Capítulo 1: Construindo um Sistema de Q&A para Documentos

Visão Geral do Projeto

Construa um sistema completo de perguntas e respostas para PDFs com:

• Upload e processamento de PDF
• Interface de chat com histórico
• Citações de fontes

Arquitetura

Passo 1: Configuração

# requirements.txt
langchain==0.1.0
openai==1.3.0
chromadb==0.4.0
streamlit==1.28.0
pypdf==3.17.0

Passo 2: Processamento de Documentos

# document_processor.py
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import Chroma

def process_pdf(pdf_path, collection_name="docs"):
    # Load
    loader = PyPDFLoader(pdf_path)
    documents = loader.load()
    
    # Split
    splitter = RecursiveCharacterTextSplitter(
        chunk_size=1000,
        chunk_overlap=200
    )
    chunks = splitter.split_documents(documents)
    
    # Embed and store
    embeddings = OpenAIEmbeddings()
    vectorstore = Chroma.from_documents(
        chunks, 
        embeddings,
        collection_name=collection_name,
        persist_directory="./chroma_db"
    )
    
    return vectorstore

Passo 3: RAG Chain

# rag_chain.py
from langchain.chat_models import ChatOpenAI
from langchain.chains import ConversationalRetrievalChain
from langchain.memory import ConversationBufferMemory

def create_qa_chain(vectorstore):
    llm = ChatOpenAI(model="gpt-4", temperature=0)
    
    memory = ConversationBufferMemory(
        memory_key="chat_history",
        return_messages=True,
        output_key="answer"
    )
    
    chain = ConversationalRetrievalChain.from_llm(
        llm=llm,
        retriever=vectorstore.as_retriever(search_kwargs={"k": 3}),
        memory=memory,
        return_source_documents=True
    )
    
    return chain

Passo 4: Interface Streamlit

# app.py
import streamlit as st
from document_processor import process_pdf
from rag_chain import create_qa_chain

st.title("📄 Q&A de Documentos")

# File upload
uploaded_file = st.file_uploader("Envie um PDF", type="pdf")

if uploaded_file:
    # Process PDF
    with st.spinner("Processando..."):
        vectorstore = process_pdf(uploaded_file)
        chain = create_qa_chain(vectorstore)
    
    # Chat interface
    if "messages" not in st.session_state:
        st.session_state.messages = []
    
    for msg in st.session_state.messages:
        st.chat_message(msg["role"]).write(msg["content"])
    
    if prompt := st.chat_input("Pergunte sobre seu documento"):
        st.session_state.messages.append({"role": "user", "content": prompt})
        st.chat_message("user").write(prompt)
        
        response = chain({"question": prompt})
        answer = response["answer"]
        
        st.session_state.messages.append({"role": "assistant", "content": answer})
        st.chat_message("assistant").write(answer)
        
        # Show sources
        with st.expander("Fontes"):
            for doc in response["source_documents"]:
                st.write(f"Página {doc.metadata.get('page', 'N/A')}")
                st.write(doc.page_content[:200] + "...")

Executando o App

streamlit run app.py

Melhorias

• Adicionar autenticação
• Suportar múltiplos PDFs
• Exportar conversas
• Adicionar reranking