1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-11-26 06:33:22 +08:00
youtube-dl/youtube_dl/extractor/skysports.py

34 lines
1.1 KiB
Python
Raw Normal View History

# coding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
class SkySportsIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?skysports\.com/watch/video/(?P<id>[0-9]+)'
_TEST = {
'url': 'http://www.skysports.com/watch/video/10328419/bale-its-our-time-to-shine',
'md5': 'c44a1db29f27daf9a0003e010af82100',
'info_dict': {
'id': '10328419',
'ext': 'flv',
'title': 'Bale: Its our time to shine',
'description': 'md5:9fd1de3614d525f5addda32ac3c482c9',
},
'add_ie': ['Ooyala'],
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
return {
'_type': 'url_transparent',
'id': video_id,
'url': 'ooyala:%s' % self._search_regex(
r'data-video-id="([^"]+)"', webpage, 'ooyala id'),
'title': self._og_search_title(webpage),
'description': self._og_search_description(webpage),
'ie_key': 'Ooyala',
}