initial openlane project

This commit is contained in:
qananasikq
2026-04-21 23:01:33 +03:00
parent 55203d33ea
commit cccf6b3916
70 changed files with 4176 additions and 6804 deletions

View File

@@ -3,7 +3,7 @@ from __future__ import annotations
import unittest
from unittest.mock import MagicMock, patch
from iaai_scraper.worker import tasks
from openlane_scraper.worker import tasks
class TestWorkerTaskLockHelpers(unittest.TestCase):
@@ -52,8 +52,10 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
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("openlane_scraper.worker.tasks.time.sleep"), \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 7,
"status": "success",
"cars_upserted": 2,
"cars_failed": 0,
"images_upserted": 4,
@@ -71,7 +73,7 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
tasks.sync_listing_task.push_request(id="task-123")
try:
result = tasks.sync_listing_task.run(make="Toyota")
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
@@ -117,8 +119,10 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
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("openlane_scraper.worker.tasks.time.sleep"), \
patch.object(tasks, "_run_browser_job", return_value={
"run_id": 9,
"status": "success",
"cars_upserted": 3,
"cars_failed": 0,
"images_upserted": 5,
@@ -136,7 +140,7 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
tasks.sync_listing_task.push_request(id="task-456")
try:
result = tasks.sync_listing_task.run(make="Honda")
result = tasks.sync_listing_task.run()
finally:
tasks.sync_listing_task.pop_request()
@@ -145,278 +149,41 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
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"},
]
def test_bootstrap_forces_full_scan(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, "_set_full_scan_done") as set_done, \
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, "_release_lock_if_owner"), \
patch.object(tasks.sync_listing_task, "update_state"), \
patch("iaai_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, "IAAIScraper", 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, "_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("iaai_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, "IAAIScraper", return_value=scraper_ctx):
tasks.sync_listing_task.push_request(id="task-792")
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)
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, "_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("iaai_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, "IAAIScraper", 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("openlane_scraper.worker.tasks.time.sleep"), \
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,
"run_id": 11,
"status": "success",
"cars_upserted": 5,
"cars_failed": 0,
"images_upserted": 10,
"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()
"elapsed_seconds": 3.0,
}):
persistence = MagicMock()
get_persistence.return_value = persistence
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")
tasks.sync_listing_task.push_request(id="task-bootstrap")
try:
result = tasks.sync_listing_task.run(make="Toyota")
result = tasks.sync_listing_task.run(only_new=True)
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()
self.assertEqual(result["status"], "success")
# Bootstrap should set full_scan_done to True on success.
set_done.assert_called_once_with(redis_client, True)
if __name__ == "__main__":
unittest.main()
unittest.main()