import streamlit as st
import os

from embeddings.vector_store import build_vector_db
from run_system import answer_query

st.set_page_config(page_title="Engineering Knowledge Bot", layout="wide")
st.title("Engineering Knowledge Bot")

if not os.path.exists("vector_db/index.faiss"):
    with st.spinner("Building vector database..."):
        build_vector_db()

query = st.text_input("Ask an engineering question:")

if query:
    with st.spinner("Thinking with Llama 3.3 70B Instruct..."):
        answer, ctx = answer_query(query)

    st.markdown("## Answer")
    st.write(answer)

    st.markdown("---")
    st.markdown("### Retrieved Context")
    for c in ctx:
        st.markdown(f"**Source:** {c['metadata']['source']}")
        st.write(c["text"])