"""Тесты для backend/app/workers/tasks/scrape_cadastre.py.

Покрывает:
- Fix H — per-quarter freshness skip в enqueue_cadastre_harvest.
- Bug #1236 — генерик-фейл инкрементит targets_failed только на финальной
  попытке + регистрирует cleanup_cadastre_zombies в beat_schedule.
"""

from __future__ import annotations

import datetime as _dt
from types import SimpleNamespace
from typing import Any
from unittest.mock import MagicMock, patch

import pytest


def _make_mock_job(quarters: list[str], raw_targets_extra: dict[str, Any] | None = None) -> dict:
    """Mock cadastre_jobs row."""
    raw_targets = {"quarters": quarters}
    if raw_targets_extra:
        raw_targets.update(raw_targets_extra)
    return {
        "job_id": 99,
        "status": "pending",
        "raw_targets": raw_targets,
    }


def _make_mock_db(job: dict, fresh_quarters: list[str] | None = None) -> MagicMock:
    """Mock DB session.

    1. db.execute("SELECT * FROM cadastre_jobs ...").mappings().first() → job
    2. db.execute("SELECT cad_number FROM cad_quarter_stats ...").scalars().all() → fresh
    3. db.execute(UPDATE) → no-op
    """
    db = MagicMock()
    fresh_quarters = fresh_quarters or []

    def execute_router(stmt: Any, params: Any = None) -> MagicMock:
        sql = str(stmt).upper()
        result = MagicMock()
        if "FROM CADASTRE_JOBS" in sql and "WHERE JOB_ID" in sql and "SELECT *" in str(stmt):
            result.mappings.return_value.first.return_value = job
        elif "FROM CAD_QUARTER_STATS" in sql:
            result.scalars.return_value.all.return_value = fresh_quarters
        else:
            # UPDATE / other queries
            result.scalars.return_value.all.return_value = []
            result.mappings.return_value.first.return_value = None
        return result

    db.execute = execute_router
    return db


def test_enqueue_skips_fresh_quarters_with_default_threshold() -> None:
    """Fix H: фреш кварталы (cad_quarter_stats.fetched_at < 1h) пропускаются."""
    from app.workers.tasks import scrape_cadastre

    quarters = ["66:41:0001001", "66:41:0001002", "66:41:0001003"]
    fresh = ["66:41:0001001", "66:41:0001002"]  # 2 свежих, 1 stale

    job = _make_mock_job(quarters)
    db = _make_mock_db(job, fresh_quarters=fresh)

    with (
        patch.object(scrape_cadastre, "SessionLocal", return_value=db),
        patch.object(scrape_cadastre.bulk_harvest_quarter_task, "apply_async") as mock_apply,
    ):
        result = scrape_cadastre.enqueue_cadastre_harvest.run(job_id=99)

    # Только 1 stale quarter должен попасть в apply_async
    assert mock_apply.call_count == 1
    assert mock_apply.call_args[1]["args"] == ["66:41:0001003", 99]
    assert result["enqueued"] == 1
    assert result["skipped_fresh"] == 2


def test_enqueue_no_skip_when_skip_fresh_hours_zero() -> None:
    """skip_fresh_hours=0 → отключает freshness check, все quarters enqueued."""
    from app.workers.tasks import scrape_cadastre

    quarters = ["66:41:0001001", "66:41:0001002"]

    job = _make_mock_job(quarters, raw_targets_extra={"skip_fresh_hours": 0})
    db = _make_mock_db(job, fresh_quarters=["66:41:0001001"])  # один свежий

    with (
        patch.object(scrape_cadastre, "SessionLocal", return_value=db),
        patch.object(scrape_cadastre.bulk_harvest_quarter_task, "apply_async") as mock_apply,
    ):
        result = scrape_cadastre.enqueue_cadastre_harvest.run(job_id=99)

    # Все 2 quarters должны быть enqueued (skip отключен)
    assert mock_apply.call_count == 2
    assert result["enqueued"] == 2
    assert result["skipped_fresh"] == 0


def test_enqueue_freshness_threshold_passed_in_sql() -> None:
    """skip_fresh_hours=2 → threshold = NOW() - 2h в SQL query."""
    from app.workers.tasks import scrape_cadastre

    quarters = ["66:41:0001001"]
    captured_params: list[dict[str, Any]] = []

    job = _make_mock_job(quarters, raw_targets_extra={"skip_fresh_hours": 2})

    def execute_router(stmt: Any, params: Any = None) -> MagicMock:
        sql = str(stmt).upper()
        result = MagicMock()
        if "FROM CADASTRE_JOBS" in sql and "SELECT *" in str(stmt):
            result.mappings.return_value.first.return_value = job
        elif "FROM CAD_QUARTER_STATS" in sql:
            if params:
                captured_params.append(params)
            result.scalars.return_value.all.return_value = []
        else:
            result.scalars.return_value.all.return_value = []
        return result

    db = MagicMock()
    db.execute = execute_router

    with (
        patch.object(scrape_cadastre, "SessionLocal", return_value=db),
        patch.object(scrape_cadastre.bulk_harvest_quarter_task, "apply_async"),
    ):
        scrape_cadastre.enqueue_cadastre_harvest.run(job_id=99)

    assert len(captured_params) == 1
    threshold: _dt.datetime = captured_params[0]["t"]
    now = _dt.datetime.now(_dt.UTC)
    # Threshold должен быть приблизительно 2h назад (±5s tolerance)
    delta = now - threshold
    assert _dt.timedelta(hours=1, minutes=59) < delta < _dt.timedelta(hours=2, minutes=1)


# ── Bug #1236: permanent non-WAF fail counter ──────────────────────────────


def _make_counter_db() -> tuple[MagicMock, list[dict[str, Any]]]:
    """Mock db, capturing UPDATE cadastre_jobs param dicts in chronological order."""
    captured: list[dict[str, Any]] = []
    db = MagicMock()

    def execute_router(stmt: Any, params: Any = None) -> MagicMock:
        sql = str(stmt).upper()
        result = MagicMock()
        if "UPDATE CADASTRE_JOBS" in sql:
            captured.append({"sql": sql, "params": params or {}})
            result.rowcount = 1
        elif "SELECT TARGETS_TOTAL" in sql:
            # _maybe_finish_job: эмулируем done+failed < total → finish не дёрнется,
            # тест проверяет ТОЛЬКО факт вызова + targets_failed_inc.
            result.mappings.return_value.first.return_value = {
                "targets_total": 10,
                "targets_done": 0,
                "targets_failed": 1,
                "status": "running",
            }
        else:
            result.mappings.return_value.first.return_value = None
            result.scalars.return_value.all.return_value = []
        return result

    db.execute = execute_router
    return db, captured


def test_bulk_harvest_quarter_final_fail_increments_targets_failed() -> None:
    """Bug #1236: на финальной (retries == max_retries) попытке non-WAF фейл
    инкрементит targets_failed_inc=1 и вызывает _maybe_finish_job."""
    from app.workers.tasks import scrape_cadastre

    db, captured = _make_counter_db()
    # Эмулируем `self` Celery task: финальная попытка autoretry.
    self_mock = SimpleNamespace(
        max_retries=3,
        request=SimpleNamespace(retries=3),
    )

    finish_called: list[int] = []

    def fake_finish(_db: Any, job_id: int) -> None:
        finish_called.append(job_id)

    with (
        patch.object(scrape_cadastre, "SessionLocal", return_value=db),
        patch.object(scrape_cadastre, "_maybe_finish_job", side_effect=fake_finish),
        patch.object(
            scrape_cadastre,
            "_is_job_cancelled",
            return_value=False,
        ),
        patch(
            "app.services.cadastre.bulk_harvest.harvest_quarter",
            side_effect=RuntimeError("boom"),
        ),
    ):
        with pytest.raises(RuntimeError):
            # Вызываем underlying function через .run (не apply_async).
            scrape_cadastre.bulk_harvest_quarter_task.__wrapped__.__func__(
                self_mock, "66:41:0001001", 42
            )

    # В except-ветке ровно один UPDATE с targets_failed_inc=1 и last_error.
    fail_updates = [
        c for c in captured if c["params"].get("failed_inc") == 1 and "last_error" in c["params"]
    ]
    assert len(fail_updates) == 1, f"expected single fail-update, got: {captured}"
    assert fail_updates[0]["params"]["id"] == 42
    assert finish_called == [42], "_maybe_finish_job must be called on final attempt"


def test_bulk_harvest_quarter_intermediate_fail_does_not_increment() -> None:
    """Bug #1236: промежуточная (retries < max_retries) попытка НЕ трогает
    targets_failed_inc и НЕ вызывает _maybe_finish_job — Celery re-enqueue'ит."""
    from app.workers.tasks import scrape_cadastre

    db, captured = _make_counter_db()
    self_mock = SimpleNamespace(
        max_retries=3,
        request=SimpleNamespace(retries=1),  # ещё две попытки впереди
    )

    finish_called: list[int] = []

    with (
        patch.object(scrape_cadastre, "SessionLocal", return_value=db),
        patch.object(
            scrape_cadastre,
            "_maybe_finish_job",
            side_effect=lambda _db, jid: finish_called.append(jid),
        ),
        patch.object(scrape_cadastre, "_is_job_cancelled", return_value=False),
        patch(
            "app.services.cadastre.bulk_harvest.harvest_quarter",
            side_effect=RuntimeError("transient"),
        ),
    ):
        with pytest.raises(RuntimeError):
            scrape_cadastre.bulk_harvest_quarter_task.__wrapped__.__func__(
                self_mock, "66:41:0001002", 43
            )

    fail_updates = [c for c in captured if c["params"].get("failed_inc")]
    assert fail_updates == [], "intermediate retry must NOT bump targets_failed"
    assert finish_called == [], "intermediate retry must NOT call _maybe_finish_job"


# ── Bug #1236: beat schedule registers cadastre-zombie-cleanup ─────────────


def test_beat_schedule_includes_cadastre_zombie_cleanup() -> None:
    """Bug #1236: cleanup_cadastre_zombies должен быть зарегистрирован в beat,
    иначе застрявший в 'running' job никто не переведёт в 'paused'."""
    from app.workers import beat_schedule

    # build_beat_schedule читает DB → мокаем _build_beat_schedule_from_db,
    # чтобы тест не зависел от живого Postgres.
    with patch.object(beat_schedule, "_build_beat_schedule_from_db", return_value={}):
        schedule = beat_schedule.build_beat_schedule()

    assert "cadastre-zombie-cleanup" in schedule, (
        f"cadastre-zombie-cleanup отсутствует в beat: {sorted(schedule.keys())}"
    )
    entry = schedule["cadastre-zombie-cleanup"]
    assert entry["task"] == "tasks.cadastre.cleanup_zombies"