Files
mobile.de/tests/test_mobilede_scraper.py
2026-05-11 12:46:07 +03:00

183 lines
7.4 KiB
Python

from __future__ import annotations
import unittest
from datetime import datetime, timezone
from unittest.mock import patch
from mobilede_scraper.mobile_de.models import MobileDeListing, MobileDeSearchPage
from mobilede_scraper.mobile_de.scraper import MobileDeScraper
class _FakePersistence:
def __init__(self, existing_ids: set[str] | None = None) -> None:
self.upsert_calls: list[list[str]] = []
self.upsert_records: list[object] = []
self.finish_payload: dict[str, object] | None = None
self.existing_ids = existing_ids or set()
def create_tables(self) -> None:
return None
def start_sync_run(self, lane: str) -> int:
return 1
def get_existing_origin_ids(self, origin_ids: list[str]) -> set[str]:
return {origin_id for origin_id in origin_ids if origin_id in self.existing_ids}
def upsert_cars_batch(self, records):
self.upsert_calls.append([record.origin_id for record in records])
self.upsert_records.extend(records)
return {
"inserted": len(records),
"updated": 0,
"images_upserted": len(records),
}
def finish_sync_run(self, run_id: int, **kwargs) -> None:
self.finish_payload = {"run_id": run_id, **kwargs}
class _FakeClient:
def __init__(self, pages: list[MobileDeSearchPage]) -> None:
self.pages = pages
self.detail_calls: list[str] = []
def build_make_model_param(self, make_id: str, model_id: str | None) -> str:
return make_id if not model_id else f"{make_id};{model_id}"
def iter_search_pages(self, **kwargs):
del kwargs
yield from self.pages
def fetch_detail(self, listing_id: str) -> dict[str, object]:
self.detail_calls.append(str(listing_id))
return {
"shortTitle": "BMW X1",
"subTitle": "xDrive20i",
"make": {"localized": "BMW"},
"model": {"localized": "X1"},
"price": {"grs": {"amount": "19999", "currency": "EUR"}},
"contact": {"countryCode": "DE"},
"attributes": [
{"tag": "firstRegistration", "value": "05/2016"},
{"tag": "mileage", "value": "71 500 km"},
{"tag": "category", "value": "OffRoad"},
{"tag": "color", "value": "Серый"},
{"tag": "wheelDrive", "value": "xDrive"},
{"tag": "cubicCapacity", "value": "1 998 ccm"},
],
"images": ["https://img.example.test/1.jpg"],
}
class TestMobileDeScraperStreamingSync(unittest.TestCase):
def test_sync_search_upserts_each_page_during_run(self) -> None:
pages = [
MobileDeSearchPage(
url="https://example.test/page-1",
page_number=1,
total_results=3,
listings=[
MobileDeListing(id="1", url="https://example.test/1", title="Car 1"),
MobileDeListing(id="2", url="https://example.test/2", title="Car 2"),
],
),
MobileDeSearchPage(
url="https://example.test/page-2",
page_number=2,
total_results=3,
listings=[
MobileDeListing(id="3", url="https://example.test/3", title="Car 3"),
],
),
]
persistence = _FakePersistence()
scraper = MobileDeScraper(
client=_FakeClient(pages),
persistence=persistence,
)
result = scraper.sync_search(max_pages=2)
self.assertEqual(len(persistence.upsert_calls), 2)
self.assertEqual(persistence.upsert_calls[0], ["mobile.de:1", "mobile.de:2"])
self.assertEqual(persistence.upsert_calls[1], ["mobile.de:3"])
self.assertEqual(int(result["upsert"]["inserted"]), 3)
self.assertEqual(int(result["listing_count"]), 3)
self.assertEqual(int(result["unique_listing_count"]), 3)
self.assertIsNotNone(persistence.finish_payload)
self.assertEqual(int(persistence.finish_payload["cars_upserted"]), 3)
def test_sync_search_applies_one_seen_at_to_all_records_in_run(self) -> None:
seen_at = datetime(2026, 5, 5, 12, 30, tzinfo=timezone.utc)
pages = [
MobileDeSearchPage(
url="https://example.test/page-1",
page_number=1,
total_results=2,
listings=[
MobileDeListing(id="1", url="https://example.test/1", title="Car 1"),
MobileDeListing(id="2", url="https://example.test/2", title="Car 2"),
],
),
]
persistence = _FakePersistence()
scraper = MobileDeScraper(client=_FakeClient(pages), persistence=persistence)
scraper.sync_search(max_pages=1, seen_at=seen_at)
records = persistence.upsert_calls
self.assertEqual(records, [["mobile.de:1", "mobile.de:2"]])
self.assertTrue(all(record.first_seen_at == seen_at for record in persistence.upsert_records))
self.assertTrue(all(record.last_seen_at == seen_at for record in persistence.upsert_records))
self.assertTrue(all(record.is_sold is False for record in persistence.upsert_records))
self.assertTrue(all(record.sold_at is None for record in persistence.upsert_records))
def test_sync_search_selectively_enriches_new_records_with_missing_fields(self) -> None:
pages = [
MobileDeSearchPage(
url="https://example.test/page-1",
page_number=1,
total_results=2,
listings=[
MobileDeListing(
id="1",
url="https://example.test/1",
title="BMW X1",
subtitle="xDrive20i",
raw={"attr": {"fr": "", "ml": "", "c": "", "ecol": "", "cc": ""}},
),
MobileDeListing(
id="2",
url="https://example.test/2",
title="BMW 320",
raw={"attr": {"fr": "09/2016", "ml": "115 000 km", "c": "EstateCar", "ecol": "Серый", "cc": "1 998 ccm"}},
),
],
),
]
client = _FakeClient(pages)
persistence = _FakePersistence(existing_ids={"mobile.de:2"})
scraper = MobileDeScraper(client=client, persistence=persistence)
with (
patch("mobilede_scraper.mobile_de.scraper.MOBILEDE_SELECTIVE_DETAIL_ENRICH_ENABLED", True),
patch("mobilede_scraper.mobile_de.scraper.MOBILEDE_SELECTIVE_DETAIL_ENRICH_MAX_PER_RUN", 1),
patch("mobilede_scraper.mobile_de.scraper.MOBILEDE_SELECTIVE_DETAIL_ENRICH_MAX_PER_PAGE", 1),
):
result = scraper.sync_search(max_pages=1)
self.assertEqual(client.detail_calls, ["1"])
self.assertEqual(result["detail_enriched"], 1)
self.assertEqual(result["detail_enrich_failed"], 0)
record_by_id = {record.origin_id: record for record in persistence.upsert_records}
self.assertEqual(record_by_id["mobile.de:1"].drive, "4WD")
self.assertEqual(record_by_id["mobile.de:1"].engine_volume, 1998)
self.assertEqual(record_by_id["mobile.de:1"].body_type, "SUV")
self.assertEqual(record_by_id["mobile.de:1"].year, 2016)
self.assertEqual(record_by_id["mobile.de:2"].engine_volume, 1998)
if __name__ == "__main__":
unittest.main()