# =============================================================================
# QWHPI — Quebec Weekly Housing Price Index
# Author  : Simon-Pierre Boucher
# Contact : contact@spboucher.ai
# File    : engine/tests/test_features_clean.py
# Purpose : Unit tests for cleaning flags and feature engineering.
# =============================================================================
"""Cleaning + feature unit tests on synthetic rows."""

from __future__ import annotations

import datetime as dt
import sys
from pathlib import Path

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

import polars as pl

from qwhpi.clean import add_flags
from qwhpi.features import build_features


def synthetic(n: int = 8) -> tuple[pl.DataFrame, pl.DataFrame]:
    tx = pl.DataFrame({
        "id": [f"t{i}" for i in range(n)],
        "date": [dt.date(2023, 1, 9)] * n,
        "amount": [300000.0, 300000.0, 55000.0, 20_000_000.0,
                   400000.0, 500000.0, 350000.0, 360000.0],
        "street": ["1 rue A", "1 rue A", "2 rue B", "3 rue C",
                   "4 rue D", "5 rue E", "6 rue F", "7 rue G"],
        "zipCode": ["G1A 1A1"] * n,
        "city": ["X"] * n,
        "lat": [46.8] * n,
        "lng": [-71.2] * n,
        "propertyType": ["unifamilial"] * 6 + ["condo", "indéterminé"],
        "yearBuilt": [1990, 1990, 1990, 1990, 2030, None, 2000, 2000],
        "floorArea": [120.0, 120.0, 120.0, 120.0, 5000.0, None, 80.0, 80.0],
        "buildingType": ["single-story"] * 6 + [None, None],
        "previousValue": [250000.0] * n,
        "totalArValue": [280000.0, 280000.0, 280000.0, 280000.0,
                         380000.0, 100.0, 330000.0, 330000.0],
        "ownerType": ["physical_person"] * n,
        "week": [dt.date(2023, 1, 9)] * n,
        "log_amount": [12.6] * n,
    })
    geo = pl.DataFrame({
        "id": [f"t{i}" for i in range(n)],
        "geo_code": ["23027"] * n,
        "municipality": ["Québec"] * n,
        "munic_type": ["V"] * n,
        "mrc_code": ["230"] * n,
        "mrc": ["Québec"] * n,
        "region_code": ["03"] * n,
        "region": ["Capitale-Nationale"] * n,
        "join_method": ["within"] * n,
    })
    return tx, geo


def test_flags():
    tx, geo = synthetic()
    df = add_flags(tx, geo)
    by_id = {r["id"]: r for r in df.iter_rows(named=True)}
    assert by_id["t1"]["duplicate_flag"] is True          # exact dup of t0
    assert by_id["t0"]["duplicate_flag"] is False          # first kept
    assert by_id["t3"]["exclude_reason"] == "price_above_10m"
    assert by_id["t5"]["nonmarket_ratio"] is True          # ratio 5000x
    assert by_id["t4"]["yb_suspect"] is True               # built 2030
    assert by_id["t4"]["fa_suspect"] is True               # 5000 m2
    assert by_id["t4"]["yearBuilt"] is None                # nulled, kept
    assert by_id["t7"]["is_undetermined_type"] is True
    # nothing is deleted
    assert df.height == tx.height


def test_features_age_bins_and_imputation():
    tx, geo = synthetic()
    feat = build_features(add_flags(tx, geo))
    by_id = {r["id"]: r for r in feat.iter_rows(named=True)}
    assert by_id["t0"]["age_bin"] == "04_21-40"            # age 33
    # missing yearBuilt -> conditional-median age imputation, NO missing bin
    # (time-correlated missingness rule — see features.py docstring)
    assert by_id["t5"]["yb_missing"] is True
    assert by_id["t5"]["age_filled"] is not None
    assert by_id["t5"]["age_bin"] != "99_missing"
    assert by_id["t5"]["fa_missing"] is True
    assert by_id["t5"]["floor_area_filled"] is not None    # imputed
    assert by_id["t0"]["loc_fine"] == "G1A"