app.py

import gradio as gr
import os
import json
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
from ppt_parser import transfer_to_structure
from functools import lru_cache

# ✅ Get Hugging Face token from Space Secrets
hf_token = os.getenv("HF_TOKEN")

# ✅ Load summarization model (BART)
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")

# ✅ Load Mistral model (memoized to avoid reloading)
@lru_cache(maxsize=1)
def load_mistral():
    tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1", token=hf_token)
    model = AutoModelForCausalLM.from_pretrained(
        "mistralai/Mistral-7B-Instruct-v0.1",
        torch_dtype=torch.float16,
        device_map="auto",
        token=hf_token
    )
    return pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512)

mistral_pipe = load_mistral()

# ✅ Global variable to hold extracted content
extracted_text = ""

def extract_text_from_pptx_json(parsed_json: dict) -> str:
    text = ""
    for slide in parsed_json.values():
        for shape in slide.values():
            if shape.get('type') == 'group':
                for group_shape in shape.get('group_content', {}).values():
                    if group_shape.get('type') == 'text':
                        for para_key, para in group_shape.items():
                            if para_key.startswith("paragraph_"):
                                text += para.get("text", "") + "\n"
            elif shape.get('type') == 'text':
                for para_key, para in shape.items():
                    if para_key.startswith("paragraph_"):
                        text += para.get("text", "") + "\n"
    return text.strip()

def handle_pptx_upload(pptx_file):
    global extracted_text
    tmp_path = pptx_file.name
    parsed_json_str, _ = transfer_to_structure(tmp_path, "images")
    parsed_json = json.loads(parsed_json_str)
    extracted_text = extract_text_from_pptx_json(parsed_json)
    return extracted_text or "No readable text found in slides."

def summarize_text():
    global extracted_text
    if not extracted_text:
        return "Please upload and extract text from a PPTX file first."
    summary = summarizer(extracted_text, max_length=200, min_length=50, do_sample=False)[0]['summary_text']
    return summary

def clarify_concept(question):
    global extracted_text
    if not extracted_text:
        return "Please upload and extract text from a PPTX file first."
    prompt = f"[INST] Use the following context to answer the question:\n\n{extracted_text}\n\nQuestion: {question} [/INST]"
    response = mistral_pipe(prompt)[0]["generated_text"]
    return response.replace(prompt, "").strip()

# ✅ Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("## 🧠 AI-Powered Study Assistant for PowerPoint Lectures (Mistral 7B)")

    pptx_input = gr.File(label="📂 Upload PPTX File", file_types=[".pptx"])
    extract_btn = gr.Button("📜 Extract & Summarize")

    extracted_output = gr.Textbox(label="📄 Extracted Text", lines=10, interactive=False)
    summary_output = gr.Textbox(label="📝 Summary", interactive=False)

    extract_btn.click(handle_pptx_upload, inputs=[pptx_input], outputs=[extracted_output])
    extract_btn.click(summarize_text, outputs=[summary_output])

    question = gr.Textbox(label="❓ Ask a Question")
    ask_btn = gr.Button("💬 Ask Mistral")
    ai_answer = gr.Textbox(label="🤖 Mistral Answer", lines=4)

    ask_btn.click(clarify_concept, inputs=[question], outputs=[ai_answer])

if __name__ == "__main__":
    demo.launch()