[keek] fix info extraction

This commit is contained in:
remitamine 2015-09-24 22:38:00 +01:00 committed by Sergey M․
parent dd467d33d0
commit e13b9e7885

View file

@ -4,43 +4,27 @@ from .common import InfoExtractor
class KeekIE(InfoExtractor): class KeekIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?keek\.com/(?:!|\w+/keeks/)(?P<id>\w+)' _VALID_URL = r'https?://(?:www\.)?keek\.com/keek/(?P<id>\w+)'
IE_NAME = 'keek' IE_NAME = 'keek'
_TEST = { _TEST = {
'url': 'https://www.keek.com/ytdl/keeks/NODfbab', 'url': 'https://www.keek.com/keek/NODfbab',
'md5': '09c5c109067536c1cec8bac8c21fea05', 'md5': '9b0636f8c0f7614afa4ea5e4c6e57e83',
'info_dict': { 'info_dict': {
'id': 'NODfbab', 'id': 'NODfbab',
'ext': 'mp4', 'ext': 'mp4',
'uploader': 'youtube-dl project', 'title': 'test chars: "\'/\\ä<>This is a test video for youtube-dl.For more information, contact phihag@phihag.de . - Video - Videos on Keek',
'uploader_id': 'ytdl',
'title': 'test chars: "\'/\\\u00e4<>This is a test video for youtube-dl.For more information, contact phihag@phihag.de .',
}, },
} }
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
video_url = 'http://cdn.keek.com/keek/video/%s' % video_id
thumbnail = 'http://cdn.keek.com/keek/thumbnail/%s/w100/h75' % video_id
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
raw_desc = self._html_search_meta('description', webpage)
if raw_desc:
uploader = self._html_search_regex(
r'Watch (.*?)\s+\(', raw_desc, 'uploader', fatal=False)
uploader_id = self._html_search_regex(
r'Watch .*?\(@(.+?)\)', raw_desc, 'uploader_id', fatal=False)
else:
uploader = None
uploader_id = None
return { return {
'id': video_id, 'id': video_id,
'url': video_url, 'url': self._og_search_video_url(webpage),
'ext': 'mp4', 'ext': 'mp4',
'title': self._og_search_title(webpage), 'title': self._og_search_title(webpage),
'thumbnail': thumbnail, 'thumbnail': self._og_search_thumbnail(webpage),
'uploader': uploader,
'uploader_id': uploader_id,
} }