app.py

import io
import os
import streamlit as st
from dotenv import load_dotenv
from PyPDF2 import PdfReader
from langchain_community.embeddings import OpenAIEmbeddings
from langchain_community.vectorstores import FAISS
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.llms import OpenAI
import openai
import random

# Load environment variables
load_dotenv()
openai_api_key = os.getenv('OPENAI_API_KEY')
openai.api_key = openai_api_key

# Initialize Streamlit session states
if 'vectorDB' not in st.session_state:
    st.session_state.vectorDB = None

# Function to extract text from a PDF file
def get_pdf_text(pdf):
    text = ""
    pdf_reader = PdfReader(pdf)
    for page in pdf_reader.pages:
        text += page.extract_text()
    return text

def get_text_chunks(text: str):
    # This function will split the text into smaller chunks
    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=1000,
        chunk_overlap=100,
        length_function=len,
        is_separator_regex=False,
    )
    chunks = text_splitter.split_text(text)
    return chunks

def get_vectorstore(text_chunks):
    # This function will create a vector database as well as create and store the embedding of the text chunks into the VectorDB 
    embeddings = OpenAIEmbeddings()
    vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
    return vectorstore

def search_vectorstore(vectorDB, query):
    # This function searches for specific data or content in the vector database
    # Specify the search type
    search_type = 'similarity'
    search_results = vectorDB.search(query, search_type=search_type)
    return search_results

def generate_quiz_questions(search_results, num_questions):
    # Generate quiz questions with options using GPT-3.5-turbo-16k based on the search results
    st.header(f"Quiz Generator: {quiz_name}")
    st.subheader(f"Topic: {quiz_topic}")
    # Process PDF and create vector database
    if st.button('Generate Quiz'):
        st.session_state['vectorDB'] = (pdf_content)

    # Placeholder for quiz questions
    quiz_questions = []

    # Generate questions using GPT-3.5-turbo-16k with the correct endpoint
    for i in range(num_questions):
        prompt = f"As a highly intelligent and powerful quiz generator, your task is to Generate a different multiple-choice question and correct answer in a comprehensive and accurate format for the same related to:\n- {random.choice(search_results)} and the response generated by you should be comprehensive and relevant to the topic specified, you should not generate the same content multiple times in a single response, generate the explaination for each correct answer"

        # Prepare messages
        messages = [
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt},
        ]

        # Generate questions using GPT-3.5-turbo-16k with the correct endpoint
        response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo-16k",
        messages=messages,
        max_tokens=200,
        n=num_questions,
        stop=None,
        temperature=0.7,
        )

        # Extract the generated question from the response
        question = response['choices'][0]['message']['content'].strip()
        correct_answer = f"Correct Answer: {random.choice(search_results)}"

        
        # Generate options (3 incorrect + 1 correct)
        correct_option = f"Correct: {random.choice(search_results)}"
        incorrect_options = [f"Incorrect: {random.choice(search_results)}" for _ in range(3)]
        options = incorrect_options + [correct_option]

        # Shuffle the options to randomize the order
        random.shuffle(options)

        # Append question and options to the quiz_questions list
        quiz_questions.append((question, options, correct_answer))

    # Display quiz questions with options
    st.write('Quiz Questions:')
    for i, (question, options, correct_answer) in enumerate(quiz_questions, 1):
        st.write(f"{i}. {question}")
        correct_option = [opt for opt in options if opt.startswith('Correct')][0]

if __name__ == '__main__':
    st.set_page_config(page_title="CB Quiz Generator", page_icon="📝")
    st.title('🤖CB Quiz Generator🧠')

    # User inputs
    quiz_name = st.text_input('Enter Quiz Name:')
    quiz_topic = st.text_input('Enter Quiz Topic:')
    num_questions = st.number_input('Enter Number of Questions:', min_value=1, value=5, step=1)
    pdf_content = st.file_uploader("Upload PDF Content for Questions:", type='pdf')

    # Generate quiz if all inputs are provided
    if quiz_name and quiz_topic and num_questions and pdf_content:
        # Process PDF and create vector database
        vectorDB = get_vectorstore(get_text_chunks(get_pdf_text(pdf_content)))

        # Searching for a specific query in the vector database
        if quiz_topic:
            search_results = search_vectorstore(vectorDB, quiz_topic)

            # Generate quiz questions with options using GPT-3.5-turbo-16k based on the search results
            generate_quiz_questions(search_results, num_questions)
        else:
            st.error('Please provide a topic to search for!')