[utils] Check ext with trailing slash against the list of known extensions

This commit is contained in:
Sergey M․ 2015-11-22 17:27:13 +06:00
parent 5035536e3f
commit 9cb9a5df77
2 changed files with 19 additions and 1 deletions

View File

@ -242,6 +242,9 @@ def test_unified_dates(self):
def test_determine_ext(self): def test_determine_ext(self):
self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None) self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
def test_find_xpath_attr(self): def test_find_xpath_attr(self):
testxml = '''<root> testxml = '''<root>

View File

@ -922,9 +922,24 @@ def unified_strdate(date_str, day_first=True):
def determine_ext(url, default_ext='unknown_video'): def determine_ext(url, default_ext='unknown_video'):
if url is None: if url is None:
return default_ext return default_ext
guess = url.partition('?')[0].rpartition('.')[2].rstrip('/') guess = url.partition('?')[0].rpartition('.')[2]
if re.match(r'^[A-Za-z0-9]+$', guess): if re.match(r'^[A-Za-z0-9]+$', guess):
return guess return guess
elif guess.rstrip('/') in (
'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'aac',
'flv', 'f4v', 'f4a', 'f4b',
'webm', 'ogg', 'ogv', 'oga', 'ogx', 'spx', 'opus',
'mkv', 'mka', 'mk3d',
'avi', 'divx',
'mov',
'asf', 'wmv', 'wma',
'3gp', '3g2',
'mp3',
'flac',
'ape',
'wav',
'f4f', 'f4m', 'm3u8', 'smil'):
return guess.rstrip('/')
else: else:
return default_ext return default_ext