#!/usr/bin/env python3
# Author: Simon-Pierre Boucher — contact@spboucher.ai
#
"""Step 1 — Extract the house sample from data/raw/louka.db.

Output: data/processed/houses.parquet (n = 17,087 rows expected).
"""

import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parents[1]))

from src import config
from src.data import load_houses


def main():
    print(f"Reading {config.DB_PATH} ...")
    df = load_houses(config.DB_PATH)
    print(f"  {len(df):,} houses with a valid price and description")
    print(f"  price: min=${df['price'].min():,.0f}  "
          f"median=${df['price'].median():,.0f}  max=${df['price'].max():,.0f}")

    config.DATA_PROCESSED.mkdir(parents=True, exist_ok=True)
    df.to_parquet(config.HOUSES_PARQUET, index=False)
    print(f"  -> {config.HOUSES_PARQUET}")


if __name__ == "__main__":
    main()