import streamlit as st
import joblib
import pandas as pd
import os

# Debug: List files in current directory to see what's available
st.write("Current directory files:", os.listdir("."))

# Load model and columns from the same src folder
model = joblib.load("churn_predictor_xgb.pkl")
columns = joblib.load("churn_model_columns.pkl")

st.set_page_config(page_title="RUCTURO", page_icon="🟣", layout="centered")

# Premium futuristic CSS
st.markdown("""
<style>
    .main {
        background: linear-gradient(to bottom, #0F172A, #1E293B);
        color: #F1F5F9;
    }
    h1, h2, h3 {
        color: #6D28D9;
        text-shadow: 0 0 10px #22D3EE;
        font-family: 'Orbitron', sans-serif;
    }
    .stButton>button {
        background: linear-gradient(to right, #6D28D9, #22D3EE);
        color: white;
        border: none;
        border-radius: 12px;
        padding: 12px 24px;
        font-size: 18px;
        box-shadow: 0 0 20px #22D3EE;
        transition: all 0.3s;
    }
    .stButton>button:hover {
        transform: scale(1.05);
        box-shadow: 0 0 30px #6D28D9;
    }
    .risk-high { color: #F87171; font-size: 36px; font-weight: bold; text-shadow: 0 0 10px #F87171; }
    .risk-medium { color: #FBBF24; font-size: 32px; font-weight: bold; }
    .risk-low { color: #10B981; font-size: 32px; font-weight: bold; }
    @import url('https://fonts.googleapis.com/css2?family=Orbitron:wght@700&display=swap');
</style>
""", unsafe_allow_html=True)

st.title("🟣 RUCTURO")
st.markdown("### Premium AI-Powered Customer Churn Predictor for SaaS Tools")
st.markdown("Predict churn risk instantly and get actionable retention insights.")

# Input form
with st.form("churn_form"):
    col1, col2 = st.columns(2)
    with col1:
        tenure = st.slider("Tenure (months)", 0, 72, 12)
        monthly = st.slider("Monthly Charges ($)", 18, 120, 70)
        senior = st.radio("Senior Citizen", [0, 1], format_func=lambda x: "Yes" if x else "No")
        contract = st.selectbox("Contract Type", ["Month-to-month", "One year", "Two year"])
        internet = st.selectbox("Internet Service", ["DSL", "Fiber optic", "No"])
        tech_support = st.selectbox("Tech Support", ["Yes", "No"])
    with col2:
        online_security = st.selectbox("Online Security", ["Yes", "No"])
        payment = st.selectbox("Payment Method", ["Electronic check", "Mailed check", "Bank transfer (automatic)", "Credit card (automatic)"])
        paperless = st.selectbox("Paperless Billing", ["Yes", "No"])
        num_services = st.slider("Number of Services (approx)", 0, 10, 5)
        has_internet = st.radio("Has Internet", [0, 1], format_func=lambda x: "Yes" if x else "No")

    submitted = st.form_submit_button("Predict Churn Risk")

if submitted:
    data = {
        'tenure': tenure,
        'MonthlyCharges': monthly,
        'TotalCharges': monthly * (tenure + 1),
        'SeniorCitizen': senior,
        'Num_Services': num_services,
        'Has_Internet': has_internet,
        'TotalCharges_per_Tenure': monthly,
        'Charges_Increase': 0,
        'Is_Month_to_Month': 1 if contract == "Month-to-month" else 0,
        'Is_Fiber_Optic': 1 if internet == "Fiber optic" else 0,
        'Has_No_TechSupport': 1 if tech_support == "No" else 0,
        'PaperlessBilling_Yes': 1 if paperless == "Yes" else 0,
    }

    df = pd.DataFrame([data])
    df = pd.get_dummies(df, columns=['Contract', 'InternetService', 'TechSupport', 'OnlineSecurity', 'PaymentMethod'])
    df = df.reindex(columns=columns, fill_value=0)

    prob = model.predict_proba(df)[0, 1]

    st.markdown("---")
    st.markdown(f"### Churn Probability: **{prob:.1%}**")

    if prob >= 0.4:
        st.markdown('<p class="risk-high">🟥 HIGH RISK – Immediate Action Needed</p>', unsafe_allow_html=True)
        st.warning("• Short tenure + month-to-month contract\n• Fiber optic service\n• No tech support\n• Electronic check payment\n**Recommendation**: Offer discount, upgrade, or dedicated support")
    elif prob >= 0.2:
        st.markdown('<p class="risk-medium">🟧 Medium Risk</p>', unsafe_allow_html=True)
        st.info("Monitor closely – consider proactive engagement")
    else:
        st.markdown('<p class="risk-low">🟩 Low Risk – Strong Retention</p>', unsafe_allow_html=True)
        st.success("Excellent loyalty signals – keep up the great service!")

st.markdown("---")
st.markdown("Powered by XGBoost • Built for SaaS teams • Premium futuristic design • © 2025")