import gradio as gr
import pandas as pd
import numpy as np
import pickle
import os

# Load model
model_filename = "apartment_rf_model.pkl"
with open(model_filename, mode="rb") as f:
    model_data = pickle.load(f)

model = model_data["model"]
features = model_data["features"]
log_target = model_data["log_target"]

DEFAULT_POP_DENS_CITY    = 4729.0
DEFAULT_POP_DENS_OUTSIDE = 1328.0
DEFAULT_FRG_PCT          = 25.0
DEFAULT_LAT_CITY         = 47.380402
DEFAULT_LON_CITY         = 8.530496
DEFAULT_LAT_OUTSIDE      = 47.424900
DEFAULT_LON_OUTSIDE      = 8.638663


def predict_price(
    rooms, area, tax_income,
    luxurious, furnished, temporary, zurich_city,
    attika, loft, seesicht,
    kreis
):
    # Use data-derived defaults for hidden features
    pop_dens = DEFAULT_POP_DENS_CITY if zurich_city else DEFAULT_POP_DENS_OUTSIDE
    lat      = DEFAULT_LAT_CITY      if zurich_city else DEFAULT_LAT_OUTSIDE
    lon      = DEFAULT_LON_CITY      if zurich_city else DEFAULT_LON_OUTSIDE
    frg_pct  = DEFAULT_FRG_PCT

    # Derived features
    log_area             = np.log1p(area)
    log_rooms            = np.log1p(rooms)
    rooms_area_ratio     = rooms / (area + 1)
    room_per_m2          = area / rooms if rooms > 0 else area
    income_density_score = tax_income * pop_dens / 1e6
    log_income_dens      = np.log1p(income_density_score)
    log_pop_dens         = np.log1p(pop_dens)
    area_rooms_interact  = area * rooms

    # Kreis dummies
    kreis_cols = {f"Kreis {i}": 0 for i in range(1, 13)}
    if kreis in kreis_cols:
        kreis_cols[kreis] = 1

    input_dict = {
        "rooms": rooms,
        "area": area,
        "log_area": log_area,
        "log_rooms": log_rooms,
        "pop_dens": pop_dens,
        "log_pop_dens": log_pop_dens,
        "frg_pct": frg_pct,
        "tax_income": tax_income,
        "income_density_score": income_density_score,
        "log_income_dens": log_income_dens,
        "rooms_area_ratio": rooms_area_ratio,
        "area_rooms_interact": area_rooms_interact,
        "lat": lat,
        "lon": lon,
        "luxurious": int(luxurious),
        "furnished": int(furnished),
        "temporary": int(temporary),
        "zurich_city": int(zurich_city),
        "room_per_m2": room_per_m2,
        **kreis_cols,
        "(ATTIKA)": int(attika),
        "(LOFT)": int(loft),
        "(SEESICHT)": int(seesicht),
        "(LUXURIÖS)": 0,
        "(POOL)": 0,
        "(EXKLUSIV)": 0,
    }

    input_df = pd.DataFrame([input_dict])[features]
    prediction_log = model.predict(input_df)[0]
    prediction = np.expm1(prediction_log)

    return f"Estimated Monthly Rent: **CHF {prediction:,.0f}**"


demo = gr.Interface(
    fn=predict_price,
    inputs=[
        gr.Number(label="Number of Rooms", value=3.0),
        gr.Number(label="Living Area (m²)", value=80),
        gr.Number(label="Municipal Tax Income (CHF)", value=80000),
        gr.Checkbox(label="Luxurious"),
        gr.Checkbox(label="Furnished"),
        gr.Checkbox(label="Temporary"),
        gr.Checkbox(label="Located in Zurich City"),
        gr.Checkbox(label="Attika (Penthouse)"),
        gr.Checkbox(label="Loft"),
        gr.Checkbox(label="Lake View (Seesicht)"),
        gr.Dropdown(
            label="Zurich District (Kreis)",
            choices=["None"] + [f"Kreis {i}" for i in range(1, 13)],
            value="None"
        ),
    ],
    outputs=gr.Markdown(),
    submit_btn="Predict Rent",
    examples=[
        [3.0, 80, 70000, False, False, False, True, False, False, False, "Kreis 3"],
        [4.5, 120, 120000, True, False, False, True, True, False, True, "Kreis 1"],
        [2.0, 55, 55000, False, True, False, False, False, False, False, "None"],
    ],
    title="Zurich Apartment Rent Predictor",
    description=(
        "Predict the monthly rental price for an apartment in the Canton of Zurich.\n\n"
        "This model uses a **Gradient Boosting Regressor** trained on ~800 Zurich apartment listings, "
        "with features including location (district), apartment characteristics, "
        "and the **Income-Density Score** (a neighbourhood affluence proxy)."
    ),
    theme=gr.themes.Soft(),
)

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