import streamlit as st
import numpy as np
import joblib
import pandas as pd
from huggingface_hub import hf_hub_download


# ----------------------------------------------------
# Load trained model
# ----------------------------------------------------
MODEL_PATH = hf_hub_download(
    repo_id="raj2261992/predictive_maintenance_model",
    filename="engine_condition_xgboost_v1.joblib"
)

model = joblib.load(MODEL_PATH)


# ----------------------------------------------------
# Page Config
# ----------------------------------------------------
st.set_page_config(
    page_title="Engine Health Prediction",
    layout="centered"
)

st.title("🛠️ Predictive Maintenance – Engine Health")
st.markdown("Enter live sensor values to predict engine condition.")

st.divider()


# ----------------------------------------------------
# Sidebar Inputs
# ----------------------------------------------------
st.sidebar.header("Sensor Inputs")

engine_rpm = st.sidebar.number_input("Engine RPM", min_value=0.0, value=1500.0)

lub_oil_pressure = st.sidebar.number_input(
    "Lub Oil Pressure", min_value=0.0, value=3.5
)

fuel_pressure = st.sidebar.number_input(
    "Fuel Pressure", min_value=0.0, value=4.0
)

coolant_pressure = st.sidebar.number_input(
    "Coolant Pressure", min_value=0.0, value=2.0
)

lub_oil_temp = st.sidebar.number_input(
    "Lub Oil Temperature", min_value=0.0, value=80.0
)

coolant_temp = st.sidebar.number_input(
    "Coolant Temperature", min_value=0.0, value=75.0
)


# ----------------------------------------------------
# Prediction
# ----------------------------------------------------
if st.button("Predict Engine Condition"):

    # IMPORTANT: Must be DataFrame (not NumPy)
    input_data = pd.DataFrame([{
        "Engine rpm": float(engine_rpm),
        "Lub oil pressure": float(lub_oil_pressure),
        "Fuel pressure": float(fuel_pressure),
        "Coolant pressure": float(coolant_pressure),
        "lub oil temp": float(lub_oil_temp),
        "Coolant temp": float(coolant_temp),
    }])

    # Debug display (optional)
    st.write("### Input Data")
    st.write(input_data)

    # Predict probability
    prob = model.predict_proba(input_data)[0][1]

    threshold = 0.45
    prediction = int(prob >= threshold)

    st.subheader("Prediction Result")

    if prediction == 0:
        st.success("Engine Condition: NORMAL")
    else:
        st.error("Engine Condition: FAULTY / AT RISK")

    st.metric("Failure Probability", f"{prob:.2%}")

    st.divider()

    st.write("### Input Summary")
    st.table(input_data)


# ----------------------------------------------------
# Footer
# ----------------------------------------------------
st.markdown("---")
st.caption("Predictive Maintenance Dashboard | Built with Streamlit")