gendesign/tradein-mvp/backend/tests/test_estimator_imv_blend.py

"""Unit tests for #651 IMV blend (killer accuracy fix) + #652 DKP corridor.

Тестируем чистую blend-трансформацию `_apply_imv_blend` (без БД) и DB-helpers
с замоканной Session — premium-blend поднимает медиану + расширяет диапазон,
no-anchor / sub-threshold случаи — no-op.
"""

import os

# Settings требует DATABASE_URL при инициализации (fail-fast, C-3).
os.environ.setdefault("DATABASE_URL", "postgresql://test:test@localhost/test_db")

from unittest.mock import MagicMock

from app.services.estimator import (
    _apply_imv_blend,
    _fetch_dkp_corridor,
    _fetch_house_imv_anchor,
)

# ── #651: pure blend transform ────────────────────────────────────────────────


def test_blend_premium_raises_median_and_extends_range() -> None:
    """IMV ≫ median → median поднимается blend'ом, range_high расширяется."""
    # median 50М, IMV recommended 100М (≈2x — premium/view unit underestimate),
    # IMV higher 110М. weight 0.5, threshold 1.15.
    area = 80.0
    median_price = 50_000_000
    range_high = 60_000_000
    median_ppm2 = median_price / area

    new_median, new_range_high, new_ppm2, blended, anchor_used = _apply_imv_blend(
        median_price=median_price,
        range_high=range_high,
        median_ppm2=median_ppm2,
        area=area,
        anchor_total=100_000_000,
        anchor_higher=110_000_000,
        weight=0.5,
        threshold=1.15,
    )

    assert blended is True
    # blend = 50М*0.5 + 100М*0.5 = 75М
    assert new_median == 75_000_000
    # range_high расширен до IMV higher (110М)
    assert new_range_high == 110_000_000
    # ppm2 пересчитан консистентно
    assert new_ppm2 == 75_000_000 / area
    assert anchor_used == 100_000_000


def test_blend_no_op_when_anchor_below_median() -> None:
    """A < median → медиану НЕ понижаем (однонаправленность), но диапазон может расшириться."""
    area = 50.0
    median_price = 10_000_000
    range_high = 11_000_000
    median_ppm2 = median_price / area

    new_median, new_range_high, new_ppm2, blended, anchor_used = _apply_imv_blend(
        median_price=median_price,
        range_high=range_high,
        median_ppm2=median_ppm2,
        area=area,
        anchor_total=8_000_000,  # ниже медианы
        anchor_higher=8_500_000,
        weight=0.5,
        threshold=1.15,
    )

    assert blended is False
    assert new_median == median_price  # не понижаем
    assert new_ppm2 == median_ppm2
    assert anchor_used == 8_000_000
    # range_high не сужается (anchor ниже текущего high)
    assert new_range_high == range_high


def test_blend_no_op_below_threshold() -> None:
    """A выше median, но в пределах threshold → blend не срабатывает (медиана та же)."""
    area = 40.0
    median_price = 10_000_000
    range_high = 11_000_000
    median_ppm2 = median_price / area

    new_median, _, new_ppm2, blended, _ = _apply_imv_blend(
        median_price=median_price,
        range_high=range_high,
        median_ppm2=median_ppm2,
        area=area,
        anchor_total=11_000_000,  # ×1.1 < threshold 1.15
        anchor_higher=None,
        weight=0.5,
        threshold=1.15,
    )

    assert blended is False
    assert new_median == median_price
    assert new_ppm2 == median_ppm2


def test_blend_no_op_when_no_anchor() -> None:
    """anchor_total=None → graceful no-op (common case без IMV-данных)."""
    new_median, new_range_high, new_ppm2, blended, anchor_used = _apply_imv_blend(
        median_price=12_000_000,
        range_high=14_000_000,
        median_ppm2=200_000.0,
        area=60.0,
        anchor_total=None,
        anchor_higher=None,
        weight=0.5,
        threshold=1.15,
    )

    assert blended is False
    assert anchor_used is None
    assert new_median == 12_000_000
    assert new_range_high == 14_000_000
    assert new_ppm2 == 200_000.0


def test_blend_extends_range_to_anchor_when_no_higher() -> None:
    """Нет higher_price → диапазон расширяется до самого якоря."""
    _, new_range_high, _, blended, _ = _apply_imv_blend(
        median_price=50_000_000,
        range_high=60_000_000,
        median_ppm2=625_000.0,
        area=80.0,
        anchor_total=100_000_000,
        anchor_higher=None,
        weight=0.5,
        threshold=1.15,
    )
    assert blended is True
    assert new_range_high == 100_000_000  # до якоря, т.к. higher отсутствует


# ── #651: house_imv anchor lookup (mocked Session) ─────────────────────────────


def test_fetch_house_imv_anchor_none_when_no_house_id() -> None:
    """target_house_id=None → None без обращения к БД."""
    mock_db = MagicMock()
    assert _fetch_house_imv_anchor(mock_db, target_house_id=None, rooms=2, area=50.0) is None
    mock_db.execute.assert_not_called()


def test_fetch_house_imv_anchor_returns_row() -> None:
    """Возвращает dict из house_imv_evaluations при наличии строки."""
    mock_db = MagicMock()
    mock_db.execute.return_value.mappings.return_value.first.return_value = {
        "recommended_price": 31_950_000,
        "lower_price": 30_352_500,
        "higher_price": 33_547_500,
        "market_count": 800,
        "rooms": 2,
        "area_m2": 80.0,
    }
    res = _fetch_house_imv_anchor(mock_db, target_house_id=11308, rooms=2, area=80.0)
    assert res is not None
    assert res["recommended_price"] == 31_950_000
    assert res["higher_price"] == 33_547_500


# ── #652: DKP corridor (mocked Session) ────────────────────────────────────────


def test_dkp_corridor_none_without_street() -> None:
    """Адрес без распознаваемой улицы → None."""
    mock_db = MagicMock()
    assert _fetch_dkp_corridor(mock_db, address="", rooms=2, area=50.0) is None


def test_dkp_corridor_aggregates_ppm2() -> None:
    """Считает low/median/high ₽/м² по сделкам."""
    mock_db = MagicMock()
    mock_db.execute.return_value.mappings.return_value.all.return_value = [
        {"price_per_m2": 100_000},
        {"price_per_m2": 120_000},
        {"price_per_m2": 140_000},
    ]
    res = _fetch_dkp_corridor(
        mock_db, address="улица Ленина, 5", rooms=2, area=50.0
    )
    assert res is not None
    assert res["count"] == 3
    assert res["low_ppm2"] == 100_000
    assert res["median_ppm2"] == 120_000
    assert res["high_ppm2"] == 140_000