Spaces:
Running
Running
File size: 19,862 Bytes
d0dd276 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 |
import json
import os
import httpx
from app.models.schemas import ChatCompletionRequest
from dataclasses import dataclass
from typing import Optional, Dict, Any, List
import httpx
import secrets
import string
import app.config.settings as settings
from app.utils.logging import log
def generate_secure_random_string(length):
all_characters = string.ascii_letters + string.digits
secure_random_string = ''.join(secrets.choice(all_characters) for _ in range(length))
return secure_random_string
@dataclass
class GeneratedText:
text: str
finish_reason: Optional[str] = None
class GeminiResponseWrapper:
def __init__(self, data: Dict[Any, Any]):
self._data = data
self._text = self._extract_text()
self._finish_reason = self._extract_finish_reason()
self._prompt_token_count = self._extract_prompt_token_count()
self._candidates_token_count = self._extract_candidates_token_count()
self._total_token_count = self._extract_total_token_count()
self._thoughts = self._extract_thoughts()
self._function_call = self._extract_function_call()
self._json_dumps = json.dumps(self._data, indent=4, ensure_ascii=False)
self._model = "gemini"
def _extract_thoughts(self) -> Optional[str]:
try:
for part in self._data['candidates'][0]['content']['parts']:
if 'thought' in part:
return part['text']
return ""
except (KeyError, IndexError):
return ""
def _extract_text(self) -> str:
try:
text=""
for part in self._data['candidates'][0]['content']['parts']:
if 'thought' not in part and 'text' in part:
text += part['text']
return text
except (KeyError, IndexError):
return ""
def _extract_function_call(self) -> Optional[Dict[str, Any]]:
try:
parts = self._data.get('candidates', [{}])[0].get('content', {}).get('parts', [])
# 使用列表推导式查找所有包含 'functionCall' 的 part,并提取其值
function_calls = [
part['functionCall']
for part in parts
if isinstance(part, dict) and 'functionCall' in part
]
# 如果列表不为空,则返回列表;否则返回 None
return function_calls if function_calls else None
except (KeyError, IndexError, TypeError):
return None
def _extract_finish_reason(self) -> Optional[str]:
try:
return self._data['candidates'][0].get('finishReason')
except (KeyError, IndexError):
return None
def _extract_prompt_token_count(self) -> Optional[int]:
try:
return self._data['usageMetadata'].get('promptTokenCount')
except (KeyError):
return None
def _extract_candidates_token_count(self) -> Optional[int]:
try:
return self._data['usageMetadata'].get('candidatesTokenCount')
except (KeyError):
return None
def _extract_total_token_count(self) -> Optional[int]:
try:
return self._data['usageMetadata'].get('totalTokenCount')
except (KeyError):
return None
def set_model(self,model) -> Optional[str]:
self._model = model
@property
def data(self) -> Dict[Any, Any]:
return self._data
@property
def text(self) -> str:
return self._text
@property
def finish_reason(self) -> Optional[str]:
return self._finish_reason
@property
def prompt_token_count(self) -> Optional[int]:
return self._prompt_token_count
@property
def candidates_token_count(self) -> Optional[int]:
return self._candidates_token_count
@property
def total_token_count(self) -> Optional[int]:
return self._total_token_count
@property
def thoughts(self) -> Optional[str]:
return self._thoughts
@property
def json_dumps(self) -> str:
return self._json_dumps
@property
def model(self) -> str:
return self._model
@property
def function_call(self) -> Optional[Dict[str, Any]]:
return self._function_call
class GeminiClient:
AVAILABLE_MODELS = []
extra_models_str = os.environ.get("EXTRA_MODELS", "")
EXTRA_MODELS = [model.strip() for model in extra_models_str.split(",") if model.strip()]
def __init__(self, api_key: str):
self.api_key = api_key
# 请求参数处理
def _convert_request_data(self, request, contents, safety_settings, system_instruction):
model = request.model
format_type = getattr(request, 'format_type', None)
if format_type and (format_type == "gemini"):
api_version = "v1alpha" if "think" in request.model else "v1beta"
if request.payload:
# 将 Pydantic 模型转换为字典, 假设 Pydantic V2+
data = request.payload.model_dump(exclude_none=True)
# # 注入搜索提示
# if settings.search["search_mode"] and request.model and request.model.endswith("-search"):
# data.insert(len(data)-2,{'role': 'user', 'parts': [{'text':settings.search["search_prompt"]}]})
# # 注入随机字符串
# if settings.RANDOM_STRING:
# data.insert(1,{'role': 'user', 'parts': [{'text': generate_secure_random_string(settings.RANDOM_STRING_LENGTH)}]})
# data.insert(len(data)-1,{'role': 'user', 'parts': [{'text': generate_secure_random_string(settings.RANDOM_STRING_LENGTH)}]})
# log('INFO', "伪装消息成功")
else:
api_version, data = self._convert_openAI_request(request, contents, safety_settings, system_instruction)
# 联网模式
if settings.search["search_mode"] and request.model.endswith("-search"):
log('INFO', "开启联网搜索模式", extra={'key': self.api_key[:8], 'model':request.model})
data.setdefault("tools", []).append({"google_search": {}})
model= request.model.removesuffix("-search")
return api_version, model, data
def _convert_openAI_request(self, request: ChatCompletionRequest, contents, safety_settings, system_instruction):
config_params = {
"temperature": request.temperature,
"maxOutputTokens": request.max_tokens,
"topP": request.top_p,
"topK": request.top_k,
"stopSequences": request.stop if isinstance(request.stop, list) else [request.stop] if request.stop is not None else None,
"candidateCount": request.n,
}
if request.thinking_budget:
config_params["thinkingConfig"] = {
"thinkingBudget": request.thinking_budget
}
generationConfig = {k: v for k, v in config_params.items() if v is not None}
api_version = "v1alpha" if "think" in request.model else "v1beta"
data = {
"contents": contents,
"generationConfig": generationConfig,
"safetySettings": safety_settings,
}
# --- 函数调用处理 ---
# 1. 添加 tools (函数声明)
function_declarations = []
if request.tools:
# 显式提取 Gemini API 所需的字段,避免包含 'id' 等无效字段
function_declarations = []
for tool in request.tools:
if tool.get("type") == "function":
func_def = tool.get("function")
if func_def:
# 只包含 Gemini API 接受的字段
declaration = {
"name": func_def.get("name"),
"description": func_def.get("description"),
}
# 获取 parameters 并移除可能存在的 $schema 字段
parameters = func_def.get("parameters")
if isinstance(parameters, dict) and "$schema" in parameters:
parameters = parameters.copy()
del parameters["$schema"]
if parameters is not None:
declaration["parameters"] = parameters
# 移除值为 None 的键,以保持 payload 清洁
declaration = {k: v for k, v in declaration.items() if v is not None}
if declaration.get("name"): # 确保 name 存在
function_declarations.append(declaration)
if function_declarations:
data["tools"] = [{"function_declarations": function_declarations}]
# 2. 添加 tool_config (基于 tool_choice)
tool_config = None
if request.tool_choice:
choice = request.tool_choice
mode = None
allowed_functions = None
if isinstance(choice, str):
if choice == "none":
mode = "NONE"
elif choice == "auto":
mode = "AUTO"
elif isinstance(choice, dict) and choice.get("type") == "function":
func_name = choice.get("function", {}).get("name")
if func_name:
mode = "ANY" # 'ANY' 模式用于强制调用特定函数
allowed_functions = [func_name]
# 如果成功解析出有效的 mode,构建 tool_config
if mode:
config = {"mode": mode}
if allowed_functions:
config["allowed_function_names"] = allowed_functions
tool_config = {"function_calling_config": config}
# 3. 添加 tool_config 到 data
if tool_config:
data["tool_config"] = tool_config
if system_instruction:
data["system_instruction"] = system_instruction
return api_version, data
# 流式请求
async def stream_chat(self, request, contents, safety_settings, system_instruction):
# 真流式请求处理逻辑
extra_log = {'key': self.api_key[:8], 'request_type': 'stream', 'model': request.model}
log('INFO', "流式请求开始", extra=extra_log)
api_version, model, data = self._convert_request_data(request, contents, safety_settings, system_instruction)
url = f"https://generativelanguage.googleapis.com/{api_version}/models/{model}:streamGenerateContent?key={self.api_key}&alt=sse"
headers = {
"Content-Type": "application/json",
}
async with httpx.AsyncClient() as client:
async with client.stream("POST", url, headers=headers, json=data, timeout=600) as response:
response.raise_for_status()
buffer = b"" # 用于累积可能不完整的 JSON 数据
try:
async for line in response.aiter_lines():
if not line.strip(): # 跳过空行 (SSE 消息分隔符)
continue
if line.startswith("data: "):
line = line[len("data: "):].strip() # 去除 "data: " 前缀
# 检查是否是结束标志,如果是,结束循环
if line == "[DONE]":
break
buffer += line.encode('utf-8')
try:
# 尝试解析整个缓冲区
data = json.loads(buffer.decode('utf-8'))
# 解析成功,清空缓冲区
buffer = b""
yield GeminiResponseWrapper(data)
except json.JSONDecodeError:
# JSON 不完整,继续累积到 buffer
continue
except Exception as e:
log('ERROR', f"流式处理期间发生错误",
extra={'key': self.api_key[:8], 'request_type': 'stream', 'model': request.model})
raise e
except Exception as e:
raise e
finally:
log('info', "流式请求结束")
# 非流式处理
async def complete_chat(self, request, contents, safety_settings, system_instruction):
api_version, model, data = self._convert_request_data(request, contents, safety_settings, system_instruction)
url = f"https://generativelanguage.googleapis.com/{api_version}/models/{model}:generateContent?key={self.api_key}"
headers = {
"Content-Type": "application/json",
}
try:
async with httpx.AsyncClient() as client:
response = await client.post(url, headers=headers, json=data, timeout=600)
response.raise_for_status() # 检查 HTTP 错误状态
return GeminiResponseWrapper(response.json())
except Exception as e:
raise
# OpenAI 格式请求转换为 gemini 格式请求
def convert_messages(self, messages, use_system_prompt=False, model=None):
gemini_history = []
errors = []
system_instruction_text = ""
system_instruction_parts = [] # 用于收集系统指令文本
# 处理系统指令
if use_system_prompt:
# 遍历消息列表,查找开头的连续 system 消息
for i, message in enumerate(messages):
# 必须是 system 角色且内容是字符串
if message.get('role') == 'system' and isinstance(message.get('content'), str):
system_instruction_parts.append(message.get('content'))
else:
break # 遇到第一个非 system 或内容非字符串的消息就停止
# 将收集到的系统指令合并为一个字符串
system_instruction_text = "\n".join(system_instruction_parts)
system_instruction = {"parts": [{"text": system_instruction_text}]} if system_instruction_text else None
# 转换主要消息
for i, message in enumerate(messages):
role = message.get('role')
content = message.get('content')
if isinstance(content, str):
if role == 'tool':
role_to_use = 'function'
tool_call_id = message.get('tool_call_id')
prefix = "call_"
if tool_call_id.startswith(prefix):
# 假设 tool_call_id = f"call_{function_name}" (response.py中的处理)
function_name = tool_call_id[len(prefix):]
else:
continue
function_response_part = {
"functionResponse": {
"name": function_name,
"response": {"content": content}
}
}
gemini_history.append({"role": role_to_use, "parts": [function_response_part]})
continue
elif role in ['user', 'system']:
role_to_use = 'user'
elif role == 'assistant':
role_to_use = 'model'
else:
errors.append(f"Invalid role: {role}")
continue
# Gemini 的一个重要规则:连续的同角色消息需要合并
# 如果 gemini_history 已有内容,并且最后一条消息的角色和当前要添加的角色相同
if gemini_history and gemini_history[-1]['role'] == role_to_use:
gemini_history[-1]['parts'].append({"text": content})
else:
gemini_history.append({"role": role_to_use, "parts": [{"text": content}]})
elif isinstance(content, list):
parts = []
for item in content:
if item.get('type') == 'text':
parts.append({"text": item.get('text')})
elif item.get('type') == 'image_url':
image_data = item.get('image_url', {}).get('url', '')
if image_data.startswith('data:image/'):
try:
mime_type, base64_data = image_data.split(';')[0].split(':')[1], image_data.split(',')[1]
parts.append({
"inline_data": {
"mime_type": mime_type,
"data": base64_data
}
})
except (IndexError, ValueError):
errors.append(
f"Invalid data URI for image: {image_data}")
else:
errors.append(
f"Invalid image URL format for item: {item}")
if parts:
if role in ['user', 'system']:
role_to_use = 'user'
elif role == 'assistant':
role_to_use = 'model'
else:
errors.append(f"Invalid role: {role}")
continue
if gemini_history and gemini_history[-1]['role'] == role_to_use:
gemini_history[-1]['parts'].extend(parts)
else:
gemini_history.append(
{"role": role_to_use, "parts": parts})
if errors:
return errors
# --- 后处理 ---
# 注入搜索提示
if settings.search["search_mode"] and model and model.endswith("-search"):
gemini_history.insert(len(gemini_history)-2,{'role': 'user', 'parts': [{'text':settings.search["search_prompt"]}]})
# 注入随机字符串
if settings.RANDOM_STRING:
gemini_history.insert(1,{'role': 'user', 'parts': [{'text': generate_secure_random_string(settings.RANDOM_STRING_LENGTH)}]})
gemini_history.insert(len(gemini_history)-1,{'role': 'user', 'parts': [{'text': generate_secure_random_string(settings.RANDOM_STRING_LENGTH)}]})
log('INFO', "伪装消息成功")
return gemini_history, system_instruction
@staticmethod
async def list_available_models(api_key) -> list:
url = "https://generativelanguage.googleapis.com/v1beta/models?key={}".format(
api_key)
async with httpx.AsyncClient() as client:
response = await client.get(url)
response.raise_for_status()
data = response.json()
models = []
for model in data.get("models", []):
models.append(model["name"])
if model["name"].startswith("models/gemini-2") and settings.search["search_mode"]:
models.append(model["name"] + "-search")
models.extend(GeminiClient.EXTRA_MODELS)
return models
|