app.py

from langchain.document_loaders import PyPDFLoader
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.chains import RetrievalQAWithSourcesChain
from langchain.chat_models import ChatOpenAI
from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    HumanMessagePromptTemplate,
)
import os
import chainlit as cl
import tempfile
from dotenv import load_dotenv

load_dotenv()

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

system_template = """ Try to find detailed information

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

messages = [
    SystemMessagePromptTemplate.from_template(system_template),
    HumanMessagePromptTemplate.from_template("{question}"),
]

prompt = ChatPromptTemplate.from_messages(messages)

@cl.on_chat_start
async def init():
    files = None

    # Wait for the user to upload a file
    while files is None:
        files = await cl.AskFileMessage(
            content="Please upload a file to start chatting!", accept=["pdf"]
        ).send()

    file = files[0]

    msg = cl.Message(content=f"Processing `{file.name}`...")
    await msg.send()

    with tempfile.NamedTemporaryFile(delete=False) as temp:
        temp.write(file.content)
        temp_path = temp.name

    # Load the PDF using PyPDFLoader into an array of documents, where each document contains the page content and metadata with page number.
    loader = PyPDFLoader(temp_path)
    pages = loader.load_and_split()

    # Combine the page content into a single text variable.
    text = ' '.join([page.page_content for page in pages])

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

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

    # Create a Chroma vector store
    embeddings = OpenAIEmbeddings()
    docsearch = await cl.make_async(Chroma.from_texts)(
        texts, embeddings, metadatas=metadatas
    )

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

    # 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"`{file.name}` processed. You can now ask questions!"
    await msg.update()

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

@cl.on_message
async def process_response(message):
    chain = cl.user_session.get("chain")

    if chain is None:
        await cl.Message(content="The system is not initialized. Please upload a PDF file first.").send()
        return

    # Use the chain to process the user's question
    response = await chain.acall({
        "question": message.content
    })

    answer = response["answer"]
    sources = response["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")

    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"

    await cl.Message(content=answer, elements=source_elements).send()