dubizzle project
This commit is contained in:
319
tests/test_scraper.py
Normal file
319
tests/test_scraper.py
Normal file
@@ -0,0 +1,319 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
from concurrent.futures import TimeoutError as FuturesTimeoutError
|
||||
from types import SimpleNamespace
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
from dubizzle_scraper.core.config import Settings
|
||||
from dubizzle_scraper.core.exceptions import AntiBotDetectedError, SiteStructureChangedError
|
||||
from dubizzle_scraper.scraper import DUBIZZLEScraper
|
||||
from dubizzle_scraper.storage.schemas import CarRecord
|
||||
|
||||
|
||||
def make_db_record(origin_id: str) -> dict[str, object]:
|
||||
return CarRecord(
|
||||
parser_id=f"dubizzle:{origin_id}",
|
||||
brand="Toyota",
|
||||
model="Camry",
|
||||
origin_url=f"https://www.dubizzle.com/VehicleDetail/{origin_id}~US",
|
||||
origin_id=origin_id,
|
||||
slug=f"toyota-camry-{origin_id}",
|
||||
).model_dump(mode="json")
|
||||
|
||||
|
||||
class TestScraperSync(unittest.TestCase):
|
||||
def _make_scraper(self) -> DUBIZZLEScraper:
|
||||
s = Settings()
|
||||
s.log_level = "CRITICAL"
|
||||
s.database.url = "sqlite://"
|
||||
return DUBIZZLEScraper(s)
|
||||
|
||||
def test_sync_vehicle_uses_db_record(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.persistence.create_tables = MagicMock()
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=1)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.upsert_car = MagicMock(return_value={"action": "inserted", "images_upserted": 0})
|
||||
scraper.scrape_vehicle_detail = MagicMock(return_value={"db_record": make_db_record("111")})
|
||||
|
||||
result = scraper.sync_vehicle("https://www.dubizzle.com/VehicleDetail/111~US")
|
||||
|
||||
self.assertEqual(result["status"], "success")
|
||||
self.assertIn("trace_id", result)
|
||||
scraper.persistence.upsert_car.assert_called_once()
|
||||
|
||||
def test_only_new_routing_legacy_and_streaming(self) -> None:
|
||||
# Legacy path: only_new без listing_url.
|
||||
scraper = self._make_scraper()
|
||||
scraper.persistence.create_tables = MagicMock()
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=2)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.get_existing_urls_and_ids = MagicMock(return_value=(
|
||||
{"https://www.dubizzle.com/VehicleDetail/111~US"}, {"dubizzle:222"},
|
||||
))
|
||||
scraper.collect_listing = MagicMock(return_value={
|
||||
"vehicle_urls": [
|
||||
"https://www.dubizzle.com/VehicleDetail/111~US",
|
||||
"https://www.dubizzle.com/VehicleDetail/222~US",
|
||||
"https://www.dubizzle.com/VehicleDetail/333~US",
|
||||
]
|
||||
})
|
||||
scraper.sync_batch = MagicMock(return_value={
|
||||
"cars_upserted": 1, "cars_failed": 0, "images_upserted": 0, "failures": [],
|
||||
})
|
||||
result = scraper.sync_listing(only_new=True)
|
||||
self.assertEqual(result["skipped_existing"], 2)
|
||||
self.assertEqual(result["cars_upserted"], 1)
|
||||
|
||||
# Streaming path: only_new + listing_url.
|
||||
scraper2 = self._make_scraper()
|
||||
scraper2.persistence.create_tables = MagicMock()
|
||||
scraper2.persistence.start_sync_run = MagicMock(return_value=6)
|
||||
scraper2.persistence.finish_sync_run = MagicMock()
|
||||
scraper2.collect_listing = MagicMock(side_effect=AssertionError("legacy path should not be used"))
|
||||
scraper2._sync_listing_streaming = MagicMock(return_value={
|
||||
"listing": {"vehicles_collected": 10, "early_stopped": False, "truncated_by_time_budget": False},
|
||||
"total": 10, "skipped_existing": 0, "cars_upserted": 10, "cars_failed": 0,
|
||||
"images_upserted": 20, "failures": [], "all_listing_origin_urls": set(),
|
||||
})
|
||||
result2 = scraper2.sync_listing(
|
||||
only_new=True,
|
||||
listing_url="https://www.dubizzle.com/Vehiclelisting/Cars?Make=TOYOTA",
|
||||
year_min=2020, year_max=2027,
|
||||
)
|
||||
self.assertEqual(result2["cars_upserted"], 10)
|
||||
scraper2._sync_listing_streaming.assert_called_once()
|
||||
scraper2.collect_listing.assert_not_called()
|
||||
|
||||
def test_segmented_sync_calls_per_segment_and_resumes(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.persistence.create_tables = MagicMock()
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=3)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
|
||||
call_args_log: list[dict] = []
|
||||
def _fake_sync_listing(**kwargs):
|
||||
call_args_log.append(kwargs)
|
||||
return {
|
||||
"status": "success", "cars_upserted": 5, "cars_failed": 0,
|
||||
"images_upserted": 10, "skipped_existing": 0,
|
||||
"listing": {"vehicles_collected": 50}, "failures": [],
|
||||
}
|
||||
|
||||
scraper.sync_listing = MagicMock(side_effect=_fake_sync_listing)
|
||||
segments = [
|
||||
{"make": "TOYOTA", "year_min": 2020, "year_max": 2027},
|
||||
{"make": "FORD", "year_min": None, "year_max": None},
|
||||
{"make": "HONDA", "year_min": None, "year_max": None},
|
||||
]
|
||||
|
||||
# Resume: пропускаем TOYOTA, начинаем сразу с FORD.
|
||||
result = scraper.sync_listing_segmented(
|
||||
segments=segments, start_segment=1,
|
||||
)
|
||||
|
||||
self.assertEqual(result["segments_completed"], 2) # FORD + HONDA
|
||||
self.assertEqual(result["cars_upserted"], 10)
|
||||
# URL содержит бренд.
|
||||
self.assertIn("FORD", call_args_log[0]["listing_url"])
|
||||
self.assertIn("HONDA", call_args_log[1]["listing_url"])
|
||||
|
||||
def test_segmented_sync_does_not_mark_full_scan_completed_when_segment_failed(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.sync_listing = MagicMock(side_effect=[
|
||||
{
|
||||
"status": "failed",
|
||||
"full_scan_completed": False,
|
||||
"cars_upserted": 0,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 0,
|
||||
"skipped_existing": 0,
|
||||
"listing": {"vehicles_collected": 0},
|
||||
"failures": [{"vehicle_url": "segment", "error": "resume failed"}],
|
||||
},
|
||||
{
|
||||
"status": "success",
|
||||
"full_scan_completed": True,
|
||||
"cars_upserted": 1,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 0,
|
||||
"skipped_existing": 0,
|
||||
"listing": {"vehicles_collected": 10},
|
||||
"failures": [],
|
||||
},
|
||||
])
|
||||
|
||||
result = scraper.sync_listing_segmented(
|
||||
segments=[
|
||||
{"make": "EAGLE", "year_min": None, "year_max": None},
|
||||
{"make": "FORD", "year_min": None, "year_max": None},
|
||||
]
|
||||
)
|
||||
|
||||
self.assertFalse(result["full_scan_completed"])
|
||||
self.assertEqual(result["status"], "partial_success")
|
||||
|
||||
def test_build_segment_listing_url(self) -> None:
|
||||
base = "https://www.dubizzle.com/Vehiclelisting/Cars"
|
||||
self.assertEqual(
|
||||
DUBIZZLEScraper._build_segment_listing_url(base, "TOYOTA"),
|
||||
"https://www.dubizzle.com/Vehiclelisting/Cars?Make=TOYOTA",
|
||||
)
|
||||
self.assertEqual(
|
||||
DUBIZZLEScraper._build_segment_listing_url(base, "LAND ROVER"),
|
||||
"https://www.dubizzle.com/Vehiclelisting/Cars?Make=LAND%20ROVER",
|
||||
)
|
||||
self.assertEqual(DUBIZZLEScraper._build_segment_listing_url(base, None), base)
|
||||
self.assertEqual(DUBIZZLEScraper._build_segment_listing_url(base, ""), base)
|
||||
|
||||
def test_guard_and_protection_detection(self) -> None:
|
||||
with self.assertRaises(AntiBotDetectedError):
|
||||
DUBIZZLEScraper._raise_if_blocked_or_incomplete(
|
||||
{"dom_hints": {"has_captcha_text": True, "has_antibot_text": False},
|
||||
"access_notes": {}, "vehicle_summary": {}},
|
||||
"https://www.dubizzle.com/VehicleDetail/999~US",
|
||||
)
|
||||
with self.assertRaises(SiteStructureChangedError):
|
||||
DUBIZZLEScraper._raise_if_blocked_or_incomplete(
|
||||
{"dom_hints": {"has_captcha_text": False, "has_antibot_text": False},
|
||||
"access_notes": {"possible_captcha": False, "possible_antibot": False},
|
||||
"vehicle_summary": {}},
|
||||
"https://www.dubizzle.com/VehicleDetail/999~US",
|
||||
)
|
||||
self.assertTrue(DUBIZZLEScraper._is_protection_or_network_error(RuntimeError("NS_ERROR_NET_INTERRUPT")))
|
||||
self.assertTrue(DUBIZZLEScraper._is_protection_or_network_error(RuntimeError("captcha challenge")))
|
||||
self.assertFalse(DUBIZZLEScraper._is_protection_or_network_error(RuntimeError("plain validation error")))
|
||||
|
||||
def test_close_resets_browser_state(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
http_pool = MagicMock()
|
||||
scraper._http_pool = http_pool
|
||||
scraper.context = MagicMock()
|
||||
scraper.browser = MagicMock()
|
||||
scraper.playwright = MagicMock()
|
||||
scraper.close()
|
||||
http_pool.clear.assert_called_once()
|
||||
self.assertIsNone(scraper.context)
|
||||
self.assertIsNone(scraper.browser)
|
||||
|
||||
def test_recover_empty_listing_page(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
page = MagicMock()
|
||||
page_result = SimpleNamespace(
|
||||
vehicle_links=[SimpleNamespace(href="https://www.dubizzle.com/VehicleDetail/123~US")],
|
||||
)
|
||||
scraper.listing_collector.collect_current_page = MagicMock(return_value=page_result)
|
||||
scraper.listing_collector.open_cars_listing = MagicMock()
|
||||
|
||||
# Страница > 1: reload.
|
||||
_, urls = scraper._recover_empty_listing_page(
|
||||
page, page_number=5, all_raw_urls=[], seen_urls=set(),
|
||||
)
|
||||
page.reload.assert_called_once()
|
||||
self.assertEqual(len(urls), 1)
|
||||
|
||||
# Страница 1: переоткрытие листинга.
|
||||
page.reset_mock()
|
||||
_, urls = scraper._recover_empty_listing_page(
|
||||
page, page_number=1, all_raw_urls=[], seen_urls=set(),
|
||||
)
|
||||
scraper.listing_collector.open_cars_listing.assert_called_once()
|
||||
page.reload.assert_not_called()
|
||||
|
||||
def test_sync_listing_always_starts_from_page_one(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.settings.celery.batch_size = 1
|
||||
|
||||
page = MagicMock()
|
||||
page_result = SimpleNamespace(
|
||||
page_number=1,
|
||||
vehicle_links=[SimpleNamespace(href="https://www.dubizzle.com/VehicleDetail/999~US", lot_number="999")],
|
||||
next_page_detected=False,
|
||||
)
|
||||
|
||||
scraper._get_page_with_warmup = MagicMock(return_value=page)
|
||||
# Pagination-resume убран: _reopen_listing_and_resume не должен вызываться.
|
||||
scraper._reopen_listing_and_resume = MagicMock(
|
||||
side_effect=AssertionError("pagination resume must not be used")
|
||||
)
|
||||
scraper.listing_collector.open_cars_listing = MagicMock()
|
||||
scraper.listing_collector.apply_filters = MagicMock(return_value={
|
||||
"make": None,
|
||||
"model": None,
|
||||
"year_min": None,
|
||||
"year_max": None,
|
||||
})
|
||||
scraper.listing_collector.collect_current_page = MagicMock(return_value=page_result)
|
||||
scraper._extract_page_urls = MagicMock(return_value=["https://www.dubizzle.com/VehicleDetail/999~US"])
|
||||
scraper.sync_batch = MagicMock(return_value={
|
||||
"cars_upserted": 1,
|
||||
"cars_failed": 0,
|
||||
"images_upserted": 0,
|
||||
"failures": [],
|
||||
})
|
||||
|
||||
result = scraper._sync_listing_streaming(
|
||||
make=None,
|
||||
model=None,
|
||||
lane="dubizzle_cars",
|
||||
limit=None,
|
||||
effective_only_new=False,
|
||||
started_at=0.0,
|
||||
listing_url="https://www.dubizzle.com/Vehiclelisting/Cars?Make=EAGLE",
|
||||
)
|
||||
|
||||
scraper.listing_collector.open_cars_listing.assert_called_once()
|
||||
scraper._reopen_listing_and_resume.assert_not_called()
|
||||
scraper.sync_batch.assert_called_once()
|
||||
self.assertEqual(result["cars_upserted"], 1)
|
||||
self.assertEqual(result["total"], 1)
|
||||
|
||||
def test_sync_batch_timeout_does_not_duplicate_already_processed_urls(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.persistence.upsert_cars_batch = MagicMock(return_value={
|
||||
"inserted": 1,
|
||||
"updated": 0,
|
||||
"images_upserted": 0,
|
||||
})
|
||||
|
||||
urls = [
|
||||
"https://www.dubizzle.com/VehicleDetail/111~US",
|
||||
"https://www.dubizzle.com/VehicleDetail/222~US",
|
||||
"https://www.dubizzle.com/VehicleDetail/333~US",
|
||||
]
|
||||
first_record = CarRecord.model_validate(make_db_record("111"))
|
||||
|
||||
class _FakeExecutor:
|
||||
def __init__(self, *args, **kwargs):
|
||||
pass
|
||||
|
||||
def __enter__(self):
|
||||
return self
|
||||
|
||||
def __exit__(self, exc_type, exc, tb):
|
||||
return False
|
||||
|
||||
def map(self, fn, iterable, timeout=None): # noqa: ARG002
|
||||
yield 0, first_record
|
||||
raise FuturesTimeoutError()
|
||||
|
||||
with patch("dubizzle_scraper.core.runtime_config.RuntimeFiltersConfig.is_empty", return_value=True), \
|
||||
patch("dubizzle_scraper.scraper.ThreadPoolExecutor", _FakeExecutor), \
|
||||
patch.object(scraper, "_browser_fallback_parallel", return_value={
|
||||
"records": [],
|
||||
"failures": [],
|
||||
"cars_failed": 0,
|
||||
"protection_events": 0,
|
||||
}) as fallback_mock:
|
||||
result = scraper.sync_batch(urls)
|
||||
|
||||
self.assertEqual(result["cars_upserted"], 1)
|
||||
fallback_urls = fallback_mock.call_args.args[0]
|
||||
self.assertEqual(len(fallback_urls), 2)
|
||||
self.assertEqual({u for u, _ in fallback_urls}, {urls[1], urls[2]})
|
||||
self.assertNotIn(urls[0], {u for u, _ in fallback_urls})
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
Reference in New Issue
Block a user