LangGraph - 技术栈

1.简单调用

python 复制代码

import operator
import os
from typing import TypedDict, Literal

from langchain_community.chat_models.tongyi import ChatTongyi
from langchain_core.messages import SystemMessage, HumanMessage, ToolMessage, AnyMessage
from langchain_core.tools import tool
from langgraph.graph import StateGraph, START, END
from langgraph.prebuilt import ToolNode
from typing import Annotated


# 1. 定义工具
@tool
def multiply(a: int, b: int) -> int:
    """求两个数的乘积
    参数：a:第一个int类型
    参数：b:第二个int类型
    """
    return a * b

@tool
def divide(a: int, b: int) -> int:
    """求两个数的商"""
    return a / b

# 工具列表 + 工具字典
tools = [multiply, divide]
tools_by_name = {tool.name: tool for tool in tools}
'''结构：key是工具的名字，value是真正的tool
{
    "multiply": multiply工具函数,
    "divide": divide工具函数
}  
'''

# 2. 模型 + 绑定工具
model = ChatTongyi(
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    model_name="qwen-plus-2025-12-01"
)
model = model.bind_tools(tools)

# ------------------------------
# 3. 定义状态（关键修复）
# ------------------------------
class State(TypedDict):
    # 消息：自动追加
    messages: Annotated[list[AnyMessage], operator.add]
    llm_calls: int

# ------------------------------
# 4. LLM 节点（修复逻辑）
# ------------------------------
def llm_call(state: State):
    """模型思考：是否调用工具"""
    return {
        "messages": [model.invoke(state["messages"])],
        "llm_calls": state.get("llm_calls", 0) + 1
    }

'''
tool_calls 结构：
[
    {   
        "name": "工具名字",
        "args": { 参数字典 },
        "id": "工具调用唯一ID"
    }
]
'''

# ------------------------------
# 5. 工具执行节点（你写的那个！）
# ------------------------------
def tool_node(state: State):
    """执行工具，返回 ToolMessage"""
    result = []
    last_message = state["messages"][-1]

    for tool_call in last_message.tool_calls:
        # 遍历模型让我调用的每一个工具
        tool = tools_by_name[tool_call["name"]]  # 通过名字得到真正的函数
        observation = tool.invoke(tool_call["args"])
        # 把结果变成 ToolMessage
        result.append(
            ToolMessage(
                content=observation,
                tool_call_id=tool_call["id"]
            )
        )

    # 返回结果 → 自动追加到 messages 里
    return {"messages": result}

# ------------------------------
# 6. 条件判断：是否继续调用工具
# ------------------------------
def should_continue(state: State) -> Literal["tool_node", END]:
    messages = state["messages"]
    last_message = messages[-1]

    # 如果有工具调用 → 去执行工具
    if last_message.tool_calls:
        return "tool_node"

    # 没有 → 结束
    return END

# ------------------------------
# 7. 构建流程图（全部修复）
# ------------------------------
workflow = StateGraph(State)

# 添加节点
workflow.add_node("llm_call", llm_call)
workflow.add_node("tool_node", tool_node)

# 流程
workflow.add_edge(START, "llm_call")

# 条件边：模型 → 工具 or 结束
workflow.add_conditional_edges(
    "llm_call",
    should_continue
)

# 工具执行完 → 回到模型
workflow.add_edge("tool_node", "llm_call")

# 编译
graph = workflow.compile()

# ------------------------------
# 8. 运行！
# ------------------------------
if __name__ == "__main__":
    final_state = graph.invoke({
        "messages": [HumanMessage(content="123456789 乘 987654321 等于多少？")],
        "llm_calls": 0
    })

    # 打印所有消息
    print("\n===== 完整对话记录 =====")
    for msg in final_state["messages"]:
        msg.pretty_print()

2.加记忆

加入记忆很简单，就是在正常的流程里面首先设置一个 checkpointer，然后编译的时候指定即可，然后调用的时候，config: RunnableConfig = {"configurable": {"thread_id": "1"}} 这种是固定的形式

python 复制代码

from langgraph.graph import StateGraph, START, END
from langgraph.checkpoint.memory import InMemorySaver
from langchain_core.runnables import RunnableConfig
from typing import Annotated
from typing_extensions import TypedDict
from operator import add

class State(TypedDict):
    foo: str
    bar: Annotated[list[str], add]

def node_a(state: State):
    return {"foo": "a", "bar": ["a"]}

def node_b(state: State):
    return {"foo": "b", "bar": ["b"]}


workflow = StateGraph(State)
workflow.add_node(node_a)
workflow.add_node(node_b)
workflow.add_edge(START, "node_a")
workflow.add_edge("node_a", "node_b")
workflow.add_edge("node_b", END)

checkpointer = InMemorySaver()
graph = workflow.compile(checkpointer=checkpointer)

config: RunnableConfig = {"configurable": {"thread_id": "1"}}
graph.invoke({"foo": "", "bar":[]}, config)