[hgtv.com:show] Relax video config regex and update test (closes #13279, closes #13461)

This commit is contained in:
Sergey M․ 2017-06-23 00:42:42 +07:00
parent 0f4a5a73e7
commit fa3ea7223a
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -7,14 +7,19 @@ from .common import InfoExtractor
class HGTVComShowIE(InfoExtractor):
IE_NAME = 'hgtv.com:show'
_VALID_URL = r'https?://(?:www\.)?hgtv\.com/shows/[^/]+/(?P<id>[^/?#&]+)'
_TEST = {
'url': 'http://www.hgtv.com/shows/flip-or-flop/flip-or-flop-full-episodes-videos',
_TESTS = [{
# data-module="video"
'url': 'http://www.hgtv.com/shows/flip-or-flop/flip-or-flop-full-episodes-season-4-videos',
'info_dict': {
'id': 'flip-or-flop-full-episodes-videos',
'id': 'flip-or-flop-full-episodes-season-4-videos',
'title': 'Flip or Flop Full Episodes',
},
'playlist_mincount': 15,
}
}, {
# data-deferred-module="video"
'url': 'http://www.hgtv.com/shows/good-bones/episodes/an-old-victorian-house-gets-a-new-facelift',
'only_matching': True,
}]
def _real_extract(self, url):
display_id = self._match_id(url)
@ -23,7 +28,7 @@ class HGTVComShowIE(InfoExtractor):
config = self._parse_json(
self._search_regex(
r'(?s)data-module=["\']video["\'][^>]*>.*?<script[^>]+type=["\']text/x-config["\'][^>]*>(.+?)</script',
r'(?s)data-(?:deferred)?-module=["\']video["\'][^>]*>.*?<script[^>]+type=["\']text/x-config["\'][^>]*>(.+?)</script',
webpage, 'video config'),
display_id)['channels'][0]