fix listing parsing
This commit is contained in:
@@ -10,6 +10,7 @@ from iaai_scraper.browser.listing import ListingCollector
|
||||
class _FakePage:
|
||||
def __init__(self, counts: dict[str, int]) -> None:
|
||||
self._counts = counts
|
||||
self._evaluate_result = False
|
||||
|
||||
class _Locator:
|
||||
def __init__(self, count_value: int) -> None:
|
||||
@@ -21,6 +22,9 @@ class _FakePage:
|
||||
def locator(self, selector: str) -> "_FakePage._Locator":
|
||||
return _FakePage._Locator(self._counts.get(selector, 0))
|
||||
|
||||
def evaluate(self, _script: str):
|
||||
return self._evaluate_result
|
||||
|
||||
|
||||
class TestListingUnit(unittest.TestCase):
|
||||
def test_has_next_page_true_for_known_selector(self) -> None:
|
||||
@@ -31,6 +35,30 @@ class TestListingUnit(unittest.TestCase):
|
||||
page = _FakePage({})
|
||||
self.assertFalse(ListingCollector._has_next_page(page))
|
||||
|
||||
def test_has_next_page_true_for_numeric_pagination_fallback(self) -> None:
|
||||
page = _FakePage({})
|
||||
page._evaluate_result = True
|
||||
self.assertTrue(ListingCollector._has_next_page(page))
|
||||
|
||||
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()
|
||||
|
||||
Reference in New Issue
Block a user