Issue #20046: Locale alias table no longer contains entities which can be

calculated.  Generalized support of the euro modifier.

Lib/test/test_locale.py

@@ -384,6 +384,7 @@ class NormalizeTest(unittest.TestCase):
     def test_english(self):
         self.check('en', 'en_US.ISO8859-1')
         self.check('EN', 'en_US.ISO8859-1')
+        self.check('en.iso88591', 'en_US.ISO8859-1')
         self.check('en_US', 'en_US.ISO8859-1')
         self.check('en_us', 'en_US.ISO8859-1')
         self.check('en_GB', 'en_GB.ISO8859-1')
@@ -392,7 +393,10 @@ class NormalizeTest(unittest.TestCase):
         self.check('en_US:UTF-8', 'en_US.UTF-8')
         self.check('en_US.ISO8859-1', 'en_US.ISO8859-1')
         self.check('en_US.US-ASCII', 'en_US.ISO8859-1')
+        self.check('en_US.88591', 'en_US.ISO8859-1')
+        self.check('en_US.885915', 'en_US.ISO8859-15')
         self.check('english', 'en_EN.ISO8859-1')
+        self.check('english_uk.ascii', 'en_GB.ISO8859-1')
 
     def test_hyphenated_encoding(self):
         self.check('az_AZ.iso88599e', 'az_AZ.ISO8859-9E')
@@ -412,10 +416,12 @@ class NormalizeTest(unittest.TestCase):
     def test_euro_modifier(self):
         self.check('de_DE@euro', 'de_DE.ISO8859-15')
         self.check('en_US.ISO8859-15@euro', 'en_US.ISO8859-15')
+        self.check('de_DE.utf8@euro', 'de_DE.UTF-8')
 
     def test_latin_modifier(self):
         self.check('be_BY.UTF-8@latin', 'be_BY.UTF-8@latin')
         self.check('sr_RS.UTF-8@latin', 'sr_RS.UTF-8@latin')
+        self.check('sr_RS.UTF-8@latn', 'sr_RS.UTF-8@latin')
 
     def test_valencia_modifier(self):
         self.check('ca_ES.UTF-8@valencia', 'ca_ES.UTF-8@valencia')
@@ -436,6 +442,39 @@ class NormalizeTest(unittest.TestCase):
         self.check('sd_IN', 'sd_IN.UTF-8')
         self.check('sd', 'sd_IN.UTF-8')
 
+    def test_euc_encoding(self):
+        self.check('ja_jp.euc', 'ja_JP.eucJP')
+        self.check('ja_jp.eucjp', 'ja_JP.eucJP')
+        self.check('ko_kr.euc', 'ko_KR.eucKR')
+        self.check('ko_kr.euckr', 'ko_KR.eucKR')
+        self.check('zh_cn.euc', 'zh_CN.eucCN')
+        self.check('zh_tw.euc', 'zh_TW.eucTW')
+        self.check('zh_tw.euctw', 'zh_TW.eucTW')
+
+    def test_japanese(self):
+        self.check('ja', 'ja_JP.eucJP')
+        self.check('ja.jis', 'ja_JP.JIS7')
+        self.check('ja.sjis', 'ja_JP.SJIS')
+        self.check('ja_jp', 'ja_JP.eucJP')
+        self.check('ja_jp.ajec', 'ja_JP.eucJP')
+        self.check('ja_jp.euc', 'ja_JP.eucJP')
+        self.check('ja_jp.eucjp', 'ja_JP.eucJP')
+        self.check('ja_jp.iso-2022-jp', 'ja_JP.JIS7')
+        self.check('ja_jp.iso2022jp', 'ja_JP.JIS7')
+        self.check('ja_jp.jis', 'ja_JP.JIS7')
+        self.check('ja_jp.jis7', 'ja_JP.JIS7')
+        self.check('ja_jp.mscode', 'ja_JP.SJIS')
+        self.check('ja_jp.pck', 'ja_JP.SJIS')
+        self.check('ja_jp.sjis', 'ja_JP.SJIS')
+        self.check('ja_jp.ujis', 'ja_JP.eucJP')
+        self.check('ja_jp.utf8', 'ja_JP.UTF-8')
+        self.check('japan', 'ja_JP.eucJP')
+        self.check('japanese', 'ja_JP.eucJP')
+        self.check('japanese-euc', 'ja_JP.eucJP')
+        self.check('japanese.euc', 'ja_JP.eucJP')
+        self.check('japanese.sjis', 'ja_JP.SJIS')
+        self.check('jp_jp', 'ja_JP.eucJP')
+
 
 class TestMiscellaneous(unittest.TestCase):
     def test_getpreferredencoding(self):

Misc/NEWS

@@ -44,6 +44,9 @@ Core and Builtins
 Library
 -------
 
+- Issue #20046: Locale alias table no longer contains entities which can be
+  calculated.  Generalized support of the euro modifier.
+
 - Issue #20027: Fixed locale aliases for devanagari locales.
 
 - Issue #20067: Tkinter variables now work when wantobjects is false.

Tools/i18n/makelocalealias.py

@@ -7,6 +7,7 @@
 
 """
 import locale
+import sys
 
 # Location of the alias file
 LOCALE_ALIAS = '/usr/share/X11/locale/locale.alias'
@@ -65,9 +66,35 @@ def print_differences(data, olddata):
                   (k, olddata[k], data[k]))
         # Additions are not mentioned
 
+def optimize(data):
+    locale_alias = locale.locale_alias
+    locale.locale_alias = data.copy()
+    for k, v in data.items():
+        del locale.locale_alias[k]
+        if locale.normalize(k) != v:
+            locale.locale_alias[k] = v
+    newdata = locale.locale_alias
+    errors = check(data)
+    locale.locale_alias = locale_alias
+    if errors:
+        sys.exit(1)
+    return newdata
+
+def check(data):
+    # Check that all alias definitions from the X11 file
+    # are actually mapped to the correct alias locales.
+    errors = 0
+    for k, v in data.items():
+        if locale.normalize(k) != v:
+            print('ERROR: %a -> %a != %a' % (k, locale.normalize(k), v),
+                  file=sys.stderr)
+            errors += 1
+    return errors
+
 if __name__ == '__main__':
     data = locale.locale_alias.copy()
     data.update(parse(LOCALE_ALIAS))
+    data = optimize(data)
     print_differences(data, locale.locale_alias)
     print()
     print('locale_alias = {')