_ 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         'info_dict': {
  12             'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG',
  13             'ext': 'mp4',
  14             'title': '30 for 30 Shorts: Judging Jewell',
  15             'description': None,
  16         },
  17         'params': {
  18             # m3u8 download
  19             'skip_download': True,
  20         },
  21     }, {
  22         # intl video, from http://www.espnfc.us/video/mls-highlights/150/video/2743663/must-see-moments-best-of-the-mls-season
  23         'url': 'http://espn.go.com/video/clip?id=2743663',
  24         'info_dict': {
  25             'id': '50NDFkeTqRHB0nXBOK-RGdSG5YQPuxHg',
  26             'ext': 'mp4',
  27             'title': 'Must-See Moments: Best of the MLS season',
  28         },
  29         'params': {
  30             # m3u8 download
  31             'skip_download': True,
  32         },
  33     }, {
  34         'url': 'https://espn.go.com/video/iframe/twitter/?cms=espn&id=10365079',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://espn.go.com/nba/recap?gameId=400793786',
  38         'only_matching': True,
  39     }, {
  40         'url': 'http://espn.go.com/blog/golden-state-warriors/post/_/id/593/how-warriors-rapidly-regained-a-winning-edge',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://espn.go.com/sports/endurance/story/_/id/12893522/dzhokhar-tsarnaev-sentenced-role-boston-marathon-bombings',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://espn.go.com/nba/playoffs/2015/story/_/id/12887571/john-wall-washington-wizards-no-swelling-left-hand-wrist-game-5-return',
  47         'only_matching': True,
  48     }]
  49
  50     def _real_extract(self, url):
  51         video_id = self._match_id(url)
  52
  53         webpage = self._download_webpage(url, video_id)
  54
  55         video_id = self._search_regex(
  56             r'class="video-play-button"[^>]+data-id="(\d+)',
  57             webpage, 'video id')
  58
  59         cms = 'espn'
  60         if 'data-source="intl"' in webpage:
  61             cms = 'intl'
  62         player_url = 'https://espn.go.com/video/iframe/twitter/?id=%s&cms=%s' % (video_id, cms)
  63         player = self._download_webpage(
  64             player_url, video_id)
  65
  66         pcode = self._search_regex(
  67             r'["\']pcode=([^"\']+)["\']', player, 'pcode')
  68
  69         title = remove_end(
  70             self._og_search_title(webpage),
  71             '- ESPN Video').strip()
  72
  73         return {
  74             '_type': 'url_transparent',
  75             'url': 'ooyalaexternal:%s:%s:%s' % (cms, video_id, pcode),
  76             'ie_key': 'OoyalaExternal',
  77             'title': title,
  78         }