initial openlane project
This commit is contained in:
@@ -6,10 +6,10 @@ from pathlib import Path
|
||||
|
||||
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, SyncRun
|
||||
from iaai_scraper.storage.schemas import CarRecord, ImageRecord
|
||||
from openlane_scraper.core.config import Settings
|
||||
from openlane_scraper.storage.db import PersistenceService
|
||||
from openlane_scraper.storage.models import Car, Image, SyncRun
|
||||
from openlane_scraper.storage.schemas import CarRecord, ImageRecord
|
||||
|
||||
|
||||
class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
@@ -31,18 +31,18 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
@staticmethod
|
||||
def _record(origin_id: str, *, price: int = 1000) -> CarRecord:
|
||||
return CarRecord(
|
||||
parser_id=f"iaai:{origin_id}",
|
||||
parser_id=f"openlane:{origin_id}",
|
||||
brand="Toyota",
|
||||
model="Camry",
|
||||
year=2014,
|
||||
price=price,
|
||||
origin_url=f"https://www.iaai.com/VehicleDetail/{origin_id}~US",
|
||||
origin_id=origin_id,
|
||||
origin_url=f"https://app.openlane.com/vehicles/{origin_id}",
|
||||
origin_id=f"openlane:{origin_id}",
|
||||
slug=f"toyota-camry-{origin_id}",
|
||||
images=[
|
||||
ImageRecord(
|
||||
fullres_image="https://vis.iaai.com/resizer?imageKeys=1&width=845&height=633",
|
||||
preview_image="https://vis.iaai.com/resizer?imageKeys=1&width=400&height=300",
|
||||
fullres_image="https://images.openlane.com/full/1.jpg",
|
||||
preview_image="https://images.openlane.com/thumb/1.jpg",
|
||||
order_index=0,
|
||||
)
|
||||
],
|
||||
@@ -79,8 +79,8 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
second = self._record("888")
|
||||
second.images = [
|
||||
ImageRecord(
|
||||
fullres_image="https://vis.iaai.com/resizer?imageKeys=2&width=845&height=633",
|
||||
preview_image="https://vis.iaai.com/resizer?imageKeys=2&width=400&height=300",
|
||||
fullres_image="https://images.openlane.com/full/2.jpg",
|
||||
preview_image="https://images.openlane.com/thumb/2.jpg",
|
||||
order_index=0,
|
||||
)
|
||||
]
|
||||
@@ -90,7 +90,7 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
images = session.execute(select(Image)).scalars().all()
|
||||
|
||||
self.assertEqual(len(images), 1)
|
||||
self.assertIn("imageKeys=2", images[0].fullres_image)
|
||||
self.assertIn("full/2", images[0].fullres_image)
|
||||
|
||||
def test_start_sync_run_marks_stale_running_runs_as_failed(self) -> None:
|
||||
first_run_id = self.persistence.start_sync_run("lane-a")
|
||||
@@ -108,11 +108,11 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
|
||||
def test_upsert_falls_back_to_origin_url_to_prevent_duplicates(self) -> None:
|
||||
first = self._record("OLD-ID")
|
||||
first.origin_url = "https://www.iaai.com/VehicleDetail/45089484~US"
|
||||
first.origin_url = "https://app.openlane.com/vehicles/45089484"
|
||||
self.persistence.upsert_car(first)
|
||||
|
||||
second = self._record("NEW-ID")
|
||||
second.origin_url = "https://www.iaai.com/VehicleDetail/45089484~US"
|
||||
second.origin_url = "https://app.openlane.com/vehicles/45089484"
|
||||
result = self.persistence.upsert_car(second)
|
||||
|
||||
self.assertEqual(result["action"], "updated")
|
||||
@@ -120,7 +120,7 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
|
||||
cars = session.execute(select(Car)).scalars().all()
|
||||
|
||||
self.assertEqual(len(cars), 1)
|
||||
self.assertEqual(cars[0].origin_id, "NEW-ID")
|
||||
self.assertEqual(cars[0].origin_id, "openlane:NEW-ID")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
|
||||
@@ -1,68 +0,0 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from iaai_scraper.browser.pace import HumanPacer
|
||||
from iaai_scraper.core.config import Settings
|
||||
from iaai_scraper.browser.listing import ListingCollector
|
||||
|
||||
|
||||
class _FakePage:
|
||||
def __init__(self, counts: dict[str, int]) -> None:
|
||||
self._counts = counts
|
||||
self._evaluate_result = False
|
||||
self._evaluate_values: list[object] = []
|
||||
|
||||
class _Locator:
|
||||
def __init__(self, count_value: int) -> None:
|
||||
self._count_value = count_value
|
||||
|
||||
def count(self) -> int:
|
||||
return self._count_value
|
||||
|
||||
def locator(self, selector: str) -> "_FakePage._Locator":
|
||||
return _FakePage._Locator(self._counts.get(selector, 0))
|
||||
|
||||
def evaluate(self, _script: str):
|
||||
if self._evaluate_values:
|
||||
return self._evaluate_values.pop(0)
|
||||
return self._evaluate_result
|
||||
|
||||
|
||||
class TestListingUnit(unittest.TestCase):
|
||||
def test_pagination_detection_and_page_number(self) -> None:
|
||||
# Есть кнопка Next → True.
|
||||
self.assertTrue(ListingCollector._has_next_page(_FakePage({"a[aria-label*='Next']": 1})))
|
||||
# Нет селекторов → False.
|
||||
self.assertFalse(ListingCollector._has_next_page(_FakePage({})))
|
||||
# Числовая пагинация через JS → True только если evaluate явно нашёл next.
|
||||
page = _FakePage({})
|
||||
page._evaluate_result = True
|
||||
self.assertTrue(ListingCollector._has_next_page(page))
|
||||
# Номер текущей страницы.
|
||||
page2 = _FakePage({})
|
||||
page2._evaluate_values = [5]
|
||||
self.assertEqual(ListingCollector._get_current_page_number(page2), 5)
|
||||
|
||||
def test_extract_vehicle_links_from_html_finds_detail_urls(self) -> None:
|
||||
collector = ListingCollector(Settings(), HumanPacer(Settings()))
|
||||
html = """
|
||||
<div>
|
||||
<h4><a href=\"/VehicleDetail/45184893~US\">Car 1</a></h4>
|
||||
<script>window.__data = {\"href\":\"\\/VehicleDetail\\/45171480~US\"}</script>
|
||||
</div>
|
||||
"""
|
||||
|
||||
links = collector._extract_vehicle_links_from_html(html)
|
||||
|
||||
self.assertEqual(
|
||||
links,
|
||||
[
|
||||
("https://www.iaai.com/VehicleDetail/45184893~US", "45184893"),
|
||||
("https://www.iaai.com/VehicleDetail/45171480~US", "45171480"),
|
||||
],
|
||||
)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
198
tests/test_mapper.py
Normal file
198
tests/test_mapper.py
Normal file
@@ -0,0 +1,198 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from openlane_scraper.openlane.mapper import (
|
||||
map_openlane_record,
|
||||
map_openlane_records,
|
||||
_parse_mileage,
|
||||
_parse_engine_volume_cc,
|
||||
_generate_slug,
|
||||
)
|
||||
|
||||
|
||||
class TestMapOpenLaneRecord(unittest.TestCase):
|
||||
def _sample_record(self, **overrides) -> dict:
|
||||
base = {
|
||||
"id": "12345",
|
||||
"make": "Toyota",
|
||||
"model": "Camry",
|
||||
"year": 2020,
|
||||
"mileage": 45000,
|
||||
"price": 18500,
|
||||
"color": "White",
|
||||
"body_type": "sedan",
|
||||
"transmission": "automatic",
|
||||
"drivetrain": "fwd",
|
||||
"vin": "1HGBH41JXMN109186",
|
||||
"url": "https://app.openlane.com/vehicles/12345",
|
||||
"images": [
|
||||
{"full": "https://img.openlane.com/1_full.jpg", "thumbnail": "https://img.openlane.com/1_thumb.jpg"},
|
||||
{"full": "https://img.openlane.com/2_full.jpg", "thumbnail": "https://img.openlane.com/2_thumb.jpg"},
|
||||
],
|
||||
}
|
||||
base.update(overrides)
|
||||
return base
|
||||
|
||||
def test_basic_mapping(self) -> None:
|
||||
record = self._sample_record()
|
||||
result = map_openlane_record(record)
|
||||
self.assertIsNotNone(result)
|
||||
self.assertEqual(result.brand, "TOYOTA")
|
||||
self.assertEqual(result.model, "CAMRY")
|
||||
self.assertEqual(result.year, 2020)
|
||||
self.assertEqual(result.price, 18500)
|
||||
self.assertEqual(result.mileage, 45000)
|
||||
self.assertEqual(result.color, "white")
|
||||
self.assertEqual(result.body_type, "SEDAN")
|
||||
self.assertEqual(result.gearbox, "AT")
|
||||
self.assertEqual(result.drive, "FWD")
|
||||
self.assertEqual(result.origin, "OPENLANE")
|
||||
self.assertEqual(result.origin_id, "openlane:12345")
|
||||
self.assertEqual(result.origin_url, "https://app.openlane.com/vehicles/12345")
|
||||
self.assertEqual(result.evaluation, "1HGBH41JXMN109186")
|
||||
self.assertEqual(len(result.images), 2)
|
||||
self.assertEqual(result.images[0].fullres_image, "https://img.openlane.com/1_full.jpg")
|
||||
self.assertEqual(result.images[0].preview_image, "https://img.openlane.com/1_thumb.jpg")
|
||||
|
||||
def test_missing_id_returns_none(self) -> None:
|
||||
record = {"make": "Toyota", "model": "Camry"}
|
||||
self.assertIsNone(map_openlane_record(record))
|
||||
|
||||
def test_missing_brand_returns_none(self) -> None:
|
||||
record = {"id": "123", "model": "Camry"}
|
||||
self.assertIsNone(map_openlane_record(record))
|
||||
|
||||
def test_missing_model_returns_none(self) -> None:
|
||||
record = {"id": "123", "make": "Toyota"}
|
||||
self.assertIsNone(map_openlane_record(record))
|
||||
|
||||
def test_vin_as_id_fallback(self) -> None:
|
||||
record = self._sample_record()
|
||||
del record["id"]
|
||||
record["vin"] = "WBAPH5C52BA012345"
|
||||
result = map_openlane_record(record)
|
||||
self.assertIsNotNone(result)
|
||||
self.assertEqual(result.origin_id, "openlane:WBAPH5C52BA012345")
|
||||
|
||||
def test_existing_openlane_prefix_is_normalized(self) -> None:
|
||||
record = self._sample_record(id="openlane:777")
|
||||
result = map_openlane_record(record)
|
||||
self.assertIsNotNone(result)
|
||||
self.assertEqual(result.origin_id, "openlane:777")
|
||||
|
||||
def test_whitespace_id_is_normalized(self) -> None:
|
||||
record = self._sample_record(id=" 888 ")
|
||||
result = map_openlane_record(record)
|
||||
self.assertIsNotNone(result)
|
||||
self.assertEqual(result.origin_id, "openlane:888")
|
||||
|
||||
def test_image_url_strings(self) -> None:
|
||||
record = self._sample_record(images=["https://a.com/1.jpg", "https://a.com/2.jpg"])
|
||||
result = map_openlane_record(record)
|
||||
self.assertEqual(len(result.images), 2)
|
||||
self.assertEqual(result.images[0].fullres_image, "https://a.com/1.jpg")
|
||||
|
||||
def test_image_deduplication(self) -> None:
|
||||
record = self._sample_record(images=[
|
||||
{"full": "https://a.com/same.jpg"},
|
||||
{"full": "https://a.com/same.jpg"},
|
||||
{"full": "https://a.com/other.jpg"},
|
||||
])
|
||||
result = map_openlane_record(record)
|
||||
self.assertEqual(len(result.images), 2)
|
||||
|
||||
def test_vehicle_images_api_format(self) -> None:
|
||||
record = self._sample_record(images=[
|
||||
{
|
||||
"url": "https://a.com/original.jpg",
|
||||
"large_resolution_url": "https://a.com/large.jpg",
|
||||
"low_resolution_url": "https://a.com/medium.jpg",
|
||||
}
|
||||
])
|
||||
result = map_openlane_record(record)
|
||||
self.assertEqual(len(result.images), 1)
|
||||
self.assertEqual(result.images[0].fullres_image, "https://a.com/original.jpg")
|
||||
self.assertEqual(result.images[0].preview_image, "https://a.com/medium.jpg")
|
||||
|
||||
def test_default_url_generation(self) -> None:
|
||||
record = self._sample_record()
|
||||
del record["url"]
|
||||
result = map_openlane_record(record)
|
||||
self.assertEqual(result.origin_url, "https://app.openlane.com/vehicles/12345")
|
||||
|
||||
def test_nested_vehicle_fields(self) -> None:
|
||||
record = {
|
||||
"id": "999",
|
||||
"vehicle": {
|
||||
"make": "BMW",
|
||||
"model": "X5",
|
||||
"year": 2022,
|
||||
"mileage": 10000,
|
||||
},
|
||||
}
|
||||
result = map_openlane_record(record)
|
||||
self.assertIsNotNone(result)
|
||||
self.assertEqual(result.brand, "BMW")
|
||||
self.assertEqual(result.model, "X5")
|
||||
self.assertEqual(result.year, 2022)
|
||||
|
||||
def test_map_records_filters_invalid(self) -> None:
|
||||
records = [
|
||||
self._sample_record(id="1"),
|
||||
{"no_id": True},
|
||||
self._sample_record(id="2"),
|
||||
]
|
||||
results = map_openlane_records(records)
|
||||
self.assertEqual(len(results), 2)
|
||||
|
||||
def test_sale_type_mapping(self) -> None:
|
||||
record = self._sample_record(sale_type="buy_now")
|
||||
result = map_openlane_record(record)
|
||||
self.assertEqual(result.selling_type, "STOCK")
|
||||
|
||||
record2 = self._sample_record(sale_type="tender")
|
||||
result2 = map_openlane_record(record2)
|
||||
self.assertEqual(result2.selling_type, "TENDER")
|
||||
|
||||
|
||||
class TestParseMileage(unittest.TestCase):
|
||||
def test_integer(self) -> None:
|
||||
self.assertEqual(_parse_mileage(45000), 45000)
|
||||
|
||||
def test_string_with_commas(self) -> None:
|
||||
self.assertEqual(_parse_mileage("45,000 miles"), 45000)
|
||||
|
||||
def test_none_returns_zero(self) -> None:
|
||||
self.assertEqual(_parse_mileage(None), 0)
|
||||
|
||||
def test_float(self) -> None:
|
||||
self.assertEqual(_parse_mileage(45000.5), 45000)
|
||||
|
||||
|
||||
class TestParseEngineVolume(unittest.TestCase):
|
||||
def test_liters_float(self) -> None:
|
||||
self.assertEqual(_parse_engine_volume_cc(2.5), 2500)
|
||||
|
||||
def test_cc_int(self) -> None:
|
||||
self.assertEqual(_parse_engine_volume_cc(2500), 2500)
|
||||
|
||||
def test_string_liters(self) -> None:
|
||||
self.assertEqual(_parse_engine_volume_cc("2.5L"), 2500)
|
||||
|
||||
def test_none(self) -> None:
|
||||
self.assertIsNone(_parse_engine_volume_cc(None))
|
||||
|
||||
|
||||
class TestGenerateSlug(unittest.TestCase):
|
||||
def test_basic(self) -> None:
|
||||
slug = _generate_slug(2020, "TOYOTA", "CAMRY", "openlane:123")
|
||||
self.assertEqual(slug, "2020-toyota-camry-openlane-123")
|
||||
|
||||
def test_no_year(self) -> None:
|
||||
slug = _generate_slug(None, "BMW", "X5", "openlane:456")
|
||||
self.assertEqual(slug, "bmw-x5-openlane-456")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -1,94 +0,0 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from iaai_scraper.parsing.mapper import CarMapper
|
||||
|
||||
|
||||
class TestCarMapper(unittest.TestCase):
|
||||
def setUp(self) -> None:
|
||||
self.mapper = CarMapper()
|
||||
|
||||
def test_deduplicates_images_by_image_key(self) -> None:
|
||||
urls = [
|
||||
"https://vis.iaai.com/resizer?imageKeys=1&width=200&height=150",
|
||||
"https://vis.iaai.com/resizer?imageKeys=1&width=845&height=633",
|
||||
"https://vis.iaai.com/resizer?imageKeys=2&width=400&height=300",
|
||||
]
|
||||
|
||||
record = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/123~US",
|
||||
vehicle_summary={"make": "Honda", "model": "Civic", "image_urls": urls},
|
||||
payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
|
||||
)
|
||||
|
||||
self.assertEqual(len(record.images), 2)
|
||||
self.assertIn("width=845", record.images[0].fullres_image)
|
||||
self.assertIn("height=633", record.images[0].fullres_image)
|
||||
|
||||
def test_no_damage_marker_is_not_damaged(self) -> None:
|
||||
record = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/123~US",
|
||||
vehicle_summary={"make": "Ford", "model": "Focus"},
|
||||
payload_insights={
|
||||
"vehicle_core": {},
|
||||
"pricing": {},
|
||||
"damage": {"primary": "normal wear"},
|
||||
"auction": {},
|
||||
"images": {},
|
||||
},
|
||||
)
|
||||
|
||||
self.assertFalse(record.is_damaged)
|
||||
|
||||
def test_unknown_empty_values_and_normalization(self) -> None:
|
||||
record = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/999~US",
|
||||
vehicle_summary={"make": " ", "model": None, "drive": "???", "gearbox": "unknown"},
|
||||
payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
|
||||
)
|
||||
self.assertEqual(record.brand, "UNKNOWN")
|
||||
self.assertEqual(record.model, "UNKNOWN")
|
||||
self.assertEqual(record.drive, "NA")
|
||||
self.assertEqual(record.gearbox, "NA")
|
||||
|
||||
# Нормализация регистра и пробелов.
|
||||
record2 = self.mapper.map_to_car_record(
|
||||
vehicle_url="https://www.iaai.com/VehicleDetail/888~US",
|
||||
vehicle_summary={"make": "Honda", "model": "Civic", "drive": " Front Wheel Drive ", "gearbox": " AUTOMATIC "},
|
||||
payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
|
||||
)
|
||||
self.assertEqual(record2.drive, "FWD")
|
||||
self.assertEqual(record2.gearbox, "AT")
|
||||
|
||||
def test_price_and_currency_parsing(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)
|
||||
|
||||
record2 = 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(record2.currency, "EUR")
|
||||
self.assertEqual(record2.price, 4500)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
129
tests/test_openlane.py
Normal file
129
tests/test_openlane.py
Normal file
@@ -0,0 +1,129 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import json
|
||||
import tempfile
|
||||
import unittest
|
||||
from pathlib import Path
|
||||
|
||||
from openlane_scraper.core.config import OpenLaneConfig
|
||||
from openlane_scraper.openlane.checkpoint import OpenLaneCheckpoint, OpenLaneCheckpointStore
|
||||
from openlane_scraper.openlane.client import OpenLaneClient
|
||||
from openlane_scraper.openlane.writer import OpenLaneResultWriter
|
||||
|
||||
|
||||
class TestOpenLaneCheckpointStore(unittest.TestCase):
|
||||
def test_save_and_load_checkpoint(self) -> None:
|
||||
with tempfile.TemporaryDirectory() as tmp_dir:
|
||||
path = Path(tmp_dir) / "checkpoint.json"
|
||||
store = OpenLaneCheckpointStore(path)
|
||||
checkpoint = OpenLaneCheckpoint(
|
||||
max_pages=512,
|
||||
completed_pages=[1, 2, 3],
|
||||
failed_pages=[7],
|
||||
total_records=99,
|
||||
last_saved_at="2026-04-20T00:00:00+00:00",
|
||||
)
|
||||
|
||||
store.save(checkpoint)
|
||||
loaded = store.load(max_pages=512)
|
||||
|
||||
self.assertEqual(loaded.max_pages, 512)
|
||||
self.assertEqual(loaded.completed_pages, [1, 2, 3])
|
||||
self.assertEqual(loaded.failed_pages, [7])
|
||||
self.assertEqual(loaded.total_records, 99)
|
||||
self.assertEqual(loaded.last_saved_at, "2026-04-20T00:00:00+00:00")
|
||||
|
||||
|
||||
class TestOpenLaneWriter(unittest.TestCase):
|
||||
def test_append_and_finalize_outputs(self) -> None:
|
||||
with tempfile.TemporaryDirectory() as tmp_dir:
|
||||
jsonl_path = Path(tmp_dir) / "openlane.jsonl"
|
||||
aggregated_path = Path(tmp_dir) / "openlane_aggregated.json"
|
||||
writer = OpenLaneResultWriter(jsonl_path, aggregated_path)
|
||||
|
||||
written = writer.append_page(1, [{"id": 1}, {"id": 2}])
|
||||
self.assertEqual(written, 2)
|
||||
|
||||
summary = writer.finalize(max_pages=10, completed_pages=[1], failed_pages=[2])
|
||||
|
||||
self.assertEqual(summary["total_records"], 2)
|
||||
self.assertEqual(summary["total_pages_completed"], 1)
|
||||
self.assertEqual(summary["total_pages_failed"], 1)
|
||||
self.assertTrue(aggregated_path.exists())
|
||||
|
||||
payload = json.loads(aggregated_path.read_text(encoding="utf-8"))
|
||||
self.assertEqual(payload["items"][0]["page"], 1)
|
||||
self.assertEqual(payload["items"][0]["record"]["id"], 1)
|
||||
|
||||
|
||||
class TestOpenLaneConfig(unittest.TestCase):
|
||||
def test_retry_schedule_defaults(self) -> None:
|
||||
config = OpenLaneConfig()
|
||||
self.assertGreaterEqual(len(config.retry_schedule_seconds), 1)
|
||||
self.assertEqual(tuple(float(x) for x in config.retry_schedule_seconds), config.retry_schedule_seconds)
|
||||
|
||||
def test_storage_state_defaults(self) -> None:
|
||||
config = OpenLaneConfig()
|
||||
self.assertTrue(config.storage_state_file)
|
||||
self.assertIn("artifacts/openlane", config.storage_state_file)
|
||||
self.assertIsInstance(config.persist_storage_state, bool)
|
||||
|
||||
|
||||
class TestOpenLaneClient(unittest.TestCase):
|
||||
def test_extract_records_from_common_shapes(self) -> None:
|
||||
self.assertEqual(
|
||||
OpenLaneClient._extract_records({"results": [{"id": 1}, {"id": 2}]}),
|
||||
[{"id": 1}, {"id": 2}],
|
||||
)
|
||||
self.assertEqual(
|
||||
OpenLaneClient._extract_records({"data": {"items": [{"id": 3}]}}),
|
||||
[{"id": 3}],
|
||||
)
|
||||
self.assertEqual(
|
||||
OpenLaneClient._extract_records({"vehicles": [{"id": 4}]}),
|
||||
[{"id": 4}],
|
||||
)
|
||||
self.assertEqual(OpenLaneClient._extract_records({"data": {"foo": "bar"}}), [])
|
||||
|
||||
def test_extract_vehicle_images_from_common_shapes(self) -> None:
|
||||
payload = {
|
||||
"vehicle_images": [
|
||||
{"url": "https://img/1.png", "low_resolution_url": "https://img/1_small.png"},
|
||||
]
|
||||
}
|
||||
self.assertEqual(
|
||||
OpenLaneClient._extract_vehicle_images(payload),
|
||||
[{"url": "https://img/1.png", "low_resolution_url": "https://img/1_small.png"}],
|
||||
)
|
||||
|
||||
payload_nested = {
|
||||
"data": {
|
||||
"images": [
|
||||
{"url": "https://img/2.png"},
|
||||
]
|
||||
}
|
||||
}
|
||||
self.assertEqual(
|
||||
OpenLaneClient._extract_vehicle_images(payload_nested),
|
||||
[{"url": "https://img/2.png"}],
|
||||
)
|
||||
|
||||
self.assertEqual(OpenLaneClient._extract_vehicle_images({"data": {"foo": 1}}), [])
|
||||
|
||||
def test_extract_vehicle_images_batch_helper_normalization(self) -> None:
|
||||
# Косвенно проверяем нормализацию входа под батчевый метод
|
||||
ids = [" 123 ", "123", 456, "", None]
|
||||
normalized = []
|
||||
seen = set()
|
||||
for v in ids:
|
||||
s = str(v).strip()
|
||||
if not s or s in seen:
|
||||
continue
|
||||
seen.add(s)
|
||||
normalized.append(s)
|
||||
|
||||
self.assertEqual(normalized, ["123", "456", "None"])
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -1,54 +0,0 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from iaai_scraper.parsing.parser import VehicleParser
|
||||
|
||||
|
||||
class TestVehicleParserUnit(unittest.TestCase):
|
||||
def setUp(self) -> None:
|
||||
self.parser = VehicleParser()
|
||||
|
||||
def test_parse_dom_pairs_and_title(self) -> None:
|
||||
dom_text = """
|
||||
Stock #:
|
||||
45089484
|
||||
Primary Damage:
|
||||
Front End
|
||||
Odometer:
|
||||
50,123 mi (Actual)
|
||||
"""
|
||||
result = self.parser._parse_dom_key_value_pairs(dom_text)
|
||||
self.assertEqual(result.get("lot_number"), "45089484")
|
||||
self.assertEqual(result.get("primary_damage"), "Front End")
|
||||
self.assertEqual(result.get("odometer"), "50,123 mi (Actual)")
|
||||
|
||||
parsed = self.parser._parse_title_for_year_make_model("2014 TOYOTA CAMRY for sale", "")
|
||||
self.assertEqual(parsed["year"], "2014")
|
||||
self.assertEqual(parsed["make"], "TOYOTA")
|
||||
self.assertEqual(parsed["model"], "CAMRY")
|
||||
|
||||
def test_extract_image_urls_filters_and_deduplicates(self) -> None:
|
||||
vehicle_url = "https://www.iaai.com/VehicleDetail/45089484~US"
|
||||
payloads = [{"imageUrls": [
|
||||
"https://vis.iaai.com/resizer?imageKeys=45089484~SID1&width=845&height=633",
|
||||
"https://vis.iaai.com/resizer?imageKeys=45089484~SID1&width=845&height=633",
|
||||
"https://vis.iaai.com/resizer?imageKeys=99999999~SID2&width=845&height=633",
|
||||
]}]
|
||||
urls = self.parser._extract_image_urls(payloads, "", vehicle_url)
|
||||
self.assertEqual(len(urls), 1)
|
||||
self.assertIn("45089484", urls[0])
|
||||
|
||||
def test_dom_hints_and_access_notes_detect_antibot(self) -> None:
|
||||
hints = self.parser._dom_hints("Please verify you are human. CAPTCHA. Incapsula access denied.")
|
||||
self.assertTrue(hints["has_captcha_text"])
|
||||
self.assertTrue(hints["has_antibot_text"])
|
||||
|
||||
summary = {"vin": "", "image_urls": [], "note": "Incapsula access denied. Verify you are human."}
|
||||
notes = self.parser._build_access_notes(summary, [])
|
||||
self.assertTrue(notes["possible_captcha"])
|
||||
self.assertTrue(notes["possible_antibot"])
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -1,273 +0,0 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
from types import SimpleNamespace
|
||||
from unittest.mock import MagicMock
|
||||
|
||||
from iaai_scraper.core.config import Settings
|
||||
from iaai_scraper.core.exceptions import AntiBotDetectedError, SiteStructureChangedError
|
||||
from iaai_scraper.scraper import IAAIScraper
|
||||
from iaai_scraper.storage.schemas import CarRecord
|
||||
|
||||
|
||||
def make_db_record(origin_id: str) -> dict[str, object]:
|
||||
return CarRecord(
|
||||
parser_id=f"iaai:{origin_id}",
|
||||
brand="Toyota",
|
||||
model="Camry",
|
||||
origin_url=f"https://www.iaai.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) -> IAAIScraper:
|
||||
s = Settings()
|
||||
s.log_level = "CRITICAL"
|
||||
s.database.url = "sqlite://"
|
||||
return IAAIScraper(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.iaai.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.iaai.com/VehicleDetail/111~US"}, {"iaai:222"},
|
||||
))
|
||||
scraper.collect_listing = MagicMock(return_value={
|
||||
"vehicle_urls": [
|
||||
"https://www.iaai.com/VehicleDetail/111~US",
|
||||
"https://www.iaai.com/VehicleDetail/222~US",
|
||||
"https://www.iaai.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.iaai.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.iaai.com/Vehiclelisting/Cars"
|
||||
self.assertEqual(
|
||||
IAAIScraper._build_segment_listing_url(base, "TOYOTA"),
|
||||
"https://www.iaai.com/Vehiclelisting/Cars?Make=TOYOTA",
|
||||
)
|
||||
self.assertEqual(
|
||||
IAAIScraper._build_segment_listing_url(base, "LAND ROVER"),
|
||||
"https://www.iaai.com/Vehiclelisting/Cars?Make=LAND%20ROVER",
|
||||
)
|
||||
self.assertEqual(IAAIScraper._build_segment_listing_url(base, None), base)
|
||||
self.assertEqual(IAAIScraper._build_segment_listing_url(base, ""), base)
|
||||
|
||||
def test_guard_and_protection_detection(self) -> None:
|
||||
with self.assertRaises(AntiBotDetectedError):
|
||||
IAAIScraper._raise_if_blocked_or_incomplete(
|
||||
{"dom_hints": {"has_captcha_text": True, "has_antibot_text": False},
|
||||
"access_notes": {}, "vehicle_summary": {}},
|
||||
"https://www.iaai.com/VehicleDetail/999~US",
|
||||
)
|
||||
with self.assertRaises(SiteStructureChangedError):
|
||||
IAAIScraper._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.iaai.com/VehicleDetail/999~US",
|
||||
)
|
||||
self.assertTrue(IAAIScraper._is_protection_or_network_error(RuntimeError("NS_ERROR_NET_INTERRUPT")))
|
||||
self.assertTrue(IAAIScraper._is_protection_or_network_error(RuntimeError("captcha challenge")))
|
||||
self.assertFalse(IAAIScraper._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.iaai.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.iaai.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.iaai.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="iaai_cars",
|
||||
limit=None,
|
||||
effective_only_new=False,
|
||||
started_at=0.0,
|
||||
listing_url="https://www.iaai.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)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
68
tests/test_scraper_images.py
Normal file
68
tests/test_scraper_images.py
Normal file
@@ -0,0 +1,68 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from openlane_scraper.scraper import OpenLaneScraper
|
||||
from openlane_scraper.storage.schemas import ImageRecord
|
||||
|
||||
|
||||
class TestScraperImageHelpers(unittest.TestCase):
|
||||
def test_build_image_records_supports_image_large_and_image(self) -> None:
|
||||
payload = [
|
||||
{
|
||||
"image": "https://img.cdn/vehicle-medium-1.jpg",
|
||||
"image_large": "https://img.cdn/vehicle-large-1.jpg",
|
||||
},
|
||||
{
|
||||
"image": "https://img.cdn/vehicle-medium-2.jpg",
|
||||
"large_resolution_url": "https://img.cdn/vehicle-large-2.jpg",
|
||||
"low_resolution_url": "https://img.cdn/vehicle-low-2.jpg",
|
||||
},
|
||||
]
|
||||
|
||||
records = OpenLaneScraper._build_image_records(payload)
|
||||
|
||||
self.assertEqual(len(records), 2)
|
||||
self.assertEqual(records[0].fullres_image, "https://img.cdn/vehicle-large-1.jpg")
|
||||
self.assertEqual(records[0].preview_image, "https://img.cdn/vehicle-medium-1.jpg")
|
||||
self.assertEqual(records[1].fullres_image, "https://img.cdn/vehicle-large-2.jpg")
|
||||
self.assertEqual(records[1].preview_image, "https://img.cdn/vehicle-low-2.jpg")
|
||||
|
||||
def test_merge_image_records_keeps_unique_and_reindexes(self) -> None:
|
||||
existing = [
|
||||
ImageRecord(
|
||||
fullres_image="https://img.cdn/1.jpg",
|
||||
preview_image="https://img.cdn/1_small.jpg",
|
||||
order_index=10,
|
||||
),
|
||||
ImageRecord(
|
||||
fullres_image="https://img.cdn/2.jpg",
|
||||
preview_image="https://img.cdn/2_small.jpg",
|
||||
order_index=11,
|
||||
),
|
||||
]
|
||||
fetched = [
|
||||
ImageRecord(
|
||||
fullres_image="https://img.cdn/2.jpg",
|
||||
preview_image="https://img.cdn/2_small_new.jpg",
|
||||
order_index=0,
|
||||
),
|
||||
ImageRecord(
|
||||
fullres_image="https://img.cdn/3.jpg",
|
||||
preview_image="https://img.cdn/3_small.jpg",
|
||||
order_index=1,
|
||||
),
|
||||
]
|
||||
|
||||
merged = OpenLaneScraper._merge_image_records(existing, fetched)
|
||||
|
||||
self.assertEqual([img.fullres_image for img in merged], [
|
||||
"https://img.cdn/2.jpg",
|
||||
"https://img.cdn/3.jpg",
|
||||
"https://img.cdn/1.jpg",
|
||||
])
|
||||
self.assertEqual([img.order_index for img in merged], [0, 1, 2])
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -2,8 +2,7 @@ from __future__ import annotations
|
||||
|
||||
import unittest
|
||||
|
||||
from iaai_scraper.core.utils import deep_find_key
|
||||
from iaai_scraper.core.config import parse_listing_segments, IAAI_DEFAULT_MAKES, _LARGE_MAKES, _YEAR_SPLITS
|
||||
from openlane_scraper.core.utils import deep_find_key
|
||||
|
||||
|
||||
class TestDeepFindKey(unittest.TestCase):
|
||||
@@ -21,55 +20,5 @@ class TestDeepFindKey(unittest.TestCase):
|
||||
self.assertEqual(deep_find_key(deep_payload, {"target"}, max_depth=8), ["value"])
|
||||
|
||||
|
||||
class TestParseListingSegments(unittest.TestCase):
|
||||
def test_empty_and_invalid_return_empty(self) -> None:
|
||||
self.assertEqual(parse_listing_segments(""), [])
|
||||
self.assertEqual(parse_listing_segments(" "), [])
|
||||
self.assertEqual(parse_listing_segments("invalid"), [])
|
||||
|
||||
def test_auto_segments_complete_coverage(self) -> None:
|
||||
"""auto: все бренды покрыты, крупные разбиты по годам, годы без дыр."""
|
||||
segs = parse_listing_segments("auto")
|
||||
|
||||
# Точное число сегментов.
|
||||
expected = len(_LARGE_MAKES) * len(_YEAR_SPLITS) + (len(IAAI_DEFAULT_MAKES) - len(_LARGE_MAKES))
|
||||
self.assertEqual(len(segs), expected)
|
||||
|
||||
# Все бренды из списка присутствуют.
|
||||
makes_in_segments = {s["make"] for s in segs}
|
||||
for make in IAAI_DEFAULT_MAKES:
|
||||
self.assertIn(make, makes_in_segments)
|
||||
|
||||
# Крупные бренды разбиты на 3 сегмента с годами.
|
||||
toyota = [s for s in segs if s["make"] == "TOYOTA"]
|
||||
self.assertEqual(len(toyota), 3)
|
||||
for s in toyota:
|
||||
self.assertIsNotNone(s["year_min"])
|
||||
|
||||
# Мелкие бренды без годов.
|
||||
lexus = [s for s in segs if s["make"] == "LEXUS"]
|
||||
self.assertEqual(len(lexus), 1)
|
||||
self.assertIsNone(lexus[0]["year_min"])
|
||||
|
||||
# Годовые диапазоны покрывают 1950-2027.
|
||||
years = set()
|
||||
for yr_min, yr_max in _YEAR_SPLITS:
|
||||
years.update(range(yr_min, yr_max + 1))
|
||||
for year in range(1950, 2027):
|
||||
self.assertIn(year, years)
|
||||
|
||||
def test_json_input_formats(self) -> None:
|
||||
# Массив строк.
|
||||
segs = parse_listing_segments('["toyota", "ford"]')
|
||||
self.assertEqual(len(segs), 2)
|
||||
self.assertEqual(segs[0]["make"], "TOYOTA")
|
||||
|
||||
# Массив объектов с годами.
|
||||
segs = parse_listing_segments('[{"make":"BMW","year_min":2020,"year_max":2025}]')
|
||||
self.assertEqual(segs[0]["make"], "BMW")
|
||||
self.assertEqual(segs[0]["year_min"], 2020)
|
||||
self.assertEqual(segs[0]["year_max"], 2025)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
|
||||
@@ -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()
|
||||
|
||||
Reference in New Issue
Block a user