class YahooIE(InfoExtractor):
IE_DESC = 'Yahoo screen and movies'
- _VALID_URL = r'https?://(?:screen|movies)\.yahoo\.com/.*?-(?P<id>[0-9]+)(?:-[a-z]+)?\.html'
+ _VALID_URL = r'(?P<url>https?://(?:screen|movies)\.yahoo\.com/.*?-(?P<id>[0-9]+)(?:-[a-z]+)?\.html)'
_TESTS = [
{
'url': 'http://screen.yahoo.com/julian-smith-travis-legg-watch-214727115.html',
'title': 'The World Loves Spider-Man',
'description': '''People all over the world are celebrating the release of \"The Amazing Spider-Man 2.\" We're taking a look at the enthusiastic response Spider-Man has received from viewers all over the world.''',
}
- }
+ },
+ {
+ 'url': 'https://screen.yahoo.com/community/community-sizzle-reel-203225340.html?format=embed',
+ 'md5': '60e8ac193d8fb71997caa8fce54c6460',
+ 'info_dict': {
+ 'id': '4fe78544-8d48-39d8-97cd-13f205d9fcdb',
+ 'ext': 'mp4',
+ 'title': "Yahoo Saves 'Community'",
+ 'description': 'md5:4d4145af2fd3de00cbb6c1d664105053',
+ }
+ },
]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
+ url = mobj.group('url')
webpage = self._download_webpage(url, video_id)
items_json = self._search_regex(
# The 'meta' field is not always in the video webpage, we request it
# from another page
long_id = info['id']
- return self._get_info(long_id, video_id)
+ return self._get_info(long_id, video_id, webpage)
- def _get_info(self, long_id, video_id):
+ def _get_info(self, long_id, video_id, webpage):
query = ('SELECT * FROM yahoo.media.video.streams WHERE id="%s"'
' AND plrs="86Gj0vCaSzV_Iuf6hNylf2" AND region="US"'
' AND protocol="http"' % long_id)
'title': meta['title'],
'formats': formats,
'description': clean_html(meta['description']),
- 'thumbnail': meta.get('thumbnail'),
+ 'thumbnail': meta['thumbnail'] if meta.get('thumbnail') else self._og_search_thumbnail(webpage),
}
video_id = mobj.group('id')
webpage = self._download_webpage(url, video_id)
long_id = self._search_regex(r'contentId: \'(.+?)\',', webpage, 'long id')
- return self._get_info(long_id, video_id)
+ return self._get_info(long_id, video_id, webpage)
class YahooSearchIE(SearchInfoExtractor):