import gradio as gr
import joblib
import pandas as pd
import numpy as np

pipeline = joblib.load("model.joblib")

NUM = ["lead_time","stays_in_weekend_nights","stays_in_week_nights","adults","children","babies",
       "previous_cancellations","previous_bookings_not_canceled","booking_changes",
       "days_in_waiting_list","adr","required_car_parking_spaces","total_of_special_requests",
       "total_nights","total_guests","revenue_est","risky_repeat"]
CAT = ["hotel","meal","market_segment","distribution_channel",
       "reserved_room_type","deposit_type","customer_type","lead_cat","arrival_season"]

DEFAULTS = {
    "stays_in_weekend_nights":2,"stays_in_week_nights":3,"adults":2,"children":0,"babies":0,
    "previous_bookings_not_canceled":0,"booking_changes":0,"days_in_waiting_list":0,
    "required_car_parking_spaces":0,"meal":"BB","distribution_channel":"TA/TO",
    "reserved_room_type":"A","customer_type":"Transient","arrival_season":"summer",
    "total_guests":2,
}

def predict(hotel, lead_time, deposit_type, market_segment,
            total_of_special_requests, previous_cancellations, adr, total_nights):
    row = DEFAULTS.copy()
    row.update({
        "hotel": hotel, "lead_time": float(lead_time),
        "deposit_type": deposit_type, "market_segment": market_segment,
        "total_of_special_requests": float(total_of_special_requests),
        "previous_cancellations": float(previous_cancellations),
        "adr": float(adr), "total_nights": float(total_nights),
        "revenue_est": float(adr) * max(float(total_nights), 1),
        "risky_repeat": int(float(previous_cancellations) > 0),
        "lead_cat": pd.cut([float(lead_time)], bins=[-1,7,30,90,180,9999],
                    labels=["last_minute","short","medium","long","very_long"])[0],
    })
    X = pd.DataFrame([row])[NUM + CAT]
    prob = float(pipeline.predict_proba(X)[0, 1])
    label = "Low risk" if prob < 0.30 else ("Medium risk" if prob < 0.60 else "High risk")
    return {label: prob}

demo = gr.Interface(
    fn=predict,
    inputs=[
        gr.Dropdown(["City Hotel","Resort Hotel"], label="Hotel type", value="City Hotel"),
        gr.Slider(0, 500, value=30, step=1, label="Lead time (days)"),
        gr.Dropdown(["No Deposit","Non Refund","Refundable"], label="Deposit type", value="No Deposit"),
        gr.Dropdown(["Direct","Online TA","Offline TA/TO","Corporate","Groups"],
                    label="Market segment", value="Online TA"),
        gr.Slider(0, 5, value=1, step=1, label="Special requests"),
        gr.Slider(0, 20, value=0, step=1, label="Previous cancellations"),
        gr.Slider(0, 500, value=100, step=5, label="ADR (€)"),
        gr.Slider(1, 30, value=3, step=1, label="Total nights"),
    ],
    outputs=gr.Label(label="Cancellation risk"),
    title="Hotel Booking Cancellation Risk",
    description="Predict cancellation probability from booking parameters. Educational demo only.",
    examples=[
        ["City Hotel",200,"Non Refund","Online TA",0,3,95,2],
        ["Resort Hotel",7,"No Deposit","Direct",3,0,150,7],
    ],
    flagging_mode="never",
)

if __name__ == "__main__":
    demo.launch()