[vimeo] Improve regex for the config js dict (fixes #3955, fixes #3974)

The javascript code contains assignments of empty dicts with the same variable name
This commit is contained in:
Jaime Marquínez Ferrándiz 2014-10-17 15:49:16 +02:00
parent 355d074ff9
commit 4698f0d858

View File

@ -157,6 +157,18 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
'duration': 62,
}
},
{
# from https://www.ouya.tv/game/Pier-Solar-and-the-Great-Architects/
'url': 'https://player.vimeo.com/video/98044508',
'note': 'The js code contains assignments to the same variable as the config',
'info_dict': {
'id': '98044508',
'ext': 'mp4',
'title': 'Pier Solar OUYA Official Trailer',
'uploader': 'Tulio Gonçalves',
'uploader_id': 'user28849593',
},
},
]
def _verify_video_password(self, url, video_id, webpage):
@ -244,7 +256,7 @@ def _real_extract(self, url):
# We try to find out to which variable is assigned the config dic
m_variable_name = re.search('(\w)\.video\.id', webpage)
if m_variable_name is not None:
config_re = r'%s=({.+?});' % re.escape(m_variable_name.group(1))
config_re = r'%s=({[^}].+?});' % re.escape(m_variable_name.group(1))
else:
config_re = [r' = {config:({.+?}),assets:', r'(?:[abc])=({.+?});']
config = self._search_regex(config_re, webpage, 'info section',