add helper scripts

This commit is contained in:
qananasikq
2026-04-24 20:47:18 +03:00
commit cec1288652
70 changed files with 14249 additions and 0 deletions

723
tests/test_worker_tasks.py Normal file
View File

@@ -0,0 +1,723 @@
from __future__ import annotations
import json
from dataclasses import replace
import unittest
from unittest.mock import MagicMock, patch
from dubizzle_scraper.worker import tasks
from dubizzle_scraper.core.config import settings as base_settings
class TestWorkerTaskLockHelpers(unittest.TestCase):
def test_lock_acquire_refresh_release(self) -> None:
redis_client = MagicMock()
# Acquire.
redis_client.set.return_value = True
self.assertTrue(tasks._acquire_lock(redis_client, "lock:key", "owner-token", 120))
redis_client.set.assert_called_once_with("lock:key", "owner-token", nx=True, ex=120)
# Refresh.
redis_client.eval.return_value = 1
self.assertTrue(tasks._refresh_lock_if_owner(redis_client, "lock:key", "owner-token", 120))
# Release.
redis_client.eval.reset_mock()
tasks._release_lock_if_owner(redis_client, "lock:key", "owner-token")
args = redis_client.eval.call_args[0]
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.sync_listing_task, "update_state"), \
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(self) -> None:
# Нет запущенных задач → удаляет.
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
self.assertTrue(tasks._clear_orphan_sync_listing_lock(redis_client, celery_app))
redis_client.delete.assert_called_once_with(tasks.SYNC_LISTING_LOCK_KEY)
# Задача активна → не удаляет.
redis_client2 = MagicMock()
redis_client2.get.return_value = "owner-token"
inspector2 = MagicMock()
inspector2.active.return_value = {"worker@node": [{"name": tasks.SYNC_LISTING_TASK_NAME}]}
inspector2.reserved.return_value = {"worker@node": []}
inspector2.scheduled.return_value = {"worker@node": []}
celery_app2 = MagicMock()
celery_app2.control.inspect.return_value = inspector2
self.assertFalse(tasks._clear_orphan_sync_listing_lock(redis_client2, celery_app2))
redis_client2.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.sync_listing_task, "update_state"), \
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_checkpoint_save_load_roundtrip(self) -> None:
storage: dict[str, str] = {}
def _fake_set(key, value, ex=None):
storage[key] = value
return True
redis_client = MagicMock()
redis_client.set.side_effect = _fake_set
redis_client.get.side_effect = lambda key: storage.get(key)
tasks._save_last_completed_segment(redis_client, 12)
self.assertEqual(tasks._load_last_completed_segment(redis_client), 12)
self.assertEqual(redis_client.set.call_args.kwargs["ex"], tasks.SYNC_LISTING_CHECKPOINT_TTL_SECONDS)
def test_load_checkpoint_clears_invalid_payload(self) -> None:
redis_client = MagicMock()
redis_client.get.return_value = "{not-a-number"
self.assertIsNone(tasks._load_last_completed_segment(redis_client))
redis_client.delete.assert_called_once_with(tasks.SYNC_LISTING_CHECKPOINT_KEY)
def test_load_checkpoint_empty_when_missing(self) -> None:
redis_client = MagicMock()
redis_client.get.return_value = None
self.assertIsNone(tasks._load_last_completed_segment(redis_client))
redis_client.delete.assert_not_called()
def test_sync_listing_resumes_from_next_segment_during_bootstrap(self) -> None:
segments = [
{"make": "ACURA"},
{"make": "AUDI"},
{"make": "BMW"},
{"make": "EAGLE"},
]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(
base_settings.discovery,
always_full_scan=False,
mode="listing",
),
celery=replace(
base_settings.celery,
parallel_segments=False,
),
),
), \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=False), \
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", side_effect=lambda fn: fn()), \
patch.object(tasks.sync_listing_task, "update_state"), \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.side_effect = lambda key: (
"1" if key == tasks.SYNC_LISTING_CHECKPOINT_KEY else None
)
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
sync_segmented_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_segmented = sync_segmented_mock
scraper_ctx.__exit__.return_value = None
with patch.object(tasks, "DUBIZZLEScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-resume-seg")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
# last_completed=1 → start_segment=2 (AUDI завершён, возобновляем с BMW).
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_segment"], 2)
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_page"], 1)
release_lock.assert_called_once()
def test_sync_listing_ignores_checkpoint_after_full_scan_completed(self) -> None:
segments = [{"make": "ACURA"}, {"make": "AUDI"}]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(
base_settings.discovery,
always_full_scan=False,
mode="listing",
),
celery=replace(
base_settings.celery,
parallel_segments=False,
),
),
), \
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()), \
patch.object(tasks.sync_listing_task, "update_state"), \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
# Оставшийся чекпоинт не должен использоваться.
redis_client.get.side_effect = lambda key: (
"0" if key == tasks.SYNC_LISTING_CHECKPOINT_KEY else None
)
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
sync_segmented_mock = MagicMock(return_value={
"run_id": 14, "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_segmented = sync_segmented_mock
scraper_ctx.__exit__.return_value = None
with patch.object(tasks, "DUBIZZLEScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-792")
try:
result = tasks.sync_listing_task.run(limit=1)
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_segment"], 0)
self.assertIsNone(sync_segmented_mock.call_args.kwargs["progress_callback"])
clear_checkpoint.assert_called()
release_lock.assert_called_once()
def test_sync_listing_checkpoint_beyond_segments_restarts_from_zero(self) -> None:
segments = [{"make": "ACURA"}, {"make": "AUDI"}]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(
base_settings.discovery,
always_full_scan=False,
mode="listing",
),
celery=replace(
base_settings.celery,
parallel_segments=False,
),
),
), \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=False), \
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", side_effect=lambda fn: fn()), \
patch.object(tasks.sync_listing_task, "update_state"), \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.side_effect = lambda key: (
"99" if key == tasks.SYNC_LISTING_CHECKPOINT_KEY else None
)
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
sync_segmented_mock = MagicMock(return_value={
"run_id": 15, "status": "success", "full_scan_completed": True,
"cars_upserted": 0, "cars_failed": 0, "images_upserted": 0,
"skipped_existing": 0, "elapsed_seconds": 1.0, "failures": [],
})
scraper_ctx = MagicMock()
scraper_ctx.__enter__.return_value.sync_listing_segmented = sync_segmented_mock
scraper_ctx.__exit__.return_value = None
with patch.object(tasks, "DUBIZZLEScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-beyond")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_segment"], 0)
release_lock.assert_called_once()
def test_sync_listing_task_clears_checkpoint_on_hourly_run(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", return_value={
"run_id": 13,
"status": "partial_success",
"full_scan_completed": True,
"cars_upserted": 2,
"cars_failed": 1,
"images_upserted": 3,
"skipped_existing": 0,
"elapsed_seconds": 4.0,
"failures": [{"vehicle_url": "v", "error": "e"}],
}), \
patch.object(tasks.sync_listing_task, "update_state"):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
tasks.sync_listing_task.push_request(id="task-791")
try:
result = tasks.sync_listing_task.run(make="Toyota")
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "partial_success")
# Hourly-ветка (full_scan_done=True) всегда удаляет оставшийся чекпоинт
# до браузерного job + после. Главное — вызов произошёл.
clear_checkpoint.assert_called()
release_lock.assert_called_once()
def test_try_set_followup_pending_deduplicates(self) -> None:
redis_client = MagicMock()
redis_client.set.side_effect = [True, False]
self.assertTrue(tasks._try_set_followup_pending(redis_client, ttl_seconds=120))
self.assertFalse(tasks._try_set_followup_pending(redis_client, ttl_seconds=120))
def test_bump_bootstrap_failure_streak_opens_circuit_breaker(self) -> None:
redis_client = MagicMock()
redis_client.incr.return_value = tasks.SYNC_LISTING_BOOTSTRAP_FAILURE_STREAK_LIMIT
streak, should_enqueue = tasks._bump_bootstrap_failure_streak(
redis_client,
reason="max_retries_exceeded",
)
self.assertEqual(streak, tasks.SYNC_LISTING_BOOTSTRAP_FAILURE_STREAK_LIMIT)
self.assertFalse(should_enqueue)
redis_client.expire.assert_called_once_with(
tasks.SYNC_LISTING_BOOTSTRAP_FAILURE_STREAK_KEY,
tasks.SYNC_LISTING_BOOTSTRAP_FAILURE_STREAK_TTL_SECONDS,
)
def test_bump_bootstrap_failure_streak_allows_retry_before_limit(self) -> None:
redis_client = MagicMock()
redis_client.incr.return_value = 1
streak, should_enqueue = tasks._bump_bootstrap_failure_streak(
redis_client,
reason="bootstrap_not_completed",
)
self.assertEqual(streak, 1)
self.assertTrue(should_enqueue)
def test_sync_listing_task_does_not_enqueue_followup_after_bootstrap_error_limit(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=False), \
patch.object(tasks, "_start_lock_heartbeat") as start_heartbeat, \
patch.object(tasks, "_release_lock_if_owner") as release_lock, \
patch.object(tasks, "_try_set_followup_pending", return_value=True), \
patch.object(tasks, "_bump_bootstrap_failure_streak", return_value=(tasks.SYNC_LISTING_BOOTSTRAP_FAILURE_STREAK_LIMIT, False)) as bump_streak, \
patch.object(tasks, "_clear_followup_pending") as clear_pending, \
patch.object(tasks.sync_listing_task, "update_state"), \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 99,
"status": "failed",
"full_scan_completed": False,
"cars_upserted": 0,
"cars_failed": 0,
"images_upserted": 0,
"skipped_existing": 0,
"elapsed_seconds": 1.0,
"failures": [{"vehicle_url": "listing", "error": "bad resume"}],
"listing": {"vehicles_collected": 0},
}):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
with patch.object(tasks.sync_listing_task, "app", new=MagicMock()) as task_app:
tasks.sync_listing_task.push_request(id="task-900")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "failed")
bump_streak.assert_called_once()
clear_pending.assert_called()
task_app.send_task.assert_not_called()
def test_sync_listing_task_soft_timeout_deduplicates_continuation(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", return_value=(MagicMock(), MagicMock())), \
patch.object(tasks, "_release_lock_if_owner"), \
patch.object(tasks, "_run_browser_job", side_effect=tasks.SoftTimeLimitExceeded()), \
patch.object(tasks, "_try_set_followup_pending", return_value=False) as set_pending, \
patch.object(tasks.sync_listing_task, "update_state"):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
with patch.object(tasks.sync_listing_task, "app", new=MagicMock()) as task_app:
tasks.sync_listing_task.push_request(id="task-soft-timeout")
try:
result = tasks.sync_listing_task.run(make="Toyota")
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "timed_out")
set_pending.assert_called_once()
task_app.send_task.assert_not_called()
def test_sync_listing_task_stops_immediate_bootstrap_continuation_after_limit(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=False), \
patch.object(tasks, "_start_lock_heartbeat", return_value=(MagicMock(), MagicMock())), \
patch.object(tasks, "_release_lock_if_owner"), \
patch.object(tasks, "_try_set_followup_pending", return_value=True), \
patch.object(tasks, "_bump_bootstrap_continuation_streak", return_value=(999, False)), \
patch.object(tasks, "_set_full_scan_done") as set_full_scan_done, \
patch.object(tasks, "_clear_sync_checkpoint") as clear_checkpoint, \
patch.object(tasks.sync_listing_task, "update_state"), \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 101,
"status": "partial_success",
"full_scan_completed": False,
"cars_upserted": 2,
"cars_failed": 0,
"images_upserted": 1,
"skipped_existing": 0,
"elapsed_seconds": 1.0,
"failures": [],
"listing": {"vehicles_collected": 2},
}):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
with patch.object(tasks.sync_listing_task, "app", new=MagicMock()) as task_app:
tasks.sync_listing_task.push_request(id="task-breaker-stop")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "partial_success")
# Сначала bootstrap помечается незавершённым, затем breaker переключает на hourly.
self.assertTrue(any(call.args == (redis_client, False) for call in set_full_scan_done.call_args_list))
self.assertTrue(any(call.args == (redis_client, True) for call in set_full_scan_done.call_args_list))
clear_checkpoint.assert_called_once()
task_app.send_task.assert_not_called()
def test_sync_listing_task_always_full_scan_forces_bootstrap_even_after_done(self) -> None:
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(base_settings.discovery, always_full_scan=True),
),
), \
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, "_set_full_scan_done") as set_full_scan_done, \
patch.object(tasks.sync_listing_task, "update_state"), \
patch.object(tasks, "_run_browser_job") as run_job, \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=[]):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.return_value = None
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
run_job.return_value = {
"run_id": 17,
"status": "success",
"full_scan_completed": True,
"cars_upserted": 1,
"cars_failed": 0,
"images_upserted": 0,
"skipped_existing": 0,
"elapsed_seconds": 1.0,
"failures": [],
}
tasks.sync_listing_task.push_request(id="task-always-full")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
set_full_scan_done.assert_called_with(redis_client, False)
release_lock.assert_called_once()
def test_sync_listing_task_always_full_scan_uses_segmented_resume_path(self) -> None:
segments = [{"make": "TOYOTA"}, {"make": "FORD"}, {"make": "HONDA"}]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(base_settings.discovery, always_full_scan=True),
),
), \
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", side_effect=lambda fn: fn()), \
patch.object(tasks.sync_listing_task, "update_state"), \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.get.side_effect = lambda key: (
"0" if key == tasks.SYNC_LISTING_CHECKPOINT_KEY else None
)
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
sync_segmented_mock = MagicMock(return_value={
"run_id": 18,
"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_segmented = sync_segmented_mock
scraper_ctx.__enter__.return_value.sync_listing = MagicMock()
scraper_ctx.__exit__.return_value = None
with patch.object(tasks, "DUBIZZLEScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-always-full-resume")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_segment"], 1)
self.assertEqual(sync_segmented_mock.call_args.kwargs["start_page"], 1)
release_lock.assert_called_once()
def test_sync_listing_task_always_full_scan_resets_segment_progress_and_dispatches_again(self) -> None:
segments = [{"make": "TOYOTA"}, {"make": "FORD"}]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(base_settings.discovery, always_full_scan=True),
celery=replace(base_settings.celery, parallel_segments=True),
),
), \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=True), \
patch.object(tasks, "_is_segmented_scan_in_progress", return_value=False), \
patch.object(tasks, "_start_lock_heartbeat") as start_heartbeat, \
patch.object(tasks, "_release_lock_if_owner") as release_lock, \
patch.object(tasks, "_reset_segments_progress") as reset_progress, \
patch.object(tasks, "_set_segmented_scan_active") as set_active, \
patch.object(tasks.sync_listing_task, "update_state"):
get_persistence.return_value = MagicMock()
redis_client = MagicMock()
redis_client.smembers.return_value = {"0", "1"}
redis_client.get.return_value = None
get_redis.return_value = redis_client
start_heartbeat.return_value = (MagicMock(), MagicMock())
with patch.object(tasks.sync_listing_task, "app", new=MagicMock()) as task_app, \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
tasks.sync_listing_task.push_request(id="task-always-hourly")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "success")
self.assertEqual(result["segments_dispatched"], 2)
reset_progress.assert_called_once_with(redis_client, len(segments))
set_active.assert_called_once_with(redis_client, total=len(segments))
self.assertEqual(task_app.send_task.call_count, 2)
release_lock.assert_called_once()
def test_sync_listing_task_skips_duplicate_parallel_dispatch_while_segments_active(self) -> None:
segments = [{"make": "TOYOTA"}, {"make": "FORD"}]
with patch.object(tasks, "_get_persistence") as get_persistence, \
patch.object(tasks, "_get_redis") as get_redis, \
patch.object(
tasks,
"Settings",
return_value=replace(
base_settings,
discovery=replace(base_settings.discovery, always_full_scan=True),
celery=replace(base_settings.celery, parallel_segments=True),
),
), \
patch.object(tasks, "_acquire_lock", return_value=True), \
patch.object(tasks, "_is_full_scan_done", return_value=True), \
patch.object(tasks, "_is_segmented_scan_in_progress", 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.sync_listing_task, "update_state"):
get_persistence.return_value = MagicMock()
get_redis.return_value = MagicMock()
start_heartbeat.return_value = (MagicMock(), MagicMock())
with patch.object(tasks.sync_listing_task, "app", new=MagicMock()) as task_app, \
patch("dubizzle_scraper.worker.tasks.parse_listing_segments", return_value=segments):
tasks.sync_listing_task.push_request(id="task-always-skip-duplicate")
try:
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
self.assertEqual(result["status"], "skipped")
self.assertEqual(result["reason"], "segmented_scan_in_progress")
task_app.send_task.assert_not_called()
release_lock.assert_called_once()
if __name__ == "__main__":
unittest.main()