gendesign/backend/tests/workers/test_supply_layers_refresh.py

"""Тесты Celery-таски supply_layers_refresh (#950 EPIC 6 Step5).

Чистые тесты (mock session + mock сервиса, без живой сети/БД). Покрывают:
  • контракт UPSERT-SQL: точный ON CONFLICT target (uq_supply_layers_logical column
    list), DO UPDATE по ВСЕМ мутабельным колонкам (database-expert: иначе stale),
    ключевые колонки НЕ обновляются, CAST(:x AS type) не :x::type (psycopg v3);
  • district-enumeration: UNION objective_lots + domrf_kn_objects (region_cd=66),
    DISTINCT non-NULL, форма запроса;
  • coherent single snapshot_date: один date.today() прокинут КАЖДОЙ строке;
  • per-row SAVEPOINT (db.begin_nested) на каждой строке;
  • счётчики upserted/skipped; resilience: сбой compute одного района не валит батч.
"""

from __future__ import annotations

import os
from datetime import date
from typing import Any
from unittest.mock import MagicMock, call, patch

os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")

from app.services.site_finder.supply_layers import SupplyLayerRow


def _make_mock_db(
    district_rows: list[dict[str, Any]],
) -> tuple[MagicMock, list[dict[str, Any]], list[Any]]:
    """Mock DB session с поддержкой db.begin_nested() (SAVEPOINT).

    Первый db.execute (без params, либо с region_cd) трактуем как districts-запрос →
    возвращаем district_rows через .mappings().all(). Остальные execute (UPSERT) —
    захватываем params в captured_upserts.

    Returns: (db, captured_upserts, executed_sql) — captured_upserts содержит params
    каждого UPSERT-execute; executed_sql — все stmt-объекты (для проверки SQL).
    """
    captured_upserts: list[dict[str, Any]] = []
    executed_sql: list[Any] = []
    state = {"districts_served": False}

    def execute_router(stmt: Any, params: Any = None) -> MagicMock:
        executed_sql.append(stmt)
        result = MagicMock()
        # districts-запрос: первый execute (region_cd-only params или до UPSERT'ов)
        if not state["districts_served"] and "supply_layers" not in str(stmt).lower():
            state["districts_served"] = True
            result.mappings.return_value.all.return_value = district_rows
            return result
        if params is not None:
            captured_upserts.append(params)
        return result

    db = MagicMock()
    db.execute = execute_router
    db.begin_nested.return_value.__enter__ = MagicMock(return_value=None)
    db.begin_nested.return_value.__exit__ = MagicMock(return_value=False)
    return db, captured_upserts, executed_sql


def _row(
    *,
    layer: int,
    district: str,
    source: str,
    confidence: str = "high",
    units: int | None = 10,
    area: float | None = 500.0,
    obj_class: str | None = None,
    dev_group_name: str | None = None,
    expected: date | None = None,
    snapshot_date: date | None = None,
) -> SupplyLayerRow:
    return SupplyLayerRow(
        layer=layer,  # type: ignore[arg-type]
        district_name=district,
        complex_id=None,
        obj_class=obj_class,
        dev_group_name=dev_group_name,
        units_estimate=units,
        area_estimate=area,
        expected_online_date=expected,
        source=source,  # type: ignore[arg-type]
        confidence=confidence,  # type: ignore[arg-type]
        method="test",
        snapshot_date=snapshot_date,
    )


# ── UPSERT-SQL контракт ──────────────────────────────────────────────────────


def test_upsert_sql_conflict_target_and_updated_columns() -> None:
    """ON CONFLICT таргетит точный логический ключ (#970: + dev_group_name); DO UPDATE
    обновляет ВСЕ мутабельные колонки + computed_at; ключевые колонки НЕ в SET; нет :x::type."""
    from app.workers.tasks.supply_layers_refresh import _UPSERT_SQL

    sql = str(_UPSERT_SQL)
    upper = sql.upper()
    # нормализуем whitespace — ON CONFLICT column-list перенесён на неск. строк (м.128).
    flat = " ".join(upper.split())

    assert "INSERT INTO SUPPLY_LAYERS" in upper
    # точный ON CONFLICT target = uq_supply_layers_logical column-list (м.128, с dev_group_name)
    assert (
        "ON CONFLICT ( LAYER, DISTRICT_NAME, COMPLEX_ID, OBJ_CLASS, "
        "DEV_GROUP_NAME, SOURCE, SNAPSHOT_DATE )" in flat
    ), f"ON CONFLICT target mismatch (must include dev_group_name): {flat}"
    # dev_group_name присутствует в INSERT column-list и в VALUES как CAST(...AS text)
    assert "DEV_GROUP_NAME," in upper
    assert "CAST(:DEV_GROUP_NAME AS TEXT)" in upper
    assert "DO UPDATE SET" in upper
    # ВСЕ мутабельные колонки обновляются (database-expert: иначе stale-данные)
    for col in (
        "UNITS_ESTIMATE = EXCLUDED.UNITS_ESTIMATE",
        "AREA_ESTIMATE = EXCLUDED.AREA_ESTIMATE",
        "EXPECTED_ONLINE_DATE = EXCLUDED.EXPECTED_ONLINE_DATE",
        "CONFIDENCE = EXCLUDED.CONFIDENCE",
        "METHOD = EXCLUDED.METHOD",
        "COMPUTED_AT = NOW()",
    ):
        assert col in upper, f"missing updated column: {col}"
    # ключевые колонки НЕ обновляются в SET-части (после DO UPDATE SET) — dev_group_name
    # теперь тоже КЛЮЧ (#970) → не должен апдейтиться.
    set_clause = upper.split("DO UPDATE SET", 1)[1]
    for key_col in (
        "LAYER = EXCLUDED",
        "DISTRICT_NAME = EXCLUDED",
        "COMPLEX_ID = EXCLUDED",
        "OBJ_CLASS = EXCLUDED",
        "DEV_GROUP_NAME = EXCLUDED",
        "SOURCE = EXCLUDED",
        "SNAPSHOT_DATE = EXCLUDED",
    ):
        assert key_col not in set_clause, f"key column must not be updated: {key_col}"
    # psycopg v3: CAST(:x AS type), НИКОГДА :x::type
    assert "::" not in sql
    assert "CAST(:LAYER AS SMALLINT)" in upper
    assert "CAST(:SNAPSHOT_DATE AS DATE)" in upper


def test_no_double_colon_cast_in_sql() -> None:
    """psycopg v3: ни один bind не использует :name::type (recurring bug class)."""
    import re

    from app.workers.tasks.supply_layers_refresh import _DISTRICTS_SQL, _UPSERT_SQL

    for stmt in (_UPSERT_SQL, _DISTRICTS_SQL):
        sql = str(stmt)
        assert not re.search(r":[a-z_]+::[a-z]", sql), f"found :x::type cast in: {sql}"


# ── district enumeration ─────────────────────────────────────────────────────


def test_districts_query_unions_both_sources() -> None:
    """Запрос районов = UNION objective_lots.district + domrf_kn_objects.district_name
    (region_cd=66), DISTINCT non-NULL — не хардкодит имена."""
    from app.workers.tasks.supply_layers_refresh import _DISTRICTS_SQL

    sql = str(_DISTRICTS_SQL)
    upper = sql.upper()

    assert "FROM OBJECTIVE_LOTS" in upper
    assert "FROM DOMRF_KN_OBJECTS" in upper
    assert "UNION" in upper
    assert "DISTINCT" in upper
    assert "IS NOT NULL" in upper
    # region-фильтр domrf через CAST(:x AS integer)
    assert "REGION_CD = CAST(:REGION_CD AS INTEGER)" in upper


def test_list_districts_filters_null_and_returns_names() -> None:
    """_list_districts отдаёт только non-NULL имена; на пустом источнике → []."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db, _, _ = _make_mock_db(
        [{"district_name": "Кировский"}, {"district_name": None}, {"district_name": "Чкаловский"}]
    )
    assert task_mod._list_districts(db) == ["Кировский", "Чкаловский"]

    db_empty, _, _ = _make_mock_db([])
    assert task_mod._list_districts(db_empty) == []


def test_list_districts_graceful_on_query_error() -> None:
    """Сбой districts-запроса → [] (батч no-op, не падает)."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db = MagicMock()
    db.execute.side_effect = RuntimeError("db down")
    assert task_mod._list_districts(db) == []


# ── per-row UPSERT / SAVEPOINT ───────────────────────────────────────────────


def test_upsert_rows_uses_savepoint_per_row() -> None:
    """Каждая строка UPSERT-ится внутри db.begin_nested() (SAVEPOINT, backend.md)."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db, captured, _ = _make_mock_db([])
    # пометим districts как обслуженные, чтобы execute шёл в UPSERT-ветку
    db.execute("SELECT FROM supply_layers warmup", None)
    snap = date(2026, 6, 3)
    rows = [
        _row(layer=1, district="Кировский", source="objective", snapshot_date=snap),
        _row(layer=2, district="Кировский", source="domrf_declared", snapshot_date=snap),
    ]
    upserted, skipped = task_mod._upsert_rows(db, rows)

    assert upserted == 2
    assert skipped == 0
    # begin_nested вызван по разу на строку
    assert db.begin_nested.call_count == 2
    # params каждого UPSERT — это row.as_dict()
    assert captured == [rows[0].as_dict(), rows[1].as_dict()]


def test_upsert_rows_savepoint_isolates_bad_row() -> None:
    """Битая строка (execute бросает) → skipped++, остальные продолжают (SAVEPOINT)."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    captured: list[dict[str, Any]] = []
    calls = {"n": 0}

    def execute_router(stmt: Any, params: Any = None) -> MagicMock:
        calls["n"] += 1
        if calls["n"] == 2:  # вторая строка — битая
            raise RuntimeError("CHECK violation")
        captured.append(params)
        return MagicMock()

    db = MagicMock()
    db.execute = execute_router
    db.begin_nested.return_value.__enter__ = MagicMock(return_value=None)
    db.begin_nested.return_value.__exit__ = MagicMock(return_value=False)

    rows = [
        _row(layer=1, district="A", source="objective", snapshot_date=date(2026, 6, 3)),
        _row(layer=2, district="B", source="domrf_declared", snapshot_date=date(2026, 6, 3)),
        _row(layer=3, district="C", source="domrf_multiphase", snapshot_date=date(2026, 6, 3)),
    ]
    upserted, skipped = task_mod._upsert_rows(db, rows)

    assert upserted == 2
    assert skipped == 1
    # первая и третья строки прошли, вторая — нет
    assert [p["district_name"] for p in captured] == ["A", "C"]


def test_upsert_rows_l3_same_district_diff_dev_group_distinct_keys() -> None:
    """#970 (REGRESSION GUARD, worker-level): две L3-строки ОДНОГО района с РАЗНЫМИ
    dev_group_name проходят как ДВЕ строки с РАЗЛИЧНЫМИ ключами UPSERT — не схлопываются.

    Без dev_group_name в ключе (complex_id/obj_class у L3 = NULL) обе дали бы один ключ
    → ON CONFLICT перезатёр бы их (~95.6% потерь L3 на prod). Проверяем, что param-dict
    каждой строки несёт свой dev_group_name и ключевые кортежи различны.
    """
    from app.workers.tasks import supply_layers_refresh as task_mod

    db, captured, _ = _make_mock_db([])
    db.execute("SELECT FROM supply_layers warmup", None)  # пометить districts обслуженными
    snap = date(2026, 6, 3)
    rows = [
        _row(
            layer=3,
            district="Академический",
            source="domrf_multiphase",
            confidence="low",
            dev_group_name="Брусника",
            snapshot_date=snap,
        ),
        _row(
            layer=3,
            district="Академический",
            source="domrf_multiphase",
            confidence="low",
            dev_group_name="Атомстройкомплекс",
            snapshot_date=snap,
        ),
    ]
    upserted, skipped = task_mod._upsert_rows(db, rows)

    assert upserted == 2
    assert skipped == 0
    # оба dev_group дошли до UPSERT-params (binding передаёт dev_group_name)
    assert {p["dev_group_name"] for p in captured} == {"Брусника", "Атомстройкомплекс"}

    # ключевые кортежи (под uq_supply_layers_logical м.128) различны → нет коллапса
    def _key(p: dict[str, Any]) -> tuple:
        return (
            p["layer"],
            p["district_name"],
            p["complex_id"],
            p["obj_class"],
            p["dev_group_name"],
            p["source"],
            p["snapshot_date"],
        )

    keys = [_key(p) for p in captured]
    assert len(set(keys)) == 2, f"L3 dev_group keys collapsed at worker level: {keys}"


# ── end-to-end task: coherent snapshot, counters, resilience ─────────────────


def test_task_coherent_single_snapshot_date() -> None:
    """run_date = date.today() ОДИН раз → передан snapshot_date КАЖДОЙ строке всех
    районов; ВСЕ UPSERT-params несут один и тот же snapshot_date."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db, captured, _ = _make_mock_db(
        [{"district_name": "Кировский"}, {"district_name": "Чкаловский"}]
    )
    today = date(2026, 6, 3)

    def fake_compute(_db: Any, *, district: str, snapshot_date: date) -> list[SupplyLayerRow]:
        # сервис обязан получить наш единый snapshot_date
        assert snapshot_date == today
        return [_row(layer=1, district=district, source="objective", snapshot_date=snapshot_date)]

    with (
        patch.object(task_mod, "SessionLocal", return_value=db),
        patch.object(task_mod, "date") as mock_date,
        patch.object(task_mod, "compute_all_layers", side_effect=fake_compute) as mock_compute,
    ):
        mock_date.today.return_value = today
        result = task_mod.supply_layers_refresh.run()

    # compute вызван на каждый район с единым snapshot_date
    assert mock_compute.call_count == 2
    for c in mock_compute.call_args_list:
        assert c == call(db, district=c.kwargs["district"], snapshot_date=today)
    # все UPSERT-params несут один snapshot_date
    assert captured, "expected at least one upsert"
    assert {p["snapshot_date"] for p in captured} == {today}
    assert result["snapshot_date"] == today.isoformat()
    assert result["districts"] == 2
    assert result["rows"] == 2
    assert result["upserted"] == 2
    assert result["skipped"] == 0
    db.commit.assert_called_once()
    db.close.assert_called_once()


def test_task_resilient_to_one_district_failure() -> None:
    """Сбой compute_all_layers одного района не валит батч: остальные районы
    обрабатываются, commit вызывается."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db, captured, _ = _make_mock_db([{"district_name": "Good"}, {"district_name": "Bad"}])

    def fake_compute(_db: Any, *, district: str, snapshot_date: date) -> list[SupplyLayerRow]:
        if district == "Bad":
            raise RuntimeError("service blew up")
        return [_row(layer=1, district=district, source="objective", snapshot_date=snapshot_date)]

    with (
        patch.object(task_mod, "SessionLocal", return_value=db),
        patch.object(task_mod, "compute_all_layers", side_effect=fake_compute),
    ):
        result = task_mod.supply_layers_refresh.run()

    # только "Good" дал строку; "Bad" проглочен (logged) и пропущен
    assert result["districts"] == 2
    assert result["rows"] == 1
    assert result["upserted"] == 1
    assert [p["district_name"] for p in captured] == ["Good"]
    db.commit.assert_called_once()
    db.close.assert_called_once()


def test_task_closes_db_on_infra_error() -> None:
    """Инфраструктурный сбой (напр. districts-enum) пробрасывается, db закрывается."""
    from app.workers.tasks import supply_layers_refresh as task_mod

    db = MagicMock()
    db.execute.side_effect = None
    with (
        patch.object(task_mod, "SessionLocal", return_value=db),
        patch.object(task_mod, "_list_districts", side_effect=RuntimeError("boom")),
    ):
        try:
            task_mod.supply_layers_refresh.run()
            raise AssertionError("expected RuntimeError to propagate")
        except RuntimeError as e:
            assert "boom" in str(e)

    db.close.assert_called_once()


# ── beat registration ────────────────────────────────────────────────────────


def test_task_registered_in_beat_weekly_monday() -> None:
    """Задача зарегистрирована в beat как еженедельная (понедельник)."""
    from app.workers.beat_schedule import build_beat_schedule

    schedule = build_beat_schedule()
    assert "supply-layers-refresh-weekly" in schedule
    entry = schedule["supply-layers-refresh-weekly"]
    assert entry["task"] == "tasks.supply_layers_refresh.supply_layers_refresh"
    # crontab day_of_week = понедельник (1)
    assert 1 in entry["schedule"].day_of_week