import json
from .common import InfoExtractor
+from ..compat import (
+ compat_str,
+ compat_urllib_parse,
+ compat_urllib_request,
+)
from ..utils import (
ExtractorError,
- compat_urllib_request,
- compat_urllib_parse,
- compat_str,
+ orderedSet,
unescapeHTML,
unified_strdate,
- orderedSet)
+)
class VKIE(InfoExtractor):
'id': '162222515',
'ext': 'flv',
'title': 'ProtivoGunz - Хуёвая песня',
- 'uploader': 're:Noize MC.*',
+ 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
'duration': 195,
'upload_date': '20120212',
},
'act': 'login',
'role': 'al_frame',
'expire': '1',
- 'email': username,
- 'pass': password,
+ 'email': username.encode('cp1251'),
+ 'pass': password.encode('cp1251'),
}
request = compat_urllib_request.Request('https://login.vk.com/?act=login',
- compat_urllib_parse.urlencode(login_form).encode('utf-8'))
+ compat_urllib_parse.urlencode(login_form).encode('utf-8'))
login_page = self._download_webpage(request, None, note='Logging in as %s' % username)
if re.search(r'onLoginFailed', login_page):
if not video_id:
video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
- info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id
+ info_url = 'http://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id
info_page = self._download_webpage(info_url, video_id)
ERRORS = {
r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
- 'Video %s has been removed from public access due to rightholder complaint.',
+ 'Video %s has been removed from public access due to rightholder complaint.',
+
r'<!>Please log in or <':
- 'Video %s is only available for registered users, '
- 'use --username and --password options to provide account credentials.',
- '<!>Unknown error':
- 'Video %s does not exist.'
+ 'Video %s is only available for registered users, '
+ 'use --username and --password options to provide account credentials.',
+
+ r'<!>Unknown error':
+ 'Video %s does not exist.',
+
+ r'<!>Видео временно недоступно':
+ 'Video %s is temporarily unavailable.',
}
for error_re, error_msg in ERRORS.items():
self.to_screen('Youtube video detected')
return self.url_result(m_yt.group(1), 'Youtube')
- m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.*?});', info_page)
+ m_rutube = re.search(
+ r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page)
+ if m_rutube is not None:
+ self.to_screen('rutube video detected')
+ rutube_url = self._proto_relative_url(
+ m_rutube.group(1).replace('\\', ''))
+ return self.url_result(rutube_url)
+
+ m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page)
if m_opts:
- m_opts_url = re.search(r"url\s*:\s*'([^']+)", m_opts.group(1))
+ m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1))
if m_opts_url:
opts_url = m_opts_url.group(1)
if opts_url.startswith('//'):
opts_url = 'http:' + opts_url
return self.url_result(opts_url)
- data_json = self._search_regex(r'var vars = ({.*?});', info_page, 'vars')
+ data_json = self._search_regex(r'var\s+vars\s*=\s*({.+?});', info_page, 'vars')
data = json.loads(data_json)
# Extract upload date
upload_date = None
- mobj = re.search(r'id="mv_date_wrap".*?Added ([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page)
+ mobj = re.search(r'id="mv_date(?:_views)?_wrap"[^>]*>([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page)
if mobj is not None:
- x = mobj.group(1) + ' ' + mobj.group(2)
+ mobj.group(1) + ' ' + mobj.group(2)
upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2))
formats = [{
_TEMPLATE_URL = 'https://vk.com/videos'
_TEST = {
'url': 'http://vk.com/videos205387401',
+ 'info_dict': {
+ 'id': '205387401',
+ },
'playlist_mincount': 4,
}