app.py

import streamlit as st
import PyPDF2
from transformers import AutoTokenizer, AutoModel
import torch
import numpy as np
import faiss
from gtts import gTTS
import os

# Initialize the model and tokenizer
model_name = "sentence-transformers/all-MiniLM-L6-v2"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModel.from_pretrained(model_name)

# Function to get embeddings
def get_embedding(text):
    inputs = tokenizer(text, return_tensors='pt')
    with torch.no_grad():
        embeddings = model(**inputs).last_hidden_state.mean(dim=1).numpy()
    return embeddings

# Initialize FAISS index
embeddings_dimension = 384  # for MiniLM
index = faiss.IndexFlatL2(embeddings_dimension)

# Title of the app
st.title("Study Assistant for Grade 9")

# File uploader widget
uploaded_file = st.file_uploader("Upload your PDF file", type=["pdf"])

if uploaded_file is not None:
    # Read the uploaded PDF file
    pdf_reader = PyPDF2.PdfReader(uploaded_file)
    text = ""

    # Extract text from each page
    for page in pdf_reader.pages:
        text += page.extract_text() if page.extract_text() else ""

    st.subheader("Extracted Text:")
    st.write(text)

    # Generate embedding for the extracted text
    embeddings = get_embedding(text)
    index.add(embeddings)  # Add embedding to the FAISS index

    st.success("Text extracted and embeddings generated!")

# Subject selection and query input
subject = st.selectbox("Select Subject", ["Math", "Science", "English"])
query = st.text_input("Type your query")

if st.button("Submit"):
    if query:
        # Get embedding for the query
        query_embedding = get_embedding(query)
        
        # Search for the nearest neighbors in the FAISS index
        D, I = index.search(query_embedding, k=5)  # Retrieve top 5 matches
        
        st.subheader("Top Matches:")
        for idx in I[0]:
            if idx < len(embeddings):  # Ensure index is valid
                st.write(f"Match Index: {idx}, Distance: {D[0][idx]}")  # Display match details

        # Convert response to speech
        response_text = f"You asked about '{query}' in {subject}. Here are your top matches."
        tts = gTTS(text=response_text, lang='en')
        tts.save("response.mp3")

        # Display audio controls (Streamlit doesn't support direct playback)
        st.audio("response.mp3")

        st.success("Response generated!")