Improve OpenGraph property matching

This commit is contained in:
Philipp Hagemeister 2013-07-13 20:39:47 +02:00
parent 36034aecc2
commit 3c4e6d8337

View file

@ -259,16 +259,16 @@ class InfoExtractor(object):
# Helper functions for extracting OpenGraph info # Helper functions for extracting OpenGraph info
@staticmethod @staticmethod
def _og_regex(property): def _og_regex(prop):
return r'<meta.+?property=[\'"]og:%s[\'"].+?content=(?:"(.+?)"|\'(.+?)\')' % property return r'<meta.+?property=[\'"]og:%s[\'"].+?content=(?:"(.+?)"|\'(.+?)\')' % re.escape(prop)
def _og_search_property(self, property, html, name=None, **kargs): def _og_search_property(self, prop, html, name=None, **kargs):
if name is None: if name is None:
name = 'OpenGraph %s' % property name = 'OpenGraph %s' % prop
return self._html_search_regex(self._og_regex(property), html, name, flags=re.DOTALL, **kargs) return self._html_search_regex(self._og_regex(prop), html, name, flags=re.DOTALL, **kargs)
def _og_search_thumbnail(self, html, **kargs): def _og_search_thumbnail(self, html, **kargs):
return self._og_search_property('image', html, 'thumbnail url', fatal=False, **kargs) return self._og_search_property('image', html, u'thumbnail url', fatal=False, **kargs)
def _og_search_description(self, html, **kargs): def _og_search_description(self, html, **kargs):
return self._og_search_property('description', html, fatal=False, **kargs) return self._og_search_property('description', html, fatal=False, **kargs)