1
0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2025-01-03 06:23:02 +00:00

Compare commits

...

2 Commits

Author SHA1 Message Date
dirkf
dd9aa74bee [test] Avoid name TestIE which causes a pytest warning
See: 060ac76257
2023-02-14 16:36:40 +00:00
dirkf
42b098dd79 [InfoExtractor] Handle unquoted values in OpenGraph searches 2023-02-14 02:53:16 +00:00
2 changed files with 5 additions and 3 deletions
test
youtube_dl/extractor

View File

@ -35,13 +35,13 @@ class InfoExtractorTestRequestHandler(compat_http_server.BaseHTTPRequestHandler)
assert False
class TestIE(InfoExtractor):
class DummyIE(InfoExtractor):
pass
class TestInfoExtractor(unittest.TestCase):
def setUp(self):
self.ie = TestIE(FakeYDL())
self.ie = DummyIE(FakeYDL())
def test_ie_key(self):
self.assertEqual(get_info_extractor(YoutubeIE.ie_key()), YoutubeIE)
@ -62,6 +62,7 @@ class TestInfoExtractor(unittest.TestCase):
<meta name="og:test1" content='foo > < bar'/>
<meta name="og:test2" content="foo >//< bar"/>
<meta property=og-test3 content='Ill-formatted opengraph'/>
<meta property=og:test4 content=unquoted-value/>
'''
self.assertEqual(ie._og_search_title(html), 'Foo')
self.assertEqual(ie._og_search_description(html), 'Some video\'s description ')
@ -74,6 +75,7 @@ class TestInfoExtractor(unittest.TestCase):
self.assertEqual(ie._og_search_property(('test0', 'test1'), html), 'foo > < bar')
self.assertRaises(RegexNotFoundError, ie._og_search_property, 'test0', html, None, fatal=True)
self.assertRaises(RegexNotFoundError, ie._og_search_property, ('test0', 'test00'), html, None, fatal=True)
self.assertEqual(ie._og_search_property('test4', html), 'unquoted-value')
def test_html_search_meta(self):
ie = self.ie

View File

@ -1087,7 +1087,7 @@ class InfoExtractor(object):
# Helper functions for extracting OpenGraph info
@staticmethod
def _og_regexes(prop):
content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))'
content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\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'