app.py

# Import necessary libraries
import PyPDF2
from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
from sentence_transformers import SentenceTransformer
import torch
from tkinter import Tk, filedialog

# Step 1: Ask for the PDF file via file upload dialog
def upload_file():
    """Open a file dialog to select a PDF file."""
    root = Tk()
    root.withdraw()  # Hide the root window
    file_path = filedialog.askopenfilename(
        title="Select PDF File",
        filetypes=[("PDF Files", "*.pdf")],
    )
    if not file_path:
        raise ValueError("No file was selected. Please upload a PDF.")
    return file_path

print("Please select the PDF file containing the chapter.")
try:
    file_path = upload_file()
    print(f"File selected: {file_path}")
except Exception as e:
    print(f"Error: {e}")
    exit()

# Step 2: Extract text from the PDF
def extract_text_from_pdf(file_path):
    text = ""
    with open(file_path, 'rb') as file:
        pdf_reader = PyPDF2.PdfReader(file)
        for page in pdf_reader.pages:
            text += page.extract_text()
    return text

chapter_text = extract_text_from_pdf(file_path)
print("Text extracted from the PDF successfully!")

# Step 3: Split the text into smaller passages
def split_text_into_chunks(text, chunk_size=500):
    """Split the text into chunks of size chunk_size."""
    words = text.split()
    chunks = []
    for i in range(0, len(words), chunk_size):
        chunk = " ".join(words[i:i + chunk_size])
        chunks.append(chunk)
    return chunks

passages = split_text_into_chunks(chapter_text)
print(f"Chapter split into {len(passages)} passages for RAG processing.")

# Step 4: Initialize the RAG model and tokenizer
device = "cuda" if torch.cuda.is_available() else "cpu"

# Load the RAG model, tokenizer, and retriever
tokenizer = RagTokenizer.from_pretrained("facebook/rag-token-nq")
retriever = RagRetriever.from_pretrained(
    "facebook/rag-token-nq",
    index_name="custom",
    passages=passages,  # Set passages as the custom index
    use_dummy_dataset=True,  # Dummy dataset required for custom index
)
model = RagSequenceForGeneration.from_pretrained("facebook/rag-token-nq").to(device)

# Step 5: Encode passages into embeddings for retrieval
sentence_model = SentenceTransformer("sentence-transformers/all-mpnet-base-v2")
passage_embeddings = sentence_model.encode(passages, convert_to_tensor=True)
retriever.index.set_passages(passages, passage_embeddings.cpu().detach().numpy())
print("Passages indexed successfully!")

# Step 6: Define a function to generate answers
def generate_answer(question, passages):
    inputs = tokenizer.prepare_seq2seq_batch(
        questions=[question], return_tensors="pt"
    ).to(device)
    generated_ids = model.generate(**inputs)
    answer = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
    return answer

# Step 7: Interactive Question-Answering
print("\nChapter is ready. You can now ask questions!")
while True:
    user_question = input("\nEnter your question (or type 'exit' to quit): ")
    if user_question.lower() == "exit":
        print("Exiting the application. Thank you!")
        break
    try:
        answer = generate_answer(user_question, passages)
        print(f"Answer: {answer}")
    except Exception as e:
        print(f"An error occurred: {e}")