compat_urllib_parse,
compat_str,
unescapeHTML,
- unified_strdate)
+ unified_strdate,
+ orderedSet)
class VKIE(InfoExtractor):
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)
+ video_ids = orderedSet(m.group(1) for m in re.finditer(r'href="/video([0-9_]+)"', page))
self._downloader.to_screen('[vk] User videos %s: Found %i videos' % (page_id, len(video_ids)))