Files
iaai-parser/tests/test_worker_tasks.py
2026-04-15 11:10:32 +03:00

236 lines
9.9 KiB
Python

from __future__ import annotations
import json
import unittest
from unittest.mock import MagicMock, patch
from iaai_scraper.worker import tasks
class TestWorkerTaskLockHelpers(unittest.TestCase):
def test_acquire_lock_returns_true_on_success(self) -> None:
redis_client = MagicMock()
redis_client.set.return_value = True
acquired = tasks._acquire_lock(redis_client, "lock:key", "owner-token", 120)
self.assertTrue(acquired)
redis_client.set.assert_called_once_with("lock:key", "owner-token", nx=True, ex=120)
def test_refresh_lock_if_owner_extends_ttl(self) -> None:
redis_client = MagicMock()
redis_client.eval.return_value = 1
refreshed = tasks._refresh_lock_if_owner(redis_client, "lock:key", "owner-token", 120)
self.assertTrue(refreshed)
redis_client.eval.assert_called_once()
def test_release_lock_if_owner_uses_owner_token(self) -> None:
redis_client = MagicMock()
tasks._release_lock_if_owner(redis_client, "lock:key", "owner-token")
redis_client.eval.assert_called_once()
args = redis_client.eval.call_args[0]
self.assertEqual(args[1], 1)
self.assertEqual(args[2], "lock:key")
self.assertEqual(args[3], "owner-token")
def test_sync_listing_task_skips_when_lock_not_acquired(self) -> None:
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(tasks, "_acquire_lock", return_value=False):
persistence = MagicMock()
get_persistence.return_value = persistence
get_redis.return_value = MagicMock()
tasks.sync_listing_task.push_request(id="task-123")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
persistence.create_tables.assert_called_once()
self.assertEqual(result["status"], "skipped")
self.assertEqual(result["reason"], "sync_already_running")
def test_sync_listing_task_releases_owned_lock(self) -> None:
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=True), \
patch.object(tasks, "_start_lock_heartbeat") as start_heartbeat, \
patch.object(tasks, "_release_lock_if_owner") as release_lock, \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 7,
"cars_upserted": 2,
"cars_failed": 0,
"images_upserted": 4,
"skipped_existing": 1,
"elapsed_seconds": 1.25,
}):
persistence = MagicMock()
get_persistence.return_value = persistence
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
stop_event = MagicMock()
heartbeat_thread = MagicMock()
start_heartbeat.return_value = (stop_event, heartbeat_thread)
tasks.sync_listing_task.push_request(id="task-123")
try:
result = tasks.sync_listing_task.run(make="Toyota")
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
stop_event.set.assert_called_once()
heartbeat_thread.join.assert_called_once()
release_lock.assert_called_once()
def test_clear_orphan_sync_listing_lock_deletes_when_no_tasks_running(self) -> None:
redis_client = MagicMock()
redis_client.get.return_value = "owner-token"
redis_client.ttl.return_value = 120
celery_app = MagicMock()
inspector = MagicMock()
inspector.active.return_value = {"worker@node": []}
inspector.reserved.return_value = {"worker@node": []}
inspector.scheduled.return_value = {"worker@node": []}
celery_app.control.inspect.return_value = inspector
cleared = tasks._clear_orphan_sync_listing_lock(redis_client, celery_app)
self.assertTrue(cleared)
redis_client.delete.assert_called_once_with(tasks.SYNC_LISTING_LOCK_KEY)
def test_clear_orphan_sync_listing_lock_keeps_when_task_detected(self) -> None:
redis_client = MagicMock()
redis_client.get.return_value = "owner-token"
celery_app = MagicMock()
inspector = MagicMock()
inspector.active.return_value = {
"worker@node": [{"name": tasks.SYNC_LISTING_TASK_NAME}]
}
inspector.reserved.return_value = {"worker@node": []}
inspector.scheduled.return_value = {"worker@node": []}
celery_app.control.inspect.return_value = inspector
cleared = tasks._clear_orphan_sync_listing_lock(redis_client, celery_app)
self.assertFalse(cleared)
redis_client.delete.assert_not_called()
def test_sync_listing_task_recovers_orphan_lock_and_runs(self) -> None:
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(tasks, "_acquire_lock", side_effect=[False, True]) as acquire_lock, \
patch.object(tasks, "_clear_orphan_sync_listing_lock", return_value=True) as clear_orphan, \
patch.object(tasks, "_is_full_scan_done", return_value=True), \
patch.object(tasks, "_start_lock_heartbeat") as start_heartbeat, \
patch.object(tasks, "_release_lock_if_owner") as release_lock, \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 9,
"cars_upserted": 3,
"cars_failed": 0,
"images_upserted": 5,
"skipped_existing": 0,
"elapsed_seconds": 2.0,
}):
persistence = MagicMock()
get_persistence.return_value = persistence
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
stop_event = MagicMock()
heartbeat_thread = MagicMock()
start_heartbeat.return_value = (stop_event, heartbeat_thread)
tasks.sync_listing_task.push_request(id="task-456")
try:
result = tasks.sync_listing_task.run(make="Honda")
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
clear_orphan.assert_called_once()
self.assertEqual(acquire_lock.call_count, 2)
release_lock.assert_called_once()
def test_sync_listing_checkpoint_roundtrip(self) -> None:
redis_client = MagicMock()
storage: dict[str, str] = {}
redis_client.set.side_effect = lambda key, value: storage.__setitem__(key, value)
redis_client.get.side_effect = lambda key: storage.get(key)
tasks._save_sync_checkpoint(
redis_client,
task_id="task-1",
page_number=12,
make=None,
model=None,
lane="iaai_cars",
)
checkpoint = tasks._load_sync_checkpoint(redis_client)
self.assertIsNotNone(checkpoint)
self.assertEqual(checkpoint["last_successful_page"], 12)
self.assertEqual(checkpoint["status"], "in_progress")
def test_sync_listing_task_resumes_from_checkpoint_page(self) -> None:
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=True), \
patch.object(tasks, "_start_lock_heartbeat") as start_heartbeat, \
patch.object(tasks, "_release_lock_if_owner") as release_lock, \
patch.object(tasks, "_clear_sync_checkpoint") as clear_checkpoint, \
patch.object(tasks, "_run_browser_job", side_effect=lambda fn: fn()):
persistence = MagicMock()
get_persistence.return_value = persistence
redis_client = MagicMock()
redis_client.get.side_effect = lambda key: json.dumps({
"status": "in_progress",
"last_successful_page": 9,
"make": None,
"model": None,
"lane": "iaai_cars",
}) if key == tasks.SYNC_LISTING_CHECKPOINT_KEY else None
get_redis.return_value = redis_client
stop_event = MagicMock()
heartbeat_thread = MagicMock()
start_heartbeat.return_value = (stop_event, heartbeat_thread)
sync_listing_mock = MagicMock(return_value={
"run_id": 11,
"status": "success",
"full_scan_completed": True,
"cars_upserted": 1,
"cars_failed": 0,
"images_upserted": 0,
"skipped_existing": 0,
"elapsed_seconds": 1.0,
"failures": [],
})
scraper_ctx = MagicMock()
scraper_ctx.__enter__.return_value.sync_listing = sync_listing_mock
scraper_ctx.__exit__.return_value = None
with patch.object(tasks, "IAAIScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-789")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
self.assertEqual(sync_listing_mock.call_args.kwargs["start_page"], 10)
clear_checkpoint.assert_called()
release_lock.assert_called_once()
if __name__ == "__main__":
unittest.main()