app.py

import streamlit as st
from transformers import pipeline
from PyPDF2 import PdfReader
from sentence_transformers import SentenceTransformer, util
import smtplib
from email.mime.text import MIMEText
from email.mime.multipart import MIMEMultipart
import os
import random

# Load Hugging Face model (e.g., FLAN-T5)
@st.cache_resource
def load_model():
    return pipeline("text2text-generation", model="google/flan-t5-base")

# Extract text from PDF files in the dataset folder
def extract_text_from_pdfs(folder_path):
    pdf_texts = []
    for filename in os.listdir(folder_path):
        if filename.endswith(".pdf"):
            pdf_path = os.path.join(folder_path, filename)
            reader = PdfReader(pdf_path)
            text = " ".join([page.extract_text() for page in reader.pages])
            pdf_texts.append(text)
    return pdf_texts

# Embed PDF texts for similarity matching
@st.cache_resource
def create_embeddings(texts):
    embedder = SentenceTransformer('all-MiniLM-L6-v2')
    embeddings = embedder.encode(texts, convert_to_tensor=True)
    return embedder, embeddings

# Query matching function
def get_relevant_content(query, texts, embedder, embeddings):
    query_embedding = embedder.encode(query, convert_to_tensor=True)
    scores = util.pytorch_cos_sim(query_embedding, embeddings)
    best_idx = scores.argmax().item()
    return texts[best_idx]

# Email functionality
def send_email(to_email, subject, body):
    try:
        sender_email = "your-email@example.com"  # Replace with your email
        sender_password = "your-email-password"  # Replace with your email password
        smtp_server = "smtp.gmail.com"
        smtp_port = 587

        msg = MIMEMultipart()
        msg['From'] = sender_email
        msg['To'] = to_email
        msg['Subject'] = subject
        msg.attach(MIMEText(body, 'plain'))

        with smtplib.SMTP(smtp_server, smtp_port) as server:
            server.starttls()
            server.login(sender_email, sender_password)
            server.send_message(msg)
        return True
    except Exception as e:
        st.error(f"Failed to send email: {e}")
        return False

# CAPTCHA generation
def generate_captcha():
    num1 = random.randint(1, 9)
    num2 = random.randint(1, 9)
    return num1, num2, num1 + num2

# Streamlit UI
def main():
    st.title("Educational Assistant Chatbot")
    st.markdown("Welcome! Let's create your student profile to provide tailored guidance.")

    # Hardcoded dataset folder and email
    dataset_folder = "./pdf_dataset"
    email_to_send = "application@aspireec.pk"

    # Load dataset on startup
    if "pdf_texts" not in st.session_state:
        st.session_state.pdf_texts = extract_text_from_pdfs(dataset_folder)
        st.session_state.embedder, st.session_state.embeddings = create_embeddings(st.session_state.pdf_texts)

    # Initialize session state for profile submission
    if "profile_submitted" not in st.session_state:
        st.session_state.profile_submitted = False

    # Show profile creation form only if profile is not submitted
    if not st.session_state.profile_submitted:
        st.markdown("### Create a Student Profile")
        name = st.text_input("Name (required):")
        email = st.text_input("Email (required):")
        contact_number = st.text_input("Contact Number (optional):")
        study_level = st.selectbox("Level of Study:", ["Undergraduate", "Postgraduate", "PhD"])
        field_of_interest = st.text_input("Field of Interest:")
        career_goal = st.text_area("Career Goals:")
        visa_query = st.text_area("Visa Concerns or Questions:")

        # CAPTCHA
        if "captcha_result" not in st.session_state:
            num1, num2, st.session_state.captcha_result = generate_captcha()
        st.markdown(f"**CAPTCHA: {num1} + {num2} = ?**")
        captcha_input = st.text_input("Enter CAPTCHA Result:")

        if st.button("Submit Profile"):
            if not name or not email:
                st.error("Name and Email are required fields.")
            elif int(captcha_input) != st.session_state.captcha_result:
                st.error("Invalid CAPTCHA. Please try again.")
            else:
                profile = f"""
                Name: {name}
                Email: {email}
                Contact Number: {contact_number}
                Level of Study: {study_level}
                Field of Interest: {field_of_interest}
                Career Goals: {career_goal}
                Visa Queries: {visa_query}
                """
                email_sent = send_email(email_to_send, "New Student Profile Submission", profile)
                if email_sent:
                    st.session_state.profile_submitted = True
                    st.success(f"Profile submitted successfully to {email_to_send}!")
                else:
                    st.error("Failed to submit the profile. Please try again later.")

    # Show question input only if profile is submitted
    if st.session_state.profile_submitted:
        st.markdown("### Ask a Question")
        user_query = st.text_input("Your Query:")
        if st.button("Ask"):
            if "embedder" not in st.session_state or "pdf_texts" not in st.session_state:
                st.error("Dataset is not loaded. Please restart the application.")
            else:
                relevant_content = get_relevant_content(
                    user_query, st.session_state.pdf_texts, st.session_state.embedder, st.session_state.embeddings
                )
                model = load_model()
                response = model(f"Question: {user_query} Context: {relevant_content}", max_length=200)
                st.success(response[0]['generated_text'])

if __name__ == "__main__":
    main()