app.py

import gradio as gr
from fastapi import FastAPI
from risk_model import predict_risk, retrain_model, get_history_df

app = FastAPI()
gradio_app = gr.Blocks()

with gradio_app:
    gr.Markdown("## 🔥 Heating Mantle Safety Risk Predictor")

    with gr.Row():
        temp = gr.Number(label="Max Temperature (°C)", value=100)
        duration = gr.Number(label="Duration (min)", value=30)

    with gr.Row():
        predict_btn = gr.Button("🔍 Predict")
        retrain_btn = gr.Button("🔁 Retrain Model")

    result = gr.Textbox(label="Risk Prediction")
    score = gr.Textbox(label="Confidence (%)")
    retrain_output = gr.Textbox(label="Retrain Status")

    history_table = gr.Dataframe(headers=["Temperature", "Duration", "Risk", "Confidence"], label="📈 Prediction History")

    def classify(temp, duration):
        if temp <= 0 or duration <= 0:
            return "Invalid Input", "Use values > 0", get_history_df()
        risk, confidence = predict_risk(temp, duration)
        emoji = "🟢" if risk == "Low" else "🟠" if risk == "Moderate" else "🔴"
        return f"{emoji} {risk}", f"{confidence}%", get_history_df()

    predict_btn.click(classify, inputs=[temp, duration], outputs=[result, score, history_table])
    retrain_btn.click(retrain_model, outputs=[retrain_output])

# Mount Gradio onto FastAPI
@app.get("/")
def read_root():
    return {"message": "Heating Mantle Risk API is running!"}

app = gr.mount_gradio_app(app, gradio_app, path="/predict-ui")