初识LangChainLangChain 小试牛刀 LangServe 使用方法 windows下使用Jupyter n

LangChain 小试牛刀

1、使用Jupyter notebook进行学习

注：本篇使用的是windows10系统，cmd命令均在jupyter -> 新建terminals中运行，可通过Setting->Theme->Dark修改主题方便使用

安装Anaconda后找到Jupyter快捷方式并查看属性，复制目标输入框内的内容

局部截取_20250705_105450.png

D:\Anaconda3\python.exe D:\Anaconda3\cwp.py D:\Anaconda3 D:\Anaconda3\python.exe D:\Anaconda3\Scripts\jupyter-notebook-script.py 【此处替换你的目录】

例如：

D:\Anaconda3\python.exe D:\Anaconda3\cwp.py D:\Anaconda3 D:\Anaconda3\python.exe D:\Anaconda3\Scripts\jupyter-notebook-script.py E:\langchain

在命令行中执行，将会从E:\langchain下启动Jupyter

2、安装依赖包因为我使用的python3，下方均使用pip3进行安装

新建Terminal窗口，并运行安装命令局部截取_20250705_110728.png

pip3 install langchain

也可以在ipynb中安装，前面加上!即可，之后不再赘述

!pip3 install langchain

3、禁用LangSmith

新建Notebook 局部截取_20250705_110551.png

编写ipynb代码禁用LangSmith（前期学习不建议使用）

import getpass  
import os  
# tracing_context块中的代码调用可禁用smith上传
from langsmith import tracing_context

4、调用语言模型（这里使用火山方舟API）

pip3 install -qU langchain-openai

from langchain_openai import ChatOpenAI

# 设置 API Key 和 API Base
os.environ["OPENAI_API_KEY"] = getpass.getpass() # 本地使用可直接写入明文
os.environ["OPENAI_API_BASE"] = "https://ark.cn-beijing.volces.com/api/v3"

model = ChatOpenAI(model="ep-***") # 模型推理点名称

测试调用

from langchain_core.messages import HumanMessage, SystemMessage

messages = [
    SystemMessage(content="Translate the following from English into Chinese"),
    HumanMessage(content="hi!"),
]
with tracing_context(enabled=False):
    result = model.invoke(messages)
    print(result)

局部截取_20250705_132126.png

5、使用Parser

使用parser提取结果（1）

from langchain_core.output_parsers import StrOutputParser
from langchain_core.messages import HumanMessage, SystemMessage

parser = StrOutputParser()

messages = [
    SystemMessage(content="Translate the following from English into Chinese"),
    HumanMessage(content="hi!"),
]

with tracing_context(enabled=False):
    result = model.invoke(messages)
    parserResult = parser.invoke(result)
    print(parserResult)

局部截取_20250705_134714.png

或使用链式调用parser（2）

from langchain_core.output_parsers import StrOutputParser
from langchain_core.messages import HumanMessage, SystemMessage

parser = StrOutputParser()

messages = [
    SystemMessage(content="Translate the following from English into Chinese"),
    HumanMessage(content="hi!"),
]

with tracing_context(enabled=False):
    chain = model | parser
    parserResult = chain.invoke(messages)
    print(parserResult)

局部截取_20250705_134714.png

6、使用提示词模板

from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
parser = StrOutputParser()

system_template = "Translate the following into {language}:"

prompt_template = ChatPromptTemplate.from_messages(
    [("system", system_template), ("user", "{text}")]
)

with tracing_context(enabled=False):
    # 用于生成提示词模板
    prompt_result = prompt_template.invoke({"language": "chinese", "text": "hi"})
    print(prompt_result.to_messages())
    
    # 链式调用
    chain = prompt_template | model | parser
    
    # invoke传入参数key需要与prompt_template定义的参数key保持一致
    result = chain.invoke({"language": "chinese", "text": "hello world!"})
    print(result)

局部截取_20250705_140516.png

7、使用LangServe部署一个应用服务器

安装依赖包

pip3 install "langserve[all]"

新建python文件serve.py

import os
import getpass
from fastapi import FastAPI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langserve import add_routes

# 1. Create prompt template
system_template = "Translate the following into {language}:"
prompt_template = ChatPromptTemplate.from_messages([
    ('system', system_template),
    ('user', '{text}')
])


# 设置 API Key 和 API Base
os.environ["OPENAI_API_KEY"] = getpass.getpass() # 本地使用可直接写入明文
os.environ["OPENAI_API_BASE"] = "https://ark.cn-beijing.volces.com/api/v3"  # 假设是这个地址，需确认
os.environ["OPENAI_API_TYPE"] = "openai"  # 通常不用设置，除非特殊情况

# 2. Create model
from langchain_openai import ChatOpenAI
model = ChatOpenAI(model="ep-***") # 模型推理点名称

# 3. Create parser
parser = StrOutputParser()

# 4. Create chain
chain = prompt_template | model | parser


# 4. App definition
app = FastAPI(
  title="LangChain Server",
  version="1.0",
  description="A simple API server using LangChain's Runnable interfaces",
)

# 5. Adding chain route
add_routes(
    app,
    chain,
    path="/chain",
)

if __name__ == "__main__":
    import uvicorn

    uvicorn.run(app, host="localhost", port=8000)

执行

python serve.py

访问 http://localhost:8000/chain/playground/ 即可使用

http://localhost:8000/docs 查看接口文档

局部截取_20250705_143512.png

8、服务启动后，即可在其他应用代码中直接请求调用

from langserve import RemoteRunnable

remote_chain = RemoteRunnable("http://localhost:8000/chain")
result = remote_chain.invoke({"language": "chinese", "text": "hello world"})
print(result)

或者使用requests

import requests

url = "http://localhost:8000/chain/invoke"
payload = {
    "input": {
        "language": "chinese",
        "text": "hello world"
    }
}
headers = {
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers).json()
print(response['output'])

初识LangChain