compat_urllib_parse,
compat_str,
unescapeHTML,
- unified_strdate)
+ unified_strdate,
+ orderedSet)
class VKIE(InfoExtractor):
},
'skip': 'Requires vk account credentials',
},
- {
- # VIDEO NOW REMOVED
- # please update if you find a video whose URL follows the same pattern
- 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
- 'md5': 'd82c22e449f036282d1d3f7f4d276869',
- 'info_dict': {
- 'id': '166094326',
- 'ext': 'mp4',
- 'uploader': 'Киномания - лучшее из мира кино',
- 'title': 'Запах женщины (1992)',
- 'duration': 9392,
- 'upload_date': '20130914'
- },
- 'skip': 'Requires vk account credentials',
- },
{
'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d',
'md5': '4d7a5ef8cf114dfa09577e57b2993202',
},
'skip': 'Only works from Russia',
},
+ {
+ # removed video, just testing that we match the pattern
+ 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
+ 'only_matching': True,
+ },
]
def _login(self):
class VKUserVideosIE(InfoExtractor):
IE_NAME = 'vk.com:user-videos'
IE_DESC = 'vk.com:All of a user\'s videos'
- _VALID_URL = r'https?://(?:m\.)?vk\.com/videos([0-9]+)(?:m\?.*)?'
+ _VALID_URL = r'https?://vk\.com/videos(?P<id>[0-9]+)(?:m\?.*)?'
_TEMPLATE_URL = 'https://vk.com/videos'
_TEST = {
'url': 'http://vk.com/videos205387401',
'playlist_mincount': 4,
}
- def extract_videos_from_page(self, page):
- ids_in_page = []
- for mobj in re.finditer(r'href="/video([0-9_]+)"', page):
- if mobj.group(1) not in ids_in_page:
- ids_in_page.append(mobj.group(1))
- return ids_in_page
-
def _real_extract(self, url):
- # Extract page id
- mobj = re.match(self._VALID_URL, url)
- if mobj is None:
- raise ExtractorError('Invalid URL: %s' % url)
-
- # Download page and get video ids
- page_id = mobj.group(1)
+ page_id = self._match_id(url)
page = self._download_webpage(url, page_id)
- video_ids = self.extract_videos_from_page(page)
-
- self._downloader.to_screen('[vk] User videos %s: Found %i videos' % (page_id, len(video_ids)))
-
- url_entries = [self.url_result('http://vk.com/video' + video_id, 'VK', video_id=video_id)
- for video_id in video_ids]
- return self.playlist_result(url_entries, page_id)
\ No newline at end of file
+ video_ids = orderedSet(
+ m.group(1) for m in re.finditer(r'href="/video([0-9_]+)"', page))
+ url_entries = [
+ self.url_result(
+ 'http://vk.com/video' + video_id, 'VK', video_id=video_id)
+ for video_id in video_ids]
+ return self.playlist_result(url_entries, page_id)