[utils] Transliterate "þ" as "th" (#20897)
Despite visual similarity "þ" is unrelated to "p". It is normally transliterated as "th": $ echo þ-Þ | iconv -t ASCII//TRANSLIT th-THpull/21061/head
parent
4eec112740
commit
fd35d8cdfd
|
@ -183,7 +183,7 @@ class TestUtil(unittest.TestCase):
|
|||
|
||||
self.assertEqual(sanitize_filename(
|
||||
'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
|
||||
'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
|
||||
'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
|
||||
|
||||
def test_sanitize_ids(self):
|
||||
self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
|
||||
|
|
|
@ -125,8 +125,8 @@ KNOWN_EXTENSIONS = (
|
|||
|
||||
# needed for sanitizing filenames in restricted mode
|
||||
ACCENT_CHARS = dict(zip('ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ',
|
||||
itertools.chain('AAAAAA', ['AE'], 'CEEEEIIIIDNOOOOOOO', ['OE'], 'UUUUUYP', ['ss'],
|
||||
'aaaaaa', ['ae'], 'ceeeeiiiionooooooo', ['oe'], 'uuuuuypy')))
|
||||
itertools.chain('AAAAAA', ['AE'], 'CEEEEIIIIDNOOOOOOO', ['OE'], 'UUUUUY', ['TH', 'ss'],
|
||||
'aaaaaa', ['ae'], 'ceeeeiiiionooooooo', ['oe'], 'uuuuuy', ['th'], 'y')))
|
||||
|
||||
DATE_FORMATS = (
|
||||
'%d %B %Y',
|
||||
|
|
Loading…
Reference in New Issue