app.py

import os
import time

import openai
import streamlit as st
from dotenv import load_dotenv

from extract import extract_text_from_pdfs
from generate import generate_response
from preprocess import preprocess_text
from retrieve import create_vectorizer, retrieve

# Load environment variables from .env file
load_dotenv()

# Set OpenAI API key
openai.api_key = os.getenv('api_key')

# Initialize session state
if "messages" not in st.session_state:
    st.session_state.messages = []

if "pdf_files" not in st.session_state:
    st.session_state.pdf_files = []

if "processed_texts" not in st.session_state:
    st.session_state.processed_texts = []

st.title("RAG-based PDF Query System")

# File uploader for PDF files
uploaded_files = st.file_uploader("Upload PDFs", type=["pdf"], accept_multiple_files=True)

if uploaded_files:
    if "uploaded_files" not in st.session_state or uploaded_files != st.session_state.uploaded_files:
        st.session_state.uploaded_files = uploaded_files
        st.session_state.messages = []  # Clear previous messages
        st.session_state.pdf_files = []
        st.session_state.processed_texts = []

        # Initialize status container
        with st.status("Processing the uploaded PDFs...", state="running") as status:
            # Save uploaded files to disk
            for uploaded_file in uploaded_files:
                with open(uploaded_file.name, "wb") as f:
                    f.write(uploaded_file.getbuffer())
                st.session_state.pdf_files.append(uploaded_file.name)

            # Extract text from PDFs
            num_files = len(st.session_state.pdf_files)
            texts = []
            for i, pdf_file in enumerate(st.session_state.pdf_files):
                st.write(f"Extracting text from file {i + 1} of {num_files}...")
                text = extract_text_from_pdfs([pdf_file])
                texts.extend(text)
                time.sleep(0.1)  # Simulate time taken for processing

            # Preprocess text
            st.write("Preprocessing text...")
            st.session_state.processed_texts = preprocess_text(texts)
            time.sleep(0.1)  # Simulate time taken for processing

            # Create vectorizer and transform texts
            st.write("Creating vectorizer and transforming texts...")
            st.session_state.vectorizer, st.session_state.X = create_vectorizer(st.session_state.processed_texts)
            time.sleep(0.1)  # Simulate time taken for processing

            # Update status to complete
            status.update(label="Processing complete!", state="complete")

else:
    st.stop()

# Chat interface
st.write("### Ask a question about the uploaded PDFs")

# Display chat messages
for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.write(message["content"])

# Chat input
prompt = st.chat_input("Ask something about the uploaded PDFs")
if prompt:
    # Add user message to session state
    st.session_state.messages.append({"role": "user", "content": prompt})

    # Retrieve relevant texts
    top_indices = retrieve(prompt, st.session_state.X, st.session_state.vectorizer)
    retrieved_texts = [" ".join(st.session_state.processed_texts[i]) for i in top_indices]

    # Generate response
    response = generate_response(retrieved_texts, prompt)
    st.session_state.messages.append({"role": "assistant", "content": response})

    # Display user message
    with st.chat_message("user"):
        st.write(prompt)

    # Display assistant message
    with st.chat_message("assistant"):
        st.write(response)

# Clean up uploaded files
for pdf_file in st.session_state.pdf_files:
    if os.path.exists(pdf_file):
        os.remove(pdf_file)