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

# -----------------------------
# Load pipeline model
# -----------------------------
model_path = hf_hub_download(
    repo_id="Rajanan/model-predictive-engine-maintenance-v1",
    filename="best_engine_failure_model_v1.joblib",
    repo_type="model"
)

model = joblib.load(model_path)

# -----------------------------
# UI
# -----------------------------
st.title("Predictive Engine Maintenance System")

st.markdown("""
Predict whether an engine requires maintenance based on sensor inputs.
""")

# -----------------------------
# User Inputs
# -----------------------------
engine_rpm = st.number_input("Engine RPM", min_value=0, max_value=3000, value=800)

lub_oil_pressure = st.number_input("Lub Oil Pressure", min_value=0.0, max_value=10.0, value=3.0)

fuel_pressure = st.number_input("Fuel Pressure", min_value=0.0, max_value=25.0, value=6.0)

coolant_pressure = st.number_input("Coolant Pressure", min_value=0.0, max_value=10.0, value=2.0)

lub_oil_temp = st.number_input("Lub Oil Temperature (°C)", min_value=50.0, max_value=120.0, value=75.0)

coolant_temp = st.number_input("Coolant Temperature (°C)", min_value=50.0, max_value=200.0, value=80.0)

# -----------------------------
# Create input DataFrame
# -----------------------------
input_data = pd.DataFrame([{
    "Engine rpm": engine_rpm,
    "Lub oil pressure": lub_oil_pressure,
    "Fuel pressure": fuel_pressure,
    "Coolant pressure": coolant_pressure,
    "lub oil temp": lub_oil_temp,
    "Coolant temp": coolant_temp
}])

input_data["pressure_diff"] = input_data["Fuel pressure"] - input_data["Coolant pressure"]
input_data["temp_diff"] = input_data["Coolant temp"] - input_data["lub oil temp"]
input_data["rpm_pressure_ratio"] = input_data["Engine rpm"] / (input_data["Fuel pressure"] + 1e-6)

input_data = input_data.drop(columns=["lub oil temp", "Coolant pressure"])

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

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

    # SAME threshold used in training
    threshold = 0.45

    if prob < threshold:
        st.success(" Engine operating normally")
    elif prob < 0.65:
        st.warning(" Maintenance recommended (early risk detected)")
    else:
        st.error("High risk of engine failure! Immediate attention required")