import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# -------------------------
# HuggingFace model to use
# -------------------------
MODEL_NAME = "tiiuae/falcon-7b-instruct"  # you can change to any hosted model

# -------------------------
# Load model and tokenizer
# -------------------------
print("Loading model...")
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForCausalLM.from_pretrained(MODEL_NAME)
device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)

# -------------------------
# System prompt
# -------------------------
SYSTEM_PROMPT = (
    "You are a helpful, creative AI assistant. "
    "Your creator is Austin. Answer clearly and politely."
)

# -------------------------
# Chat function
# -------------------------
def chat_with_ai(user_input, history=[]):
    full_prompt = SYSTEM_PROMPT + "\n"
    for i, (u, r) in enumerate(history):
        full_prompt += f"User: {u}\nAI: {r}\n"
    full_prompt += f"User: {user_input}\nAI:"

    inputs = tokenizer(full_prompt, return_tensors="pt").to(device)
    outputs = model.generate(**inputs, max_new_tokens=200)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    response = response.split("AI:")[-1].strip()

    history.append((user_input, response))
    return response, history

# -------------------------
# Build Gradio GUI
# -------------------------
with gr.Blocks() as demo:
    gr.Markdown("# Austin's AI Chatbot")
    gr.Markdown("This chatbot was created by **Austin**. Chat with it below!")

    chatbot = gr.Chatbot()
    user_input = gr.Textbox(placeholder="Type your message here...")
    submit_btn = gr.Button("Send")
    history_state = gr.State([])

    submit_btn.click(
        chat_with_ai,
        inputs=[user_input, history_state],
        outputs=[chatbot, history_state]
    )

# Launch app
demo.launch()