1 from __future__ import unicode_literals
3 from .common import InfoExtractor
5 from ..compat import compat_urllib_parse
6 from ..utils import qualities
9 class TheSceneIE(InfoExtractor):
10 _VALID_URL = r'https://thescene\.com/watch/[^/]+/(?P<id>[^/#?]+)'
13 'url': 'https://thescene.com/watch/vogue/narciso-rodriguez-spring-2013-ready-to-wear',
15 'id': '520e8faac2b4c00e3c6e5f43',
17 'title': 'Narciso Rodriguez: Spring 2013 Ready-to-Wear',
18 'display_id': 'narciso-rodriguez-spring-2013-ready-to-wear',
22 def _real_extract(self, url):
23 display_id = self._match_id(url)
24 webpage = self._download_webpage(url, display_id)
25 player_url = compat_urllib_parse.urljoin(
27 self._html_search_regex(
28 r'id=\'js-player-script\'[^>]+src=\'(.+?)\'', webpage, 'player url'))
30 self.to_screen(player_url)
31 player = self._download_webpage(player_url, player_url)
32 info = self._parse_json(self._search_regex(r'(?m)var\s+video\s+=\s+({.+?});$', player, 'info json'), display_id)
34 qualities_order = qualities(['low', 'high'])
36 'format_id': '{0}-{1}'.format(f['type'].split('/')[0], f['quality']),
38 'quality': qualities_order(f['quality']),
39 } for f in info['sources'][0]]
40 self._sort_formats(formats)
44 'title': info['title'],
46 'thumbnail': info.get('poster_frame'),
47 'display_id': display_id,