Prepare mobile de parser release

This commit is contained in:
qananasikq
2026-04-28 12:08:24 +03:00
commit 0f2d5de535
76 changed files with 17855 additions and 0 deletions

141
tests/test_fast_sync.py Normal file
View File

@@ -0,0 +1,141 @@
from __future__ import annotations
from dataclasses import dataclass
from iaai_scraper.browser.fast_client import FastListingVehicle
from iaai_scraper.core.config import Settings
from iaai_scraper.core.runtime_config import RuntimeConfig, RuntimeFiltersConfig
from iaai_scraper.fast_sync import FastSyncEngine
from iaai_scraper.parsing.fast_mapper import FastCarMapper
def _listing_vehicle(inventory_id: str, *, status: str = "RS") -> FastListingVehicle:
return FastListingVehicle(
inventory_id=inventory_id,
tenant="US",
auction_id="1_1",
auction_date="2026-04-08T08:30:00+00:00",
inventory_status=status,
currency="USD",
timed_auction_closed=False,
timed_auction_indicator=False,
prebid_indicator=True,
buynow_indicator=False,
)
def _detail_payload(inventory_id: str, *, make: str = "ACURA", model: str = "RSX", bid: int = 500) -> dict:
return {
"inventoryView": {
"attributes": {
"Id": inventory_id,
"Year": "2002",
"Make": make,
"Model": model,
"Series": "BASE",
"Currency": "USD",
"ODOValue": "219187",
"ExteriorColor": "GRAY",
"DriveLineTypeDesc": "FWD",
"Transmission": "Automatic",
"BodyStyleName": "COUPE",
"EngineSize": "2.0L I-4",
"VehicleGrade": "50",
"PrimaryDamageDesc": "NORMAL WEAR & TEAR",
},
"imageDimensions": {
"keys": {"$values": [{"k": f"{inventory_id}~I1", "w": 1600, "h": 1200, "i": 0}]}
},
},
"auctionInformation": {
"prebidInformation": {"decimalHighBidAmount": str(bid), "highBidAmount": f"${bid}"},
"biddingInformation": {"buyNowPrice": "$0"},
},
}
class FakeFastClient:
def __init__(self, listing: list[FastListingVehicle], details: dict[str, dict]) -> None:
self.listing = listing
self.details = details
self.detail_calls = 0
self.persist_calls = 0
def iter_listing_vehicles(self, *, listing_start_url=None, make=None, max_pages=None): # noqa: ANN001, ANN202
del listing_start_url, make, max_pages
return iter(self.listing)
def fetch_vehicle_detail_payload(self, inventory_id: str) -> dict:
self.detail_calls += 1
return self.details[inventory_id]
def persist_session_state(self) -> None:
self.persist_calls += 1
@dataclass
class FakePersistence:
inserted: int = 0
images: int = 0
def get_existing_urls_and_ids(self, origin_urls, origin_ids): # noqa: ANN001, ANN202
del origin_urls, origin_ids
return set(), set()
def upsert_cars_batch(self, records): # noqa: ANN001, ANN202
self.inserted += len(records)
self.images += sum(len(record.images) for record in records)
return {"inserted": len(records), "updated": 0, "images_upserted": sum(len(record.images) for record in records)}
def mark_sold_not_in_listing_by_urls(self, active_origin_urls, lane="iaai"): # noqa: ANN001, ANN202
del active_origin_urls, lane
return 0
def test_fast_sync_engine_collects_details_and_bulk_upserts() -> None:
listing = [_listing_vehicle("45078011~US"), _listing_vehicle("45268167~US")]
details = {
"45078011~US": _detail_payload("45078011~US", make="ACURA", model="RSX", bid=500),
"45268167~US": _detail_payload("45268167~US", make="BMW", model="X5", bid=900),
}
client = FakeFastClient(listing, details)
persistence = FakePersistence()
engine = FastSyncEngine(
client=client, # type: ignore[arg-type]
mapper=FastCarMapper(),
persistence=persistence, # type: ignore[arg-type]
batch_size=10,
fetch_concurrency=2,
)
result = engine.sync_listing(runtime_config=RuntimeConfig(), only_new=False)
assert result["status"] == "success"
assert result["cars_upserted"] == 2
assert result["images_upserted"] == 2
assert result["listing"]["mode"] == "fast_hidden_payload"
assert client.detail_calls == 2
assert client.persist_calls == 1
def test_fast_sync_engine_applies_runtime_filters_before_db() -> None:
listing = [_listing_vehicle("45078011~US"), _listing_vehicle("45268167~US")]
details = {
"45078011~US": _detail_payload("45078011~US", make="ACURA", model="RSX", bid=500),
"45268167~US": _detail_payload("45268167~US", make="BMW", model="X5", bid=900),
}
runtime = RuntimeConfig(filters=RuntimeFiltersConfig.from_dict({"brands": ["BMW"]}))
persistence = FakePersistence()
engine = FastSyncEngine(
client=FakeFastClient(listing, details), # type: ignore[arg-type]
mapper=FastCarMapper(),
persistence=persistence, # type: ignore[arg-type]
batch_size=10,
fetch_concurrency=2,
)
result = engine.sync_listing(runtime_config=runtime, only_new=False)
assert result["cars_upserted"] == 1
assert result["cars_filtered"] == 1
assert persistence.inserted == 1