import gradio as gr
import numpy as np
from joblib import load
from tensorflow.keras.models import load_model
import tensorflow as tf
import requests

# Load models from Hugging Face (update with your model URLs)
def load_huggingface_model(model_url):
    response = requests.get(model_url)
    with open("model_temp.joblib" if model_url.endswith(".joblib") else "model_temp.h5", "wb") as f:
        f.write(response.content)

    return load("model_temp.joblib") if model_url.endswith(".joblib") else load_model("model_temp.h5")

# URLs of your models on Hugging Face
MODEL_URLS = {
    'Deep Learning': 'https://huggingface.co/your-username/best_DeepLearning_model/resolve/main/best_DeepLearning_model.h5',
    'Support Vector Regression': 'https://huggingface.co/your-username/best_SVR_model/resolve/main/best_SVR_model.joblib',
    'Random Forest Regression': 'https://huggingface.co/your-username/best_RandomForestRegressor_model/resolve/main/best_RandomForestRegressor_model.joblib',
    'Decision Tree Regression': 'https://huggingface.co/your-username/best_DecisionTreeRegressor_model/resolve/main/best_DecisionTreeRegressor_model.joblib'
}

# Prediction function
def multi_inputs(input1, input2, input3, input4, input5, input6, input7, input8, input9, input10, input11, input12):
    try:
        model = load_huggingface_model(MODEL_URLS[input12])

        # Convert numeric inputs to float
        input4, input5, input9 = float(input4), float(input5), float(input9)

        # Create input list with proper types
        inputs_to_transform = [
            str(input1), str(input2), str(input3), input4, input5,
            float(input6), float(input7), float(input8), input9,
            str(input10), str(input11)
        ]

        # Reshape input for prediction
        input_data = np.array([inputs_to_transform], dtype=np.float32)

        if input12 == 'Deep Learning':
            tensor_input = tf.convert_to_tensor(input_data)
            prediction = model.predict(tensor_input, verbose=0)
            return float(prediction[0][0])
        else:
            prediction = model.predict(input_data)
            return float(prediction[0])

    except Exception as e:
        return f"Error in prediction: {str(e)}"

# Gradio Interface
interface = gr.Interface(
    fn=multi_inputs,
    inputs=[
        gr.Radio(['male', 'female'], label='Gender'),
        gr.Dropdown(['Race1', 'Race2'], label='Race'),
        gr.Dropdown([20, 30, 40], label='Age'),
        gr.Number(label='Height (cm)'),
        gr.Number(label='Weight (kg)'),
        gr.Radio([0.0, 1.0], label='Diabetes'),
        gr.Radio([0.0, 1.0], label='Simvastatin'),
        gr.Radio([0.0, 1.0], label='Amiodarone'),
        gr.Number(label='INR'),
        gr.Dropdown(['Cyp2C9_A', 'Cyp2C9_B'], label='Cyp2C9 genotypes'),
        gr.Dropdown(['VKORC1_A', 'VKORC1_B'], label='VKORC1 genotypes'),
        gr.Dropdown(['Deep Learning', 'Support Vector Regression', 'Random Forest Regression', 'Decision Tree Regression'], label='Model Selection')
    ],
    outputs="number",
    live=False
)

if __name__ == '__main__':
    interface.launch(debug=True)