from sklearn.preprocessing import MinMaxScaler
import gradio as gr
import pandas as pd
import numpy as np
# Change to regression if needed
from pycaret.regression import load_model, predict_model, setup


# Load your dataframe
df = pd.read_csv("data.csv", index_col=0)  # Replace with your actual data file

clf1 = setup(data=df, target='Mu (kN⋅m)', normalize=False)
# Use the same scaler that was used during training (important!)
scaler = MinMaxScaler()
input_columns = df.columns[:-1].tolist()  # First 4 columns as input
target_column = df.columns[-1]  # Last column as target
scaler.fit(df[input_columns])

# Min-Max scaling per column
min_max_dict = {
    col: (df[col].min(), df[col].max()) for col in input_columns
}

# Load your PyCaret model
# Replace with your actual saved model


def pycaret_predict_function(*inputs):
    # Convert input tuple to DataFrame
    dfx = pd.DataFrame([inputs], columns=input_columns)
    dfx_scaled = pd.DataFrame(scaler.transform(dfx), columns=input_columns)
    print("Input DataFrame for prediction:", dfx_scaled)
    model = load_model("TVAESynthesizer_best")
    prediction = predict_model(model, data=dfx_scaled)
    return prediction["prediction_label"].values[0]

    try:
        pass
    except KeyError as e:
        return 'Error: Prediction label not found. Please check the model output: ' + str(e)


# Create Gradio inputs dynamically based on scaled range
input_components = [
    gr.Slider(
        minimum=min_max_dict[col][0],
        maximum=min_max_dict[col][1],
        step=(min_max_dict[col][1] - min_max_dict[col][0]) / 100,
        label=col,
    ) for col in input_columns
]

# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("## 🔮 Predict the Target Variable using PyCaret Model")

    with gr.Row():
        inputs = [component.render() for component in input_components]

    output = gr.Textbox(label=f"Predicted: {target_column}")

    predict_btn = gr.Button("Predict")

    predict_btn.click(fn=pycaret_predict_function,
                      inputs=inputs, outputs=output)

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