diff options
Diffstat (limited to 'youtube_dl/extractor/vk.py')
| -rw-r--r-- | youtube_dl/extractor/vk.py | 94 |
1 files changed, 79 insertions, 15 deletions
diff --git a/youtube_dl/extractor/vk.py b/youtube_dl/extractor/vk.py index 38ff3c1a9..8f677cae3 100644 --- a/youtube_dl/extractor/vk.py +++ b/youtube_dl/extractor/vk.py @@ -21,7 +21,17 @@ from ..utils import ( class VKIE(InfoExtractor): IE_NAME = 'vk.com' - _VALID_URL = r'https?://(?:m\.)?vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:.+?\?.*?z=)?video(?P<videoid>[^s].*?)(?:\?|%2F|$))' + _VALID_URL = r'''(?x) + https?:// + (?: + (?:m\.)?vk\.com/video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)| + (?: + (?:m\.)?vk\.com/(?:.+?\?.*?z=)?video| + (?:www\.)?biqle\.ru/watch/ + ) + (?P<videoid>[^s].*?)(?:\?(?:.*\blist=(?P<list_id>[\da-f]+))?|%2F|$) + ) + ''' _NETRC_MACHINE = 'vk' _TESTS = [ @@ -110,10 +120,44 @@ class VKIE(InfoExtractor): 'skip': 'Only works from Russia', }, { + # video (removed?) only available with list id + 'url': 'https://vk.com/video30481095_171201961?list=8764ae2d21f14088d4', + 'md5': '091287af5402239a1051c37ec7b92913', + 'info_dict': { + 'id': '171201961', + 'ext': 'mp4', + 'title': 'ТюменцевВВ_09.07.2015', + 'uploader': 'Anton Ivanov', + 'duration': 109, + 'upload_date': '20150709', + 'view_count': int, + }, + }, + { + # youtube embed + 'url': 'https://vk.com/video276849682_170681728', + 'info_dict': { + 'id': 'V3K4mi0SYkc', + 'ext': 'mp4', + 'title': "DSWD Awards 'Children's Joy Foundation, Inc.' Certificate of Registration and License to Operate", + 'description': 'md5:bf9c26cfa4acdfb146362682edd3827a', + 'duration': 179, + 'upload_date': '20130116', + 'uploader': "Children's Joy Foundation", + 'uploader_id': 'thecjf', + 'view_count': int, + }, + }, + { # removed video, just testing that we match the pattern 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a', 'only_matching': True, }, + { + # vk wrapper + 'url': 'http://www.biqle.ru/watch/847655_160197695', + 'only_matching': True, + } ] def _login(self): @@ -121,20 +165,25 @@ class VKIE(InfoExtractor): if username is None: return - login_form = { - 'act': 'login', - 'role': 'al_frame', - 'expire': '1', + login_page = self._download_webpage( + 'https://vk.com', None, 'Downloading login page') + + login_form = self._hidden_inputs(login_page) + + login_form.update({ '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')) - login_page = self._download_webpage(request, None, note='Logging in as %s' % username) + request = compat_urllib_request.Request( + 'https://login.vk.com/?act=login', + 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): - raise ExtractorError('Unable to login, incorrect username and/or password', expected=True) + raise ExtractorError( + 'Unable to login, incorrect username and/or password', expected=True) def _real_initialize(self): self._login() @@ -146,9 +195,20 @@ class VKIE(InfoExtractor): 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&module=video&video=%s' % video_id + info_url = 'https://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id + + # Some videos (removed?) can only be downloaded with list id specified + list_id = mobj.group('list_id') + if list_id: + info_url += '&list=%s' % list_id + info_page = self._download_webpage(info_url, video_id) + if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page): + raise ExtractorError( + 'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.', + expected=True) + 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.', @@ -162,16 +222,20 @@ class VKIE(InfoExtractor): r'<!>Видео временно недоступно': 'Video %s is temporarily unavailable.', + + r'<!>Access denied': + 'Access denied to video %s.', } for error_re, error_msg in ERRORS.items(): if re.search(error_re, info_page): raise ExtractorError(error_msg % video_id, expected=True) - m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page) - if m_yt is not None: - self.to_screen('Youtube video detected') - return self.url_result(m_yt.group(1), 'Youtube') + youtube_url = self._search_regex( + r'<iframe[^>]+src="((?:https?:)?//www.youtube.com/embed/[^"]+)"', + info_page, 'youtube iframe', default=None) + if youtube_url: + return self.url_result(youtube_url, 'Youtube') m_rutube = re.search( r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page) |