c3560b43fd
- Added documentation for audit tracking (IP address, invocation method). - Updated database model descriptions for enhanced WorkOrder and Conversation fields. - Documented the new UnifiedConfig system. - Reflected enhanced logging transparency for knowledge base parsing. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
260 lines
9.0 KiB
Python
260 lines
9.0 KiB
Python
|
||
# -*- coding: utf-8 -*-
|
||
"""
|
||
大模型客户端 - 统一的LLM接口
|
||
支持多种大模型提供商
|
||
"""
|
||
|
||
import logging
|
||
import asyncio
|
||
import json
|
||
from typing import Dict, Any, Optional, List
|
||
from abc import ABC, abstractmethod
|
||
from dataclasses import dataclass
|
||
|
||
from src.config.unified_config import get_config
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
@dataclass
|
||
class LLMConfig:
|
||
"""LLM配置"""
|
||
provider: str
|
||
api_key: str
|
||
base_url: Optional[str] = None
|
||
model: str = "gpt-3.5-turbo"
|
||
temperature: float = 0.7
|
||
max_tokens: int = 2000
|
||
|
||
class BaseLLMClient(ABC):
|
||
"""LLM客户端基类"""
|
||
|
||
@abstractmethod
|
||
async def generate(self, prompt: str, **kwargs) -> str:
|
||
"""生成文本"""
|
||
pass
|
||
|
||
@abstractmethod
|
||
async def chat(self, messages: List[Dict[str, str]], **kwargs) -> str:
|
||
"""对话生成"""
|
||
pass
|
||
|
||
class OpenAIClient(BaseLLMClient):
|
||
"""OpenAI客户端 - 支持OpenAI和兼容OpenAI API的模型（如千问）"""
|
||
|
||
def __init__(self, config: LLMConfig):
|
||
self.config = config
|
||
self.client = None
|
||
self._init_client()
|
||
|
||
def _init_client(self):
|
||
"""初始化客户端"""
|
||
try:
|
||
import openai
|
||
self.client = openai.AsyncOpenAI(
|
||
api_key=self.config.api_key,
|
||
base_url=self.config.base_url
|
||
)
|
||
except ImportError:
|
||
logger.warning("OpenAI库未安装，将使用模拟客户端")
|
||
self.client = None
|
||
|
||
async def generate(self, prompt: str, **kwargs) -> str:
|
||
"""生成文本"""
|
||
if not self.client:
|
||
raise ImportError("OpenAI client not initialized. Please install the 'openai' package.")
|
||
|
||
try:
|
||
response = await self.client.chat.completions.create(
|
||
model=self.config.model,
|
||
messages=[{"role": "user", "content": prompt}],
|
||
temperature=kwargs.get("temperature", self.config.temperature),
|
||
max_tokens=kwargs.get("max_tokens", self.config.max_tokens)
|
||
)
|
||
return response.choices[0].message.content
|
||
except Exception as e:
|
||
logger.error(f"OpenAI API调用失败: {e}")
|
||
raise e
|
||
|
||
async def chat(self, messages: List[Dict[str, str]], **kwargs) -> str:
|
||
"""对话生成"""
|
||
if not self.client:
|
||
raise ImportError("OpenAI client not initialized. Please install the 'openai' package.")
|
||
|
||
try:
|
||
response = await self.client.chat.completions.create(
|
||
model=self.config.model,
|
||
messages=messages,
|
||
temperature=kwargs.get("temperature", self.config.temperature),
|
||
max_tokens=kwargs.get("max_tokens", self.config.max_tokens)
|
||
)
|
||
return response.choices[0].message.content
|
||
except Exception as e:
|
||
logger.error(f"OpenAI Chat API调用失败: {e}")
|
||
raise e
|
||
|
||
def _simulate_response(self, prompt: str) -> str:
|
||
"""模拟响应"""
|
||
if "千问" in self.config.model or "qwen" in self.config.model.lower():
|
||
return f"【千问模型模拟响应】根据您的问题，我建议采取以下措施：{prompt[:50]}... 这是一个智能化的解决方案。"
|
||
return f"模拟LLM响应: {prompt[:100]}..."
|
||
|
||
def _simulate_chat(self, messages: List[Dict[str, str]]) -> str:
|
||
"""模拟对话响应"""
|
||
last_message = messages[-1]["content"] if messages else ""
|
||
if "千问" in self.config.model or "qwen" in self.config.model.lower():
|
||
return f"【千问模型模拟对话】我理解您的问题：{last_message[:50]}... 让我为您提供专业的建议。"
|
||
return f"模拟对话响应: {last_message[:100]}..."
|
||
|
||
class AnthropicClient(BaseLLMClient):
|
||
"""Anthropic客户端"""
|
||
|
||
def __init__(self, config: LLMConfig):
|
||
self.config = config
|
||
self.client = None
|
||
self._init_client()
|
||
|
||
def _init_client(self):
|
||
"""初始化客户端"""
|
||
try:
|
||
import anthropic
|
||
self.client = anthropic.AsyncAnthropic(
|
||
api_key=self.config.api_key
|
||
)
|
||
except ImportError:
|
||
logger.warning("Anthropic库未安装，将使用模拟客户端")
|
||
self.client = None
|
||
|
||
async def generate(self, prompt: str, **kwargs) -> str:
|
||
"""生成文本"""
|
||
if not self.client:
|
||
return self._simulate_response(prompt)
|
||
|
||
try:
|
||
response = await self.client.messages.create(
|
||
model=self.config.model,
|
||
max_tokens=kwargs.get("max_tokens", self.config.max_tokens),
|
||
temperature=kwargs.get("temperature", self.config.temperature),
|
||
messages=[{"role": "user", "content": prompt}]
|
||
)
|
||
return response.content[0].text
|
||
except Exception as e:
|
||
logger.error(f"Anthropic API调用失败: {e}")
|
||
return self._simulate_response(prompt)
|
||
|
||
async def chat(self, messages: List[Dict[str, str]], **kwargs) -> str:
|
||
"""对话生成"""
|
||
if not self.client:
|
||
return self._simulate_chat(messages)
|
||
|
||
try:
|
||
response = await self.client.messages.create(
|
||
model=self.config.model,
|
||
max_tokens=kwargs.get("max_tokens", self.config.max_tokens),
|
||
temperature=kwargs.get("temperature", self.config.temperature),
|
||
messages=messages
|
||
)
|
||
return response.content[0].text
|
||
except Exception as e:
|
||
logger.error(f"Anthropic Chat API调用失败: {e}")
|
||
return self._simulate_chat(messages)
|
||
|
||
def _simulate_response(self, prompt: str) -> str:
|
||
"""模拟响应"""
|
||
return f"模拟Anthropic响应: {prompt[:100]}..."
|
||
|
||
def _simulate_chat(self, messages: List[Dict[str, str]]) -> str:
|
||
"""模拟对话响应"""
|
||
last_message = messages[-1]["content"] if messages else ""
|
||
return f"模拟Anthropic对话: {last_message[:100]}..."
|
||
|
||
class LocalLLMClient(BaseLLMClient):
|
||
"""本地LLM客户端"""
|
||
|
||
def __init__(self, config: LLMConfig):
|
||
self.config = config
|
||
self.client = None
|
||
self._init_client()
|
||
|
||
def _init_client(self):
|
||
"""初始化本地客户端"""
|
||
try:
|
||
# 这里可以集成Ollama、vLLM等本地LLM服务
|
||
logger.info("本地LLM客户端初始化")
|
||
except Exception as e:
|
||
logger.warning(f"本地LLM客户端初始化失败: {e}")
|
||
|
||
async def generate(self, prompt: str, **kwargs) -> str:
|
||
"""生成文本"""
|
||
# 实现本地LLM调用
|
||
return f"本地LLM响应: {prompt[:100]}..."
|
||
|
||
async def chat(self, messages: List[Dict[str, str]], **kwargs) -> str:
|
||
"""对话生成"""
|
||
last_message = messages[-1]["content"] if messages else ""
|
||
return f"本地LLM对话: {last_message[:100]}..."
|
||
|
||
class LLMClientFactory:
|
||
"""LLM客户端工厂"""
|
||
|
||
@staticmethod
|
||
def create_client(config: LLMConfig) -> BaseLLMClient:
|
||
"""创建LLM客户端"""
|
||
provider = config.provider.lower()
|
||
|
||
# qwen 使用 OpenAI 兼容的 API
|
||
if provider in ["openai", "qwen"]:
|
||
return OpenAIClient(config)
|
||
elif provider == "anthropic":
|
||
return AnthropicClient(config)
|
||
elif provider == "local":
|
||
return LocalLLMClient(config)
|
||
else:
|
||
raise ValueError(f"不支持的LLM提供商: {config.provider}")
|
||
|
||
class LLMManager:
|
||
"""LLM管理器"""
|
||
|
||
def __init__(self, config=None):
|
||
if config:
|
||
self.config = config
|
||
else:
|
||
# If no config is provided, fetch it from the unified config system
|
||
self.config = get_config().llm
|
||
|
||
self.client = LLMClientFactory.create_client(self.config)
|
||
self.usage_stats = {
|
||
"total_requests": 0,
|
||
"total_tokens": 0,
|
||
"error_count": 0
|
||
}
|
||
|
||
|
||
async def generate(self, prompt: str, **kwargs) -> str:
|
||
"""生成文本"""
|
||
try:
|
||
self.usage_stats["total_requests"] += 1
|
||
response = await self.client.generate(prompt, **kwargs)
|
||
self.usage_stats["total_tokens"] += len(response)
|
||
return response
|
||
except Exception as e:
|
||
self.usage_stats["error_count"] += 1
|
||
logger.error(f"LLM生成失败: {e}")
|
||
raise
|
||
|
||
async def chat(self, messages: List[Dict[str, str]], **kwargs) -> str:
|
||
"""对话生成"""
|
||
try:
|
||
self.usage_stats["total_requests"] += 1
|
||
response = await self.client.chat(messages, **kwargs)
|
||
self.usage_stats["total_tokens"] += len(response)
|
||
return response
|
||
except Exception as e:
|
||
self.usage_stats["error_count"] += 1
|
||
logger.error(f"LLM对话失败: {e}")
|
||
raise
|
||
|
||
def get_usage_stats(self) -> Dict[str, Any]:
|
||
"""获取使用统计"""
|
||
return self.usage_stats.copy()
|