Update model translations

This commit is contained in:
qananasikq
2026-07-01 14:35:49 +03:00
commit 3904cd003e
53 changed files with 7629 additions and 0 deletions

148
tests/test_encar.py Normal file
View File

@@ -0,0 +1,148 @@
import unittest
from unittest.mock import MagicMock
from encar_scraper.encar import EncarMapper, EncarScraper, ENCAR_DETAIL_URL_TEMPLATE
class TestEncarMapper(unittest.TestCase):
def test_map_to_car_record(self):
payload = {
"Id": 41421262,
"Manufacturer": "Hyundai",
"Model": "G90 (RS4)",
"FormYear": "2023",
"Price": 8590,
"Mileage": 37923,
"Photos": [
{"location": "/carpicture01/pic4141/41412663_001.jpg", "ordering": 1},
{"location": "/carpicture01/pic4141/41412663_003.jpg", "ordering": 3},
],
}
record = EncarMapper().map_to_car_record(
ENCAR_DETAIL_URL_TEMPLATE.format(vehicle_id="41421262"),
payload,
)
self.assertEqual(record.origin_id, "encar:41421262")
self.assertEqual(record.parser_id, "encar:41421262")
self.assertEqual(record.brand, "Hyundai")
# Скобки удаляются — на оригинальном сайте Encar их нет.
self.assertEqual(record.model, "G90")
self.assertEqual(record.year, 2023)
self.assertEqual(record.price, 85900000) # 8590 만원 × 10000
self.assertEqual(record.mileage, 37923)
self.assertEqual(record.country, "KR")
self.assertEqual(record.currency, "KRW")
self.assertEqual(len(record.images), 2)
self.assertTrue(record.images[0].fullres_image.endswith("_001.jpg"))
def test_extract_vehicle_id_from_url(self):
url = ENCAR_DETAIL_URL_TEMPLATE.format(vehicle_id="41421262")
self.assertEqual(EncarMapper()._extract_vehicle_id(url), "41421262")
def test_map_to_car_record_uses_base_model_without_trim(self):
payload = {
"Id": 41854374,
"Manufacturer": "현대",
"Model": "Genesis",
"Badge": "BH330 Grand",
"BadgeDetail": "Prime팩",
"FormYear": "2011",
"Price": 1200,
"Mileage": 100000,
"Photos": [
{"location": "/carpicture01/pic4185/41854374_001.jpg", "ordering": 1},
],
}
record = EncarMapper().map_to_car_record(
ENCAR_DETAIL_URL_TEMPLATE.format(vehicle_id="41854374"),
payload,
)
self.assertEqual(record.brand, "Hyundai")
self.assertEqual(record.model, "Genesis")
self.assertEqual(record.origin_id, "encar:41854374")
def test_map_to_car_record_resolves_numeric_id_from_photos(self):
payload = {
"Id": "car-BOqLLkjkgbpStlVJEJvbRG",
"Manufacturer": "Kia",
"Model": "K3",
"FormYear": "2018",
"Price": 990,
"Mileage": 75000,
"Photos": [
{"location": "/carpicture01/pic4187/41878507_001.jpg", "ordering": 1},
],
}
record = EncarMapper().map_to_car_record("", payload)
self.assertEqual(record.origin_id, "encar:41878507")
self.assertEqual(record.parser_id, "encar:41878507")
self.assertIn("carid=41878507", record.origin_url)
def test_translate_model_keeps_english_text(self):
mapper = EncarMapper()
self.assertEqual(mapper._translate_model("All New Carnival"), "All New Carnival")
self.assertEqual(mapper._translate_model("New Kia Ray"), "New Kia Ray")
self.assertEqual(mapper._translate_model("K5 Hybrid 3rd gen"), "K5 Hybrid 3rd gen")
def test_translate_model_handles_korean_fragments(self):
mapper = EncarMapper()
cases = {
# Скобки и их содержимое отбрасываются как на сайте Encar.
"5시리즈 (G30)": "5 Series",
"S-클래스 W223": "S-Class W223",
"Sonata 디 엣지(DN8)": "Sonata The Edge",
"렉스턴 Sport 칸": "Rexton Sport Khan",
"티볼리 Air": "Tivoli Air",
"코란도 투리스모": "Korando Turismo",
"Kia Carnival 4세대": "Kia Carnival 4th gen",
"Cadillac 에스컬레이드 5세대": "Cadillac Escalade 5th gen",
"K7 프리미어": "K7 Premier",
}
for raw, expected in cases.items():
with self.subTest(raw=raw):
self.assertEqual(mapper._translate_model(raw), expected)
def test_translate_model_drops_untranslated_korean(self):
mapper = EncarMapper()
cases = {
# "레이" (Ray) не должен резать слово "슈팅브레이크" изнутри.
"슈팅브레이크": "",
"The New G70 슈팅브레이크": "The New G70",
"G70 (슈팅브레이크)": "G70",
}
for raw, expected in cases.items():
with self.subTest(raw=raw):
self.assertEqual(mapper._translate_model(raw), expected)
class TestEncarScraper(unittest.TestCase):
def test_collect_listing_limits_results(self):
scraper = EncarScraper()
scraper._fetch_listing_page = MagicMock(side_effect=[
{"SearchResults": [{"Id": 1}, {"Id": 2}]},
])
result = scraper.collect_listing(limit=1, page_size=2)
self.assertEqual(result["items_collected"], 1)
self.assertEqual(result["vehicle_urls"], [ENCAR_DETAIL_URL_TEMPLATE.format(vehicle_id="1")])
def test_scrape_vehicle_detail_raises_without_id(self):
scraper = EncarScraper()
with self.assertRaises(ValueError):
scraper.scrape_vehicle_detail("https://www.encar.com/invalid")
if __name__ == "__main__":
unittest.main()