app.py

# ✅ Install dependencies

# 📚 Imports
import fitz  # PyMuPDF
import requests
import gradio as gr
import tempfile
import os
import io

# 🔑 Enter your OpenRouter API key here
OPENROUTER_API_KEY = "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"  

# Global variable to store the extracted text
pdf_text = ""

# 📄 Extract text from PDF
def extract_text_from_pdf(file_obj):
    global pdf_text
    
    if file_obj is None:
        return "Please upload a PDF file first."
    
    try:
        # Get the file path from the file object
        # In Gradio, the file object has a name attribute that contains the path
        file_path = file_obj.name
        
        # Now open the file with PyMuPDF
        doc = fitz.open(file_path)
        text = ""
        for page in doc:
            text += page.get_text()
        doc.close()
        
        # Store the text for later use
        pdf_text = text
        
        # Return preview of the extracted text
        preview = text[:500] + "..." if len(text) > 500 else text
        return f"✅ PDF uploaded and processed successfully. Preview:\n\n{preview}"
    
    except Exception as e:
        return f"❌ Error processing PDF: {str(e)}"

# 💬 Ask the open-source LLM (Mistral-7B via OpenRouter)
def ask_open_source_llm(question, model_choice="nvidia/llama-3.1-nemotron-nano-8b-v1:free"):
    global pdf_text
    
    if not pdf_text:
        return "⚠️ Please upload a PDF document first."
    
    # Limit text to prevent token overflow
    limited_text = pdf_text[:3000]  # First 3000 characters
    
    # Create prompt based on question
    if not question:
        prompt = f"Summarize the following document:\n\n{limited_text}"
    else:
        prompt = f"The document says:\n\n{limited_text}\n\nNow answer this: {question}"
    
    # Call the API
    url = "https://openrouter.ai/api/v1/chat/completions"
    headers = {
        "Authorization": f"Bearer {OPENROUTER_API_KEY}",
        "Content-Type": "application/json"
    }
    
    data = {
        "model": model_choice,
        "messages": [{"role": "user", "content": prompt}]
    }
    
    try:
        response = requests.post(url, headers=headers, json=data)
        if response.status_code == 200:
            return response.json()["choices"][0]["message"]["content"]
        else:
            return f"❌ Error: {response.text}"
    except Exception as e:
        return f"❌ An error occurred: {str(e)}"

# Gradio app function
def process_query(pdf_file, question, model_choice):
    # First extract text if a PDF is uploaded
    if pdf_file is not None:
        result = extract_text_from_pdf(pdf_file)
        if result.startswith("❌ Error"):
            return result
    
    # Then process the question
    if question:
        return ask_open_source_llm(question, model_choice)
    else:
        return ask_open_source_llm("Please summarize this document.", model_choice)

# Create Gradio interface
with gr.Blocks(title="PDF Document Analysis") as app:
    gr.Markdown("# 📚 PDF Document Analysis with LLM")
    gr.Markdown("Upload a PDF document and ask questions about its content.")
    
    with gr.Row():
        with gr.Column(scale=1):
            pdf_input = gr.File(label="Upload PDF Document", file_types=[".pdf"])
            model_choice = gr.Dropdown(
                choices=[
                    "nvidia/llama-3.1-nemotron-nano-8b-v1:free",
                    "mistralai/mistral-7b-instruct-v0.1:free",
                    "meta-llama/llama-2-13b-chat:free"
                ],
                label="LLM Model",
                value="nvidia/llama-3.1-nemotron-nano-8b-v1:free"
            )
            question_input = gr.Textbox(label="Ask a question (or leave empty for summary)", lines=2)
            submit_btn = gr.Button("Process", variant="primary")
        
        with gr.Column(scale=2):
            output = gr.Textbox(label="Response", lines=15)
    
    # Set up event handlers
    submit_btn.click(
        fn=process_query,
        inputs=[pdf_input, question_input, model_choice],
        outputs=output
    )
    
    gr.Markdown("### 📝 Notes")
    gr.Markdown("- For large documents, only the first 3000 characters are analyzed")
    gr.Markdown("- You can change the LLM model from the dropdown menu")
    gr.Markdown("- Leave the question field empty to get a general summary")

# Launch the app
app.launch(debug=True, share=True)