Add a test for a page that doesn't contain manifest_url itself, but

does contain an embed/vod-XXXXXXXX url for a page that does contain
manifest_url.  This test currently fails because the description on
the inner page isn't in the OpenGraph format.
This commit is contained in:
Tim Mann 2021-01-30 21:42:31 -08:00
parent 691d549938
commit f5e049fbc2

View File

@ -9,7 +9,7 @@ from .common import InfoExtractor
class Pac12IE(InfoExtractor):
_VALID_URL = r'https?://(?:[a-z]+\.)?pac-12.com/(?:embed/)?(?P<id>.*)'
_TEST = {
_TESTS = [{
'url': 'https://pac-12.com/videos/2020-pac-12-womens-basketball-media-day-arizona-cal-stanford',
'md5': 'b2e3c0cb99458c8b8e2dc22cb5ac922d',
'info_dict': {
@ -18,7 +18,16 @@ class Pac12IE(InfoExtractor):
'title': '2020 Pac-12 Women\'s Basketball Media Day - Arizona, Cal & Stanford | Pac-12',
'description': 'During the 2020 Pac-12 Women\'s Basketball Media Day, Ros Gold-Onwude moderates a discussion with Arizona\'s Adia Barnes & Aari McDonald, Cal\'s Charmin Smith & Evelien Lutje Schipholt & Stanford\'s Tara VanDerveer & Kiana Williams.',
}
}
}, {
'url': 'https://pac-12.com/article/2020/11/24/sonoran-dog-dish-presented-tums',
'md5': 'a7a8ac72273b9468924bc058cc220d37',
'info_dict': {
'id': 'vod-YLMKpNLZvR0',
'ext': 'mp4',
'title': 'Sonoran Dog | The Dish, presented by TUMS | Pac-12',
'description': 'Pac-12 Networks introduces "The Dish," presented by Tums. Jaymee Sire is bringing fans a closeup to game day treats from around the Conference with each treat connecting to a Pac-12 school, bringing the flavor and recipes fans know and love right to the dish! As Arizona and USC basketball seasons tip off, the first feature item from "The Dish" is the Sonoran Dog, a beloved treat by Trojans & Wildcat fans.',
}
}]
def _real_extract(self, url):
video_id = self._match_id(url)
@ -35,7 +44,7 @@ class Pac12IE(InfoExtractor):
return None
return self.url_result(vod_url)
video_url = re.sub(r'\\', '', video_url)
title = self._html_search_regex(r'<title>(.+?)</title>',
title = self._html_search_regex(r'<title>(.+?)</title>',
webpage, 'title')
if 'vod-' not in url and vod_url is not None:
video_id = self._match_id(vod_url)