
logger.error(f"处理输入消息时出错↙\n {error_response}")
|
| 235 |
+
return jsonify(error_response), 400
|
| 236 |
|
| 237 |
def do_request(current_api_key, attempt):
|
| 238 |
isok, time = is_within_rate_limit(current_api_key)
|
|
|
|
| 243 |
|
| 244 |
increment_request_count(current_api_key)
|
| 245 |
|
| 246 |
+
genai.configure(api_key=current_api_key)
|
| 247 |
+
|
| 248 |
+
generation_config = {
|
| 249 |
+
"temperature": temperature,
|
| 250 |
+
"max_output_tokens": max_tokens
|
| 251 |
+
}
|
| 252 |
+
|
| 253 |
+
gen_model = genai.GenerativeModel(
|
| 254 |
+
model_name=model,
|
| 255 |
+
generation_config=generation_config,
|
| 256 |
+
safety_settings=safety_settings
|
| 257 |
+
)
|
| 258 |
|
| 259 |
try:
|
| 260 |
if gemini_history:
|
|
|
|
| 298 |
yield f"data: {json.dumps(data)}\n\n"
|
| 299 |
logger.info(f"200!")
|
| 300 |
|
| 301 |
+
except Exception:
|
| 302 |
logger.error(f"流式输出时截断,请关闭流式输出或修改你的输入")
|
| 303 |
error_data = {
|
| 304 |
'error': {
|
|
|
|
| 330 |
success, response = do_request(current_api_key, attempt)
|
| 331 |
|
| 332 |
if not success:
|
| 333 |
+
logger.error(f"{MAX_RETRIES} 次尝试均失败,请调整配置或向Moonfanz反馈")
|
| 334 |
response = {
|
| 335 |
'error': {
|
| 336 |
+
'message': f'{MAX_RETRIES} 次尝试均失败,请调整配置或向Moonfanz反馈',
|
| 337 |
'type': 'internal_server_error'
|
| 338 |
}
|
| 339 |
}
|
|
|
|
| 367 |
'finish_reason': 'stop'
|
| 368 |
}],
|
| 369 |
'usage': {
|
| 370 |
+
'prompt_tokens': 0,
|
| 371 |
'completion_tokens': 0,
|
| 372 |
'total_tokens': 0
|
| 373 |
}
|
func.py
CHANGED
|
@@ -13,26 +13,6 @@ logger = logging.getLogger(__name__)
|
|
| 13 |
|
| 14 |
request_counts = {}
|
| 15 |
|
| 16 |
-
# 核心优势
|
| 17 |
-
safety_settings = [
|
| 18 |
-
{
|
| 19 |
-
"category": "HARM_CATEGORY_HARASSMENT",
|
| 20 |
-
"threshold": "BLOCK_NONE"
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"category": "HARM_CATEGORY_HATE_SPEECH",
|
| 24 |
-
"threshold": "BLOCK_NONE"
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
|
| 28 |
-
"threshold": "BLOCK_NONE"
|
| 29 |
-
},
|
| 30 |
-
{
|
| 31 |
-
"category": "HARM_CATEGORY_DANGEROUS_CONTENT",
|
| 32 |
-
"threshold": "BLOCK_NONE"
|
| 33 |
-
},
|
| 34 |
-
]
|
| 35 |
-
|
| 36 |
password = os.environ['password']
|
| 37 |
|
| 38 |
def authenticate_request(request):
|
|
@@ -54,21 +34,6 @@ def authenticate_request(request):
|
|
| 54 |
|
| 55 |
return True, None, None
|
| 56 |
|
| 57 |
-
def get_gen_model(api_key, model, temperature, max_tokens):
|
| 58 |
-
genai.configure(api_key=api_key)
|
| 59 |
-
|
| 60 |
-
generation_config = {
|
| 61 |
-
"temperature": temperature,
|
| 62 |
-
"max_output_tokens": max_tokens
|
| 63 |
-
}
|
| 64 |
-
|
| 65 |
-
gen_model = genai.GenerativeModel(
|
| 66 |
-
model_name=model,
|
| 67 |
-
generation_config=generation_config,
|
| 68 |
-
safety_settings=safety_settings
|
| 69 |
-
)
|
| 70 |
-
return gen_model
|
| 71 |
-
|
| 72 |
def process_messages_for_gemini(messages):
|
| 73 |
gemini_history = []
|
| 74 |
errors = []
|
|
|
|
| 13 |
|
| 14 |
request_counts = {}
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
password = os.environ['password']
|
| 17 |
|
| 18 |
def authenticate_request(request):
|
|
|
|
| 34 |
|
| 35 |
return True, None, None
|
| 36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
def process_messages_for_gemini(messages):
|
| 38 |
gemini_history = []
|
| 39 |
errors = []
|