blob: 17056b9df216fc1076b040e9f2ab962021c50b00 [file] [log] [blame]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001""" Locale support.
Guido van Rossum4b8c6ea2000-02-04 15:39:30 +00002
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00003 The module provides low-level access to the C lib's locale APIs
4 and adds high level number formatting APIs as well as a locale
5 aliasing engine to complement these.
6
7 The aliasing engine includes support for many commonly used locale
8 names and maps them to values suitable for passing to the C lib's
9 setlocale() function. It also includes default encodings for all
10 supported locale names.
11
12"""
13
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +000014import sys, encodings, encodings.aliases
Georg Brandl1a3284e2007-12-02 09:40:06 +000015from builtins import str as _builtin_str
Antoine Pitrou83d6a872008-07-25 21:45:08 +000016import functools
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000017
Fredrik Lundh6c86b992000-07-09 17:12:58 +000018# Try importing the _locale module.
19#
20# If this fails, fall back on a basic 'C' locale emulation.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +000021
Tim Peters1baf8292001-01-24 10:13:46 +000022# Yuck: LC_MESSAGES is non-standard: can't tell whether it exists before
23# trying the import. So __all__ is also fiddled at the end of the file.
Guido van Rossum360e4b82007-05-14 22:51:27 +000024__all__ = ["getlocale", "getdefaultlocale", "getpreferredencoding", "Error",
25 "setlocale", "resetlocale", "localeconv", "strcoll", "strxfrm",
26 "str", "atof", "atoi", "format", "format_string", "currency",
27 "normalize", "LC_CTYPE", "LC_COLLATE", "LC_TIME", "LC_MONETARY",
28 "LC_NUMERIC", "LC_ALL", "CHAR_MAX"]
Skip Montanaro17ab1232001-01-24 06:27:27 +000029
Neal Norwitz48b98de2008-03-10 04:49:25 +000030def _strcoll(a,b):
31 """ strcoll(string,string) -> int.
32 Compares two strings according to the locale.
33 """
Mark Dickinsona56c4672009-01-27 18:17:45 +000034 return (a > b) - (a < b)
Neal Norwitz48b98de2008-03-10 04:49:25 +000035
36def _strxfrm(s):
37 """ strxfrm(string) -> string.
38 Returns a string that behaves for cmp locale-aware.
39 """
40 return s
41
Marc-André Lemburg23481142000-06-08 17:49:41 +000042try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +000043
Marc-André Lemburg23481142000-06-08 17:49:41 +000044 from _locale import *
45
46except ImportError:
47
Fredrik Lundh6c86b992000-07-09 17:12:58 +000048 # Locale emulation
49
Marc-André Lemburg23481142000-06-08 17:49:41 +000050 CHAR_MAX = 127
51 LC_ALL = 6
52 LC_COLLATE = 3
53 LC_CTYPE = 0
54 LC_MESSAGES = 5
55 LC_MONETARY = 4
56 LC_NUMERIC = 1
57 LC_TIME = 2
58 Error = ValueError
59
60 def localeconv():
Fredrik Lundh6c86b992000-07-09 17:12:58 +000061 """ localeconv() -> dict.
Marc-André Lemburg23481142000-06-08 17:49:41 +000062 Returns numeric and monetary locale-specific parameters.
63 """
64 # 'C' locale default values
65 return {'grouping': [127],
66 'currency_symbol': '',
67 'n_sign_posn': 127,
Fredrik Lundh6c86b992000-07-09 17:12:58 +000068 'p_cs_precedes': 127,
69 'n_cs_precedes': 127,
70 'mon_grouping': [],
Marc-André Lemburg23481142000-06-08 17:49:41 +000071 'n_sep_by_space': 127,
72 'decimal_point': '.',
73 'negative_sign': '',
74 'positive_sign': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000075 'p_sep_by_space': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000076 'int_curr_symbol': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000077 'p_sign_posn': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000078 'thousands_sep': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000079 'mon_thousands_sep': '',
80 'frac_digits': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000081 'mon_decimal_point': '',
82 'int_frac_digits': 127}
Fredrik Lundh6c86b992000-07-09 17:12:58 +000083
Marc-André Lemburg23481142000-06-08 17:49:41 +000084 def setlocale(category, value=None):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000085 """ setlocale(integer,string=None) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000086 Activates/queries locale processing.
87 """
Martin v. Löwis103d6e72003-03-30 15:42:13 +000088 if value not in (None, '', 'C'):
Collin Winterce36ad82007-08-30 01:19:48 +000089 raise Error('_locale emulation only supports "C" locale')
Marc-André Lemburg23481142000-06-08 17:49:41 +000090 return 'C'
91
Neal Norwitz48b98de2008-03-10 04:49:25 +000092# These may or may not exist in _locale, so be sure to set them.
93if 'strxfrm' not in globals():
94 strxfrm = _strxfrm
95if 'strcoll' not in globals():
96 strcoll = _strcoll
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000097
Antoine Pitrou83d6a872008-07-25 21:45:08 +000098
99_localeconv = localeconv
100
101# With this dict, you can override some items of localeconv's return value.
102# This is useful for testing purposes.
103_override_localeconv = {}
104
105@functools.wraps(_localeconv)
106def localeconv():
107 d = _localeconv()
108 if _override_localeconv:
109 d.update(_override_localeconv)
110 return d
111
112
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000113### Number formatting APIs
114
115# Author: Martin von Loewis
Thomas Wouters477c8d52006-05-27 19:21:47 +0000116# improved by Georg Brandl
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000117
Antoine Pitrou350370c2009-03-14 00:13:13 +0000118# Iterate over grouping intervals
119def _grouping_intervals(grouping):
120 for interval in grouping:
121 # if grouping is -1, we are done
122 if interval == CHAR_MAX:
123 return
124 # 0: re-use last group ad infinitum
125 if interval == 0:
126 while True:
127 yield last_interval
128 yield interval
129 last_interval = interval
130
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000131#perform the grouping from right to left
Thomas Wouters477c8d52006-05-27 19:21:47 +0000132def _group(s, monetary=False):
133 conv = localeconv()
134 thousands_sep = conv[monetary and 'mon_thousands_sep' or 'thousands_sep']
135 grouping = conv[monetary and 'mon_grouping' or 'grouping']
136 if not grouping:
137 return (s, 0)
138 result = ""
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000139 seps = 0
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000140 if s[-1] == ' ':
Antoine Pitrou350370c2009-03-14 00:13:13 +0000141 stripped = s.rstrip()
142 right_spaces = s[len(stripped):]
143 s = stripped
144 else:
145 right_spaces = ''
146 left_spaces = ''
147 groups = []
148 for interval in _grouping_intervals(grouping):
149 if not s or s[-1] not in "0123456789":
150 # only non-digit characters remain (sign, spaces)
151 left_spaces = s
152 s = ''
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000153 break
Antoine Pitrou350370c2009-03-14 00:13:13 +0000154 groups.append(s[-interval:])
155 s = s[:-interval]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000156 if s:
Antoine Pitrou350370c2009-03-14 00:13:13 +0000157 groups.append(s)
158 groups.reverse()
159 return (
160 left_spaces + thousands_sep.join(groups) + right_spaces,
Antoine Pitrou6cf17aa2009-03-18 20:26:42 +0000161 len(thousands_sep) * (len(groups) - 1)
Antoine Pitrou350370c2009-03-14 00:13:13 +0000162 )
163
164# Strip a given amount of excess padding from the given string
165def _strip_padding(s, amount):
166 lpos = 0
167 while amount and s[lpos] == ' ':
168 lpos += 1
169 amount -= 1
170 rpos = len(s) - 1
171 while amount and s[rpos] == ' ':
172 rpos -= 1
173 amount -= 1
174 return s[lpos:rpos+1]
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000175
Thomas Wouters477c8d52006-05-27 19:21:47 +0000176def format(percent, value, grouping=False, monetary=False, *additional):
177 """Returns the locale-aware substitution of a %? specifier
178 (percent).
179
180 additional is for format strings which contain one or more
181 '*' modifiers."""
182 # this is only for one-percent-specifier strings and this should be checked
183 if percent[0] != '%':
184 raise ValueError("format() must be given exactly one %char "
185 "format specifier")
186 if additional:
187 formatted = percent % ((value,) + additional)
188 else:
189 formatted = percent % value
190 # floats and decimal ints need special action!
191 if percent[-1] in 'eEfFgG':
192 seps = 0
193 parts = formatted.split('.')
194 if grouping:
195 parts[0], seps = _group(parts[0], monetary=monetary)
196 decimal_point = localeconv()[monetary and 'mon_decimal_point'
197 or 'decimal_point']
198 formatted = decimal_point.join(parts)
Antoine Pitrou350370c2009-03-14 00:13:13 +0000199 if seps:
200 formatted = _strip_padding(formatted, seps)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000201 elif percent[-1] in 'diu':
Antoine Pitrou350370c2009-03-14 00:13:13 +0000202 seps = 0
Thomas Wouters477c8d52006-05-27 19:21:47 +0000203 if grouping:
Antoine Pitrou350370c2009-03-14 00:13:13 +0000204 formatted, seps = _group(formatted, monetary=monetary)
205 if seps:
206 formatted = _strip_padding(formatted, seps)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000207 return formatted
208
Raymond Hettinger89e12962009-01-26 02:09:03 +0000209import re, collections
Thomas Wouters477c8d52006-05-27 19:21:47 +0000210_percent_re = re.compile(r'%(?:\((?P<key>.*?)\))?'
211 r'(?P<modifiers>[-#0-9 +*.hlL]*?)[eEfFgGdiouxXcrs%]')
212
213def format_string(f, val, grouping=False):
214 """Formats a string in the same way that the % formatting would use,
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000215 but takes the current locale into account.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000216 Grouping is applied if the third parameter is true."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000217 percents = list(_percent_re.finditer(f))
218 new_f = _percent_re.sub('%s', f)
219
220 if isinstance(val, tuple):
221 new_val = list(val)
222 i = 0
223 for perc in percents:
224 starcount = perc.group('modifiers').count('*')
225 new_val[i] = format(perc.group(), new_val[i], grouping, False, *new_val[i+1:i+1+starcount])
226 del new_val[i+1:i+1+starcount]
227 i += (1 + starcount)
228 val = tuple(new_val)
Raymond Hettinger89e12962009-01-26 02:09:03 +0000229 elif isinstance(val, collections.Mapping):
Thomas Wouters477c8d52006-05-27 19:21:47 +0000230 for perc in percents:
231 key = perc.group("key")
232 val[key] = format(perc.group(), val[key], grouping)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000233 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000234 # val is a single value
235 val = format(percents[0].group(), val, grouping)
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000236
Thomas Wouters477c8d52006-05-27 19:21:47 +0000237 return new_f % val
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000238
Thomas Wouters477c8d52006-05-27 19:21:47 +0000239def currency(val, symbol=True, grouping=False, international=False):
240 """Formats val according to the currency settings
241 in the current locale."""
242 conv = localeconv()
243
244 # check for illegal values
245 digits = conv[international and 'int_frac_digits' or 'frac_digits']
246 if digits == 127:
247 raise ValueError("Currency formatting is not possible using "
248 "the 'C' locale.")
249
250 s = format('%%.%if' % digits, abs(val), grouping, monetary=True)
251 # '<' and '>' are markers if the sign must be inserted between symbol and value
252 s = '<' + s + '>'
253
254 if symbol:
255 smb = conv[international and 'int_curr_symbol' or 'currency_symbol']
256 precedes = conv[val<0 and 'n_cs_precedes' or 'p_cs_precedes']
257 separated = conv[val<0 and 'n_sep_by_space' or 'p_sep_by_space']
258
259 if precedes:
260 s = smb + (separated and ' ' or '') + s
261 else:
262 s = s + (separated and ' ' or '') + smb
263
264 sign_pos = conv[val<0 and 'n_sign_posn' or 'p_sign_posn']
265 sign = conv[val<0 and 'negative_sign' or 'positive_sign']
266
267 if sign_pos == 0:
268 s = '(' + s + ')'
269 elif sign_pos == 1:
270 s = sign + s
271 elif sign_pos == 2:
272 s = s + sign
273 elif sign_pos == 3:
274 s = s.replace('<', sign)
275 elif sign_pos == 4:
276 s = s.replace('>', sign)
277 else:
278 # the default if nothing specified;
279 # this should be the most fitting sign position
280 s = sign + s
281
282 return s.replace('<', '').replace('>', '')
Martin v. Löwisdb786872001-01-21 18:52:33 +0000283
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000284def str(val):
285 """Convert float to integer, taking the locale into account."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000286 return format("%.12g", val)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000287
Thomas Wouters477c8d52006-05-27 19:21:47 +0000288def atof(string, func=float):
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000289 "Parses a string as a float according to the locale settings."
290 #First, get rid of the grouping
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000291 ts = localeconv()['thousands_sep']
292 if ts:
Skip Montanaro249369c2004-04-10 16:39:32 +0000293 string = string.replace(ts, '')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000294 #next, replace the decimal point with a dot
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000295 dd = localeconv()['decimal_point']
296 if dd:
Skip Montanaro249369c2004-04-10 16:39:32 +0000297 string = string.replace(dd, '.')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000298 #finally, parse the string
Skip Montanaro249369c2004-04-10 16:39:32 +0000299 return func(string)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000300
301def atoi(str):
302 "Converts a string to an integer according to the locale settings."
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000303 return atof(str, int)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000304
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000305def _test():
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000306 setlocale(LC_ALL, "")
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000307 #do grouping
Thomas Wouters477c8d52006-05-27 19:21:47 +0000308 s1 = format("%d", 123456789,1)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000309 print(s1, "is", atoi(s1))
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000310 #standard formatting
Thomas Wouters477c8d52006-05-27 19:21:47 +0000311 s1 = str(3.14)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000312 print(s1, "is", atof(s1))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000313
314### Locale name aliasing engine
315
316# Author: Marc-Andre Lemburg, mal@lemburg.com
Fredrik Lundh37a09822002-10-19 20:19:10 +0000317# Various tweaks by Fredrik Lundh <fredrik@pythonware.com>
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000318
319# store away the low-level version of setlocale (it's
320# overridden below)
321_setlocale = setlocale
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000322
323def normalize(localename):
324
325 """ Returns a normalized locale code for the given locale
326 name.
327
328 The returned locale code is formatted for use with
329 setlocale().
330
331 If normalization fails, the original name is returned
332 unchanged.
333
334 If the given encoding is not known, the function defaults to
335 the default encoding for the locale code just like setlocale()
336 does.
337
338 """
339 # Normalize the locale name and extract the encoding
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000340 fullname = localename.lower()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000341 if ':' in fullname:
342 # ':' is sometimes used as encoding delimiter.
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000343 fullname = fullname.replace(':', '.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000344 if '.' in fullname:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000345 langname, encoding = fullname.split('.')[:2]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000346 fullname = langname + '.' + encoding
347 else:
348 langname = fullname
349 encoding = ''
350
351 # First lookup: fullname (possibly with encoding)
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000352 norm_encoding = encoding.replace('-', '')
353 norm_encoding = norm_encoding.replace('_', '')
354 lookup_name = langname + '.' + encoding
355 code = locale_alias.get(lookup_name, None)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000356 if code is not None:
357 return code
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000358 #print 'first lookup failed'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000359
360 # Second try: langname (without encoding)
361 code = locale_alias.get(langname, None)
362 if code is not None:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000363 #print 'langname lookup succeeded'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000364 if '.' in code:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000365 langname, defenc = code.split('.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000366 else:
367 langname = code
368 defenc = ''
369 if encoding:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000370 # Convert the encoding to a C lib compatible encoding string
371 norm_encoding = encodings.normalize_encoding(encoding)
372 #print 'norm encoding: %r' % norm_encoding
373 norm_encoding = encodings.aliases.aliases.get(norm_encoding,
374 norm_encoding)
375 #print 'aliased encoding: %r' % norm_encoding
376 encoding = locale_encoding_alias.get(norm_encoding,
377 norm_encoding)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000378 else:
379 encoding = defenc
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000380 #print 'found encoding %r' % encoding
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000381 if encoding:
382 return langname + '.' + encoding
383 else:
384 return langname
385
386 else:
387 return localename
388
389def _parse_localename(localename):
390
391 """ Parses the locale code for localename and returns the
392 result as tuple (language code, encoding).
393
394 The localename is normalized and passed through the locale
395 alias engine. A ValueError is raised in case the locale name
396 cannot be parsed.
397
398 The language code corresponds to RFC 1766. code and encoding
399 can be None in case the values cannot be determined or are
Jeremy Hyltona05e2932000-06-28 14:48:01 +0000400 unknown to this implementation.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000401
402 """
403 code = normalize(localename)
Georg Brandlb709c2c2006-01-20 09:07:35 +0000404 if '@' in code:
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000405 # Deal with locale modifiers
406 code, modifier = code.split('@')
407 if modifier == 'euro' and '.' not in code:
408 # Assume Latin-9 for @euro locales. This is bogus,
409 # since some systems may use other encodings for these
410 # locales. Also, we ignore other modifiers.
411 return code, 'iso-8859-15'
Tim Peters230a60c2002-11-09 05:08:07 +0000412
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000413 if '.' in code:
Raymond Hettinger346e67f2005-01-01 06:10:26 +0000414 return tuple(code.split('.')[:2])
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000415 elif code == 'C':
416 return None, None
Collin Winterce36ad82007-08-30 01:19:48 +0000417 raise ValueError('unknown locale: %s' % localename)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000418
419def _build_localename(localetuple):
420
421 """ Builds a locale code from the given tuple (language code,
422 encoding).
423
424 No aliasing or normalizing takes place.
425
426 """
427 language, encoding = localetuple
428 if language is None:
429 language = 'C'
430 if encoding is None:
431 return language
432 else:
433 return language + '.' + encoding
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000434
Matthias Klosef3f231f2005-09-20 07:02:49 +0000435def getdefaultlocale(envvars=('LC_ALL', 'LC_CTYPE', 'LANG', 'LANGUAGE')):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000436
437 """ Tries to determine the default locale settings and returns
438 them as tuple (language code, encoding).
439
440 According to POSIX, a program which has not called
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000441 setlocale(LC_ALL, "") runs using the portable 'C' locale.
442 Calling setlocale(LC_ALL, "") lets it use the default locale as
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000443 defined by the LANG variable. Since we don't want to interfere
Thomas Wouters7e474022000-07-16 12:04:32 +0000444 with the current locale setting we thus emulate the behavior
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000445 in the way described above.
446
447 To maintain compatibility with other platforms, not only the
448 LANG variable is tested, but a list of variables given as
449 envvars parameter. The first found to be defined will be
450 used. envvars defaults to the search path used in GNU gettext;
451 it must always contain the variable name 'LANG'.
452
453 Except for the code 'C', the language code corresponds to RFC
454 1766. code and encoding can be None in case the values cannot
455 be determined.
456
457 """
Fredrik Lundh04661322000-07-09 23:16:10 +0000458
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000459 try:
460 # check if it's supported by the _locale module
461 import _locale
462 code, encoding = _locale._getdefaultlocale()
Fredrik Lundh04661322000-07-09 23:16:10 +0000463 except (ImportError, AttributeError):
464 pass
465 else:
Fredrik Lundh663809e2000-07-10 19:32:19 +0000466 # make sure the code/encoding values are valid
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000467 if sys.platform == "win32" and code and code[:2] == "0x":
468 # map windows language identifier to language name
469 code = windows_locale.get(int(code, 0))
Fredrik Lundh663809e2000-07-10 19:32:19 +0000470 # ...add other platform-specific processing here, if
471 # necessary...
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000472 return code, encoding
Fredrik Lundh04661322000-07-09 23:16:10 +0000473
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000474 # fall back on POSIX behaviour
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000475 import os
476 lookup = os.environ.get
477 for variable in envvars:
478 localename = lookup(variable,None)
Martin v. Löwisc8ae31d2004-07-26 12:45:18 +0000479 if localename:
Matthias Klosef3f231f2005-09-20 07:02:49 +0000480 if variable == 'LANGUAGE':
481 localename = localename.split(':')[0]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000482 break
483 else:
484 localename = 'C'
485 return _parse_localename(localename)
486
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000487
488def getlocale(category=LC_CTYPE):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000489
490 """ Returns the current setting for the given locale category as
491 tuple (language code, encoding).
492
493 category may be one of the LC_* value except LC_ALL. It
494 defaults to LC_CTYPE.
495
496 Except for the code 'C', the language code corresponds to RFC
497 1766. code and encoding can be None in case the values cannot
498 be determined.
499
500 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000501 localename = _setlocale(category)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000502 if category == LC_ALL and ';' in localename:
Collin Winterce36ad82007-08-30 01:19:48 +0000503 raise TypeError('category LC_ALL is not supported')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000504 return _parse_localename(localename)
505
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000506def setlocale(category, locale=None):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000507
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000508 """ Set the locale for the given category. The locale can be
509 a string, a locale tuple (language code, encoding), or None.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000510
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000511 Locale tuples are converted to strings the locale aliasing
512 engine. Locale strings are passed directly to the C lib.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000513
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000514 category may be given as one of the LC_* values.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000515
516 """
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000517 if locale and not isinstance(locale, _builtin_str):
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000518 # convert to string
519 locale = normalize(_build_localename(locale))
520 return _setlocale(category, locale)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000521
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000522def resetlocale(category=LC_ALL):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000523
524 """ Sets the locale for category to the default setting.
525
526 The default setting is determined by calling
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000527 getdefaultlocale(). category defaults to LC_ALL.
528
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000529 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000530 _setlocale(category, _build_localename(getdefaultlocale()))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000531
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000532if sys.platform in ('win32', 'darwin', 'mac'):
533 # On Win32, this will return the ANSI code page
534 # On the Mac, it should return the system encoding;
535 # it might return "ascii" instead
536 def getpreferredencoding(do_setlocale = True):
537 """Return the charset that the user is likely using."""
538 import _locale
Tim Petersa326f472002-11-05 03:49:09 +0000539 return _locale._getdefaultlocale()[1]
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000540else:
541 # On Unix, if CODESET is available, use that.
542 try:
543 CODESET
544 except NameError:
545 # Fall back to parsing environment variables :-(
546 def getpreferredencoding(do_setlocale = True):
547 """Return the charset that the user is likely using,
548 by looking at environment variables."""
Martin v. Löwis071ef772008-03-08 11:24:24 +0000549 res = getdefaultlocale()[1]
550 if res is None:
551 # LANG not set, default conservatively to ASCII
552 res = 'ascii'
553 return res
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000554 else:
555 def getpreferredencoding(do_setlocale = True):
556 """Return the charset that the user is likely using,
557 according to the system configuration."""
558 if do_setlocale:
559 oldloc = setlocale(LC_CTYPE)
560 setlocale(LC_CTYPE, "")
561 result = nl_langinfo(CODESET)
562 setlocale(LC_CTYPE, oldloc)
563 return result
564 else:
565 return nl_langinfo(CODESET)
Tim Peters230a60c2002-11-09 05:08:07 +0000566
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000567
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000568### Database
569#
570# The following data was extracted from the locale.alias file which
571# comes with X11 and then hand edited removing the explicit encoding
572# definitions and adding some more aliases. The file is usually
573# available as /usr/lib/X11/locale/locale.alias.
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000574#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000575
576#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000577# The local_encoding_alias table maps lowercase encoding alias names
578# to C locale encoding names (case-sensitive). Note that normalize()
579# first looks up the encoding in the encodings.aliases dictionary and
580# then applies this mapping to find the correct C lib name for the
581# encoding.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000582#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000583locale_encoding_alias = {
584
585 # Mappings for non-standard encoding names used in locale names
586 '437': 'C',
587 'c': 'C',
588 'en': 'ISO8859-1',
589 'jis': 'JIS7',
590 'jis7': 'JIS7',
591 'ajec': 'eucJP',
592
593 # Mappings from Python codec names to C lib encoding names
594 'ascii': 'ISO8859-1',
595 'latin_1': 'ISO8859-1',
596 'iso8859_1': 'ISO8859-1',
597 'iso8859_10': 'ISO8859-10',
598 'iso8859_11': 'ISO8859-11',
599 'iso8859_13': 'ISO8859-13',
600 'iso8859_14': 'ISO8859-14',
601 'iso8859_15': 'ISO8859-15',
602 'iso8859_2': 'ISO8859-2',
603 'iso8859_3': 'ISO8859-3',
604 'iso8859_4': 'ISO8859-4',
605 'iso8859_5': 'ISO8859-5',
606 'iso8859_6': 'ISO8859-6',
607 'iso8859_7': 'ISO8859-7',
608 'iso8859_8': 'ISO8859-8',
609 'iso8859_9': 'ISO8859-9',
610 'iso2022_jp': 'JIS7',
611 'shift_jis': 'SJIS',
612 'tactis': 'TACTIS',
613 'euc_jp': 'eucJP',
614 'euc_kr': 'eucKR',
Marc-André Lemburgb4cebd42004-12-13 19:56:01 +0000615 'utf_8': 'UTF8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000616 'koi8_r': 'KOI8-R',
617 'koi8_u': 'KOI8-U',
618 # XXX This list is still incomplete. If you know more
619 # mappings, please file a bug report. Thanks.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000620}
621
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000622#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000623# The locale_alias table maps lowercase alias names to C locale names
624# (case-sensitive). Encodings are always separated from the locale
625# name using a dot ('.'); they should only be given in case the
626# language name is needed to interpret the given encoding alias
627# correctly (CJK codes often have this need).
628#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000629# Note that the normalize() function which uses this tables
630# removes '_' and '-' characters from the encoding part of the
631# locale name before doing the lookup. This saves a lot of
632# space in the table.
633#
634# MAL 2004-12-10:
635# Updated alias mapping to most recent locale.alias file
636# from X.org distribution using makelocalealias.py.
637#
638# These are the differences compared to the old mapping (Python 2.4
639# and older):
640#
641# updated 'bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
642# updated 'bg_bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
643# updated 'bulgarian' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
644# updated 'cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
645# updated 'cz_cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
646# updated 'czech' -> 'cs_CS.ISO8859-2' to 'cs_CZ.ISO8859-2'
647# updated 'dutch' -> 'nl_BE.ISO8859-1' to 'nl_NL.ISO8859-1'
648# updated 'et' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
649# updated 'et_ee' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
650# updated 'fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
651# updated 'fi_fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
652# updated 'iw' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
653# updated 'iw_il' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
654# updated 'japanese' -> 'ja_JP.SJIS' to 'ja_JP.eucJP'
655# updated 'lt' -> 'lt_LT.ISO8859-4' to 'lt_LT.ISO8859-13'
656# updated 'lv' -> 'lv_LV.ISO8859-4' to 'lv_LV.ISO8859-13'
657# updated 'sl' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
658# updated 'slovene' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
659# updated 'th_th' -> 'th_TH.TACTIS' to 'th_TH.ISO8859-11'
660# updated 'zh_cn' -> 'zh_CN.eucCN' to 'zh_CN.gb2312'
661# updated 'zh_cn.big5' -> 'zh_TW.eucTW' to 'zh_TW.big5'
662# updated 'zh_tw' -> 'zh_TW.eucTW' to 'zh_TW.big5'
663#
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000664# MAL 2008-05-30:
665# Updated alias mapping to most recent locale.alias file
666# from X.org distribution using makelocalealias.py.
667#
668# These are the differences compared to the old mapping (Python 2.5
669# and older):
670#
671# updated 'cs_cs.iso88592' -> 'cs_CZ.ISO8859-2' to 'cs_CS.ISO8859-2'
672# updated 'serbocroatian' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
673# updated 'sh' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
674# updated 'sh_hr.iso88592' -> 'sh_HR.ISO8859-2' to 'hr_HR.ISO8859-2'
675# updated 'sh_sp' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
676# updated 'sh_yu' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
677# updated 'sp' -> 'sp_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
678# updated 'sp_yu' -> 'sp_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
679# updated 'sr' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
680# updated 'sr@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
681# updated 'sr_sp' -> 'sr_SP.ISO8859-2' to 'sr_CS.ISO8859-2'
682# updated 'sr_yu' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
683# updated 'sr_yu.cp1251@cyrillic' -> 'sr_YU.CP1251' to 'sr_CS.CP1251'
684# updated 'sr_yu.iso88592' -> 'sr_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
685# updated 'sr_yu.iso88595' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
686# updated 'sr_yu.iso88595@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
687# updated 'sr_yu.microsoftcp1251@cyrillic' -> 'sr_YU.CP1251' to 'sr_CS.CP1251'
688# updated 'sr_yu.utf8@cyrillic' -> 'sr_YU.UTF-8' to 'sr_CS.UTF-8'
689# updated 'sr_yu@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
690
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000691locale_alias = {
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000692 'a3': 'a3_AZ.KOI8-C',
693 'a3_az': 'a3_AZ.KOI8-C',
694 'a3_az.koi8c': 'a3_AZ.KOI8-C',
695 'af': 'af_ZA.ISO8859-1',
696 'af_za': 'af_ZA.ISO8859-1',
697 'af_za.iso88591': 'af_ZA.ISO8859-1',
698 'am': 'am_ET.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000699 'am_et': 'am_ET.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000700 'american': 'en_US.ISO8859-1',
701 'american.iso88591': 'en_US.ISO8859-1',
702 'ar': 'ar_AA.ISO8859-6',
703 'ar_aa': 'ar_AA.ISO8859-6',
704 'ar_aa.iso88596': 'ar_AA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000705 'ar_ae': 'ar_AE.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000706 'ar_ae.iso88596': 'ar_AE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000707 'ar_bh': 'ar_BH.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000708 'ar_bh.iso88596': 'ar_BH.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000709 'ar_dz': 'ar_DZ.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000710 'ar_dz.iso88596': 'ar_DZ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000711 'ar_eg': 'ar_EG.ISO8859-6',
712 'ar_eg.iso88596': 'ar_EG.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000713 'ar_iq': 'ar_IQ.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000714 'ar_iq.iso88596': 'ar_IQ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000715 'ar_jo': 'ar_JO.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000716 'ar_jo.iso88596': 'ar_JO.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000717 'ar_kw': 'ar_KW.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000718 'ar_kw.iso88596': 'ar_KW.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000719 'ar_lb': 'ar_LB.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000720 'ar_lb.iso88596': 'ar_LB.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000721 'ar_ly': 'ar_LY.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000722 'ar_ly.iso88596': 'ar_LY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000723 'ar_ma': 'ar_MA.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000724 'ar_ma.iso88596': 'ar_MA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000725 'ar_om': 'ar_OM.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000726 'ar_om.iso88596': 'ar_OM.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000727 'ar_qa': 'ar_QA.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000728 'ar_qa.iso88596': 'ar_QA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000729 'ar_sa': 'ar_SA.ISO8859-6',
730 'ar_sa.iso88596': 'ar_SA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000731 'ar_sd': 'ar_SD.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000732 'ar_sd.iso88596': 'ar_SD.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000733 'ar_sy': 'ar_SY.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000734 'ar_sy.iso88596': 'ar_SY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000735 'ar_tn': 'ar_TN.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000736 'ar_tn.iso88596': 'ar_TN.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000737 'ar_ye': 'ar_YE.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000738 'ar_ye.iso88596': 'ar_YE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000739 'arabic': 'ar_AA.ISO8859-6',
740 'arabic.iso88596': 'ar_AA.ISO8859-6',
741 'az': 'az_AZ.ISO8859-9E',
742 'az_az': 'az_AZ.ISO8859-9E',
743 'az_az.iso88599e': 'az_AZ.ISO8859-9E',
744 'be': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000745 'be_by': 'be_BY.CP1251',
746 'be_by.cp1251': 'be_BY.CP1251',
747 'be_by.microsoftcp1251': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000748 'bg': 'bg_BG.CP1251',
749 'bg_bg': 'bg_BG.CP1251',
750 'bg_bg.cp1251': 'bg_BG.CP1251',
751 'bg_bg.iso88595': 'bg_BG.ISO8859-5',
752 'bg_bg.koi8r': 'bg_BG.KOI8-R',
753 'bg_bg.microsoftcp1251': 'bg_BG.CP1251',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000754 'bn_in': 'bn_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000755 'bokmal': 'nb_NO.ISO8859-1',
756 'bokm\xe5l': 'nb_NO.ISO8859-1',
757 'br': 'br_FR.ISO8859-1',
758 'br_fr': 'br_FR.ISO8859-1',
759 'br_fr.iso88591': 'br_FR.ISO8859-1',
760 'br_fr.iso885914': 'br_FR.ISO8859-14',
761 'br_fr.iso885915': 'br_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000762 'br_fr.iso885915@euro': 'br_FR.ISO8859-15',
763 'br_fr.utf8@euro': 'br_FR.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000764 'br_fr@euro': 'br_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000765 'bs': 'bs_BA.ISO8859-2',
766 'bs_ba': 'bs_BA.ISO8859-2',
767 'bs_ba.iso88592': 'bs_BA.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000768 'bulgarian': 'bg_BG.CP1251',
769 'c': 'C',
770 'c-french': 'fr_CA.ISO8859-1',
771 'c-french.iso88591': 'fr_CA.ISO8859-1',
772 'c.en': 'C',
773 'c.iso88591': 'en_US.ISO8859-1',
774 'c_c': 'C',
775 'c_c.c': 'C',
776 'ca': 'ca_ES.ISO8859-1',
777 'ca_es': 'ca_ES.ISO8859-1',
778 'ca_es.iso88591': 'ca_ES.ISO8859-1',
779 'ca_es.iso885915': 'ca_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000780 'ca_es.iso885915@euro': 'ca_ES.ISO8859-15',
781 'ca_es.utf8@euro': 'ca_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000782 'ca_es@euro': 'ca_ES.ISO8859-15',
783 'catalan': 'ca_ES.ISO8859-1',
784 'cextend': 'en_US.ISO8859-1',
785 'cextend.en': 'en_US.ISO8859-1',
786 'chinese-s': 'zh_CN.eucCN',
787 'chinese-t': 'zh_TW.eucTW',
788 'croatian': 'hr_HR.ISO8859-2',
789 'cs': 'cs_CZ.ISO8859-2',
790 'cs_cs': 'cs_CZ.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000791 'cs_cs.iso88592': 'cs_CS.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000792 'cs_cz': 'cs_CZ.ISO8859-2',
793 'cs_cz.iso88592': 'cs_CZ.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000794 'cy': 'cy_GB.ISO8859-1',
795 'cy_gb': 'cy_GB.ISO8859-1',
796 'cy_gb.iso88591': 'cy_GB.ISO8859-1',
797 'cy_gb.iso885914': 'cy_GB.ISO8859-14',
798 'cy_gb.iso885915': 'cy_GB.ISO8859-15',
799 'cy_gb@euro': 'cy_GB.ISO8859-15',
800 'cz': 'cs_CZ.ISO8859-2',
801 'cz_cz': 'cs_CZ.ISO8859-2',
802 'czech': 'cs_CZ.ISO8859-2',
803 'da': 'da_DK.ISO8859-1',
804 'da_dk': 'da_DK.ISO8859-1',
805 'da_dk.88591': 'da_DK.ISO8859-1',
806 'da_dk.885915': 'da_DK.ISO8859-15',
807 'da_dk.iso88591': 'da_DK.ISO8859-1',
808 'da_dk.iso885915': 'da_DK.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000809 'da_dk@euro': 'da_DK.ISO8859-15',
810 'danish': 'da_DK.ISO8859-1',
811 'danish.iso88591': 'da_DK.ISO8859-1',
812 'dansk': 'da_DK.ISO8859-1',
813 'de': 'de_DE.ISO8859-1',
814 'de_at': 'de_AT.ISO8859-1',
815 'de_at.iso88591': 'de_AT.ISO8859-1',
816 'de_at.iso885915': 'de_AT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000817 'de_at.iso885915@euro': 'de_AT.ISO8859-15',
818 'de_at.utf8@euro': 'de_AT.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000819 'de_at@euro': 'de_AT.ISO8859-15',
820 'de_be': 'de_BE.ISO8859-1',
821 'de_be.iso88591': 'de_BE.ISO8859-1',
822 'de_be.iso885915': 'de_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000823 'de_be.iso885915@euro': 'de_BE.ISO8859-15',
824 'de_be.utf8@euro': 'de_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000825 'de_be@euro': 'de_BE.ISO8859-15',
826 'de_ch': 'de_CH.ISO8859-1',
827 'de_ch.iso88591': 'de_CH.ISO8859-1',
828 'de_ch.iso885915': 'de_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000829 'de_ch@euro': 'de_CH.ISO8859-15',
830 'de_de': 'de_DE.ISO8859-1',
831 'de_de.88591': 'de_DE.ISO8859-1',
832 'de_de.885915': 'de_DE.ISO8859-15',
833 'de_de.885915@euro': 'de_DE.ISO8859-15',
834 'de_de.iso88591': 'de_DE.ISO8859-1',
835 'de_de.iso885915': 'de_DE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000836 'de_de.iso885915@euro': 'de_DE.ISO8859-15',
837 'de_de.utf8@euro': 'de_DE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000838 'de_de@euro': 'de_DE.ISO8859-15',
839 'de_lu': 'de_LU.ISO8859-1',
840 'de_lu.iso88591': 'de_LU.ISO8859-1',
841 'de_lu.iso885915': 'de_LU.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000842 'de_lu.iso885915@euro': 'de_LU.ISO8859-15',
843 'de_lu.utf8@euro': 'de_LU.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000844 'de_lu@euro': 'de_LU.ISO8859-15',
845 'deutsch': 'de_DE.ISO8859-1',
846 'dutch': 'nl_NL.ISO8859-1',
847 'dutch.iso88591': 'nl_BE.ISO8859-1',
848 'ee': 'ee_EE.ISO8859-4',
849 'ee_ee': 'ee_EE.ISO8859-4',
850 'ee_ee.iso88594': 'ee_EE.ISO8859-4',
851 'eesti': 'et_EE.ISO8859-1',
852 'el': 'el_GR.ISO8859-7',
853 'el_gr': 'el_GR.ISO8859-7',
854 'el_gr.iso88597': 'el_GR.ISO8859-7',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000855 'el_gr@euro': 'el_GR.ISO8859-15',
856 'en': 'en_US.ISO8859-1',
857 'en.iso88591': 'en_US.ISO8859-1',
858 'en_au': 'en_AU.ISO8859-1',
859 'en_au.iso88591': 'en_AU.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000860 'en_be': 'en_BE.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000861 'en_be@euro': 'en_BE.ISO8859-15',
862 'en_bw': 'en_BW.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000863 'en_bw.iso88591': 'en_BW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000864 'en_ca': 'en_CA.ISO8859-1',
865 'en_ca.iso88591': 'en_CA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000866 'en_gb': 'en_GB.ISO8859-1',
867 'en_gb.88591': 'en_GB.ISO8859-1',
868 'en_gb.iso88591': 'en_GB.ISO8859-1',
869 'en_gb.iso885915': 'en_GB.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000870 'en_gb@euro': 'en_GB.ISO8859-15',
871 'en_hk': 'en_HK.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000872 'en_hk.iso88591': 'en_HK.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000873 'en_ie': 'en_IE.ISO8859-1',
874 'en_ie.iso88591': 'en_IE.ISO8859-1',
875 'en_ie.iso885915': 'en_IE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000876 'en_ie.iso885915@euro': 'en_IE.ISO8859-15',
877 'en_ie.utf8@euro': 'en_IE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000878 'en_ie@euro': 'en_IE.ISO8859-15',
879 'en_in': 'en_IN.ISO8859-1',
880 'en_nz': 'en_NZ.ISO8859-1',
881 'en_nz.iso88591': 'en_NZ.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000882 'en_ph': 'en_PH.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000883 'en_ph.iso88591': 'en_PH.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000884 'en_sg': 'en_SG.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000885 'en_sg.iso88591': 'en_SG.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000886 'en_uk': 'en_GB.ISO8859-1',
887 'en_us': 'en_US.ISO8859-1',
888 'en_us.88591': 'en_US.ISO8859-1',
889 'en_us.885915': 'en_US.ISO8859-15',
890 'en_us.iso88591': 'en_US.ISO8859-1',
891 'en_us.iso885915': 'en_US.ISO8859-15',
892 'en_us.iso885915@euro': 'en_US.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000893 'en_us@euro': 'en_US.ISO8859-15',
894 'en_us@euro@euro': 'en_US.ISO8859-15',
895 'en_za': 'en_ZA.ISO8859-1',
896 'en_za.88591': 'en_ZA.ISO8859-1',
897 'en_za.iso88591': 'en_ZA.ISO8859-1',
898 'en_za.iso885915': 'en_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000899 'en_za@euro': 'en_ZA.ISO8859-15',
900 'en_zw': 'en_ZW.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000901 'en_zw.iso88591': 'en_ZW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000902 'eng_gb': 'en_GB.ISO8859-1',
903 'eng_gb.8859': 'en_GB.ISO8859-1',
904 'english': 'en_EN.ISO8859-1',
905 'english.iso88591': 'en_EN.ISO8859-1',
906 'english_uk': 'en_GB.ISO8859-1',
907 'english_uk.8859': 'en_GB.ISO8859-1',
908 'english_united-states': 'en_US.ISO8859-1',
909 'english_united-states.437': 'C',
910 'english_us': 'en_US.ISO8859-1',
911 'english_us.8859': 'en_US.ISO8859-1',
912 'english_us.ascii': 'en_US.ISO8859-1',
913 'eo': 'eo_XX.ISO8859-3',
914 'eo_eo': 'eo_EO.ISO8859-3',
915 'eo_eo.iso88593': 'eo_EO.ISO8859-3',
916 'eo_xx': 'eo_XX.ISO8859-3',
917 'eo_xx.iso88593': 'eo_XX.ISO8859-3',
918 'es': 'es_ES.ISO8859-1',
919 'es_ar': 'es_AR.ISO8859-1',
920 'es_ar.iso88591': 'es_AR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000921 'es_bo': 'es_BO.ISO8859-1',
922 'es_bo.iso88591': 'es_BO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000923 'es_cl': 'es_CL.ISO8859-1',
924 'es_cl.iso88591': 'es_CL.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000925 'es_co': 'es_CO.ISO8859-1',
926 'es_co.iso88591': 'es_CO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000927 'es_cr': 'es_CR.ISO8859-1',
928 'es_cr.iso88591': 'es_CR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000929 'es_do': 'es_DO.ISO8859-1',
930 'es_do.iso88591': 'es_DO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000931 'es_ec': 'es_EC.ISO8859-1',
932 'es_ec.iso88591': 'es_EC.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000933 'es_es': 'es_ES.ISO8859-1',
934 'es_es.88591': 'es_ES.ISO8859-1',
935 'es_es.iso88591': 'es_ES.ISO8859-1',
936 'es_es.iso885915': 'es_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000937 'es_es.iso885915@euro': 'es_ES.ISO8859-15',
938 'es_es.utf8@euro': 'es_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000939 'es_es@euro': 'es_ES.ISO8859-15',
940 'es_gt': 'es_GT.ISO8859-1',
941 'es_gt.iso88591': 'es_GT.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000942 'es_hn': 'es_HN.ISO8859-1',
943 'es_hn.iso88591': 'es_HN.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000944 'es_mx': 'es_MX.ISO8859-1',
945 'es_mx.iso88591': 'es_MX.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000946 'es_ni': 'es_NI.ISO8859-1',
947 'es_ni.iso88591': 'es_NI.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000948 'es_pa': 'es_PA.ISO8859-1',
949 'es_pa.iso88591': 'es_PA.ISO8859-1',
950 'es_pa.iso885915': 'es_PA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000951 'es_pa@euro': 'es_PA.ISO8859-15',
952 'es_pe': 'es_PE.ISO8859-1',
953 'es_pe.iso88591': 'es_PE.ISO8859-1',
954 'es_pe.iso885915': 'es_PE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000955 'es_pe@euro': 'es_PE.ISO8859-15',
956 'es_pr': 'es_PR.ISO8859-1',
957 'es_pr.iso88591': 'es_PR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000958 'es_py': 'es_PY.ISO8859-1',
959 'es_py.iso88591': 'es_PY.ISO8859-1',
960 'es_py.iso885915': 'es_PY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000961 'es_py@euro': 'es_PY.ISO8859-15',
962 'es_sv': 'es_SV.ISO8859-1',
963 'es_sv.iso88591': 'es_SV.ISO8859-1',
964 'es_sv.iso885915': 'es_SV.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000965 'es_sv@euro': 'es_SV.ISO8859-15',
966 'es_us': 'es_US.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000967 'es_us.iso88591': 'es_US.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000968 'es_uy': 'es_UY.ISO8859-1',
969 'es_uy.iso88591': 'es_UY.ISO8859-1',
970 'es_uy.iso885915': 'es_UY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000971 'es_uy@euro': 'es_UY.ISO8859-15',
972 'es_ve': 'es_VE.ISO8859-1',
973 'es_ve.iso88591': 'es_VE.ISO8859-1',
974 'es_ve.iso885915': 'es_VE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000975 'es_ve@euro': 'es_VE.ISO8859-15',
976 'estonian': 'et_EE.ISO8859-1',
977 'et': 'et_EE.ISO8859-15',
978 'et_ee': 'et_EE.ISO8859-15',
979 'et_ee.iso88591': 'et_EE.ISO8859-1',
980 'et_ee.iso885913': 'et_EE.ISO8859-13',
981 'et_ee.iso885915': 'et_EE.ISO8859-15',
982 'et_ee.iso88594': 'et_EE.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000983 'et_ee@euro': 'et_EE.ISO8859-15',
984 'eu': 'eu_ES.ISO8859-1',
985 'eu_es': 'eu_ES.ISO8859-1',
986 'eu_es.iso88591': 'eu_ES.ISO8859-1',
987 'eu_es.iso885915': 'eu_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000988 'eu_es.iso885915@euro': 'eu_ES.ISO8859-15',
989 'eu_es.utf8@euro': 'eu_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000990 'eu_es@euro': 'eu_ES.ISO8859-15',
991 'fa': 'fa_IR.UTF-8',
992 'fa_ir': 'fa_IR.UTF-8',
993 'fa_ir.isiri3342': 'fa_IR.ISIRI-3342',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000994 'fi': 'fi_FI.ISO8859-15',
995 'fi_fi': 'fi_FI.ISO8859-15',
996 'fi_fi.88591': 'fi_FI.ISO8859-1',
997 'fi_fi.iso88591': 'fi_FI.ISO8859-1',
998 'fi_fi.iso885915': 'fi_FI.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000999 'fi_fi.iso885915@euro': 'fi_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001000 'fi_fi.utf8@euro': 'fi_FI.UTF-8',
1001 'fi_fi@euro': 'fi_FI.ISO8859-15',
1002 'finnish': 'fi_FI.ISO8859-1',
1003 'finnish.iso88591': 'fi_FI.ISO8859-1',
1004 'fo': 'fo_FO.ISO8859-1',
1005 'fo_fo': 'fo_FO.ISO8859-1',
1006 'fo_fo.iso88591': 'fo_FO.ISO8859-1',
1007 'fo_fo.iso885915': 'fo_FO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001008 'fo_fo@euro': 'fo_FO.ISO8859-15',
1009 'fr': 'fr_FR.ISO8859-1',
1010 'fr_be': 'fr_BE.ISO8859-1',
1011 'fr_be.88591': 'fr_BE.ISO8859-1',
1012 'fr_be.iso88591': 'fr_BE.ISO8859-1',
1013 'fr_be.iso885915': 'fr_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001014 'fr_be.iso885915@euro': 'fr_BE.ISO8859-15',
1015 'fr_be.utf8@euro': 'fr_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001016 'fr_be@euro': 'fr_BE.ISO8859-15',
1017 'fr_ca': 'fr_CA.ISO8859-1',
1018 'fr_ca.88591': 'fr_CA.ISO8859-1',
1019 'fr_ca.iso88591': 'fr_CA.ISO8859-1',
1020 'fr_ca.iso885915': 'fr_CA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001021 'fr_ca@euro': 'fr_CA.ISO8859-15',
1022 'fr_ch': 'fr_CH.ISO8859-1',
1023 'fr_ch.88591': 'fr_CH.ISO8859-1',
1024 'fr_ch.iso88591': 'fr_CH.ISO8859-1',
1025 'fr_ch.iso885915': 'fr_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001026 'fr_ch@euro': 'fr_CH.ISO8859-15',
1027 'fr_fr': 'fr_FR.ISO8859-1',
1028 'fr_fr.88591': 'fr_FR.ISO8859-1',
1029 'fr_fr.iso88591': 'fr_FR.ISO8859-1',
1030 'fr_fr.iso885915': 'fr_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001031 'fr_fr.iso885915@euro': 'fr_FR.ISO8859-15',
1032 'fr_fr.utf8@euro': 'fr_FR.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001033 'fr_fr@euro': 'fr_FR.ISO8859-15',
1034 'fr_lu': 'fr_LU.ISO8859-1',
1035 'fr_lu.88591': 'fr_LU.ISO8859-1',
1036 'fr_lu.iso88591': 'fr_LU.ISO8859-1',
1037 'fr_lu.iso885915': 'fr_LU.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001038 'fr_lu.iso885915@euro': 'fr_LU.ISO8859-15',
1039 'fr_lu.utf8@euro': 'fr_LU.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001040 'fr_lu@euro': 'fr_LU.ISO8859-15',
1041 'fran\xe7ais': 'fr_FR.ISO8859-1',
1042 'fre_fr': 'fr_FR.ISO8859-1',
1043 'fre_fr.8859': 'fr_FR.ISO8859-1',
1044 'french': 'fr_FR.ISO8859-1',
1045 'french.iso88591': 'fr_CH.ISO8859-1',
1046 'french_france': 'fr_FR.ISO8859-1',
1047 'french_france.8859': 'fr_FR.ISO8859-1',
1048 'ga': 'ga_IE.ISO8859-1',
1049 'ga_ie': 'ga_IE.ISO8859-1',
1050 'ga_ie.iso88591': 'ga_IE.ISO8859-1',
1051 'ga_ie.iso885914': 'ga_IE.ISO8859-14',
1052 'ga_ie.iso885915': 'ga_IE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001053 'ga_ie.iso885915@euro': 'ga_IE.ISO8859-15',
1054 'ga_ie.utf8@euro': 'ga_IE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001055 'ga_ie@euro': 'ga_IE.ISO8859-15',
1056 'galego': 'gl_ES.ISO8859-1',
1057 'galician': 'gl_ES.ISO8859-1',
1058 'gd': 'gd_GB.ISO8859-1',
1059 'gd_gb': 'gd_GB.ISO8859-1',
1060 'gd_gb.iso88591': 'gd_GB.ISO8859-1',
1061 'gd_gb.iso885914': 'gd_GB.ISO8859-14',
1062 'gd_gb.iso885915': 'gd_GB.ISO8859-15',
1063 'gd_gb@euro': 'gd_GB.ISO8859-15',
1064 'ger_de': 'de_DE.ISO8859-1',
1065 'ger_de.8859': 'de_DE.ISO8859-1',
1066 'german': 'de_DE.ISO8859-1',
1067 'german.iso88591': 'de_CH.ISO8859-1',
1068 'german_germany': 'de_DE.ISO8859-1',
1069 'german_germany.8859': 'de_DE.ISO8859-1',
1070 'gl': 'gl_ES.ISO8859-1',
1071 'gl_es': 'gl_ES.ISO8859-1',
1072 'gl_es.iso88591': 'gl_ES.ISO8859-1',
1073 'gl_es.iso885915': 'gl_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001074 'gl_es.iso885915@euro': 'gl_ES.ISO8859-15',
1075 'gl_es.utf8@euro': 'gl_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001076 'gl_es@euro': 'gl_ES.ISO8859-15',
1077 'greek': 'el_GR.ISO8859-7',
1078 'greek.iso88597': 'el_GR.ISO8859-7',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001079 'gu_in': 'gu_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001080 'gv': 'gv_GB.ISO8859-1',
1081 'gv_gb': 'gv_GB.ISO8859-1',
1082 'gv_gb.iso88591': 'gv_GB.ISO8859-1',
1083 'gv_gb.iso885914': 'gv_GB.ISO8859-14',
1084 'gv_gb.iso885915': 'gv_GB.ISO8859-15',
1085 'gv_gb@euro': 'gv_GB.ISO8859-15',
1086 'he': 'he_IL.ISO8859-8',
1087 'he_il': 'he_IL.ISO8859-8',
1088 'he_il.cp1255': 'he_IL.CP1255',
1089 'he_il.iso88598': 'he_IL.ISO8859-8',
1090 'he_il.microsoftcp1255': 'he_IL.CP1255',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001091 'hebrew': 'iw_IL.ISO8859-8',
1092 'hebrew.iso88598': 'iw_IL.ISO8859-8',
1093 'hi': 'hi_IN.ISCII-DEV',
1094 'hi_in': 'hi_IN.ISCII-DEV',
1095 'hi_in.isciidev': 'hi_IN.ISCII-DEV',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001096 'hr': 'hr_HR.ISO8859-2',
1097 'hr_hr': 'hr_HR.ISO8859-2',
1098 'hr_hr.iso88592': 'hr_HR.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001099 'hrvatski': 'hr_HR.ISO8859-2',
1100 'hu': 'hu_HU.ISO8859-2',
1101 'hu_hu': 'hu_HU.ISO8859-2',
1102 'hu_hu.iso88592': 'hu_HU.ISO8859-2',
1103 'hungarian': 'hu_HU.ISO8859-2',
1104 'icelandic': 'is_IS.ISO8859-1',
1105 'icelandic.iso88591': 'is_IS.ISO8859-1',
1106 'id': 'id_ID.ISO8859-1',
1107 'id_id': 'id_ID.ISO8859-1',
1108 'in': 'id_ID.ISO8859-1',
1109 'in_id': 'id_ID.ISO8859-1',
1110 'is': 'is_IS.ISO8859-1',
1111 'is_is': 'is_IS.ISO8859-1',
1112 'is_is.iso88591': 'is_IS.ISO8859-1',
1113 'is_is.iso885915': 'is_IS.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001114 'is_is@euro': 'is_IS.ISO8859-15',
1115 'iso-8859-1': 'en_US.ISO8859-1',
1116 'iso-8859-15': 'en_US.ISO8859-15',
1117 'iso8859-1': 'en_US.ISO8859-1',
1118 'iso8859-15': 'en_US.ISO8859-15',
1119 'iso_8859_1': 'en_US.ISO8859-1',
1120 'iso_8859_15': 'en_US.ISO8859-15',
1121 'it': 'it_IT.ISO8859-1',
1122 'it_ch': 'it_CH.ISO8859-1',
1123 'it_ch.iso88591': 'it_CH.ISO8859-1',
1124 'it_ch.iso885915': 'it_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001125 'it_ch@euro': 'it_CH.ISO8859-15',
1126 'it_it': 'it_IT.ISO8859-1',
1127 'it_it.88591': 'it_IT.ISO8859-1',
1128 'it_it.iso88591': 'it_IT.ISO8859-1',
1129 'it_it.iso885915': 'it_IT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001130 'it_it.iso885915@euro': 'it_IT.ISO8859-15',
1131 'it_it.utf8@euro': 'it_IT.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001132 'it_it@euro': 'it_IT.ISO8859-15',
1133 'italian': 'it_IT.ISO8859-1',
1134 'italian.iso88591': 'it_IT.ISO8859-1',
1135 'iu': 'iu_CA.NUNACOM-8',
1136 'iu_ca': 'iu_CA.NUNACOM-8',
1137 'iu_ca.nunacom8': 'iu_CA.NUNACOM-8',
1138 'iw': 'he_IL.ISO8859-8',
1139 'iw_il': 'he_IL.ISO8859-8',
1140 'iw_il.iso88598': 'he_IL.ISO8859-8',
1141 'ja': 'ja_JP.eucJP',
1142 'ja.jis': 'ja_JP.JIS7',
1143 'ja.sjis': 'ja_JP.SJIS',
1144 'ja_jp': 'ja_JP.eucJP',
1145 'ja_jp.ajec': 'ja_JP.eucJP',
1146 'ja_jp.euc': 'ja_JP.eucJP',
1147 'ja_jp.eucjp': 'ja_JP.eucJP',
1148 'ja_jp.iso-2022-jp': 'ja_JP.JIS7',
1149 'ja_jp.iso2022jp': 'ja_JP.JIS7',
1150 'ja_jp.jis': 'ja_JP.JIS7',
1151 'ja_jp.jis7': 'ja_JP.JIS7',
1152 'ja_jp.mscode': 'ja_JP.SJIS',
1153 'ja_jp.sjis': 'ja_JP.SJIS',
1154 'ja_jp.ujis': 'ja_JP.eucJP',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001155 'japan': 'ja_JP.eucJP',
1156 'japanese': 'ja_JP.eucJP',
1157 'japanese-euc': 'ja_JP.eucJP',
1158 'japanese.euc': 'ja_JP.eucJP',
1159 'japanese.sjis': 'ja_JP.SJIS',
1160 'jp_jp': 'ja_JP.eucJP',
1161 'ka': 'ka_GE.GEORGIAN-ACADEMY',
1162 'ka_ge': 'ka_GE.GEORGIAN-ACADEMY',
1163 'ka_ge.georgianacademy': 'ka_GE.GEORGIAN-ACADEMY',
1164 'ka_ge.georgianps': 'ka_GE.GEORGIAN-PS',
1165 'ka_ge.georgianrs': 'ka_GE.GEORGIAN-ACADEMY',
1166 'kl': 'kl_GL.ISO8859-1',
1167 'kl_gl': 'kl_GL.ISO8859-1',
1168 'kl_gl.iso88591': 'kl_GL.ISO8859-1',
1169 'kl_gl.iso885915': 'kl_GL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001170 'kl_gl@euro': 'kl_GL.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001171 'km_kh': 'km_KH.UTF-8',
1172 'kn_in': 'kn_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001173 'ko': 'ko_KR.eucKR',
1174 'ko_kr': 'ko_KR.eucKR',
1175 'ko_kr.euc': 'ko_KR.eucKR',
1176 'ko_kr.euckr': 'ko_KR.eucKR',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001177 'korean': 'ko_KR.eucKR',
1178 'korean.euc': 'ko_KR.eucKR',
1179 'kw': 'kw_GB.ISO8859-1',
1180 'kw_gb': 'kw_GB.ISO8859-1',
1181 'kw_gb.iso88591': 'kw_GB.ISO8859-1',
1182 'kw_gb.iso885914': 'kw_GB.ISO8859-14',
1183 'kw_gb.iso885915': 'kw_GB.ISO8859-15',
1184 'kw_gb@euro': 'kw_GB.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001185 'ky': 'ky_KG.UTF-8',
1186 'ky_kg': 'ky_KG.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001187 'lithuanian': 'lt_LT.ISO8859-13',
1188 'lo': 'lo_LA.MULELAO-1',
1189 'lo_la': 'lo_LA.MULELAO-1',
1190 'lo_la.cp1133': 'lo_LA.IBM-CP1133',
1191 'lo_la.ibmcp1133': 'lo_LA.IBM-CP1133',
1192 'lo_la.mulelao1': 'lo_LA.MULELAO-1',
1193 'lt': 'lt_LT.ISO8859-13',
1194 'lt_lt': 'lt_LT.ISO8859-13',
1195 'lt_lt.iso885913': 'lt_LT.ISO8859-13',
1196 'lt_lt.iso88594': 'lt_LT.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001197 'lv': 'lv_LV.ISO8859-13',
1198 'lv_lv': 'lv_LV.ISO8859-13',
1199 'lv_lv.iso885913': 'lv_LV.ISO8859-13',
1200 'lv_lv.iso88594': 'lv_LV.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001201 'mi': 'mi_NZ.ISO8859-1',
1202 'mi_nz': 'mi_NZ.ISO8859-1',
1203 'mi_nz.iso88591': 'mi_NZ.ISO8859-1',
1204 'mk': 'mk_MK.ISO8859-5',
1205 'mk_mk': 'mk_MK.ISO8859-5',
1206 'mk_mk.cp1251': 'mk_MK.CP1251',
1207 'mk_mk.iso88595': 'mk_MK.ISO8859-5',
1208 'mk_mk.microsoftcp1251': 'mk_MK.CP1251',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001209 'mr_in': 'mr_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001210 'ms': 'ms_MY.ISO8859-1',
1211 'ms_my': 'ms_MY.ISO8859-1',
1212 'ms_my.iso88591': 'ms_MY.ISO8859-1',
1213 'mt': 'mt_MT.ISO8859-3',
1214 'mt_mt': 'mt_MT.ISO8859-3',
1215 'mt_mt.iso88593': 'mt_MT.ISO8859-3',
1216 'nb': 'nb_NO.ISO8859-1',
1217 'nb_no': 'nb_NO.ISO8859-1',
1218 'nb_no.88591': 'nb_NO.ISO8859-1',
1219 'nb_no.iso88591': 'nb_NO.ISO8859-1',
1220 'nb_no.iso885915': 'nb_NO.ISO8859-15',
1221 'nb_no@euro': 'nb_NO.ISO8859-15',
1222 'nl': 'nl_NL.ISO8859-1',
1223 'nl_be': 'nl_BE.ISO8859-1',
1224 'nl_be.88591': 'nl_BE.ISO8859-1',
1225 'nl_be.iso88591': 'nl_BE.ISO8859-1',
1226 'nl_be.iso885915': 'nl_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001227 'nl_be.iso885915@euro': 'nl_BE.ISO8859-15',
1228 'nl_be.utf8@euro': 'nl_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001229 'nl_be@euro': 'nl_BE.ISO8859-15',
1230 'nl_nl': 'nl_NL.ISO8859-1',
1231 'nl_nl.88591': 'nl_NL.ISO8859-1',
1232 'nl_nl.iso88591': 'nl_NL.ISO8859-1',
1233 'nl_nl.iso885915': 'nl_NL.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001234 'nl_nl.iso885915@euro': 'nl_NL.ISO8859-15',
1235 'nl_nl.utf8@euro': 'nl_NL.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001236 'nl_nl@euro': 'nl_NL.ISO8859-15',
1237 'nn': 'nn_NO.ISO8859-1',
1238 'nn_no': 'nn_NO.ISO8859-1',
1239 'nn_no.88591': 'nn_NO.ISO8859-1',
1240 'nn_no.iso88591': 'nn_NO.ISO8859-1',
1241 'nn_no.iso885915': 'nn_NO.ISO8859-15',
1242 'nn_no@euro': 'nn_NO.ISO8859-15',
1243 'no': 'no_NO.ISO8859-1',
1244 'no@nynorsk': 'ny_NO.ISO8859-1',
1245 'no_no': 'no_NO.ISO8859-1',
1246 'no_no.88591': 'no_NO.ISO8859-1',
1247 'no_no.iso88591': 'no_NO.ISO8859-1',
1248 'no_no.iso885915': 'no_NO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001249 'no_no@euro': 'no_NO.ISO8859-15',
1250 'norwegian': 'no_NO.ISO8859-1',
1251 'norwegian.iso88591': 'no_NO.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001252 'nr': 'nr_ZA.ISO8859-1',
1253 'nr_za': 'nr_ZA.ISO8859-1',
1254 'nr_za.iso88591': 'nr_ZA.ISO8859-1',
1255 'nso': 'nso_ZA.ISO8859-15',
1256 'nso_za': 'nso_ZA.ISO8859-15',
1257 'nso_za.iso885915': 'nso_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001258 'ny': 'ny_NO.ISO8859-1',
1259 'ny_no': 'ny_NO.ISO8859-1',
1260 'ny_no.88591': 'ny_NO.ISO8859-1',
1261 'ny_no.iso88591': 'ny_NO.ISO8859-1',
1262 'ny_no.iso885915': 'ny_NO.ISO8859-15',
1263 'ny_no@euro': 'ny_NO.ISO8859-15',
1264 'nynorsk': 'nn_NO.ISO8859-1',
1265 'oc': 'oc_FR.ISO8859-1',
1266 'oc_fr': 'oc_FR.ISO8859-1',
1267 'oc_fr.iso88591': 'oc_FR.ISO8859-1',
1268 'oc_fr.iso885915': 'oc_FR.ISO8859-15',
1269 'oc_fr@euro': 'oc_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001270 'pa_in': 'pa_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001271 'pd': 'pd_US.ISO8859-1',
1272 'pd_de': 'pd_DE.ISO8859-1',
1273 'pd_de.iso88591': 'pd_DE.ISO8859-1',
1274 'pd_de.iso885915': 'pd_DE.ISO8859-15',
1275 'pd_de@euro': 'pd_DE.ISO8859-15',
1276 'pd_us': 'pd_US.ISO8859-1',
1277 'pd_us.iso88591': 'pd_US.ISO8859-1',
1278 'pd_us.iso885915': 'pd_US.ISO8859-15',
1279 'pd_us@euro': 'pd_US.ISO8859-15',
1280 'ph': 'ph_PH.ISO8859-1',
1281 'ph_ph': 'ph_PH.ISO8859-1',
1282 'ph_ph.iso88591': 'ph_PH.ISO8859-1',
1283 'pl': 'pl_PL.ISO8859-2',
1284 'pl_pl': 'pl_PL.ISO8859-2',
1285 'pl_pl.iso88592': 'pl_PL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001286 'polish': 'pl_PL.ISO8859-2',
1287 'portuguese': 'pt_PT.ISO8859-1',
1288 'portuguese.iso88591': 'pt_PT.ISO8859-1',
1289 'portuguese_brazil': 'pt_BR.ISO8859-1',
1290 'portuguese_brazil.8859': 'pt_BR.ISO8859-1',
1291 'posix': 'C',
1292 'posix-utf2': 'C',
1293 'pp': 'pp_AN.ISO8859-1',
1294 'pp_an': 'pp_AN.ISO8859-1',
1295 'pp_an.iso88591': 'pp_AN.ISO8859-1',
1296 'pt': 'pt_PT.ISO8859-1',
1297 'pt_br': 'pt_BR.ISO8859-1',
1298 'pt_br.88591': 'pt_BR.ISO8859-1',
1299 'pt_br.iso88591': 'pt_BR.ISO8859-1',
1300 'pt_br.iso885915': 'pt_BR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001301 'pt_br@euro': 'pt_BR.ISO8859-15',
1302 'pt_pt': 'pt_PT.ISO8859-1',
1303 'pt_pt.88591': 'pt_PT.ISO8859-1',
1304 'pt_pt.iso88591': 'pt_PT.ISO8859-1',
1305 'pt_pt.iso885915': 'pt_PT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001306 'pt_pt.iso885915@euro': 'pt_PT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001307 'pt_pt.utf8@euro': 'pt_PT.UTF-8',
1308 'pt_pt@euro': 'pt_PT.ISO8859-15',
1309 'ro': 'ro_RO.ISO8859-2',
1310 'ro_ro': 'ro_RO.ISO8859-2',
1311 'ro_ro.iso88592': 'ro_RO.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001312 'romanian': 'ro_RO.ISO8859-2',
1313 'ru': 'ru_RU.ISO8859-5',
1314 'ru_ru': 'ru_RU.ISO8859-5',
1315 'ru_ru.cp1251': 'ru_RU.CP1251',
1316 'ru_ru.iso88595': 'ru_RU.ISO8859-5',
1317 'ru_ru.koi8r': 'ru_RU.KOI8-R',
1318 'ru_ru.microsoftcp1251': 'ru_RU.CP1251',
1319 'ru_ua': 'ru_UA.KOI8-U',
1320 'ru_ua.cp1251': 'ru_UA.CP1251',
1321 'ru_ua.koi8u': 'ru_UA.KOI8-U',
1322 'ru_ua.microsoftcp1251': 'ru_UA.CP1251',
1323 'rumanian': 'ro_RO.ISO8859-2',
1324 'russian': 'ru_RU.ISO8859-5',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001325 'rw': 'rw_RW.ISO8859-1',
1326 'rw_rw': 'rw_RW.ISO8859-1',
1327 'rw_rw.iso88591': 'rw_RW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001328 'se_no': 'se_NO.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001329 'serbocroatian': 'sr_CS.ISO8859-2',
1330 'sh': 'sr_CS.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001331 'sh_hr': 'sh_HR.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001332 'sh_hr.iso88592': 'hr_HR.ISO8859-2',
1333 'sh_sp': 'sr_CS.ISO8859-2',
1334 'sh_yu': 'sr_CS.ISO8859-2',
1335 'si': 'si_LK.UTF-8',
1336 'si_lk': 'si_LK.UTF-8',
1337 'sinhala': 'si_LK.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001338 'sk': 'sk_SK.ISO8859-2',
1339 'sk_sk': 'sk_SK.ISO8859-2',
1340 'sk_sk.iso88592': 'sk_SK.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001341 'sl': 'sl_SI.ISO8859-2',
1342 'sl_cs': 'sl_CS.ISO8859-2',
1343 'sl_si': 'sl_SI.ISO8859-2',
1344 'sl_si.iso88592': 'sl_SI.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001345 'slovak': 'sk_SK.ISO8859-2',
1346 'slovene': 'sl_SI.ISO8859-2',
1347 'slovenian': 'sl_SI.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001348 'sp': 'sr_CS.ISO8859-5',
1349 'sp_yu': 'sr_CS.ISO8859-5',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001350 'spanish': 'es_ES.ISO8859-1',
1351 'spanish.iso88591': 'es_ES.ISO8859-1',
1352 'spanish_spain': 'es_ES.ISO8859-1',
1353 'spanish_spain.8859': 'es_ES.ISO8859-1',
1354 'sq': 'sq_AL.ISO8859-2',
1355 'sq_al': 'sq_AL.ISO8859-2',
1356 'sq_al.iso88592': 'sq_AL.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001357 'sr': 'sr_CS.ISO8859-5',
1358 'sr@cyrillic': 'sr_CS.ISO8859-5',
1359 'sr@latn': 'sr_CS.ISO8859-2',
1360 'sr_cs.iso88592': 'sr_CS.ISO8859-2',
1361 'sr_cs.iso88592@latn': 'sr_CS.ISO8859-2',
1362 'sr_cs.iso88595': 'sr_CS.ISO8859-5',
1363 'sr_cs.utf8@latn': 'sr_CS.UTF-8',
1364 'sr_cs@latn': 'sr_CS.ISO8859-2',
1365 'sr_sp': 'sr_CS.ISO8859-2',
1366 'sr_yu': 'sr_CS.ISO8859-5',
1367 'sr_yu.cp1251@cyrillic': 'sr_CS.CP1251',
1368 'sr_yu.iso88592': 'sr_CS.ISO8859-2',
1369 'sr_yu.iso88595': 'sr_CS.ISO8859-5',
1370 'sr_yu.iso88595@cyrillic': 'sr_CS.ISO8859-5',
1371 'sr_yu.microsoftcp1251@cyrillic': 'sr_CS.CP1251',
1372 'sr_yu.utf8@cyrillic': 'sr_CS.UTF-8',
1373 'sr_yu@cyrillic': 'sr_CS.ISO8859-5',
1374 'ss': 'ss_ZA.ISO8859-1',
1375 'ss_za': 'ss_ZA.ISO8859-1',
1376 'ss_za.iso88591': 'ss_ZA.ISO8859-1',
1377 'st': 'st_ZA.ISO8859-1',
1378 'st_za': 'st_ZA.ISO8859-1',
1379 'st_za.iso88591': 'st_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001380 'sv': 'sv_SE.ISO8859-1',
1381 'sv_fi': 'sv_FI.ISO8859-1',
1382 'sv_fi.iso88591': 'sv_FI.ISO8859-1',
1383 'sv_fi.iso885915': 'sv_FI.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001384 'sv_fi.iso885915@euro': 'sv_FI.ISO8859-15',
1385 'sv_fi.utf8@euro': 'sv_FI.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001386 'sv_fi@euro': 'sv_FI.ISO8859-15',
1387 'sv_se': 'sv_SE.ISO8859-1',
1388 'sv_se.88591': 'sv_SE.ISO8859-1',
1389 'sv_se.iso88591': 'sv_SE.ISO8859-1',
1390 'sv_se.iso885915': 'sv_SE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001391 'sv_se@euro': 'sv_SE.ISO8859-15',
1392 'swedish': 'sv_SE.ISO8859-1',
1393 'swedish.iso88591': 'sv_SE.ISO8859-1',
1394 'ta': 'ta_IN.TSCII-0',
1395 'ta_in': 'ta_IN.TSCII-0',
1396 'ta_in.tscii': 'ta_IN.TSCII-0',
1397 'ta_in.tscii0': 'ta_IN.TSCII-0',
1398 'tg': 'tg_TJ.KOI8-C',
1399 'tg_tj': 'tg_TJ.KOI8-C',
1400 'tg_tj.koi8c': 'tg_TJ.KOI8-C',
1401 'th': 'th_TH.ISO8859-11',
1402 'th_th': 'th_TH.ISO8859-11',
1403 'th_th.iso885911': 'th_TH.ISO8859-11',
1404 'th_th.tactis': 'th_TH.TIS620',
1405 'th_th.tis620': 'th_TH.TIS620',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001406 'thai': 'th_TH.ISO8859-11',
1407 'tl': 'tl_PH.ISO8859-1',
1408 'tl_ph': 'tl_PH.ISO8859-1',
1409 'tl_ph.iso88591': 'tl_PH.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001410 'tn': 'tn_ZA.ISO8859-15',
1411 'tn_za': 'tn_ZA.ISO8859-15',
1412 'tn_za.iso885915': 'tn_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001413 'tr': 'tr_TR.ISO8859-9',
1414 'tr_tr': 'tr_TR.ISO8859-9',
1415 'tr_tr.iso88599': 'tr_TR.ISO8859-9',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001416 'ts': 'ts_ZA.ISO8859-1',
1417 'ts_za': 'ts_ZA.ISO8859-1',
1418 'ts_za.iso88591': 'ts_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001419 'tt': 'tt_RU.TATAR-CYR',
1420 'tt_ru': 'tt_RU.TATAR-CYR',
1421 'tt_ru.koi8c': 'tt_RU.KOI8-C',
1422 'tt_ru.tatarcyr': 'tt_RU.TATAR-CYR',
1423 'turkish': 'tr_TR.ISO8859-9',
1424 'turkish.iso88599': 'tr_TR.ISO8859-9',
1425 'uk': 'uk_UA.KOI8-U',
1426 'uk_ua': 'uk_UA.KOI8-U',
1427 'uk_ua.cp1251': 'uk_UA.CP1251',
1428 'uk_ua.iso88595': 'uk_UA.ISO8859-5',
1429 'uk_ua.koi8u': 'uk_UA.KOI8-U',
1430 'uk_ua.microsoftcp1251': 'uk_UA.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001431 'univ': 'en_US.utf',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001432 'universal': 'en_US.utf',
1433 'universal.utf8@ucs4': 'en_US.UTF-8',
1434 'ur': 'ur_PK.CP1256',
1435 'ur_pk': 'ur_PK.CP1256',
1436 'ur_pk.cp1256': 'ur_PK.CP1256',
1437 'ur_pk.microsoftcp1256': 'ur_PK.CP1256',
1438 'uz': 'uz_UZ.UTF-8',
1439 'uz_uz': 'uz_UZ.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001440 'uz_uz.iso88591': 'uz_UZ.ISO8859-1',
1441 'uz_uz.utf8@cyrillic': 'uz_UZ.UTF-8',
1442 'uz_uz@cyrillic': 'uz_UZ.UTF-8',
1443 've': 've_ZA.UTF-8',
1444 've_za': 've_ZA.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001445 'vi': 'vi_VN.TCVN',
1446 'vi_vn': 'vi_VN.TCVN',
1447 'vi_vn.tcvn': 'vi_VN.TCVN',
1448 'vi_vn.tcvn5712': 'vi_VN.TCVN',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001449 'vi_vn.viscii': 'vi_VN.VISCII',
1450 'vi_vn.viscii111': 'vi_VN.VISCII',
1451 'wa': 'wa_BE.ISO8859-1',
1452 'wa_be': 'wa_BE.ISO8859-1',
1453 'wa_be.iso88591': 'wa_BE.ISO8859-1',
1454 'wa_be.iso885915': 'wa_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001455 'wa_be.iso885915@euro': 'wa_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001456 'wa_be@euro': 'wa_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001457 'xh': 'xh_ZA.ISO8859-1',
1458 'xh_za': 'xh_ZA.ISO8859-1',
1459 'xh_za.iso88591': 'xh_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001460 'yi': 'yi_US.CP1255',
1461 'yi_us': 'yi_US.CP1255',
1462 'yi_us.cp1255': 'yi_US.CP1255',
1463 'yi_us.microsoftcp1255': 'yi_US.CP1255',
1464 'zh': 'zh_CN.eucCN',
1465 'zh_cn': 'zh_CN.gb2312',
1466 'zh_cn.big5': 'zh_TW.big5',
1467 'zh_cn.euc': 'zh_CN.eucCN',
1468 'zh_cn.gb18030': 'zh_CN.gb18030',
1469 'zh_cn.gb2312': 'zh_CN.gb2312',
1470 'zh_cn.gbk': 'zh_CN.gbk',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001471 'zh_hk': 'zh_HK.big5hkscs',
1472 'zh_hk.big5': 'zh_HK.big5',
1473 'zh_hk.big5hkscs': 'zh_HK.big5hkscs',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001474 'zh_tw': 'zh_TW.big5',
1475 'zh_tw.big5': 'zh_TW.big5',
1476 'zh_tw.euc': 'zh_TW.eucTW',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001477 'zh_tw.euctw': 'zh_TW.eucTW',
1478 'zu': 'zu_ZA.ISO8859-1',
1479 'zu_za': 'zu_ZA.ISO8859-1',
1480 'zu_za.iso88591': 'zu_ZA.ISO8859-1',
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001481}
1482
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001483#
Georg Brandlb709c2c2006-01-20 09:07:35 +00001484# This maps Windows language identifiers to locale strings.
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001485#
Tim Peters777f1082006-01-20 20:03:24 +00001486# This list has been updated from
Georg Brandlb709c2c2006-01-20 09:07:35 +00001487# http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/nls_238z.asp
1488# to include every locale up to Windows XP.
Fredrik Lundh37a09822002-10-19 20:19:10 +00001489#
Georg Brandl5035c1c2006-01-20 13:38:26 +00001490# NOTE: this mapping is incomplete. If your language is missing, please
1491# submit a bug report to Python bug manager, which you can find via:
1492# http://www.python.org/dev/
1493# Make sure you include the missing language identifier and the suggested
1494# locale code.
1495#
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001496
1497windows_locale = {
Georg Brandlb709c2c2006-01-20 09:07:35 +00001498 0x0436: "af_ZA", # Afrikaans
1499 0x041c: "sq_AL", # Albanian
1500 0x0401: "ar_SA", # Arabic - Saudi Arabia
1501 0x0801: "ar_IQ", # Arabic - Iraq
1502 0x0c01: "ar_EG", # Arabic - Egypt
1503 0x1001: "ar_LY", # Arabic - Libya
1504 0x1401: "ar_DZ", # Arabic - Algeria
1505 0x1801: "ar_MA", # Arabic - Morocco
1506 0x1c01: "ar_TN", # Arabic - Tunisia
1507 0x2001: "ar_OM", # Arabic - Oman
1508 0x2401: "ar_YE", # Arabic - Yemen
1509 0x2801: "ar_SY", # Arabic - Syria
1510 0x2c01: "ar_JO", # Arabic - Jordan
1511 0x3001: "ar_LB", # Arabic - Lebanon
1512 0x3401: "ar_KW", # Arabic - Kuwait
1513 0x3801: "ar_AE", # Arabic - United Arab Emirates
1514 0x3c01: "ar_BH", # Arabic - Bahrain
1515 0x4001: "ar_QA", # Arabic - Qatar
1516 0x042b: "hy_AM", # Armenian
1517 0x042c: "az_AZ", # Azeri Latin
1518 0x082c: "az_AZ", # Azeri - Cyrillic
1519 0x042d: "eu_ES", # Basque
1520 0x0423: "be_BY", # Belarusian
1521 0x0445: "bn_IN", # Begali
1522 0x201a: "bs_BA", # Bosnian
1523 0x141a: "bs_BA", # Bosnian - Cyrillic
1524 0x047e: "br_FR", # Breton - France
1525 0x0402: "bg_BG", # Bulgarian
1526 0x0403: "ca_ES", # Catalan
1527 0x0004: "zh_CHS",# Chinese - Simplified
1528 0x0404: "zh_TW", # Chinese - Taiwan
1529 0x0804: "zh_CN", # Chinese - PRC
1530 0x0c04: "zh_HK", # Chinese - Hong Kong S.A.R.
1531 0x1004: "zh_SG", # Chinese - Singapore
1532 0x1404: "zh_MO", # Chinese - Macao S.A.R.
1533 0x7c04: "zh_CHT",# Chinese - Traditional
1534 0x041a: "hr_HR", # Croatian
1535 0x101a: "hr_BA", # Croatian - Bosnia
1536 0x0405: "cs_CZ", # Czech
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001537 0x0406: "da_DK", # Danish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001538 0x048c: "gbz_AF",# Dari - Afghanistan
1539 0x0465: "div_MV",# Divehi - Maldives
1540 0x0413: "nl_NL", # Dutch - The Netherlands
1541 0x0813: "nl_BE", # Dutch - Belgium
1542 0x0409: "en_US", # English - United States
1543 0x0809: "en_GB", # English - United Kingdom
1544 0x0c09: "en_AU", # English - Australia
1545 0x1009: "en_CA", # English - Canada
1546 0x1409: "en_NZ", # English - New Zealand
1547 0x1809: "en_IE", # English - Ireland
1548 0x1c09: "en_ZA", # English - South Africa
1549 0x2009: "en_JA", # English - Jamaica
1550 0x2409: "en_CB", # English - Carribbean
1551 0x2809: "en_BZ", # English - Belize
1552 0x2c09: "en_TT", # English - Trinidad
1553 0x3009: "en_ZW", # English - Zimbabwe
1554 0x3409: "en_PH", # English - Phillippines
1555 0x0425: "et_EE", # Estonian
1556 0x0438: "fo_FO", # Faroese
1557 0x0464: "fil_PH",# Filipino
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001558 0x040b: "fi_FI", # Finnish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001559 0x040c: "fr_FR", # French - France
1560 0x080c: "fr_BE", # French - Belgium
1561 0x0c0c: "fr_CA", # French - Canada
1562 0x100c: "fr_CH", # French - Switzerland
1563 0x140c: "fr_LU", # French - Luxembourg
1564 0x180c: "fr_MC", # French - Monaco
1565 0x0462: "fy_NL", # Frisian - Netherlands
1566 0x0456: "gl_ES", # Galician
1567 0x0437: "ka_GE", # Georgian
1568 0x0407: "de_DE", # German - Germany
1569 0x0807: "de_CH", # German - Switzerland
1570 0x0c07: "de_AT", # German - Austria
1571 0x1007: "de_LU", # German - Luxembourg
1572 0x1407: "de_LI", # German - Liechtenstein
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001573 0x0408: "el_GR", # Greek
Georg Brandlb709c2c2006-01-20 09:07:35 +00001574 0x0447: "gu_IN", # Gujarati
1575 0x040d: "he_IL", # Hebrew
1576 0x0439: "hi_IN", # Hindi
1577 0x040e: "hu_HU", # Hungarian
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001578 0x040f: "is_IS", # Icelandic
Georg Brandlb709c2c2006-01-20 09:07:35 +00001579 0x0421: "id_ID", # Indonesian
1580 0x045d: "iu_CA", # Inuktitut
1581 0x085d: "iu_CA", # Inuktitut - Latin
1582 0x083c: "ga_IE", # Irish - Ireland
1583 0x0434: "xh_ZA", # Xhosa - South Africa
1584 0x0435: "zu_ZA", # Zulu
1585 0x0410: "it_IT", # Italian - Italy
1586 0x0810: "it_CH", # Italian - Switzerland
1587 0x0411: "ja_JP", # Japanese
1588 0x044b: "kn_IN", # Kannada - India
1589 0x043f: "kk_KZ", # Kazakh
1590 0x0457: "kok_IN",# Konkani
1591 0x0412: "ko_KR", # Korean
1592 0x0440: "ky_KG", # Kyrgyz
1593 0x0426: "lv_LV", # Latvian
1594 0x0427: "lt_LT", # Lithuanian
1595 0x046e: "lb_LU", # Luxembourgish
1596 0x042f: "mk_MK", # FYRO Macedonian
1597 0x043e: "ms_MY", # Malay - Malaysia
1598 0x083e: "ms_BN", # Malay - Brunei
1599 0x044c: "ml_IN", # Malayalam - India
1600 0x043a: "mt_MT", # Maltese
1601 0x0481: "mi_NZ", # Maori
1602 0x047a: "arn_CL",# Mapudungun
1603 0x044e: "mr_IN", # Marathi
1604 0x047c: "moh_CA",# Mohawk - Canada
1605 0x0450: "mn_MN", # Mongolian
1606 0x0461: "ne_NP", # Nepali
1607 0x0414: "nb_NO", # Norwegian - Bokmal
1608 0x0814: "nn_NO", # Norwegian - Nynorsk
1609 0x0482: "oc_FR", # Occitan - France
1610 0x0448: "or_IN", # Oriya - India
1611 0x0463: "ps_AF", # Pashto - Afghanistan
1612 0x0429: "fa_IR", # Persian
1613 0x0415: "pl_PL", # Polish
1614 0x0416: "pt_BR", # Portuguese - Brazil
1615 0x0816: "pt_PT", # Portuguese - Portugal
1616 0x0446: "pa_IN", # Punjabi
1617 0x046b: "quz_BO",# Quechua (Bolivia)
1618 0x086b: "quz_EC",# Quechua (Ecuador)
1619 0x0c6b: "quz_PE",# Quechua (Peru)
1620 0x0418: "ro_RO", # Romanian - Romania
1621 0x0417: "rm_CH", # Raeto-Romanese
1622 0x0419: "ru_RU", # Russian
1623 0x243b: "smn_FI",# Sami Finland
1624 0x103b: "smj_NO",# Sami Norway
1625 0x143b: "smj_SE",# Sami Sweden
1626 0x043b: "se_NO", # Sami Northern Norway
1627 0x083b: "se_SE", # Sami Northern Sweden
1628 0x0c3b: "se_FI", # Sami Northern Finland
1629 0x203b: "sms_FI",# Sami Skolt
1630 0x183b: "sma_NO",# Sami Southern Norway
1631 0x1c3b: "sma_SE",# Sami Southern Sweden
1632 0x044f: "sa_IN", # Sanskrit
1633 0x0c1a: "sr_SP", # Serbian - Cyrillic
1634 0x1c1a: "sr_BA", # Serbian - Bosnia Cyrillic
1635 0x081a: "sr_SP", # Serbian - Latin
1636 0x181a: "sr_BA", # Serbian - Bosnia Latin
1637 0x046c: "ns_ZA", # Northern Sotho
1638 0x0432: "tn_ZA", # Setswana - Southern Africa
1639 0x041b: "sk_SK", # Slovak
1640 0x0424: "sl_SI", # Slovenian
1641 0x040a: "es_ES", # Spanish - Spain
1642 0x080a: "es_MX", # Spanish - Mexico
1643 0x0c0a: "es_ES", # Spanish - Spain (Modern)
1644 0x100a: "es_GT", # Spanish - Guatemala
1645 0x140a: "es_CR", # Spanish - Costa Rica
1646 0x180a: "es_PA", # Spanish - Panama
1647 0x1c0a: "es_DO", # Spanish - Dominican Republic
1648 0x200a: "es_VE", # Spanish - Venezuela
1649 0x240a: "es_CO", # Spanish - Colombia
1650 0x280a: "es_PE", # Spanish - Peru
1651 0x2c0a: "es_AR", # Spanish - Argentina
1652 0x300a: "es_EC", # Spanish - Ecuador
1653 0x340a: "es_CL", # Spanish - Chile
1654 0x380a: "es_UR", # Spanish - Uruguay
1655 0x3c0a: "es_PY", # Spanish - Paraguay
1656 0x400a: "es_BO", # Spanish - Bolivia
1657 0x440a: "es_SV", # Spanish - El Salvador
1658 0x480a: "es_HN", # Spanish - Honduras
1659 0x4c0a: "es_NI", # Spanish - Nicaragua
1660 0x500a: "es_PR", # Spanish - Puerto Rico
1661 0x0441: "sw_KE", # Swahili
1662 0x041d: "sv_SE", # Swedish - Sweden
1663 0x081d: "sv_FI", # Swedish - Finland
1664 0x045a: "syr_SY",# Syriac
1665 0x0449: "ta_IN", # Tamil
1666 0x0444: "tt_RU", # Tatar
1667 0x044a: "te_IN", # Telugu
1668 0x041e: "th_TH", # Thai
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001669 0x041f: "tr_TR", # Turkish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001670 0x0422: "uk_UA", # Ukrainian
1671 0x0420: "ur_PK", # Urdu
1672 0x0820: "ur_IN", # Urdu - India
1673 0x0443: "uz_UZ", # Uzbek - Latin
1674 0x0843: "uz_UZ", # Uzbek - Cyrillic
1675 0x042a: "vi_VN", # Vietnamese
1676 0x0452: "cy_GB", # Welsh
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001677}
1678
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001679def _print_locale():
1680
1681 """ Test function.
1682 """
1683 categories = {}
1684 def _init_categories(categories=categories):
1685 for k,v in globals().items():
1686 if k[:3] == 'LC_':
1687 categories[k] = v
1688 _init_categories()
1689 del categories['LC_ALL']
1690
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001691 print('Locale defaults as determined by getdefaultlocale():')
1692 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001693 lang, enc = getdefaultlocale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001694 print('Language: ', lang or '(undefined)')
1695 print('Encoding: ', enc or '(undefined)')
1696 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001697
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001698 print('Locale settings on startup:')
1699 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001700 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001701 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001702 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001703 print(' Language: ', lang or '(undefined)')
1704 print(' Encoding: ', enc or '(undefined)')
1705 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001706
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001707 print()
1708 print('Locale settings after calling resetlocale():')
1709 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001710 resetlocale()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001711 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001712 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001713 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001714 print(' Language: ', lang or '(undefined)')
1715 print(' Encoding: ', enc or '(undefined)')
1716 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001717
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001718 try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001719 setlocale(LC_ALL, "")
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001720 except:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001721 print('NOTE:')
1722 print('setlocale(LC_ALL, "") does not support the default locale')
1723 print('given in the OS environment variables.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001724 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001725 print()
1726 print('Locale settings after calling setlocale(LC_ALL, ""):')
1727 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001728 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001729 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001730 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001731 print(' Language: ', lang or '(undefined)')
1732 print(' Encoding: ', enc or '(undefined)')
1733 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001734
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001735###
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001736
Tim Peters1baf8292001-01-24 10:13:46 +00001737try:
1738 LC_MESSAGES
Skip Montanaro0897f0c2002-03-25 21:40:36 +00001739except NameError:
Tim Peters1baf8292001-01-24 10:13:46 +00001740 pass
1741else:
1742 __all__.append("LC_MESSAGES")
1743
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001744if __name__=='__main__':
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001745 print('Locale aliasing:')
1746 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001747 _print_locale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001748 print()
1749 print('Number formatting:')
1750 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001751 _test()