main.py

import streamlit as st
from dotenv import load_dotenv
from ingest.loader import load_pdf
from ingest.vector_store import create_vector_store
from chains.question_generator import get_question_chain
from chains.feedback import get_feedback_chain
import random
import os

# Load environment and OpenAI API key
load_dotenv()
openai_api_key = os.getenv("OPENAI_API_KEY")

# Streamlit page setup
st.set_page_config(page_title="Study Buddy", page_icon="📚")
st.title("📚 Study Buddy Chatbot")

# Session state
if "question" not in st.session_state:
    st.session_state.question = None
    st.session_state.context = None
    st.session_state.feedback = None
    st.session_state.history = []

# Load and index content
@st.cache_resource
def setup_vector_store():
    docs = load_pdf("sample.pdf")  # updated from main.py
    return create_vector_store(docs)

db = setup_vector_store()
q_chain = get_question_chain()
f_chain = get_feedback_chain()

# Sidebar controls
st.sidebar.header("Study Settings")
topic = st.sidebar.text_input("Topic", value="GPU")
if st.sidebar.button("🔄 New Question"):
    retrieved = db.similarity_search(topic)
    context = random.choice(retrieved).page_content
    question = q_chain.invoke({"context": context}).content

    st.session_state.question = question
    st.session_state.context = context
    st.session_state.feedback = None

# Main interface
if st.session_state.question:
    st.subheader("🧠 Question")
    st.markdown(st.session_state.question)

    user_input = st.text_input("✍️ Your Answer")
    if st.button("✅ Submit Answer") and user_input:
        feedback = f_chain.invoke({
            "question": st.session_state.question,
            "user_answer": user_input,
            "context": st.session_state.context
        }).content

        st.session_state.feedback = feedback
        st.session_state.history.append({
            "question": st.session_state.question,
            "answer": user_input,
            "feedback": feedback
        })

if st.session_state.feedback:
    st.subheader("🔍 Feedback")
    st.success(st.session_state.feedback)

# Review past Q&A
if st.session_state.history:
    with st.expander("📜 Review History"):
        for idx, item in enumerate(reversed(st.session_state.history), 1):
            st.markdown(f"**Q{idx}:** {item['question']}")
            st.markdown(f"**Your Answer:** {item['answer']}")
            st.markdown(f"**Feedback:** {item['feedback']}")
            st.markdown("---")