fix scraper flow
This commit is contained in:
@@ -1,215 +1,15 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import json
|
||||
from dataclasses import replace
|
||||
import unittest
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
from iaai_scraper.worker import tasks
|
||||
|
||||
|
||||
def make_settings_stub(*, parallel_segments: bool = False) -> MagicMock:
|
||||
settings_stub = MagicMock()
|
||||
settings_stub.celery.parallel_segments = parallel_segments
|
||||
settings_stub.celery.task_soft_time_limit = 3300
|
||||
settings_stub.celery.task_time_limit = 3600
|
||||
settings_stub.celery.task_stall_timeout_seconds = 600
|
||||
settings_stub.listing.listing_segments_json = "ignored"
|
||||
settings_stub.discovery.mode = "listing"
|
||||
settings_stub.discovery.hourly_mode = "rolling_refresh"
|
||||
return settings_stub
|
||||
from iaai_scraper.core.config import settings as base_settings
|
||||
|
||||
|
||||
class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
def test_sync_listing_task_uses_hourly_sitemap_even_when_mode_is_not_sitemap(self) -> None:
|
||||
settings_stub = make_settings_stub(parallel_segments=False)
|
||||
settings_stub.discovery.mode = "listing"
|
||||
settings_stub.discovery.hourly_mode = "rolling_refresh"
|
||||
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, "_hourly_sitemap_rolling_refresh_sync", return_value={
|
||||
"status": "success",
|
||||
"run_id": None,
|
||||
"cars_upserted": 25,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 50,
|
||||
"skipped_existing": 100,
|
||||
"elapsed_seconds": None,
|
||||
"failures": [],
|
||||
"hourly_mode": "sitemap_rolling_refresh",
|
||||
"discovered_urls": 105,
|
||||
"new_urls": 5,
|
||||
"refresh_urls": 20,
|
||||
"sold_marked": 2,
|
||||
}) as hourly_sync, \
|
||||
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-hourly-force-sitemap")
|
||||
try:
|
||||
result = tasks.sync_listing_task.run()
|
||||
finally:
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
self.assertEqual(result["hourly_mode"], "sitemap_rolling_refresh")
|
||||
hourly_sync.assert_called_once()
|
||||
release_lock.assert_called_once()
|
||||
|
||||
def test_sync_listing_task_uses_hourly_sitemap_rolling_refresh_after_bootstrap(self) -> None:
|
||||
settings_stub = make_settings_stub(parallel_segments=False)
|
||||
settings_stub.discovery.mode = "sitemap"
|
||||
settings_stub.discovery.hourly_mode = "rolling_refresh"
|
||||
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, "_hourly_sitemap_rolling_refresh_sync", return_value={
|
||||
"status": "success",
|
||||
"run_id": None,
|
||||
"cars_upserted": 25,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 50,
|
||||
"skipped_existing": 100,
|
||||
"elapsed_seconds": None,
|
||||
"failures": [],
|
||||
"hourly_mode": "sitemap_rolling_refresh",
|
||||
"discovered_urls": 105,
|
||||
"new_urls": 5,
|
||||
"refresh_urls": 20,
|
||||
"sold_marked": 2,
|
||||
}) as hourly_sync, \
|
||||
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-hourly")
|
||||
try:
|
||||
result = tasks.sync_listing_task.run()
|
||||
finally:
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
self.assertEqual(result["hourly_mode"], "sitemap_rolling_refresh")
|
||||
hourly_sync.assert_called_once()
|
||||
release_lock.assert_called_once()
|
||||
|
||||
def test_sync_listing_task_can_use_hourly_sitemap_diff_when_configured(self) -> None:
|
||||
settings_stub = make_settings_stub(parallel_segments=False)
|
||||
settings_stub.discovery.mode = "sitemap"
|
||||
settings_stub.discovery.hourly_mode = "diff"
|
||||
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, "_hourly_sitemap_diff_sync", return_value={
|
||||
"status": "success",
|
||||
"run_id": None,
|
||||
"cars_upserted": 5,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 10,
|
||||
"skipped_existing": 100,
|
||||
"elapsed_seconds": None,
|
||||
"failures": [],
|
||||
"hourly_mode": "sitemap_diff",
|
||||
"discovered_urls": 105,
|
||||
"new_urls": 5,
|
||||
"sold_marked": 2,
|
||||
}) as hourly_sync, \
|
||||
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-hourly-diff")
|
||||
try:
|
||||
result = tasks.sync_listing_task.run()
|
||||
finally:
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
self.assertEqual(result["hourly_mode"], "sitemap_diff")
|
||||
hourly_sync.assert_called_once()
|
||||
release_lock.assert_called_once()
|
||||
|
||||
def test_sync_listing_task_forces_sitemap_full_scan_in_bootstrap(self) -> None:
|
||||
settings_stub = make_settings_stub(parallel_segments=False)
|
||||
settings_stub.discovery.mode = "listing"
|
||||
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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=[{"make": "HONDA"}]):
|
||||
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())
|
||||
|
||||
sync_listing_mock = MagicMock(return_value={
|
||||
"run_id": 99,
|
||||
"status": "success",
|
||||
"full_scan_completed": True,
|
||||
"cars_upserted": 10,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 20,
|
||||
"skipped_existing": 0,
|
||||
"elapsed_seconds": 1.0,
|
||||
"failures": [],
|
||||
})
|
||||
sync_listing_segmented_mock = MagicMock(side_effect=AssertionError("segmented path should not be used"))
|
||||
scraper = MagicMock()
|
||||
scraper.sync_listing = sync_listing_mock
|
||||
scraper.sync_listing_segmented = sync_listing_segmented_mock
|
||||
scraper_ctx = MagicMock()
|
||||
scraper_ctx.__enter__.return_value = scraper
|
||||
scraper_ctx.__exit__.return_value = None
|
||||
|
||||
with patch.object(tasks, "IAAIScraper", return_value=scraper_ctx):
|
||||
tasks.sync_listing_task.push_request(id="task-bootstrap-sitemap")
|
||||
try:
|
||||
result = tasks.sync_listing_task.run()
|
||||
finally:
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
sync_listing_mock.assert_called_once_with(
|
||||
make=None,
|
||||
model=None,
|
||||
lane="iaai_cars",
|
||||
limit=None,
|
||||
only_new=False,
|
||||
)
|
||||
sync_listing_segmented_mock.assert_not_called()
|
||||
release_lock.assert_called_once()
|
||||
|
||||
def test_lock_acquire_refresh_release(self) -> None:
|
||||
redis_client = MagicMock()
|
||||
|
||||
@@ -378,17 +178,15 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
self.assertIsNone(tasks._load_last_completed_segment(redis_client))
|
||||
redis_client.delete.assert_not_called()
|
||||
|
||||
def test_sync_listing_bootstrap_prefers_sitemap_over_segment_resume(self) -> None:
|
||||
def test_sync_listing_resumes_from_next_segment_during_bootstrap(self) -> None:
|
||||
segments = [
|
||||
{"make": "ACURA"},
|
||||
{"make": "AUDI"},
|
||||
{"make": "BMW"},
|
||||
{"make": "EAGLE"},
|
||||
]
|
||||
settings_stub = make_settings_stub()
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, \
|
||||
@@ -404,15 +202,13 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
get_redis.return_value = redis_client
|
||||
start_heartbeat.return_value = (MagicMock(), MagicMock())
|
||||
|
||||
sync_segmented_mock = MagicMock(side_effect=AssertionError("segmented path should not be used"))
|
||||
sync_listing_mock = MagicMock(return_value={
|
||||
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.__enter__.return_value.sync_listing = sync_listing_mock
|
||||
scraper_ctx.__exit__.return_value = None
|
||||
|
||||
with patch.object(tasks, "IAAIScraper", return_value=scraper_ctx):
|
||||
@@ -423,42 +219,21 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
sync_listing_mock.assert_called_once_with(
|
||||
make=None,
|
||||
model=None,
|
||||
lane="iaai_cars",
|
||||
limit=None,
|
||||
only_new=False,
|
||||
)
|
||||
sync_segmented_mock.assert_not_called()
|
||||
# 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_hourly_path_ignores_checkpoint_after_full_scan_completed(self) -> None:
|
||||
def test_sync_listing_ignores_checkpoint_after_full_scan_completed(self) -> None:
|
||||
segments = [{"make": "ACURA"}, {"make": "AUDI"}]
|
||||
settings_stub = make_settings_stub()
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, "_hourly_sitemap_rolling_refresh_sync", return_value={
|
||||
"status": "success",
|
||||
"run_id": None,
|
||||
"cars_upserted": 1,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 0,
|
||||
"skipped_existing": 0,
|
||||
"elapsed_seconds": None,
|
||||
"failures": [],
|
||||
"hourly_mode": "sitemap_rolling_refresh",
|
||||
"discovered_urls": 10,
|
||||
"new_urls": 1,
|
||||
"refresh_urls": 1,
|
||||
"sold_marked": 0,
|
||||
}) as hourly_sync, \
|
||||
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()
|
||||
@@ -470,24 +245,32 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
get_redis.return_value = redis_client
|
||||
start_heartbeat.return_value = (MagicMock(), MagicMock())
|
||||
|
||||
tasks.sync_listing_task.push_request(id="task-792")
|
||||
try:
|
||||
result = tasks.sync_listing_task.run()
|
||||
finally:
|
||||
tasks.sync_listing_task.pop_request()
|
||||
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(result["hourly_mode"], "sitemap_rolling_refresh")
|
||||
hourly_sync.assert_called_once()
|
||||
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_bootstrap_ignores_beyond_segment_checkpoint(self) -> None:
|
||||
def test_sync_listing_checkpoint_beyond_segments_restarts_from_zero(self) -> None:
|
||||
segments = [{"make": "ACURA"}, {"make": "AUDI"}]
|
||||
settings_stub = make_settings_stub()
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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, \
|
||||
@@ -503,15 +286,13 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
get_redis.return_value = redis_client
|
||||
start_heartbeat.return_value = (MagicMock(), MagicMock())
|
||||
|
||||
sync_segmented_mock = MagicMock(side_effect=AssertionError("segmented path should not be used"))
|
||||
sync_listing_mock = MagicMock(return_value={
|
||||
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.__enter__.return_value.sync_listing = sync_listing_mock
|
||||
scraper_ctx.__exit__.return_value = None
|
||||
|
||||
with patch.object(tasks, "IAAIScraper", return_value=scraper_ctx):
|
||||
@@ -522,14 +303,7 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
tasks.sync_listing_task.pop_request()
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
sync_listing_mock.assert_called_once_with(
|
||||
make=None,
|
||||
model=None,
|
||||
lane="iaai_cars",
|
||||
limit=None,
|
||||
only_new=False,
|
||||
)
|
||||
sync_segmented_mock.assert_not_called()
|
||||
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:
|
||||
@@ -606,38 +380,28 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
self.assertTrue(should_enqueue)
|
||||
|
||||
def test_sync_listing_task_does_not_enqueue_followup_after_bootstrap_error_limit(self) -> None:
|
||||
settings_stub = make_settings_stub(parallel_segments=False)
|
||||
with patch.object(tasks, "_get_persistence") as get_persistence, \
|
||||
patch.object(tasks, "_get_redis") as get_redis, \
|
||||
patch.object(tasks, "Settings", return_value=settings_stub), \
|
||||
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},
|
||||
},
|
||||
):
|
||||
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
|
||||
@@ -656,6 +420,177 @@ class TestWorkerTaskLockHelpers(unittest.TestCase):
|
||||
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("iaai_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("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: (
|
||||
"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, "IAAIScraper", 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()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
Reference in New Issue
Block a user