Skip to content

如何与部署进行交互

RemoteGraph 是一个接口,允许您与您的 LangGraph Platform 部署进行交互,就像与普通、本地定义的 LangGraph 图(例如 CompiledGraph)一样。本指南将展示如何初始化 RemoteGraph 并与之交互。

初始化图表

初始化 RemoteGraph 时,您必须始终指定:

  • name: 您要交互的图表的名称。这与您在部署的 langgraph.json 配置文件中使用的图表名称相同。
  • api_key: 一个有效的 LangSmith API 密钥。可以设置为环境变量 (LANGSMITH_API_KEY) 或通过 api_key 参数直接传递。如果 LangGraphClient / SyncLangGraphClient 使用 api_key 参数初始化,也可以通过 client / sync_client 参数提供 API 密钥。

此外,您必须提供以下选项之一:

  • url: 您要交互的部署的 URL。如果您传递 url 参数,则将使用提供的 URL、标头(如果提供)和默认配置值(例如,超时等)创建同步和异步客户端。
  • client: 一个 LangGraphClient 实例,用于与部署进行异步交互(例如,使用 .astream(), .ainvoke(), .aget_state(), .aupdate_state() 等)。
  • sync_client: 一个 SyncLangGraphClient 实例,用于与部署进行同步交互(例如,使用 .stream(), .invoke(), .get_state(), .update_state() 等)。

Note

如果您同时传递 clientsync_clienturl 参数,它们将优先于 url 参数。如果未提供 client / sync_client / url 参数中的任何一个,RemoteGraph 将在运行时引发 ValueError

使用 URL

from langgraph.pregel.remote import RemoteGraph

url = <DEPLOYMENT_URL>
graph_name = "agent"
remote_graph = RemoteGraph(graph_name, url=url)
import { RemoteGraph } from "@langchain/langgraph/remote";

const url = `<DEPLOYMENT_URL>`;
const graphName = "agent";
const remoteGraph = new RemoteGraph({ graphId: graphName, url });

使用客户端

from langgraph_sdk import get_client, get_sync_client
from langgraph.pregel.remote import RemoteGraph

url = <DEPLOYMENT_URL>
graph_name = "agent"
client = get_client(url=url)
sync_client = get_sync_client(url=url)
remote_graph = RemoteGraph(graph_name, client=client, sync_client=sync_client)
import { Client } from "@langchain/langgraph-sdk";
import { RemoteGraph } from "@langchain/langgraph/remote";

const client = new Client({ apiUrl: `<DEPLOYMENT_URL>` });
const graphName = "agent";
const remoteGraph = new RemoteGraph({ graphId: graphName, client });

调用图表

由于 RemoteGraph 是一个实现了与 CompiledGraph 相同方法的 Runnable,因此您可以像平常使用编译后的图表一样与它进行交互,即通过调用 .invoke(), .stream(), .get_state(), .update_state() 等(以及它们的异步对应项)。

异步调用

Note

要异步使用图表,您必须在初始化 RemoteGraph 时提供 urlclient

# 调用图表
result = await remote_graph.ainvoke({
    "messages": [{"role": "user", "content": "what's the weather in sf"}]
})

# 流式输出图表
async for chunk in remote_graph.astream({
    "messages": [{"role": "user", "content": "what's the weather in la"}]
}):
    print(chunk)
// 调用图表
const result = await remoteGraph.invoke({
    messages: [{role: "user", content: "what's the weather in sf"}]
})

// 流式输出图表
for await (const chunk of await remoteGraph.stream({
    messages: [{role: "user", content: "what's the weather in la"}]
})):
    console.log(chunk)

同步调用

Note

要同步使用图表,您必须在初始化 RemoteGraph 时提供 urlsync_client

# 调用图表
result = remote_graph.invoke({
    "messages": [{"role": "user", "content": "what's the weather in sf"}]
})

# 流式输出图表
for chunk in remote_graph.stream({
    "messages": [{"role": "user", "content": "what's the weather in la"}]
}):
    print(chunk)

按线程级别持久化

默认情况下,图表运行(即 .invoke().stream() 调用)是无状态的 - 图表的检查点和最终状态不会被持久化。如果您想持久化图表运行的输出(例如,启用人工介入功能),您可以创建一个线程并通过 config 参数提供线程 ID,就像您对普通的编译图表一样:

from langgraph_sdk import get_sync_client
url = <DEPLOYMENT_URL>
graph_name = "agent"
sync_client = get_sync_client(url=url)
remote_graph = RemoteGraph(graph_name, url=url)

# 创建一个线程(或使用现有线程)
thread = sync_client.threads.create()

# 使用线程配置调用图表
config = {"configurable": {"thread_id": thread["thread_id"]}}
result = remote_graph.invoke({
    "messages": [{"role": "user", "content": "what's the weather in sf"}]
}, config=config)

# 验证状态是否已持久化到线程
thread_state = remote_graph.get_state(config)
print(thread_state)
import { Client } from "@langchain/langgraph-sdk";
import { RemoteGraph } from "@langchain/langgraph/remote";

const url = `<DEPLOYMENT_URL>`;
const graphName = "agent";
const client = new Client({ apiUrl: url });
const remoteGraph = new RemoteGraph({ graphId: graphName, url });

// 创建一个线程(或使用现有线程)
const thread = await client.threads.create();

// 使用线程配置调用图表
const config = { configurable: { thread_id: thread.thread_id }};
const result = await remoteGraph.invoke({
  messages: [{ role: "user", content: "what's the weather in sf" }],
}, config);

// 验证状态是否已持久化到线程
const threadState = await remoteGraph.getState(config);
console.log(threadState);

作为子图使用

Note

如果需要将 RemoteGraph 用作具有 checkpointer 的图表的子图节点,请确保使用 UUID 作为线程 ID。

由于 RemoteGraph 的行为与普通 CompiledGraph 相同,它也可以用作另一个图表中的子图。例如:

from langgraph_sdk import get_sync_client
from langgraph.graph import StateGraph, MessagesState, START
from typing import TypedDict

url = <DEPLOYMENT_URL>
graph_name = "agent"
remote_graph = RemoteGraph(graph_name, url=url)

# 定义父图
builder = StateGraph(MessagesState)
# 直接添加远程图作为节点
builder.add_node("child", remote_graph)
builder.add_edge(START, "child")
graph = builder.compile()

# 调用父图
result = graph.invoke({
    "messages": [{"role": "user", "content": "what's the weather in sf"}]
})
print(result)

# 流式输出父图和子图
for chunk in graph.stream({
    "messages": [{"role": "user", "content": "what's the weather in sf"}]
}, subgraphs=True):
    print(chunk)
import { MessagesAnnotation, StateGraph, START } from "@langchain/langgraph";
import { RemoteGraph } from "@langchain/langgraph/remote";

const url = `<DEPLOYMENT_URL>`;
const graphName = "agent";
const remoteGraph = new RemoteGraph({ graphId: graphName, url });

// 定义父图并将远程图直接添加为节点
const graph = new StateGraph(MessagesAnnotation)
  .addNode("child", remoteGraph)
  .addEdge(START, "child")
  .compile()

// 调用父图
const result = await graph.invoke({
  messages: [{ role: "user", content: "what's the weather in sf" }]
});
console.log(result);

// 流式输出父图和子图
for await (const chunk of await graph.stream({
  messages: [{ role: "user", content: "what's the weather in la" }]
}, { subgraphs: true })) {
  console.log(chunk);
}