refactor mobile.de parser, fix country mapping, update README

This commit is contained in:
qananasikq
2026-05-04 18:12:10 +03:00
parent edddfc884c
commit 800867b4e6
46 changed files with 2887 additions and 10900 deletions

View File

@@ -0,0 +1,87 @@
from __future__ import annotations
import unittest
from mobilede_scraper.mobile_de.models import MobileDeListing, MobileDeSearchPage
from mobilede_scraper.mobile_de.scraper import MobileDeScraper
class _FakePersistence:
def __init__(self) -> None:
self.upsert_calls: list[list[str]] = []
self.finish_payload: dict[str, object] | None = None
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 set()
def upsert_cars_batch(self, records):
self.upsert_calls.append([record.origin_id for record in 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
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
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)
if __name__ == "__main__":
unittest.main()