import gradio as gr
import pickle
import numpy as np
from sklearn.preprocessing import StandardScaler

# Load the pre-trained model at startup
with open("rf_model.pkl", "rb") as f:
    MODEL = pickle.load(f)

def predict(f1, f2, f3, f4, f5):
    data = np.array([[f1, f2, f3, f4, f5]])
    scaled = StandardScaler().fit_transform(data)
    return int(MODEL.predict(scaled)[0])

if __name__ == "__main__":
    demo = gr.Interface(
        fn=predict,
        inputs=[
            gr.Slider(0, 10, value=5, label="Feature 1"),
            gr.Slider(0, 10, value=3, label="Feature 2"),
            gr.Slider(0, 10, value=7, label="Feature 3"),
            gr.Slider(0, 10, value=6, label="Feature 4"),
            gr.Slider(0, 10, value=4, label="Feature 5"),
        ],
        outputs=gr.Textbox(label="Prediction"),
        title="🐨 TaskMaster Job Scheduler",
        description="Enter five features and get a RandomForest prediction.",
    )
    demo.launch()  # on Spaces, this binds to 0.0.0.0:7860 and blocks the process