Completely refactor the LLM integration method to easily replace the LLM used by each module and optimize the retransmission mechanism.
This commit is contained in:
666ghj
@@ -1,9 +1,7 @@
|
||||
"""
|
||||
Report Engine LLM模块
|
||||
包含各种大语言模型的接口实现
|
||||
LLM module for the Report Engine.
|
||||
"""
|
||||
|
||||
from .base import BaseLLM
|
||||
from .gemini_llm import GeminiLLM
|
||||
from .base import LLMClient
|
||||
|
||||
__all__ = ["BaseLLM", "GeminiLLM"]
|
||||
__all__ = ["LLMClient"]
|
||||
|
||||
+80
-86
@@ -1,95 +1,89 @@
|
||||
"""
|
||||
Report Engine LLM基类
|
||||
定义所有LLM实现的基础接口
|
||||
Unified OpenAI-compatible LLM client for the Report Engine, with retry support.
|
||||
"""
|
||||
|
||||
from abc import ABC, abstractmethod
|
||||
from typing import Dict, Any, Optional
|
||||
import os
|
||||
import sys
|
||||
from typing import Any, Dict, Optional
|
||||
|
||||
from openai import OpenAI
|
||||
|
||||
current_dir = os.path.dirname(os.path.abspath(__file__))
|
||||
project_root = os.path.dirname(os.path.dirname(current_dir))
|
||||
utils_dir = os.path.join(project_root, "utils")
|
||||
if utils_dir not in sys.path:
|
||||
sys.path.append(utils_dir)
|
||||
|
||||
try:
|
||||
from retry_helper import with_retry, LLM_RETRY_CONFIG
|
||||
except ImportError:
|
||||
def with_retry(config=None):
|
||||
def decorator(func):
|
||||
return func
|
||||
return decorator
|
||||
|
||||
LLM_RETRY_CONFIG = None
|
||||
|
||||
|
||||
class BaseLLM(ABC):
|
||||
"""LLM基类"""
|
||||
|
||||
def __init__(self, api_key: str, model_name: Optional[str] = None):
|
||||
"""
|
||||
初始化LLM客户端
|
||||
|
||||
Args:
|
||||
api_key: API密钥
|
||||
model_name: 模型名称
|
||||
"""
|
||||
class LLMClient:
|
||||
"""Minimal wrapper around the OpenAI-compatible chat completion API."""
|
||||
|
||||
def __init__(self, api_key: str, model_name: str, base_url: Optional[str] = None):
|
||||
if not api_key:
|
||||
raise ValueError("Report Engine LLM API key is required.")
|
||||
if not model_name:
|
||||
raise ValueError("Report Engine model name is required.")
|
||||
|
||||
self.api_key = api_key
|
||||
self.base_url = base_url
|
||||
self.model_name = model_name
|
||||
|
||||
@abstractmethod
|
||||
self.provider = model_name
|
||||
timeout_fallback = os.getenv("LLM_REQUEST_TIMEOUT") or os.getenv("REPORT_ENGINE_REQUEST_TIMEOUT") or "180"
|
||||
try:
|
||||
self.timeout = float(timeout_fallback)
|
||||
except ValueError:
|
||||
self.timeout = 300.0
|
||||
|
||||
client_kwargs: Dict[str, Any] = {
|
||||
"api_key": api_key,
|
||||
"max_retries": 0,
|
||||
}
|
||||
if base_url:
|
||||
client_kwargs["base_url"] = base_url
|
||||
self.client = OpenAI(**client_kwargs)
|
||||
|
||||
@with_retry(LLM_RETRY_CONFIG)
|
||||
def invoke(self, system_prompt: str, user_prompt: str, **kwargs) -> str:
|
||||
"""
|
||||
调用LLM生成回复
|
||||
|
||||
Args:
|
||||
system_prompt: 系统提示词
|
||||
user_prompt: 用户输入
|
||||
**kwargs: 其他参数
|
||||
|
||||
Returns:
|
||||
生成的回复文本
|
||||
"""
|
||||
pass
|
||||
|
||||
@abstractmethod
|
||||
def get_model_info(self) -> Dict[str, Any]:
|
||||
"""
|
||||
获取当前模型信息
|
||||
|
||||
Returns:
|
||||
模型信息字典
|
||||
"""
|
||||
pass
|
||||
|
||||
@abstractmethod
|
||||
def get_default_model(self) -> str:
|
||||
"""
|
||||
获取默认模型名称
|
||||
|
||||
Returns:
|
||||
默认模型名称
|
||||
"""
|
||||
pass
|
||||
|
||||
def validate_response(self, response: str) -> str:
|
||||
"""
|
||||
验证和清理响应内容
|
||||
|
||||
Args:
|
||||
response: 原始响应
|
||||
|
||||
Returns:
|
||||
清理后的响应
|
||||
"""
|
||||
if not response:
|
||||
messages = [
|
||||
{"role": "system", "content": system_prompt},
|
||||
{"role": "user", "content": user_prompt},
|
||||
]
|
||||
|
||||
allowed_keys = {"temperature", "top_p", "presence_penalty", "frequency_penalty", "stream"}
|
||||
extra_params = {key: value for key, value in kwargs.items() if key in allowed_keys and value is not None}
|
||||
|
||||
timeout = kwargs.pop("timeout", self.timeout)
|
||||
|
||||
response = self.client.chat.completions.create(
|
||||
model=self.model_name,
|
||||
messages=messages,
|
||||
timeout=timeout,
|
||||
**extra_params,
|
||||
)
|
||||
|
||||
if response.choices and response.choices[0].message:
|
||||
return self.validate_response(response.choices[0].message.content)
|
||||
return ""
|
||||
|
||||
@staticmethod
|
||||
def validate_response(response: Optional[str]) -> str:
|
||||
if response is None:
|
||||
return ""
|
||||
|
||||
# 移除多余的空白字符
|
||||
response = response.strip()
|
||||
|
||||
# 确保响应不为空
|
||||
if not response:
|
||||
return "抱歉，生成的内容为空。"
|
||||
|
||||
return response
|
||||
|
||||
def estimate_tokens(self, text: str) -> int:
|
||||
"""
|
||||
估算文本的token数量（简单实现）
|
||||
|
||||
Args:
|
||||
text: 输入文本
|
||||
|
||||
Returns:
|
||||
估算的token数量
|
||||
"""
|
||||
# 简单估算：中文字符按1.5个token计算，英文单词按1个token计算
|
||||
chinese_chars = len([c for c in text if '\u4e00' <= c <= '\u9fff'])
|
||||
english_words = len(text.split()) - chinese_chars
|
||||
|
||||
return int(chinese_chars * 1.5 + english_words)
|
||||
return response.strip()
|
||||
|
||||
def get_model_info(self) -> Dict[str, Any]:
|
||||
return {
|
||||
"provider": self.provider,
|
||||
"model": self.model_name,
|
||||
"api_base": self.base_url or "default",
|
||||
}
|
||||
|
||||
@@ -1,203 +0,0 @@
|
||||
"""
|
||||
Report Engine Gemini LLM实现
|
||||
使用Gemini 2.5-pro中转API进行文本生成
|
||||
"""
|
||||
|
||||
import os
|
||||
import sys
|
||||
from typing import Optional, Dict, Any
|
||||
from openai import OpenAI
|
||||
from .base import BaseLLM
|
||||
|
||||
DEFAULT_GEMINI_BASE_URL = "https://www.chataiapi.com/v1"
|
||||
|
||||
# 导入根目录的config
|
||||
try:
|
||||
current_dir = os.path.dirname(os.path.abspath(__file__))
|
||||
root_dir = os.path.dirname(os.path.dirname(current_dir))
|
||||
if root_dir not in sys.path:
|
||||
sys.path.append(root_dir)
|
||||
import config
|
||||
except ImportError:
|
||||
config = None
|
||||
|
||||
# 添加utils目录到Python路径并导入重试模块
|
||||
try:
|
||||
if root_dir:
|
||||
utils_dir = os.path.join(root_dir, 'utils')
|
||||
if utils_dir not in sys.path:
|
||||
sys.path.append(utils_dir)
|
||||
from retry_helper import with_retry, with_graceful_retry, LLM_RETRY_CONFIG, RetryConfig
|
||||
# 创建动态重试配置生成函数
|
||||
def create_report_retry_config(config=None):
|
||||
"""创建ReportEngine专用的重试配置，适应7分钟平均生成时间"""
|
||||
return RetryConfig(
|
||||
max_retries=config.max_retries if config and hasattr(config, 'max_retries') else 8,
|
||||
initial_delay=8.0, # 初始延迟增加到8秒，适应长时间生成
|
||||
backoff_factor=2.0, # 保持2倍退避
|
||||
max_delay=config.max_retry_delay if config and hasattr(config, 'max_retry_delay') else 180.0
|
||||
)
|
||||
# 创建默认配置用于模块导入时的向后兼容
|
||||
REPORT_LLM_RETRY_CONFIG = create_report_retry_config()
|
||||
except ImportError:
|
||||
# 如果无法导入重试模块，使用空装饰器避免报错
|
||||
def with_retry(config):
|
||||
def decorator(func):
|
||||
return func
|
||||
return decorator
|
||||
LLM_RETRY_CONFIG = None
|
||||
REPORT_LLM_RETRY_CONFIG = None
|
||||
|
||||
|
||||
class GeminiLLM(BaseLLM):
|
||||
"""Report Engine Gemini LLM实现类"""
|
||||
|
||||
def __init__(self, api_key: Optional[str] = None, model_name: Optional[str] = None, base_url: Optional[str] = None, config=None):
|
||||
"""
|
||||
初始化Gemini客户端
|
||||
|
||||
Args:
|
||||
api_key: Gemini API密钥，如果不提供则从config或环境变量读取
|
||||
model_name: 模型名称，默认使用gemini-2.5-pro
|
||||
base_url: Gemini API基础地址
|
||||
config: 配置对象，用于获取超时设置
|
||||
"""
|
||||
if api_key is None:
|
||||
# 优先从根目录config读取
|
||||
if config and hasattr(config, 'GEMINI_API_KEY'):
|
||||
api_key = config.GEMINI_API_KEY
|
||||
else:
|
||||
# 备选方案：从环境变量读取
|
||||
api_key = os.getenv("GEMINI_API_KEY")
|
||||
|
||||
if not api_key:
|
||||
raise ValueError("Gemini API Key未找到！请在config.py中设置GEMINI_API_KEY或设置环境变量")
|
||||
|
||||
super().__init__(api_key, model_name)
|
||||
|
||||
# 存储配置对象
|
||||
self.config = config
|
||||
|
||||
# 从配置获取超时时间，默认15分钟（适应7分钟平均生成时间）
|
||||
timeout = config.api_timeout if config and hasattr(config, 'api_timeout') else 900.0
|
||||
|
||||
self.base_url = (
|
||||
base_url
|
||||
or (getattr(self.config, 'gemini_base_url', None) if self.config else None)
|
||||
or os.getenv('GEMINI_BASE_URL')
|
||||
or DEFAULT_GEMINI_BASE_URL
|
||||
)
|
||||
|
||||
# 创建针对此实例的重试配置
|
||||
self.retry_config = create_report_retry_config(config)
|
||||
|
||||
# 初始化OpenAI客户端，使用Gemini的中转endpoint
|
||||
# 专门为报告生成设置长超时（15分钟），适应7分钟平均生成时间
|
||||
self.client = OpenAI(
|
||||
api_key=self.api_key,
|
||||
base_url=self.base_url,
|
||||
timeout=timeout
|
||||
)
|
||||
|
||||
self.default_model = model_name or self.get_default_model()
|
||||
|
||||
def get_default_model(self) -> str:
|
||||
"""获取默认模型名称"""
|
||||
return "gemini-2.5-pro"
|
||||
|
||||
def _make_api_call(self, system_prompt: str, user_prompt: str, **kwargs) -> str:
|
||||
"""
|
||||
内部API调用方法
|
||||
|
||||
Args:
|
||||
system_prompt: 系统提示词
|
||||
user_prompt: 用户输入
|
||||
**kwargs: 其他参数
|
||||
|
||||
Returns:
|
||||
API响应内容
|
||||
"""
|
||||
# 构建消息
|
||||
messages = [
|
||||
{"role": "system", "content": system_prompt},
|
||||
{"role": "user", "content": user_prompt}
|
||||
]
|
||||
|
||||
# 设置默认参数
|
||||
params = {
|
||||
"model": self.default_model,
|
||||
"messages": messages,
|
||||
"temperature": kwargs.get("temperature", 0.7),
|
||||
"max_tokens": kwargs.get("max_tokens", 50000),
|
||||
"stream": False
|
||||
}
|
||||
|
||||
# 调用API
|
||||
response = self.client.chat.completions.create(**params)
|
||||
|
||||
# 提取回复内容
|
||||
if response.choices and response.choices[0].message:
|
||||
content = response.choices[0].message.content
|
||||
return self.validate_response(content)
|
||||
else:
|
||||
return ""
|
||||
|
||||
def invoke(self, system_prompt: str, user_prompt: str, **kwargs) -> str:
|
||||
"""
|
||||
调用Gemini API生成回复（带动态重试配置）
|
||||
|
||||
Args:
|
||||
system_prompt: 系统提示词
|
||||
user_prompt: 用户输入
|
||||
**kwargs: 其他参数，如temperature、max_tokens等
|
||||
|
||||
Returns:
|
||||
Gemini生成的回复文本
|
||||
"""
|
||||
import time
|
||||
|
||||
last_exception = None
|
||||
|
||||
for attempt in range(self.retry_config.max_retries + 1):
|
||||
try:
|
||||
result = self._make_api_call(system_prompt, user_prompt, **kwargs)
|
||||
if attempt > 0:
|
||||
print(f"Report Engine Gemini API在第 {attempt + 1} 次尝试后成功")
|
||||
return result
|
||||
|
||||
except Exception as e:
|
||||
last_exception = e
|
||||
|
||||
if attempt == self.retry_config.max_retries:
|
||||
print(f"Report Engine Gemini API在 {self.retry_config.max_retries + 1} 次尝试后仍然失败")
|
||||
print(f"最终错误: {str(e)}")
|
||||
raise e
|
||||
|
||||
# 计算延迟时间
|
||||
delay = min(
|
||||
self.retry_config.initial_delay * (self.retry_config.backoff_factor ** attempt),
|
||||
self.retry_config.max_delay
|
||||
)
|
||||
|
||||
print(f"Report Engine Gemini API第 {attempt + 1} 次尝试失败: {str(e)}")
|
||||
print(f"将在 {delay:.1f} 秒后进行第 {attempt + 2} 次尝试...")
|
||||
|
||||
time.sleep(delay)
|
||||
|
||||
# 这里不应该到达，但作为安全网
|
||||
if last_exception:
|
||||
raise last_exception
|
||||
|
||||
def get_model_info(self) -> Dict[str, Any]:
|
||||
"""
|
||||
获取当前模型信息
|
||||
|
||||
Returns:
|
||||
模型信息字典
|
||||
"""
|
||||
return {
|
||||
"provider": "Gemini",
|
||||
"model": self.default_model,
|
||||
"api_base": self.base_url,
|
||||
"purpose": "Report Generation"
|
||||
}
|
||||
Reference in New Issue
Block a user