LangChain(2):基于自己的文档构建一个问答系统

“”"
欢迎来到LangChain实战课
https://time.geekbang.org/column/intro/100617601
作者 黄佳
“”"
此笔记来自于 黄佳 的极客时间 LangChain 实战课。如有侵权请联系删除。
课程链接
课程github

pip install pypdf
pip install docx2txt
pip install qdrant-client  # qdrant 向量库
1
2
3

import os
os.environ["OPENAI_API_KEY"] = "你的OpenAI API Key"

# 1.Load 导入Document Loaders
from langchain.document_loaders import PyPDFLoader
from langchain.document_loaders import Docx2txtLoader
from langchain.document_loaders import TextLoader

# 加载Documents
base_dir = "./OneFlower"  # 文档的存放目录
documents = []
for file in os.listdir(base_dir):
    # 构建完整的文件路径
    file_path = os.path.join(base_dir, file)
    if file.endswith(".pdf"):
        loader = PyPDFLoader(file_path)
        documents.extend(loader.load())
    elif file.endswith(".docx"):
        loader = Docx2txtLoader(file_path)
        documents.extend(loader.load())
    elif file.endswith(".txt"):
        loader = TextLoader(file_path)
        documents.extend(loader.load())

# 2.Split 将Documents切分成块以便后续进行嵌入和向量存储
from langchain.text_splitter import RecursiveCharacterTextSplitter

text_splitter = RecursiveCharacterTextSplitter(chunk_size=200, chunk_overlap=10)
chunked_documents = text_splitter.split_documents(documents)

# 3.Store 将分割嵌入并存储在矢量数据库Qdrant中
from langchain.vectorstores import Qdrant
from langchain.embeddings import OpenAIEmbeddings

vectorstore = Qdrant.from_documents(
    documents=chunked_documents,  # 以分块的文档
    embedding=OpenAIEmbeddings(),  # 用OpenAI的Embedding Model做嵌入
    location=":memory:",  # in-memory 存储
    collection_name="my_documents",
)  # 指定collection_name

# 4. Retrieval 准备模型和Retrieval链
import logging  # 导入Logging工具
from langchain.chat_models import ChatOpenAI  # ChatOpenAI模型
from langchain.retrievers.multi_query import (
    MultiQueryRetriever,
)  # MultiQueryRetriever工具
from langchain.chains import RetrievalQA  # RetrievalQA链

# 设置Logging
logging.basicConfig()
logging.getLogger("langchain.retrievers.multi_query").setLevel(logging.INFO)

# 实例化一个大模型工具 - OpenAI的GPT-3.5
llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0)

# 实例化一个MultiQueryRetriever
retriever_from_llm = MultiQueryRetriever.from_llm(
    retriever=vectorstore.as_retriever(), llm=llm
)

# 实例化一个RetrievalQA链
qa_chain = RetrievalQA.from_chain_type(llm, retriever=retriever_from_llm)

# 5. Output 问答系统的UI实现
from flask import Flask, request, render_template

app = Flask(__name__)  # Flask APP


@app.route("/", methods=["GET", "POST"])
def home():
    if request.method == "POST":
        # 接收用户输入作为问题
        question = request.form.get("question")

        # RetrievalQA链 - 读入问题，生成答案
        result = qa_chain({"query": question})

        # 把大模型的回答结果返回网页进行渲染
        return render_template("index.html", result=result)

    return render_template("index.html")


if __name__ == "__main__":
    app.run(host="0.0.0.0", debug=True, port=5000)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88