Benjamin Peterson | 90f5ba5 | 2010-03-11 22:53:45 +0000 | [diff] [blame] | 1 | #!/usr/bin/env python3 |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 2 | """ |
| 3 | Convert the X11 locale.alias file into a mapping dictionary suitable |
| 4 | for locale.py. |
| 5 | |
| 6 | Written by Marc-Andre Lemburg <mal@genix.com>, 2004-12-10. |
| 7 | |
| 8 | """ |
| 9 | import locale |
Serhiy Storchaka | 8c4f57d | 2013-12-27 00:56:53 +0200 | [diff] [blame] | 10 | import sys |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 11 | _locale = locale |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 12 | |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 13 | # Location of the X11 alias file. |
Antoine Pitrou | 0c70d2d | 2010-04-11 22:35:34 +0000 | [diff] [blame] | 14 | LOCALE_ALIAS = '/usr/share/X11/locale/locale.alias' |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 15 | # Location of the glibc SUPPORTED locales file. |
| 16 | SUPPORTED = '/usr/share/i18n/SUPPORTED' |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 17 | |
| 18 | def parse(filename): |
| 19 | |
Serhiy Storchaka | 55c6cc4 | 2013-12-23 18:56:08 +0200 | [diff] [blame] | 20 | with open(filename, encoding='latin1') as f: |
| 21 | lines = list(f) |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 22 | data = {} |
| 23 | for line in lines: |
| 24 | line = line.strip() |
| 25 | if not line: |
| 26 | continue |
| 27 | if line[:1] == '#': |
| 28 | continue |
| 29 | locale, alias = line.split() |
Serhiy Storchaka | 5eb0153 | 2013-12-26 21:20:59 +0200 | [diff] [blame] | 30 | # Fix non-standard locale names, e.g. ks_IN@devanagari.UTF-8 |
| 31 | if '@' in alias: |
| 32 | alias_lang, _, alias_mod = alias.partition('@') |
| 33 | if '.' in alias_mod: |
| 34 | alias_mod, _, alias_enc = alias_mod.partition('.') |
| 35 | alias = alias_lang + '.' + alias_enc + '@' + alias_mod |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 36 | # Strip ':' |
| 37 | if locale[-1] == ':': |
| 38 | locale = locale[:-1] |
| 39 | # Lower-case locale |
| 40 | locale = locale.lower() |
| 41 | # Ignore one letter locale mappings (except for 'c') |
| 42 | if len(locale) == 1 and locale != 'c': |
| 43 | continue |
| 44 | # Normalize encoding, if given |
| 45 | if '.' in locale: |
| 46 | lang, encoding = locale.split('.')[:2] |
| 47 | encoding = encoding.replace('-', '') |
| 48 | encoding = encoding.replace('_', '') |
| 49 | locale = lang + '.' + encoding |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 50 | data[locale] = alias |
| 51 | return data |
| 52 | |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 53 | def parse_glibc_supported(filename): |
| 54 | |
| 55 | with open(filename, encoding='latin1') as f: |
| 56 | lines = list(f) |
| 57 | data = {} |
| 58 | for line in lines: |
| 59 | line = line.strip() |
| 60 | if not line: |
| 61 | continue |
| 62 | if line[:1] == '#': |
| 63 | continue |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 64 | line = line.replace('/', ' ').strip() |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 65 | line = line.rstrip('\\').rstrip() |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 66 | words = line.split() |
| 67 | if len(words) != 2: |
| 68 | continue |
| 69 | alias, alias_encoding = words |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 70 | # Lower-case locale |
| 71 | locale = alias.lower() |
| 72 | # Normalize encoding, if given |
| 73 | if '.' in locale: |
| 74 | lang, encoding = locale.split('.')[:2] |
| 75 | encoding = encoding.replace('-', '') |
| 76 | encoding = encoding.replace('_', '') |
| 77 | locale = lang + '.' + encoding |
| 78 | # Add an encoding to alias |
| 79 | alias, _, modifier = alias.partition('@') |
| 80 | alias = _locale._replace_encoding(alias, alias_encoding) |
| 81 | if modifier and not (modifier == 'euro' and alias_encoding == 'ISO-8859-15'): |
| 82 | alias += '@' + modifier |
| 83 | data[locale] = alias |
| 84 | return data |
| 85 | |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 86 | def pprint(data): |
Georg Brandl | bf82e37 | 2008-05-16 17:02:34 +0000 | [diff] [blame] | 87 | items = sorted(data.items()) |
| 88 | for k, v in items: |
Serhiy Storchaka | 55c6cc4 | 2013-12-23 18:56:08 +0200 | [diff] [blame] | 89 | print(' %-40s%a,' % ('%a:' % k, v)) |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 90 | |
| 91 | def print_differences(data, olddata): |
Georg Brandl | bf82e37 | 2008-05-16 17:02:34 +0000 | [diff] [blame] | 92 | items = sorted(olddata.items()) |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 93 | for k, v in items: |
Georg Brandl | bf82e37 | 2008-05-16 17:02:34 +0000 | [diff] [blame] | 94 | if k not in data: |
Serhiy Storchaka | 55c6cc4 | 2013-12-23 18:56:08 +0200 | [diff] [blame] | 95 | print('# removed %a' % k) |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 96 | elif olddata[k] != data[k]: |
Serhiy Storchaka | 55c6cc4 | 2013-12-23 18:56:08 +0200 | [diff] [blame] | 97 | print('# updated %a -> %a to %a' % \ |
Collin Winter | 6afaeb7 | 2007-08-03 17:06:41 +0000 | [diff] [blame] | 98 | (k, olddata[k], data[k])) |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 99 | # Additions are not mentioned |
Tim Peters | 5a9fb3c | 2005-01-07 16:01:32 +0000 | [diff] [blame] | 100 | |
Serhiy Storchaka | 8c4f57d | 2013-12-27 00:56:53 +0200 | [diff] [blame] | 101 | def optimize(data): |
| 102 | locale_alias = locale.locale_alias |
| 103 | locale.locale_alias = data.copy() |
| 104 | for k, v in data.items(): |
| 105 | del locale.locale_alias[k] |
| 106 | if locale.normalize(k) != v: |
| 107 | locale.locale_alias[k] = v |
| 108 | newdata = locale.locale_alias |
| 109 | errors = check(data) |
| 110 | locale.locale_alias = locale_alias |
| 111 | if errors: |
| 112 | sys.exit(1) |
| 113 | return newdata |
| 114 | |
| 115 | def check(data): |
| 116 | # Check that all alias definitions from the X11 file |
| 117 | # are actually mapped to the correct alias locales. |
| 118 | errors = 0 |
| 119 | for k, v in data.items(): |
| 120 | if locale.normalize(k) != v: |
| 121 | print('ERROR: %a -> %a != %a' % (k, locale.normalize(k), v), |
| 122 | file=sys.stderr) |
| 123 | errors += 1 |
| 124 | return errors |
| 125 | |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 126 | if __name__ == '__main__': |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 127 | import argparse |
| 128 | parser = argparse.ArgumentParser() |
| 129 | parser.add_argument('--locale-alias', default=LOCALE_ALIAS, |
| 130 | help='location of the X11 alias file ' |
| 131 | '(default: %a)' % LOCALE_ALIAS) |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 132 | parser.add_argument('--glibc-supported', default=SUPPORTED, |
| 133 | help='location of the glibc SUPPORTED locales file ' |
| 134 | '(default: %a)' % SUPPORTED) |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 135 | args = parser.parse_args() |
| 136 | |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 137 | data = locale.locale_alias.copy() |
Serhiy Storchaka | 8276d87 | 2014-10-02 10:38:12 +0300 | [diff] [blame] | 138 | data.update(parse_glibc_supported(args.glibc_supported)) |
Serhiy Storchaka | ea4f057 | 2014-10-01 23:42:30 +0300 | [diff] [blame] | 139 | data.update(parse(args.locale_alias)) |
Serhiy Storchaka | 5189ee5 | 2014-10-02 10:21:43 +0300 | [diff] [blame] | 140 | while True: |
| 141 | # Repeat optimization while the size is decreased. |
| 142 | n = len(data) |
| 143 | data = optimize(data) |
| 144 | if len(data) == n: |
| 145 | break |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 146 | print_differences(data, locale.locale_alias) |
Collin Winter | 6afaeb7 | 2007-08-03 17:06:41 +0000 | [diff] [blame] | 147 | print() |
| 148 | print('locale_alias = {') |
Marc-André Lemburg | bb4f1bd | 2004-12-10 21:58:14 +0000 | [diff] [blame] | 149 | pprint(data) |
Collin Winter | 6afaeb7 | 2007-08-03 17:06:41 +0000 | [diff] [blame] | 150 | print('}') |