From: Jakub Wilk <jwilk@jwilk.net>
Date: Fri, 10 May 2019 18:42:32 +0000 (+0200)
Subject: [utils] Transliterate "þ" as "th" (#20897)
X-Git-Url: http://git.oshgnacknak.de/?a=commitdiff_plain;h=fd35d8cdfdc77ca6ec6d87677fe0d00df0cbb22a;p=youtube-dl

[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-TH
---

diff --git a/test/test_utils.py b/test/test_utils.py
index ca6d832a4..9ef0e422b 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -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')
diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 71713f63a..99ee54942 100644
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -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',