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

"""Tests for the stale avito listing deactivation task (#759).

Static assertions (SQL shape, psycopg-v3 cast discipline, scheduler wiring) mirror
the pattern established in test_listing_source_snapshot.py. One behavioural test drives
the counter logic via a fake db without a live database.
"""

import inspect
import os
import re
from pathlib import Path
from typing import Any

import pytest

# Stub DATABASE_URL before any app import (same guard as test_scheduler.py /
# test_listing_source_snapshot.py — these tests are static/fake-db, no live DB).
os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")

from app.services import scheduler
from app.tasks import deactivate_stale_avito as task_mod

_SQL_DIR = Path(__file__).resolve().parents[1] / "data" / "sql"
_MIGRATION_090 = _SQL_DIR / "090_scrape_schedules_seed_deactivate_stale_avito.sql"

_DEACTIVATE_SQL = str(task_mod._DEACTIVATE_SQL.text)
_TASK_SRC = inspect.getsource(task_mod.deactivate_stale_avito_listings)


# ── SQL shape ─────────────────────────────────────────────────────────────────


def test_sql_targets_only_avito_source() -> None:
    """UPDATE must filter source='avito' — cian/yandex rows must not be touched."""
    assert "source = 'avito'" in _DEACTIVATE_SQL


def test_sql_only_deactivates_not_deletes() -> None:
    """Rows must be marked is_active=false, never DELETEd (history needed for #667)."""
    assert "SET is_active = false" in _DEACTIVATE_SQL
    assert "DELETE" not in _DEACTIVATE_SQL.upper()


def test_sql_guards_already_inactive_rows() -> None:
    """Filter AND is_active=true avoids redundant updates on rows already deactivated."""
    assert "is_active = true" in _DEACTIVATE_SQL


def test_sql_uses_last_seen_at_threshold() -> None:
    """Threshold is last_seen_at < NOW() - interval, not any other column."""
    assert "last_seen_at" in _DEACTIVATE_SQL
    assert "NOW()" in _DEACTIVATE_SQL


def test_sql_uses_cast_interval_not_double_colon() -> None:
    """psycopg v3: bind param via CAST(:ttl_days || ' days' AS interval), never ::interval."""
    assert "CAST(:ttl_days || ' days' AS interval)" in _DEACTIVATE_SQL
    # No bind-param (:name) followed by ::type anywhere.
    assert not re.search(r":\w+::", _DEACTIVATE_SQL)


def test_sql_ttl_param_named_ttl_days() -> None:
    """Bind param must be :ttl_days (wired from settings.avito_stale_ttl_days)."""
    assert ":ttl_days" in _DEACTIVATE_SQL


# ── Settings wiring ───────────────────────────────────────────────────────────


def test_settings_has_avito_stale_ttl_days() -> None:
    from app.core.config import settings

    assert hasattr(settings, "avito_stale_ttl_days")
    assert settings.avito_stale_ttl_days == 10  # default


def test_task_reads_ttl_from_settings() -> None:
    """Task must use settings.avito_stale_ttl_days, not a hardcoded literal."""
    assert "settings.avito_stale_ttl_days" in _TASK_SRC


# ── Task function contract ────────────────────────────────────────────────────


def test_task_returns_deactivated_counter() -> None:
    assert '"deactivated"' in _TASK_SRC or "'deactivated'" in _TASK_SRC


def test_task_finalises_run() -> None:
    assert "mark_done" in _TASK_SRC
    assert "mark_failed" in _TASK_SRC


def test_task_commits_and_rollbacks_on_error() -> None:
    assert "db.commit()" in _TASK_SRC
    assert "db.rollback()" in _TASK_SRC


# ── Scheduler wiring ──────────────────────────────────────────────────────────


def test_trigger_fn_exists_and_is_async() -> None:
    fn = getattr(scheduler, "trigger_deactivate_stale_avito_run", None)
    assert fn is not None, "trigger_deactivate_stale_avito_run missing from scheduler"
    assert inspect.iscoroutinefunction(fn)


def test_dispatch_branch_wired() -> None:
    """scheduler_loop dispatches source='deactivate_stale_avito' to the trigger."""
    loop_src = inspect.getsource(scheduler.scheduler_loop)
    assert 'source == "deactivate_stale_avito"' in loop_src
    assert "trigger_deactivate_stale_avito_run(db, sch)" in loop_src


def test_trigger_runs_sync_task_in_executor() -> None:
    """Sync DB-only task → run_in_executor, not a bare await."""
    trig_src = inspect.getsource(scheduler.trigger_deactivate_stale_avito_run)
    # run_in_executor call may span two lines; check both parts separately.
    assert "run_in_executor(" in trig_src
    assert "deactivate_stale_avito_listings" in trig_src
    assert "_claim_run(db, schedule_row)" in trig_src


# ── Migration 090 ─────────────────────────────────────────────────────────────


def test_migration_090_exists() -> None:
    assert _MIGRATION_090.is_file(), f"missing migration: {_MIGRATION_090}"


def test_migration_090_seeds_correct_source_and_window() -> None:
    sql = _MIGRATION_090.read_text("utf-8")
    assert "'deactivate_stale_avito'" in sql
    # Window 06:00-07:00 UTC — after avito sweep (02:00-05:00).
    assert re.search(r"\b6\b", sql), "window_start_hour 6 missing"
    assert re.search(r"\b7\b", sql), "window_end_hour 7 missing"


def test_migration_090_is_idempotent() -> None:
    sql = _MIGRATION_090.read_text("utf-8")
    assert "ON CONFLICT (source) DO NOTHING" in sql


def test_migration_090_enabled_true() -> None:
    sql = _MIGRATION_090.read_text("utf-8")
    insert_block = sql.split("INSERT INTO scrape_schedules")[1]
    # enabled=true (SAFE — pure internal DB UPDATE).
    assert "true," in insert_block


def test_migration_090_is_transactional() -> None:
    sql = _MIGRATION_090.read_text("utf-8")
    assert "BEGIN;" in sql
    assert "COMMIT;" in sql


def test_migration_090_no_psycopg_trap() -> None:
    """Migration is plain DDL/DML seed (no bind params), guard :x::type trap anyway."""
    sql = _MIGRATION_090.read_text("utf-8")
    assert not re.search(r":\w+::", sql)


# ── Behavioural test: counter logic via fake db ───────────────────────────────


class _FakeResult:
    def __init__(self, rowcount: int) -> None:
        self.rowcount = rowcount


class _FakeDB:
    """Minimal stand-in for SQLAlchemy Session."""

    def __init__(self, rowcount: int) -> None:
        self._rowcount = rowcount
        self.executed: list[Any] = []
        self.committed = False
        self.rolled_back = False

    def execute(self, stmt: Any, params: dict[str, Any] | None = None) -> _FakeResult:
        self.executed.append((stmt, params))
        return _FakeResult(self._rowcount)

    def commit(self) -> None:
        self.committed = True

    def rollback(self) -> None:
        self.rolled_back = True


def test_counter_logic_with_fake_db(monkeypatch: pytest.MonkeyPatch) -> None:
    """deactivate_stale_avito_listings maps execute() rowcount → counters + marks done."""
    marked: dict[str, Any] = {}
    monkeypatch.setattr(
        task_mod.runs_mod,
        "mark_done",
        lambda _db, run_id, counters: marked.update(run_id=run_id, counters=dict(counters)),
    )
    monkeypatch.setattr(task_mod.runs_mod, "mark_failed", lambda *a, **k: None)

    db = _FakeDB(rowcount=137)
    out = task_mod.deactivate_stale_avito_listings(db, run_id=42)  # type: ignore[arg-type]

    assert out == {"deactivated": 137}
    assert db.committed is True
    assert len(db.executed) == 1
    # Bind params must contain ttl_days wired from settings.
    _stmt, params = db.executed[0]
    assert params is not None
    assert "ttl_days" in params
    assert params["ttl_days"] == 10  # default from settings
    # run_id finalised via mark_done.
    assert marked["run_id"] == 42
    assert marked["counters"] == {"deactivated": 137}


def test_counter_logic_zero_rows(monkeypatch: pytest.MonkeyPatch) -> None:
    """rowcount=0 → deactivated=0, still commits and marks done."""
    marked: dict[str, Any] = {}
    monkeypatch.setattr(
        task_mod.runs_mod,
        "mark_done",
        lambda _db, run_id, counters: marked.update(run_id=run_id, counters=dict(counters)),
    )
    monkeypatch.setattr(task_mod.runs_mod, "mark_failed", lambda *a, **k: None)

    db = _FakeDB(rowcount=0)
    out = task_mod.deactivate_stale_avito_listings(db, run_id=1)  # type: ignore[arg-type]

    assert out == {"deactivated": 0}
    assert db.committed is True
    assert marked["counters"] == {"deactivated": 0}


def test_failure_path_rollback_and_mark_failed(monkeypatch: pytest.MonkeyPatch) -> None:
    """On execute error: rollback + mark_failed + re-raise (no silent swallow)."""
    failed: dict[str, Any] = {}
    monkeypatch.setattr(task_mod.runs_mod, "mark_done", lambda *a, **k: None)
    monkeypatch.setattr(
        task_mod.runs_mod,
        "mark_failed",
        lambda _db, run_id, err, counters: failed.update(run_id=run_id, err=err),
    )

    class _BoomDB(_FakeDB):
        def execute(self, stmt: Any, params: dict[str, Any] | None = None) -> _FakeResult:
            raise RuntimeError("db exploded")

    db = _BoomDB(rowcount=0)
    with pytest.raises(RuntimeError, match="db exploded"):
        task_mod.deactivate_stale_avito_listings(db, run_id=7)  # type: ignore[arg-type]

    assert db.rolled_back is True
    assert failed["run_id"] == 7