[vimeo:review] Fix extraction for password-protected videos

Closes #9853
This commit is contained in:
Yen Chi Hsuan 2016-06-24 16:48:37 +08:00
parent fee70322d7
commit c1ff6e1ad0
No known key found for this signature in database
GPG Key ID: 3FDDD575826C5C30
1 changed files with 50 additions and 22 deletions

View File

@ -16,6 +16,7 @@ from ..utils import (
ExtractorError, ExtractorError,
InAdvancePagedList, InAdvancePagedList,
int_or_none, int_or_none,
NO_DEFAULT,
RegexNotFoundError, RegexNotFoundError,
sanitized_Request, sanitized_Request,
smuggle_url, smuggle_url,
@ -56,6 +57,26 @@ class VimeoBaseInfoExtractor(InfoExtractor):
self._set_vimeo_cookie('vuid', vuid) self._set_vimeo_cookie('vuid', vuid)
self._download_webpage(login_request, None, False, 'Wrong login info') self._download_webpage(login_request, None, False, 'Wrong login info')
def _verify_video_password(self, url, video_id, webpage):
password = self._downloader.params.get('videopassword')
if password is None:
raise ExtractorError('This video is protected by a password, use the --video-password option', expected=True)
token, vuid = self._extract_xsrft_and_vuid(webpage)
data = urlencode_postdata({
'password': password,
'token': token,
})
if url.startswith('http://'):
# vimeo only supports https now, but the user can give an http url
url = url.replace('http://', 'https://')
password_request = sanitized_Request(url + '/password', data)
password_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
password_request.add_header('Referer', url)
self._set_vimeo_cookie('vuid', vuid)
return self._download_webpage(
password_request, video_id,
'Verifying the password', 'Wrong password')
def _extract_xsrft_and_vuid(self, webpage): def _extract_xsrft_and_vuid(self, webpage):
xsrft = self._search_regex( xsrft = self._search_regex(
r'(?:(?P<q1>["\'])xsrft(?P=q1)\s*:|xsrft\s*[=:])\s*(?P<q>["\'])(?P<xsrft>.+?)(?P=q)', r'(?:(?P<q1>["\'])xsrft(?P=q1)\s*:|xsrft\s*[=:])\s*(?P<q>["\'])(?P<xsrft>.+?)(?P=q)',
@ -344,26 +365,6 @@ class VimeoIE(VimeoBaseInfoExtractor):
if mobj: if mobj:
return mobj.group(1) return mobj.group(1)
def _verify_video_password(self, url, video_id, webpage):
password = self._downloader.params.get('videopassword')
if password is None:
raise ExtractorError('This video is protected by a password, use the --video-password option', expected=True)
token, vuid = self._extract_xsrft_and_vuid(webpage)
data = urlencode_postdata({
'password': password,
'token': token,
})
if url.startswith('http://'):
# vimeo only supports https now, but the user can give an http url
url = url.replace('http://', 'https://')
password_request = sanitized_Request(url + '/password', data)
password_request.add_header('Content-Type', 'application/x-www-form-urlencoded')
password_request.add_header('Referer', url)
self._set_vimeo_cookie('vuid', vuid)
return self._download_webpage(
password_request, video_id,
'Verifying the password', 'Wrong password')
def _verify_player_video_password(self, url, video_id): def _verify_player_video_password(self, url, video_id):
password = self._downloader.params.get('videopassword') password = self._downloader.params.get('videopassword')
if password is None: if password is None:
@ -791,12 +792,39 @@ class VimeoReviewIE(VimeoBaseInfoExtractor):
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
'uploader_id': 'user22258446', 'uploader_id': 'user22258446',
} }
}, {
'note': 'Password protected',
'url': 'https://vimeo.com/user37284429/review/138823582/c4d865efde',
'info_dict': {
'id': '138823582',
'ext': 'mp4',
'title': 'EFFICIENT PICKUP MASTERCLASS MODULE 1',
'uploader': 'TMB',
'uploader_id': 'user37284429',
},
'params': {
'videopassword': 'holygrail',
},
}] }]
def _real_initialize(self):
self._login()
def _get_config_url(self, webpage_url, video_id, video_password_verified=False):
webpage = self._download_webpage(webpage_url, video_id)
config_url = self._html_search_regex(
r'data-config-url="([^"]+)"', webpage, 'config URL',
default=NO_DEFAULT if video_password_verified else None)
if config_url is None:
self._verify_video_password(webpage_url, video_id, webpage)
config_url = self._get_config_url(
webpage_url, video_id, video_password_verified=True)
return config_url
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
config = self._download_json( config_url = self._get_config_url(url, video_id)
'https://player.vimeo.com/video/%s/config' % video_id, video_id) config = self._download_json(config_url, video_id)
info_dict = self._parse_config(config, video_id) info_dict = self._parse_config(config, video_id)
self._vimeo_sort_formats(info_dict['formats']) self._vimeo_sort_formats(info_dict['formats'])
info_dict['id'] = video_id info_dict['id'] = video_id