import streamlit as st
import PyPDF2
import os
from sentence_transformers import SentenceTransformer
import faiss
import numpy as np
from transformers import pipeline

st.set_page_config(page_title="📘 PDF RAG QA", layout="wide")

# Custom styles
st.markdown("""
    <style>
    .main {background-color: #f7faff;}
    h1 {color: #4a4a8a;}
    .stTextInput>div>div>input {border: 2px solid #d0d7ff;}
    .stButton button {background-color: #4a4a8a; color: white;}
    </style>
""", unsafe_allow_html=True)

st.title("📘 Ask Me Anything About Machine Learning")
st.caption("Using RAG (Retrieval-Augmented Generation) and a preloaded PDF")

# Load PDF from local file
PDF_FILE = "ml_large_dataset.pdf"

def load_pdf(file_path):
    with open(file_path, "rb") as f:
        reader = PyPDF2.PdfReader(f)
        return [page.extract_text() for page in reader.pages]

def chunk_text(pages, max_len=1000):
    text = " ".join(pages)
    words = text.split()
    return [' '.join(words[i:i+max_len]) for i in range(0, len(words), max_len)]

@st.cache_resource
def setup_rag():
    pages = load_pdf(PDF_FILE)
    chunks = chunk_text(pages)
    model = SentenceTransformer('all-MiniLM-L6-v2')
    embeddings = model.encode(chunks)
    index = faiss.IndexFlatL2(embeddings.shape[1])
    index.add(np.array(embeddings))
    qa = pipeline("question-answering", model="deepset/roberta-base-squad2")
    return chunks, model, index, qa

def retrieve_answer(question, chunks, model, index, qa_pipeline, k=6):
    q_embed = model.encode([question])
    _, I = index.search(np.array(q_embed), k)
    context = "\n\n".join([chunks[i] for i in I[0]])
    result = qa_pipeline(question=question, context=context)
    return result['answer']

chunks, embed_model, faiss_index, qa_model = setup_rag()

st.subheader("💬 Ask a Question")
question = st.text_input("Enter your question:", placeholder="e.g., What is supervised learning?")

if question:
    with st.spinner("🧠 Searching for the answer..."):
        answer = retrieve_answer(question, chunks, embed_model, faiss_index, qa_model)
        st.markdown("#### 📖 Answer:")
        st.write(answer)