77 lines
3.2 KiB
Python
77 lines
3.2 KiB
Python
|
import sys
|
|||
|
|
|
|||
|
|
from openai import AsyncOpenAI
|
|||
|
|
|
|||
|
|
from Config import *
|
|||
|
|
from Config.Config import ALY_LLM_API_KEY, ALY_LLM_BASE_URL, ALY_LLM_MODEL_NAME
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 保留原同步函数，添加异步版本
|
|||
|
|
async def get_llm_response(query_text: str, stream: bool = True, system_prompt: str = 'You are a helpful assistant.', chat_history: list = None, temperature: float = None):
|
|||
|
|
"""
|
|||
|
|
异步获取大模型的响应
|
|||
|
|
@param query_text: 查询文本
|
|||
|
|
@param stream: 是否使用流式输出
|
|||
|
|
@param system_prompt: 系统提示文本，默认为'You are a helpful assistant.'
|
|||
|
|
@param chat_history: 聊天历史记录，格式为[{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}, ...]
|
|||
|
|
@param temperature: 温度参数，控制输出的随机性，0-2之间，默认为None使用模型默认值
|
|||
|
|
@return: 流式响应生成器或完整响应文本
|
|||
|
|
"""
|
|||
|
|
client = AsyncOpenAI(
|
|||
|
|
api_key=ALY_LLM_API_KEY,
|
|||
|
|
base_url=ALY_LLM_BASE_URL,
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
# 构建messages列表，包含系统提示和聊天历史
|
|||
|
|
messages = [{'role': 'system', 'content': system_prompt}]
|
|||
|
|
|
|||
|
|
# 如果有聊天历史，添加到messages中
|
|||
|
|
if chat_history and isinstance(chat_history, list):
|
|||
|
|
messages.extend(chat_history)
|
|||
|
|
|
|||
|
|
# 添加当前用户查询
|
|||
|
|
messages.append({'role': 'user', 'content': query_text})
|
|||
|
|
|
|||
|
|
# 构建请求参数
|
|||
|
|
request_params = {
|
|||
|
|
'model': ALY_LLM_MODEL_NAME,
|
|||
|
|
'messages': messages,
|
|||
|
|
'stream': stream
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# 如果指定了temperature参数，添加到请求中
|
|||
|
|
if temperature is not None:
|
|||
|
|
request_params['temperature'] = temperature
|
|||
|
|
|
|||
|
|
# 创建请求
|
|||
|
|
completion = await client.chat.completions.create(**request_params)
|
|||
|
|
|
|||
|
|
if stream:
|
|||
|
|
# 流式输出模式，返回生成器
|
|||
|
|
async for chunk in completion:
|
|||
|
|
# 确保 chunk.choices 存在且不为空
|
|||
|
|
if chunk and chunk.choices and len(chunk.choices) > 0:
|
|||
|
|
# 确保 delta 存在
|
|||
|
|
delta = chunk.choices[0].delta
|
|||
|
|
if delta:
|
|||
|
|
# 确保 content 存在且不为 None 或空字符串
|
|||
|
|
content = delta.content
|
|||
|
|
if content is not None:
|
|||
|
|
# if content is not None and content.strip():
|
|||
|
|
print(content, end='', flush=True)
|
|||
|
|
yield content
|
|||
|
|
else:
|
|||
|
|
# 非流式处理
|
|||
|
|
if completion and completion.choices and len(completion.choices) > 0:
|
|||
|
|
message = completion.choices[0].message
|
|||
|
|
if message:
|
|||
|
|
content = message.content
|
|||
|
|
if content is not None:
|
|||
|
|
# if content is not None and content.strip():
|
|||
|
|
yield content
|
|||
|
|
except Exception as e:
|
|||
|
|
error_msg = f"大模型请求异常: {str(e)}"
|
|||
|
|
print(error_msg, file=sys.stderr)
|
|||
|
|
# 抛出异常而不是返回错误信息，让调用方处理
|
|||
|
|
raise Exception(error_msg)
|