2024-06-20 langchain 笔记01

# https://python.langchain.com/v0.2/docs/tutorials/llm_chain/
import os
import dotenv
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage, SystemMessage
from langchain_core.output_parsers import StrOutputParser

dotenv.load_dotenv()
base_url = os.environ["base_url"]
api_key = os.environ["api_key"]


def sample_01():
    """
    最基础的调用openai的方法
    """
    # 构建gpt
    model = ChatOpenAI(base_url=base_url, api_key=api_key, model="gpt-3.5-turbo")
    messages = [
        SystemMessage(content="Translate the following from English into Chinese"),
        HumanMessage(content="Hi!"),
    ]
    # 返回　AIMessage格式的
    ai_resp = model.invoke(messages)

    # 提取 AIMessage 中的 content 信息
    parser = StrOutputParser()
    msg = parser.invoke(ai_resp)
    print(msg)

    # 把上面两种对象使用管理串起来，变成一个chain
    chain = model | parser
    print(chain.invoke(messages))


from langchain_core.prompts import ChatPromptTemplate


def sample_02():
    model = ChatOpenAI(base_url=base_url, api_key=api_key, model="gpt-3.5-turbo")
    
    # 构建一个系统role 的prompt模板
    system_template = "Translate the following into {language}:"
    
    # 生成带变量的 system＆user prompt 对话模板
    prompt_template = ChatPromptTemplate.from_messages(
        [("system", system_template), ("user", "{text}")]
    )
    
    # 代入变量，得到最后完整的 prompt对象
    result = prompt_template.invoke({"language":"italian", "text":"hi"})
    
    print(result)
    print(result.to_messages)
    
    parser = StrOutputParser()
    
    # chain 
    chain = prompt_template | model | parser
    
    print(chain.invoke({"language": "italian", "text": "hi"}))

if __name__ == "__main__":
    sample_02()