feat: harden sync flow, proxy bridge, and data normalization
This commit is contained in:
@@ -8,7 +8,7 @@ from sqlalchemy import select
|
||||
|
||||
from iaai_scraper.core.config import Settings
|
||||
from iaai_scraper.storage.db import PersistenceService
|
||||
from iaai_scraper.storage.models import Car, Image
|
||||
from iaai_scraper.storage.models import Car, Image, SyncRun
|
||||
from iaai_scraper.storage.schemas import CarRecord, ImageRecord
|
||||
|
||||
|
||||
@@ -116,6 +116,36 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
car = session.execute(select(Car).where(Car.origin_id == "1000")).scalar_one()
|
||||
self.assertEqual(car.origin_id, "1000")
|
||||
|
||||
def test_start_sync_run_marks_stale_running_runs_as_failed(self) -> None:
|
||||
first_run_id = self.persistence.start_sync_run("lane-a")
|
||||
second_run_id = self.persistence.start_sync_run("lane-b")
|
||||
|
||||
self.assertNotEqual(first_run_id, second_run_id)
|
||||
|
||||
with self.persistence.session_scope() as session:
|
||||
first = session.get(SyncRun, first_run_id)
|
||||
second = session.get(SyncRun, second_run_id)
|
||||
|
||||
self.assertEqual(first.status, "failed")
|
||||
self.assertIsNotNone(first.finished_at)
|
||||
self.assertEqual(second.status, "running")
|
||||
|
||||
def test_upsert_falls_back_to_origin_url_to_prevent_duplicates(self) -> None:
|
||||
first = self._record("OLD-ID", content_hash="v1")
|
||||
first.origin_url = "https://www.iaai.com/VehicleDetail/45089484~US"
|
||||
self.persistence.upsert_car(first)
|
||||
|
||||
second = self._record("NEW-ID", content_hash="v2")
|
||||
second.origin_url = "https://www.iaai.com/VehicleDetail/45089484~US"
|
||||
result = self.persistence.upsert_car(second)
|
||||
|
||||
self.assertEqual(result["action"], "updated")
|
||||
with self.persistence.session_scope() as session:
|
||||
cars = session.execute(select(Car)).scalars().all()
|
||||
|
||||
self.assertEqual(len(cars), 1)
|
||||
self.assertEqual(cars[0].origin_id, "NEW-ID")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
|
||||
@@ -22,7 +22,6 @@ class TestCarMapper(unittest.TestCase):
|
||||
},
|
||||
)
|
||||
|
||||
# Длина hex-представления SHA-256
|
||||
self.assertEqual(len(record.content_hash), 64)
|
||||
|
||||
def test_content_hash_changes_when_image_set_changes(self) -> None:
|
||||
@@ -104,6 +103,37 @@ class TestCarMapper(unittest.TestCase):
|
||||
self.assertEqual(record.drive, "FWD")
|
||||
self.assertEqual(record.gearbox, "AT")
|
||||
|
||||
def test_price_parsing_dirty_formats(self) -> None:
|
||||
record = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/777~US",
|
||||
vehicle_summary={"make": "Toyota", "model": "Corolla"},
|
||||
payload_insights={
|
||||
"vehicle_core": {},
|
||||
"pricing": {"buy_now": "USD 4,500 - 5,200"},
|
||||
"damage": {},
|
||||
"auction": {},
|
||||
"images": {},
|
||||
},
|
||||
)
|
||||
|
||||
self.assertEqual(record.price, 5200)
|
||||
|
||||
def test_currency_detection_from_symbol(self) -> None:
|
||||
record = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/778~US",
|
||||
vehicle_summary={"make": "Toyota", "model": "Corolla"},
|
||||
payload_insights={
|
||||
"vehicle_core": {},
|
||||
"pricing": {"buy_now": "€4.500,00"},
|
||||
"damage": {},
|
||||
"auction": {},
|
||||
"images": {},
|
||||
},
|
||||
)
|
||||
|
||||
self.assertEqual(record.currency, "EUR")
|
||||
self.assertEqual(record.price, 4500)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
|
||||
@@ -50,6 +50,8 @@ class TestScraperSync(unittest.TestCase):
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=2)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.upsert_car = MagicMock(return_value={"action": "inserted", "images_upserted": 1})
|
||||
scraper.persistence.get_existing_origin_urls = MagicMock(return_value=set())
|
||||
scraper.persistence.get_existing_origin_ids = MagicMock(return_value=set())
|
||||
|
||||
scraper.collect_listing = MagicMock(return_value={"vehicle_urls": ["https://www.iaai.com/VehicleDetail/222~US"]})
|
||||
page = MagicMock()
|
||||
@@ -73,6 +75,8 @@ class TestScraperSync(unittest.TestCase):
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=3)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.upsert_car = MagicMock(return_value={"action": "inserted", "images_upserted": 1})
|
||||
scraper.persistence.get_existing_origin_urls = MagicMock(return_value=set())
|
||||
scraper.persistence.get_existing_origin_ids = MagicMock(return_value=set())
|
||||
scraper.collect_listing = MagicMock(return_value={
|
||||
"vehicle_urls": [
|
||||
"https://www.iaai.com/VehicleDetail/222~US",
|
||||
@@ -93,6 +97,8 @@ class TestScraperSync(unittest.TestCase):
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=4)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.upsert_car = MagicMock(return_value={"action": "skipped", "images_upserted": 0})
|
||||
scraper.persistence.get_existing_origin_urls = MagicMock(return_value=set())
|
||||
scraper.persistence.get_existing_origin_ids = MagicMock(return_value=set())
|
||||
scraper.collect_listing = MagicMock(return_value={"vehicle_urls": ["https://www.iaai.com/VehicleDetail/444~US"]})
|
||||
scraper._scrape_on_page = MagicMock(return_value={"db_record": make_db_record("444")})
|
||||
scraper._get_page = MagicMock(return_value=MagicMock())
|
||||
@@ -101,6 +107,35 @@ class TestScraperSync(unittest.TestCase):
|
||||
|
||||
self.assertEqual(result["cars_upserted"], 0)
|
||||
|
||||
def test_sync_listing_only_new_filters_existing_by_url_and_origin_id(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
|
||||
scraper.persistence.create_tables = MagicMock()
|
||||
scraper.persistence.start_sync_run = MagicMock(return_value=5)
|
||||
scraper.persistence.finish_sync_run = MagicMock()
|
||||
scraper.persistence.upsert_car = MagicMock(return_value={"action": "inserted", "images_upserted": 0})
|
||||
scraper.persistence.get_existing_origin_urls = MagicMock(return_value={"https://www.iaai.com/VehicleDetail/111~US"})
|
||||
scraper.persistence.get_existing_origin_ids = MagicMock(return_value={"222"})
|
||||
|
||||
scraper.collect_listing = MagicMock(return_value={
|
||||
"vehicle_urls": [
|
||||
"https://www.iaai.com/VehicleDetail/111~US", # exists by URL
|
||||
"https://www.iaai.com/VehicleDetail/222~US", # exists by ID
|
||||
"https://www.iaai.com/VehicleDetail/333~US", # new
|
||||
]
|
||||
})
|
||||
page = MagicMock()
|
||||
scraper._get_page = MagicMock(return_value=page)
|
||||
scraper._scrape_on_page = MagicMock(return_value={"db_record": make_db_record("333")})
|
||||
|
||||
result = scraper.sync_listing(only_new=True)
|
||||
|
||||
self.assertEqual(result["skipped_existing"], 2)
|
||||
self.assertEqual(result["cars_upserted"], 1)
|
||||
self.assertEqual(scraper._scrape_on_page.call_count, 1)
|
||||
scraper.persistence.get_existing_origin_urls.assert_called_once()
|
||||
scraper.persistence.get_existing_origin_ids.assert_called_once()
|
||||
|
||||
def test_close_resets_browser_state(self) -> None:
|
||||
scraper = self._make_scraper()
|
||||
scraper.context = MagicMock()
|
||||
|
||||
Reference in New Issue
Block a user