import gradio as gr
from huggingface_hub import InferenceClient
import threading
import time

# Use a smaller model
client = InferenceClient("distilgpt2")

def generate_with_timeout(prompt, max_new_tokens, temperature, top_p, timeout=10):
    result = []
    def target():
        try:
            response = client.text_generation(
                prompt,
                max_new_tokens=max_new_tokens,
                temperature=temperature,
                top_p=top_p,
                do_sample=True,
            )
            result.append(response)
        except Exception as e:
            result.append(str(e))

    thread = threading.Thread(target=target)
    thread.start()
    thread.join(timeout)
    
    if thread.is_alive():
        return "I'm sorry, but I'm having trouble generating a response right now. Could you try again?"
    elif result:
        return result[0]
    else:
        return "An error occurred while generating the response."

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    # Construct the prompt from history and current message
    prompt = system_message + "\n\n"
    for user_msg, bot_msg in history:
        prompt += f"Human: {user_msg}\nAI: {bot_msg}\n"
    prompt += f"Human: {message}\nAI:"

    response = generate_with_timeout(prompt, max_tokens, temperature, top_p)

    # Extract only the AI's response
    ai_response = response.split("AI:")[-1].strip()
    return ai_response

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a helpful AI assistant.", label="System message"),
        gr.Slider(minimum=1, maximum=100, value=50, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.9,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
)

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