import gradio as gr
import torch
from transformers import pipeline

# Load the Qwen2.5-72B-Instruct model
pipe = pipeline(
    "text-generation",
    model="HuggingFaceH4/Qwen2.5-72B-Instruct",
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

# Initial messages list for chat history
messages = [
    {"role": "system", "content": "You are an experienced Fashion designer who starts conversation with proper greeting, "
                                   "giving valuable and catchy fashion advice and suggestions, stays to the point and precise."}
]

# Function to reset the chat
def reset_chat():
    global messages
    messages = []  # Reset the message history
    return [], "New Chat"

# Function to handle questionnaire submission
def submit_questionnaire(name, age, location, gender, ethnicity, height, weight,
                         style_preference, color_palette, everyday_style):
    # Store questionnaire responses in a DataFrame or process as needed
    # This is just a placeholder to indicate processing
    return "Thank you for completing the questionnaire!"

# Function to handle chat
def chat(user_input, messages):
    if user_input:
        # Append user message to the conversation history
        messages.append({"role": "user", "content": user_input})

        # Prepare input for the model
        input_text = "\n".join([f"{msg['role']}: {msg['content']}" for msg in messages])

        # Generate a response using the model
        try:
            response = pipe(input_text, max_new_tokens=256)  # Using the pipeline

            # Check if response is valid and structured correctly
            if isinstance(response, list) and len(response) > 0:
                response_content = response[0]['generated_text'].strip()  # Accessing generated text
            else:
                response_content = "Sorry, I couldn't generate a response."

        except Exception as e:
            response_content = f"Error: {str(e)}"

        # Store assistant response in the chat history
        messages.append({"role": "assistant", "content": response_content})

        return messages, response_content
    return messages, ""

# Gradio Interface
with gr.Blocks() as demo:
    gr.Markdown("## Fashion Assistant Chatbot")

    # Sidebar for user inputs
    with gr.Row():
        with gr.Column():
            name = gr.Textbox(label="Name")
            age = gr.Number(label="Age", value=25, minimum=1, maximum=100)
            location = gr.Textbox(label="Location")
            gender = gr.Radio(label="Gender", choices=["Male", "Female", "Other"])
            ethnicity = gr.Radio(label="Ethnicity", choices=["Asian", "Black", "Hispanic", "White", "Other"])
            height = gr.Number(label="Height (cm)", value=170, minimum=50, maximum=250)
            weight = gr.Number(label="Weight (kg)", value=70, minimum=20, maximum=200)

        with gr.Column():
            submit_btn = gr.Button("Submit Questionnaire")
            reset_btn = gr.Button("Reset Chat")

    # Questionnaire with fashion-related questions
    style_preference = gr.Radio(label="Which style do you prefer the most?", choices=["Casual", "Formal", "Streetwear", "Athleisure", "Baggy"])
    color_palette = gr.Radio(label="What color palette do you wear often?", choices=["Neutrals", "Bright Colors", "Pastels", "Dark Shades"])
    everyday_style = gr.Radio(label="How would you describe your everyday style?", choices=["Relaxed", "Trendy", "Elegant", "Bold"])

    # Chat functionality
    chatbox = gr.Chatbot(type='messages')
    user_input = gr.Textbox(label="Your Message", placeholder="Type your message here...")

    # Connect the buttons to their respective functions
    output_message = gr.Textbox(label="Output Message")
    submit_btn.click(submit_questionnaire, inputs=[name, age, location, gender, ethnicity, height, weight,
                                                    style_preference, color_palette, everyday_style], outputs=output_message)

    reset_btn.click(reset_chat, outputs=[chatbox, output_message])  # Corrected outputs
    user_input.submit(chat, inputs=[user_input, chatbox], outputs=[chatbox, user_input])

# Run the app
demo.launch()