app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM, AutoModelForSeq2SeqLM
from PyPDF2 import PdfReader


# Models and Tokenizers Setup
models = {
    "Text Generator (Bloom)": {
        "model": AutoModelForCausalLM.from_pretrained("bigscience/bloom-560m"),
        "tokenizer": AutoTokenizer.from_pretrained("bigscience/bloom-560m"),
    },
    "PDF Summarizer (T5)": {
        "model": AutoModelForSeq2SeqLM.from_pretrained("aaliyaan/t5-small-finetuned-career"),
        "tokenizer": AutoTokenizer.from_pretrained("aaliyaan/t5-small-finetuned-career", use_fast=False),
    },
    "Broken Answer (T0pp)": {
        "model": AutoModelForSeq2SeqLM.from_pretrained("bigscience/T0pp"),
        "tokenizer": AutoTokenizer.from_pretrained("bigscience/T0pp", use_fast=False),
    },
}

# Chat Function
def chat_with_model(model_choice, user_message, chat_history, file=None):
    if model_choice == "PDF Summarizer (T5)" and file is not None:
        pdf_text = extract_text_from_pdf(file)
        user_message += f"\n\nPDF Content:\n{pdf_text}"

    if not user_message.strip():
        return chat_history

    model_info = models[model_choice]
    tokenizer = model_info["tokenizer"]
    model = model_info["model"]

    # Tokenize Input
    inputs = tokenizer(user_message, return_tensors="pt", padding=True, truncation=True, max_length=512)
    
    # Adjust max_length and parameters for the PDF summarizer
    max_length = 150
    num_beams = 5
    outputs = model.generate(
        **inputs, 
        max_length=max_length, 
        num_beams=num_beams, 
        early_stopping=True, 
        no_repeat_ngram_size=2
    )
    
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)

    # Update Chat History
    chat_history.append((user_message, response))
    return chat_history

# Function to Extract Text from PDF
def extract_text_from_pdf(file):
    from PyPDF2 import PdfReader
    reader = PdfReader(file.name)
    text = "\n".join(page.extract_text() for page in reader.pages if page.extract_text())
    return text

# Interface Setup
def create_chat_interface():
    with gr.Blocks(css=""" 
        .chatbox {
            background-color: #f7f7f8;
            border-radius: 12px;
            padding: 16px;
            font-family: 'Segoe UI', Tahoma, sans-serif;
        }
        .chat-title {
            font-size: 24px;
            font-weight: bold;
            text-align: center;
            margin-bottom: 12px;
            color: #3a9fd6;
        }
    """) as interface:
        gr.Markdown("<div class='chat-title'>GPT-Style Chat Interface</div>")

        with gr.Row():
            model_choice = gr.Dropdown(
                choices=list(models.keys()),
                value="Text Generator (Bloom)",
                label="Select Model"
            )

        chat_history = gr.Chatbot(label="Chat History", elem_classes="chatbox")

        user_message = gr.Textbox(
            placeholder="Type your message here...",
            show_label=False,
            elem_classes="chatbox",
        )

        file_input = gr.File(label="Upload PDF", visible=False, file_types=[".pdf"])

        def toggle_pdf_input(selected_model):
            return gr.update(visible=(selected_model == "PDF Summarizer (T5)"))

        model_choice.change(fn=toggle_pdf_input, inputs=model_choice, outputs=file_input)

        send_button = gr.Button("Send")

        # Link the send button to the chat function
        send_button.click(
            chat_with_model,
            inputs=[model_choice, user_message, chat_history, file_input],
            outputs=chat_history,
        )

    return interface

if __name__ == "__main__":
    interface = create_chat_interface()
    interface.launch(server_name="0.0.0.0", server_port=7860)