app.py

import os
import torch
import faiss
import wikipediaapi
from fpdf import FPDF
from transformers import AutoTokenizer, AutoModelForCausalLM
from sentence_transformers import SentenceTransformer
import gradio as gr
from googletrans import Translator

# Check if CUDA (GPU) is available and set the device
device = "cuda" if torch.cuda.is_available() else "cpu"

# Initialize Wikipedia API with a User-Agent
wiki_wiki = wikipediaapi.Wikipedia(
    language="en",
    user_agent="AdaptiveLearningApp/1.0 (Contact: [email protected])"
)

# PDF Generation from Wikipedia Content
def generate_pdf_from_wikipedia(subject, topic):
    page = wiki_wiki.page(topic)
    if not page.exists():
        return None, f"Topic '{topic}' not found on Wikipedia."
    
    pdf = FPDF()
    pdf.add_page()
    pdf.set_font("Arial", size=12)
    pdf.cell(200, 10, txt=f"{subject.upper()} - {topic.upper()}", ln=True, align="C")
    pdf.ln(10)

    # Add Wikipedia content with basic formatting
    for line in page.text.split("\n"):
        pdf.multi_cell(0, 10, line)
        pdf.ln(5)
    
    pdf_path = f"{topic.replace(' ', '_')}.pdf"
    pdf.output(pdf_path)
    return pdf_path, f"PDF for topic '{topic}' has been generated successfully."

# Chunking Text
def chunk_text(text, chunk_size=300):
    sentences = text.split(". ")
    chunks, current_chunk = [], ""
    for sentence in sentences:
        if len(current_chunk) + len(sentence) < chunk_size:
            current_chunk += sentence + ". "
        else:
            chunks.append(current_chunk.strip())
            current_chunk = sentence + ". "
    if current_chunk:
        chunks.append(current_chunk.strip())
    return chunks

# Creating Embeddings
def create_embeddings(chunks):
    embeddings = sentence_model.encode(chunks, convert_to_tensor=False)
    return embeddings

# Storing Embeddings in FAISS
def store_in_faiss(chunks, embeddings):
    dimension = embeddings[0].shape[0]
    res = faiss.StandardGpuResources()  # GPU resources
    index = faiss.IndexFlatL2(dimension)
    index = faiss.index_cpu_to_gpu(res, 0, index)
    index.add(embeddings)
    return index

# Generate Quiz using BLOOM Model
def generate_quiz(content):
    prompt = f"Generate 10 quiz questions from the following content:\n{content}"
    inputs = bloom_tokenizer(prompt, return_tensors="pt", truncation=True).to(device)
    outputs = bloom_model.generate(inputs["input_ids"], max_length=512, num_return_sequences=1)
    quiz = bloom_tokenizer.decode(outputs[0], skip_special_tokens=True)
    return quiz

# Translate Content to Urdu
def translate_to_urdu(content):
    translator = Translator()
    translation = translator.translate(content, src="en", dest="ur")
    return translation.text

# Retrieve Content by Topic
def get_content_by_topic(topic):
    page = wiki_wiki.page(topic)
    if not page.exists():
        return f"Topic '{topic}' not found on Wikipedia."
    return page.text

# Evaluate Quiz Results
def evaluate_quiz(user_answers, correct_answers):
    score = 0
    feedback = []
    for user, correct in zip(user_answers, correct_answers):
        if user.strip().lower() == correct.strip().lower():
            score += 1
            feedback.append("Correct")
        else:
            feedback.append(f"Incorrect. Correct answer: {correct}")
    return score, feedback

# Adaptive Learning App
def adaptive_learning_app(subject, topic):
    content = get_content_by_topic(topic)
    if "not found" in content:
        return None, content

    # Chunk Text and Create Embeddings
    chunks = chunk_text(content)
    embeddings = create_embeddings(chunks)
    faiss_index = store_in_faiss(chunks, embeddings)
    return content, chunks, embeddings, faiss_index, "Processing complete."

# Gradio User Interface
def main_ui():
    def process_input(subject, topic):
        global content, chunks, embeddings, faiss_index
        content, chunks, embeddings, faiss_index, message = adaptive_learning_app(subject, topic)
        return content, message

    def generate_pdf(subject, topic):
        pdf_path, message = generate_pdf_from_wikipedia(subject, topic)
        return pdf_path, message

    def interactive_quiz(content):
        quiz = generate_quiz(content)
        return quiz

    def urdu_translation(content):
        return translate_to_urdu(content)

    def submit_answers(user_answers, correct_answers):
        score, feedback = evaluate_quiz(user_answers, correct_answers)
        return f"Your Score: {score}/{len(correct_answers)}", feedback

    # Gradio Interface
    interface = gr.Blocks()
    
    with interface:
        with gr.Row():
            gr.Markdown("### Adaptive Learning App with Wikipedia Integration")
        
        with gr.Row():
            subject_input = gr.Textbox(label="Enter Subject")
            topic_input = gr.Textbox(label="Enter Topic")
            process_button = gr.Button("Process")
            course_material = gr.TextArea(label="Course Material", lines=15)
            process_button.click(
                process_input,
                inputs=[subject_input, topic_input],
                outputs=[course_material, gr.Textbox(label="Status")]
            )
        
        with gr.Row():
            pdf_button = gr.Button("Generate PDF")
            pdf_download = gr.File(label="Download PDF")
            pdf_button.click(
                generate_pdf,
                inputs=[subject_input, topic_input],
                outputs=[pdf_download, gr.Textbox(label="PDF Status")]
            )
        
        with gr.Row():
            quiz_button = gr.Button("Generate Quiz")
            quiz_view = gr.TextArea(label="Quiz Questions", lines=10)
            quiz_button.click(
                interactive_quiz,
                inputs=course_material,
                outputs=quiz_view
            )
        
        with gr.Row():
            urdu_button = gr.Button("Translate to Urdu")
            urdu_translation_view = gr.TextArea(label="Urdu Translation", lines=10)
            urdu_button.click(
                urdu_translation,
                inputs=course_material,
                outputs=urdu_translation_view
            )
        
        with gr.Row():
            user_answers = gr.Textbox(label="Your Answers (comma-separated)")
            submit_button = gr.Button("Submit Answers")
            result_output = gr.Textbox(label="Quiz Result")
            feedback_output = gr.TextArea(label="Feedback", lines=10)
            submit_button.click(
                submit_answers,
                inputs=[user_answers, quiz_view],
                outputs=[result_output, feedback_output]
            )

    interface.launch()

# Load Models
bloom_model = AutoModelForCausalLM.from_pretrained("bigscience/bloom-560m").to(device)
bloom_tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m")
sentence_model = SentenceTransformer("all-MiniLM-L6-v2", device=device)

# Run the App
main_ui()