_ Git - youtube-dl/blob - youtube_dl/extractor/espn.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import remove_end
   5
   6
   7 class ESPNIE(InfoExtractor):
   8     _VALID_URL = r'https?://espn\.go\.com/(?:[^/]+/)*(?P<id>[^/]+)'
   9     _TESTS = [{
  10         'url': 'http://espn.go.com/video/clip?id=10365079',
  11         'md5': '60e5d097a523e767d06479335d1bdc58',
  12         'info_dict': {
  13             'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG',
  14             'ext': 'mp4',
  15             'title': '30 for 30 Shorts: Judging Jewell',
  16             'description': None,
  17         },
  18         'add_ie': ['OoyalaExternal'],
  19     }, {
  20         # intl video, from http://www.espnfc.us/video/mls-highlights/150/video/2743663/must-see-moments-best-of-the-mls-season
  21         'url': 'http://espn.go.com/video/clip?id=2743663',
  22         'md5': 'f4ac89b59afc7e2d7dbb049523df6768',
  23         'info_dict': {
  24             'id': '50NDFkeTqRHB0nXBOK-RGdSG5YQPuxHg',
  25             'ext': 'mp4',
  26             'title': 'Must-See Moments: Best of the MLS season',
  27         },
  28         'add_ie': ['OoyalaExternal'],
  29     }, {
  30         'url': 'https://espn.go.com/video/iframe/twitter/?cms=espn&id=10365079',
  31         'only_matching': True,
  32     }, {
  33         'url': 'http://espn.go.com/nba/recap?gameId=400793786',
  34         'only_matching': True,
  35     }, {
  36         'url': 'http://espn.go.com/blog/golden-state-warriors/post/_/id/593/how-warriors-rapidly-regained-a-winning-edge',
  37         'only_matching': True,
  38     }, {
  39         'url': 'http://espn.go.com/sports/endurance/story/_/id/12893522/dzhokhar-tsarnaev-sentenced-role-boston-marathon-bombings',
  40         'only_matching': True,
  41     }, {
  42         'url': 'http://espn.go.com/nba/playoffs/2015/story/_/id/12887571/john-wall-washington-wizards-no-swelling-left-hand-wrist-game-5-return',
  43         'only_matching': True,
  44     }]
  45
  46     def _real_extract(self, url):
  47         video_id = self._match_id(url)
  48
  49         webpage = self._download_webpage(url, video_id)
  50
  51         video_id = self._search_regex(
  52             r'class=(["\']).*?video-play-button.*?\1[^>]+data-id=["\'](?P<id>\d+)',
  53             webpage, 'video id', group='id')
  54
  55         cms = 'espn'
  56         if 'data-source="intl"' in webpage:
  57             cms = 'intl'
  58         player_url = 'https://espn.go.com/video/iframe/twitter/?id=%s&cms=%s' % (video_id, cms)
  59         player = self._download_webpage(
  60             player_url, video_id)
  61
  62         pcode = self._search_regex(
  63             r'["\']pcode=([^"\']+)["\']', player, 'pcode')
  64
  65         title = remove_end(
  66             self._og_search_title(webpage),
  67             '- ESPN Video').strip()
  68
  69         return {
  70             '_type': 'url_transparent',
  71             'url': 'ooyalaexternal:%s:%s:%s' % (cms, video_id, pcode),
  72             'ie_key': 'OoyalaExternal',
  73             'title': title,
  74         }