[extractor/common] Allow angle brackets in attributes in _og_regexes (#7215)

This commit is contained in:
Sergey M․ 2015-10-18 09:11:02 +06:00
parent 49941c4e4f
commit 448ef1f31c
2 changed files with 5 additions and 1 deletions

View File

@ -37,12 +37,16 @@ def test_opengraph(self):
<meta property='og:image' content='http://domain.com/pic.jpg?key1=val1&amp;key2=val2'/>
<meta content='application/x-shockwave-flash' property='og:video:type'>
<meta content='Foo' property=og:foobar>
<meta name="og:test1" content='foo > < bar'/>
<meta name="og:test2" content="foo >//< bar"/>
'''
self.assertEqual(ie._og_search_title(html), 'Foo')
self.assertEqual(ie._og_search_description(html), 'Some video\'s description ')
self.assertEqual(ie._og_search_thumbnail(html), 'http://domain.com/pic.jpg?key1=val1&key2=val2')
self.assertEqual(ie._og_search_video_url(html, default=None), None)
self.assertEqual(ie._og_search_property('foobar', html), 'Foo')
self.assertEqual(ie._og_search_property('test1', html), 'foo > < bar')
self.assertEqual(ie._og_search_property('test2', html), 'foo >//< bar')
def test_html_search_meta(self):
ie = self.ie

View File

@ -645,7 +645,7 @@ def _get_tfa_info(self, note='two-factor verification code'):
# Helper functions for extracting OpenGraph info
@staticmethod
def _og_regexes(prop):
content_re = r'content=(?:"([^>]+?)"|\'([^>]+?)\'|\s*([^\s"\'=<>`]+?))'
content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))'
property_re = (r'(?:name|property)=(?:\'og:%(prop)s\'|"og:%(prop)s"|\s*og:%(prop)s\b)'
% {'prop': re.escape(prop)})
template = r'<meta[^>]+?%s[^>]+?%s'