sttFunctionCallBackend/llmagent/llm_agent.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Time    : 2025/3/16 09:46
# @Author  : old-tom
# @File    : llm_agent
# @Project : llmFunctionCallDemo
# @Desc    : 已图的方式构建代理,参考 https://github.langchain.ac.cn/langgraph/tutorials/introduction/#part-1-build-a-basic-chatbot
from typing import Annotated

from langchain_core.messages import AnyMessage
from typing_extensions import TypedDict
from langgraph.graph import StateGraph, START, END
from langgraph.graph.message import add_messages
from llmagent.llm_config import LLMConfigLoader
from langchain_openai import ChatOpenAI
from llmtools.tool_impl import tool_node, tools
from langgraph.prebuilt import tools_condition
from langgraph.checkpoint.memory import MemorySaver

memory = MemorySaver()

# 初始化LLM模型
llm_conf = LLMConfigLoader.load(item_name='siliconflow')
llm = ChatOpenAI(
    model=llm_conf.model, api_key=llm_conf.api_key,
    base_url=llm_conf.base_url, max_tokens=llm_conf.max_tokens,
    temperature=llm_conf.temperature,
    streaming=llm_conf.streaming
)
llm_with_tools = llm.bind_tools(tools)


class State(TypedDict):
    """
    图的状态机
    add_messages 函数会自动合并message到一个list中，例如HumanMessage\AIMessage
    """
    messages: Annotated[list[AnyMessage], add_messages]


graph_builder = StateGraph(State)


def chat(state: State):
    """
    LLM单轮对话
    :param state: 状态机
    LLM需要从状态机获取message
    :return:
    """
    return {"messages": [llm_with_tools.invoke(state["messages"])]}


graph_builder.add_node("chat_llm", chat)
graph_builder.add_node("tools", tool_node)
graph_builder.add_edge(START, "chat_llm")
graph_builder.add_edge("chat_llm", END)
graph_builder.add_conditional_edges("chat_llm", tools_condition)
graph_builder.add_edge("tools", "chat_llm")
graph = graph_builder.compile(checkpointer=memory)


def stream_graph_updates(user_input: str):
    config = {"configurable": {"thread_id": "1"}}
    events = graph.stream(
        {"messages": [{"role": "user", "content": user_input}]},
        config,
        stream_mode="values"
    )
    for event in events:
        event["messages"][-1].pretty_print()


while True:
    try:
        user_input = input("User: ")
        if user_input.lower() in ["quit", "exit", "q"]:
            print("Goodbye!")
            break

        stream_graph_updates(user_input)
    except:
        # fallback if input() is not available
        user_input = "What do you know about LangGraph?"
        print("User: " + user_input)
        stream_graph_updates(user_input)
        break