11. 对话式 RAG:为检索添加记忆

from langchain_openai import ChatOpenAI
from langchain_core.messages import SystemMessage, HumanMessage, AIMessage
 
llm = ChatOpenAI(model="gpt-5-mini")
 
system_prompt = (
    "Given a chat history and the latest user question "
    "which might reference context in the chat history, "
    "formulate a standalone question "
    "which can be understood without the chat history. "
    "Do NOT answer the question, just reformulate it if needed "
    "and otherwise return it as is."
)

messages = [
    SystemMessage(content=system_prompt),
    # 聊天历史
    HumanMessage(content="What is your refund policy?"),
    AIMessage(content="All physical products may be returned within 30 days of purchase for a full refund."),
    # 追问
    HumanMessage(content="Does that apply to digital products too?"),
]
 
response = llm.invoke(messages)
print(response.content)

数字产品符合退款条件吗?

from langchain.tools import tool
from langchain_openai import OpenAIEmbeddings
from langchain_chroma import Chroma
 
# 连接到第 10 章中构建的向量存储
embedding_model = OpenAIEmbeddings(model="text-embedding-3-small")
vector_store = Chroma(
    persist_directory="data/chroma_db",
    collection_name="company_docs",
    embedding_function=embedding_model,
)
 
@tool
def retrieve_context(query: str):
    """在文档中搜索与查询相关的内容。"""
    retrieved_docs = vector_store.similarity_search(query, k=3)
    serialized = "\n\n".join(
        f"Source: {doc.metadata['source']}\nContent: {doc.page_content}"
        for doc in retrieved_docs
    )
    return serialized

from langchain.agents import create_agent
from langgraph.checkpoint.memory import InMemorySaver  # 随 langchain 自动安装
 
agent = create_agent(
    model="gpt-5-mini",
    tools=[retrieve_context],
    system_prompt=(
        "You are a helpful assistant that answers questions about company policies. "
        "Use the retrieve_context tool to search for relevant information. "
        "If the retrieved context does not contain relevant information, "
        "say that you don't know. "
        "Keep the answer concise, three sentences maximum."
    ),
    checkpointer=InMemorySaver(),
)

# thread_id 是区分对话的标识符
# 使用相同的 thread_id 会延续同一段对话
thread_config = {"configurable": {"thread_id": "1"}}
 
# --- 第 1 轮:一个完整的问题 ---
response1 = agent.invoke(
    {"messages": [{"role": "user", "content": "What is your refund policy?"}]},
    thread_config,
)
print("Q: What is your refund policy?")
print("A:", response1["messages"][-1].content)
 
# --- 第 2 轮:一个依赖于第 1 轮的追问 ---
response2 = agent.invoke(
    {"messages": [{"role": "user", "content": "Does that apply to digital products too?"}]},
    thread_config,
)
print("\nQ: Does that apply to digital products too?")
print("A:", response2["messages"][-1].content)

Q: 你们的退款政策是什么?
A: 所有实体产品均可在购买后 30 天内退货并获得全额退款。
需要提供原始收据或订单确认邮件,且商品必须处于
原始包装且未使用的状态。
30 天之后,退货仅接受兑换为商店积分。
 
Q: 那也适用于数字产品吗?
A: 数字产品(软件许可证、电子书、在线课程)一旦
下载或访问链接被激活,即不可退款。
但是,如果您遇到妨碍访问的技术问题,可以在 7 天内
联系支持以获取替换或退款。

from langchain.agents import create_agent, AgentState
from langchain.agents.middleware import before_model
from langchain.messages import RemoveMessage
from langgraph.graph.message import REMOVE_ALL_MESSAGES
 
@before_model
def trim_old_messages(state: AgentState, runtime) -> dict | None:
    """在每次 LLM 调用之前移除旧消息。"""
    messages = state["messages"]
    # 如果消息足够少,则不做任何处理
    if len(messages) <= 10:
        return None
    # 仅保留系统消息(第一条)和最近的 10 条消息
    return {
        "messages": [
            RemoveMessage(id=REMOVE_ALL_MESSAGES),
            messages[0],     # 系统消息
            *messages[-10:], # 最近的 10 条消息(5 轮)
        ]
    }

agent = create_agent(
    model="gpt-5-mini",
    tools=[retrieve_context],
    system_prompt=(
        "You are a helpful assistant that answers questions about company policies. "
        "Use the retrieve_context tool to search for relevant information. "
        "If the retrieved context does not contain relevant information, "
        "say that you don't know. "
        "Keep the answer concise, three sentences maximum."
    ),
    checkpointer=InMemorySaver(),
    middleware=[trim_old_messages],  # 注册中间件
)