在我之前的文章 “Elasticsearch：与多个 PDF 聊天 | LangChain Python 应用教程（免费 LLMs 和嵌入）” 里，我详述如何使用 Streamlit，Langchain, Elasticsearch 及 OpenAI 来针对 PDF 进行聊天。在今天的文章中，我将使用 Chainlit 来展示如使用 Langchain 及 Elasticsearch 针对 PDF 文件进行查询。

为方便大家学习，我的代码在地址 GitHub - liu-xiao-guo/langchain-openai-chainlit: Chat with your documents (pdf, csv, text) using Openai model, LangChain and Chainlit 进行下载。

安装

安装 Elasticsearch 及 Kibana

如果你还没有安装好自己的 Elasticsearch 及 Kibana，那么请参考一下的文章来进行安装：

• 如何在 Linux，MacOS 及 Windows 上进行安装 Elasticsearch

• Kibana：如何在 Linux，MacOS 及 Windows 上安装 Elastic 栈中的 Kibana

在安装的时候，请选择 Elastic Stack 8.x 进行安装。在安装的时候，我们可以看到如下的安装信息：

 拷贝 Elasticsearch 证书

我们把 Elasticsearch 的证书拷贝到当前的目录下：

$ pwd
/Users/liuxg/python/elser
$ cp ~/elastic/elasticsearch-8.12.0/config/certs/http_ca.crt .
$ ls http_ca.crt 
http_ca.crt

安装 Python 依赖包

我们在当前的目录下打入如下的命令：

python3 -m venv .venv
source .venv/bin/activate

然后，我们再打入如下的命令：

$ pwd
/Users/liuxg/python/langchain-openai-chainlit
$ source .venv/bin/activate
(.venv) $ pip3 install -r requirements.txt

运行应用

有关 Chainlit 的更多知识请参考 Overview - Chainlit。这里就不再赘述。有关 pdf_qa.py 的代码如下：

pdf_qa.py

# Import necessary modules and define env variables

# from langchain.embeddings.openai import OpenAIEmbeddings
from langchain_openai import OpenAIEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.chains import RetrievalQAWithSourcesChain
from langchain_openai import ChatOpenAI
from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    HumanMessagePromptTemplate,
)
import os
import io
import chainlit as cl
import PyPDF2
from io import BytesIO

from pprint import pprint
import inspect
# from langchain.vectorstores import ElasticsearchStore
from langchain_community.vectorstores import ElasticsearchStore
from elasticsearch import Elasticsearch

from dotenv import load_dotenv

# Load environment variables from .env file
load_dotenv()

OPENAI_API_KEY= os.getenv("OPENAI_API_KEY")
ES_USER = os.getenv("ES_USER")
ES_PASSWORD = os.getenv("ES_PASSWORD")
elastic_index_name='pdf_docs'

# text_splitter and system template

text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)

system_template = """Use the following pieces of context to answer the users question.
If you don't know the answer, just say that you don't know, don't try to make up an answer.
ALWAYS return a "SOURCES" part in your answer.
The "SOURCES" part should be a reference to the source of the document from which you got your answer.

Example of your response should be:

```
The answer is foo
SOURCES: xyz
```

Begin!
----------------
{summaries}"""


messages = [
    SystemMessagePromptTemplate.from_template(system_template),
    HumanMessagePromptTemplate.from_template("{question}"),
]
prompt = ChatPromptTemplate.from_messages(messages)
chain_type_kwargs = {"prompt": prompt}


@cl.on_chat_start
async def on_chat_start():

    # Sending an image with the local file path
    elements = [
    cl.Image(name="image1", display="inline", path="./robot.jpeg")
    ]
    await cl.Message(content="Hello there, Welcome to AskAnyQuery related to Data!", elements=elements).send()
    files = None

    # Wait for the user to upload a PDF file
    while files is None:
        files = await cl.AskFileMessage(
            content="Please upload a PDF file to begin!",
            accept=["application/pdf"],
            max_size_mb=20,
            timeout=180,
        ).send()

    file = files[0]

    # print("type: ", type(file))
    # print("file: ", file)
    # pprint(vars(file))
    # print(file.content)
 
    msg = cl.Message(content=f"Processing `{file.name}`...")
    await msg.send()

    # Read the PDF file
    # pdf_stream = BytesIO(file.content)
    with open(file.path, 'rb') as f:
        pdf_content = f.read()
    pdf_stream = BytesIO(pdf_content)
    pdf = PyPDF2.PdfReader(pdf_stream)
    pdf_text = ""
    for page in pdf.pages:
        pdf_text += page.extract_text()

    # Split the text into chunks
    texts = text_splitter.split_text(pdf_text)

    # Create metadata for each chunk
    metadatas = [{"source": f"{i}-pl"} for i in range(len(texts))]

    # Create a Chroma vector store
    embeddings = OpenAIEmbeddings()
  
    url = f"https://{ES_USER}:{ES_PASSWORD}@localhost:9200"
 
    connection = Elasticsearch(
        hosts=[url], 
        ca_certs = "./http_ca.crt", 
        verify_certs = True
    )

    docsearch = None
    
    if not connection.indices.exists(index=elastic_index_name):
        print("The index does not exist, going to generate embeddings")   
        docsearch = await cl.make_async(ElasticsearchStore.from_texts)( 
                texts,
                embedding = embeddings, 
                es_url = url, 
                es_connection = connection,
                index_name = elastic_index_name, 
                es_user = ES_USER,
                es_password = ES_PASSWORD,
                metadatas=metadatas
        )
    else: 
        print("The index already existed")
        
        docsearch = ElasticsearchStore(
            es_connection=connection,
            embedding=embeddings,
            es_url = url, 
            index_name = elastic_index_name, 
            es_user = ES_USER,
            es_password = ES_PASSWORD    
        )

    # Create a chain that uses the Chroma vector store
    chain = RetrievalQAWithSourcesChain.from_chain_type(
        ChatOpenAI(temperature=0),
        chain_type="stuff",
        retriever=docsearch.as_retriever(search_kwargs={"k": 4}),
    )

    # Save the metadata and texts in the user session
    cl.user_session.set("metadatas", metadatas)
    cl.user_session.set("texts", texts)

    # Let the user know that the system is ready
    msg.content = f"Processing `{file.name}` done. You can now ask questions!"
    await msg.update()

    cl.user_session.set("chain", chain)


@cl.on_message
async def main(message:str):

    chain = cl.user_session.get("chain")  # type: RetrievalQAWithSourcesChain
    print("chain type: ", type(chain))
    cb = cl.AsyncLangchainCallbackHandler(
        stream_final_answer=True, answer_prefix_tokens=["FINAL", "ANSWER"]
    )
    cb.answer_reached = True
    
    print("message: ", message)
    pprint(vars(message))
    print(message.content)
    res = await chain.acall(message.content, callbacks=[cb])

    answer = res["answer"]
    sources = res["sources"].strip()
    source_elements = []
    
    # Get the metadata and texts from the user session
    metadatas = cl.user_session.get("metadatas")
    all_sources = [m["source"] for m in metadatas]
    texts = cl.user_session.get("texts")
    
    print("texts: ", texts)

    if sources:
        found_sources = []

        # Add the sources to the message
        for source in sources.split(","):
            source_name = source.strip().replace(".", "")
            # Get the index of the source
            try:
                index = all_sources.index(source_name)
            except ValueError:
                continue
            text = texts[index]
            found_sources.append(source_name)
            # Create the text element referenced in the message
            source_elements.append(cl.Text(content=text, name=source_name))

        if found_sources:
            answer += f"\nSources: {', '.join(found_sources)}"
        else:
            answer += "\nNo sources found"

    if cb.has_streamed_final_answer:
        cb.final_stream.elements = source_elements
        await cb.final_stream.update()
    else:
        await cl.Message(content=answer, elements=source_elements).send()

我们可以使用如下的命令来运行：

export ES_USER="elastic"
export ES_PASSWORD="xnLj56lTrH98Lf_6n76y"
export OPENAI_API_KEY="YourOpenAiKey"

chainlit run pdf_qa.py -w

(.venv) $ chainlit run pdf_qa.py -w
2024-02-14 10:58:30 - Loaded .env file
2024-02-14 10:58:33 - Your app is available at http://localhost:8000
2024-02-14 10:58:34 - Translation file for en not found. Using default translation en-US.
2024-02-14 10:58:35 - 2 changes detected

我们先选择项目自带的 pdf 文件：

Is sample PDF download critical to an organization?

Does comprehensive PDF testing have various advantages?