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