blob: 2f8cd85dfc9c7948115b12ec851c4f1c6438393b [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
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000016
Fredrik Lundh6c86b992000-07-09 17:12:58 +000017# Try importing the _locale module.
18#
19# If this fails, fall back on a basic 'C' locale emulation.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +000020
Tim Peters1baf8292001-01-24 10:13:46 +000021# Yuck: LC_MESSAGES is non-standard: can't tell whether it exists before
22# trying the import. So __all__ is also fiddled at the end of the file.
Guido van Rossum360e4b82007-05-14 22:51:27 +000023__all__ = ["getlocale", "getdefaultlocale", "getpreferredencoding", "Error",
24 "setlocale", "resetlocale", "localeconv", "strcoll", "strxfrm",
25 "str", "atof", "atoi", "format", "format_string", "currency",
26 "normalize", "LC_CTYPE", "LC_COLLATE", "LC_TIME", "LC_MONETARY",
27 "LC_NUMERIC", "LC_ALL", "CHAR_MAX"]
Skip Montanaro17ab1232001-01-24 06:27:27 +000028
Neal Norwitz48b98de2008-03-10 04:49:25 +000029def _strcoll(a,b):
30 """ strcoll(string,string) -> int.
31 Compares two strings according to the locale.
32 """
33 return cmp(a,b)
34
35def _strxfrm(s):
36 """ strxfrm(string) -> string.
37 Returns a string that behaves for cmp locale-aware.
38 """
39 return s
40
Marc-André Lemburg23481142000-06-08 17:49:41 +000041try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +000042
Marc-André Lemburg23481142000-06-08 17:49:41 +000043 from _locale import *
44
45except ImportError:
46
Fredrik Lundh6c86b992000-07-09 17:12:58 +000047 # Locale emulation
48
Marc-André Lemburg23481142000-06-08 17:49:41 +000049 CHAR_MAX = 127
50 LC_ALL = 6
51 LC_COLLATE = 3
52 LC_CTYPE = 0
53 LC_MESSAGES = 5
54 LC_MONETARY = 4
55 LC_NUMERIC = 1
56 LC_TIME = 2
57 Error = ValueError
58
59 def localeconv():
Fredrik Lundh6c86b992000-07-09 17:12:58 +000060 """ localeconv() -> dict.
Marc-André Lemburg23481142000-06-08 17:49:41 +000061 Returns numeric and monetary locale-specific parameters.
62 """
63 # 'C' locale default values
64 return {'grouping': [127],
65 'currency_symbol': '',
66 'n_sign_posn': 127,
Fredrik Lundh6c86b992000-07-09 17:12:58 +000067 'p_cs_precedes': 127,
68 'n_cs_precedes': 127,
69 'mon_grouping': [],
Marc-André Lemburg23481142000-06-08 17:49:41 +000070 'n_sep_by_space': 127,
71 'decimal_point': '.',
72 'negative_sign': '',
73 'positive_sign': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000074 'p_sep_by_space': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000075 'int_curr_symbol': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000076 'p_sign_posn': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000077 'thousands_sep': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000078 'mon_thousands_sep': '',
79 'frac_digits': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000080 'mon_decimal_point': '',
81 'int_frac_digits': 127}
Fredrik Lundh6c86b992000-07-09 17:12:58 +000082
Marc-André Lemburg23481142000-06-08 17:49:41 +000083 def setlocale(category, value=None):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000084 """ setlocale(integer,string=None) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000085 Activates/queries locale processing.
86 """
Martin v. Löwis103d6e72003-03-30 15:42:13 +000087 if value not in (None, '', 'C'):
Collin Winterce36ad82007-08-30 01:19:48 +000088 raise Error('_locale emulation only supports "C" locale')
Marc-André Lemburg23481142000-06-08 17:49:41 +000089 return 'C'
90
Neal Norwitz48b98de2008-03-10 04:49:25 +000091# These may or may not exist in _locale, so be sure to set them.
92if 'strxfrm' not in globals():
93 strxfrm = _strxfrm
94if 'strcoll' not in globals():
95 strcoll = _strcoll
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000096
97### Number formatting APIs
98
99# Author: Martin von Loewis
Thomas Wouters477c8d52006-05-27 19:21:47 +0000100# improved by Georg Brandl
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000101
102#perform the grouping from right to left
Thomas Wouters477c8d52006-05-27 19:21:47 +0000103def _group(s, monetary=False):
104 conv = localeconv()
105 thousands_sep = conv[monetary and 'mon_thousands_sep' or 'thousands_sep']
106 grouping = conv[monetary and 'mon_grouping' or 'grouping']
107 if not grouping:
108 return (s, 0)
109 result = ""
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000110 seps = 0
111 spaces = ""
112 if s[-1] == ' ':
113 sp = s.find(' ')
114 spaces = s[sp:]
115 s = s[:sp]
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000116 while s and grouping:
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000117 # if grouping is -1, we are done
Thomas Wouters477c8d52006-05-27 19:21:47 +0000118 if grouping[0] == CHAR_MAX:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000119 break
120 # 0: re-use last group ad infinitum
Thomas Wouters477c8d52006-05-27 19:21:47 +0000121 elif grouping[0] != 0:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000122 #process last group
Thomas Wouters477c8d52006-05-27 19:21:47 +0000123 group = grouping[0]
124 grouping = grouping[1:]
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000125 if result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000126 result = s[-group:] + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000127 seps += 1
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000128 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000129 result = s[-group:]
130 s = s[:-group]
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000131 if s and s[-1] not in "0123456789":
132 # the leading string is only spaces and signs
Thomas Wouters477c8d52006-05-27 19:21:47 +0000133 return s + result + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000134 if not result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000135 return s + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000136 if s:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000137 result = s + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000138 seps += 1
Thomas Wouters477c8d52006-05-27 19:21:47 +0000139 return result + spaces, seps
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000140
Thomas Wouters477c8d52006-05-27 19:21:47 +0000141def format(percent, value, grouping=False, monetary=False, *additional):
142 """Returns the locale-aware substitution of a %? specifier
143 (percent).
144
145 additional is for format strings which contain one or more
146 '*' modifiers."""
147 # this is only for one-percent-specifier strings and this should be checked
148 if percent[0] != '%':
149 raise ValueError("format() must be given exactly one %char "
150 "format specifier")
151 if additional:
152 formatted = percent % ((value,) + additional)
153 else:
154 formatted = percent % value
155 # floats and decimal ints need special action!
156 if percent[-1] in 'eEfFgG':
157 seps = 0
158 parts = formatted.split('.')
159 if grouping:
160 parts[0], seps = _group(parts[0], monetary=monetary)
161 decimal_point = localeconv()[monetary and 'mon_decimal_point'
162 or 'decimal_point']
163 formatted = decimal_point.join(parts)
164 while seps:
165 sp = formatted.find(' ')
166 if sp == -1: break
167 formatted = formatted[:sp] + formatted[sp+1:]
168 seps -= 1
169 elif percent[-1] in 'diu':
170 if grouping:
171 formatted = _group(formatted, monetary=monetary)[0]
172 return formatted
173
174import re, operator
175_percent_re = re.compile(r'%(?:\((?P<key>.*?)\))?'
176 r'(?P<modifiers>[-#0-9 +*.hlL]*?)[eEfFgGdiouxXcrs%]')
177
178def format_string(f, val, grouping=False):
179 """Formats a string in the same way that the % formatting would use,
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000180 but takes the current locale into account.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000181 Grouping is applied if the third parameter is true."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000182 percents = list(_percent_re.finditer(f))
183 new_f = _percent_re.sub('%s', f)
184
185 if isinstance(val, tuple):
186 new_val = list(val)
187 i = 0
188 for perc in percents:
189 starcount = perc.group('modifiers').count('*')
190 new_val[i] = format(perc.group(), new_val[i], grouping, False, *new_val[i+1:i+1+starcount])
191 del new_val[i+1:i+1+starcount]
192 i += (1 + starcount)
193 val = tuple(new_val)
194 elif operator.isMappingType(val):
195 for perc in percents:
196 key = perc.group("key")
197 val[key] = format(perc.group(), val[key], grouping)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000198 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000199 # val is a single value
200 val = format(percents[0].group(), val, grouping)
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000201
Thomas Wouters477c8d52006-05-27 19:21:47 +0000202 return new_f % val
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000203
Thomas Wouters477c8d52006-05-27 19:21:47 +0000204def currency(val, symbol=True, grouping=False, international=False):
205 """Formats val according to the currency settings
206 in the current locale."""
207 conv = localeconv()
208
209 # check for illegal values
210 digits = conv[international and 'int_frac_digits' or 'frac_digits']
211 if digits == 127:
212 raise ValueError("Currency formatting is not possible using "
213 "the 'C' locale.")
214
215 s = format('%%.%if' % digits, abs(val), grouping, monetary=True)
216 # '<' and '>' are markers if the sign must be inserted between symbol and value
217 s = '<' + s + '>'
218
219 if symbol:
220 smb = conv[international and 'int_curr_symbol' or 'currency_symbol']
221 precedes = conv[val<0 and 'n_cs_precedes' or 'p_cs_precedes']
222 separated = conv[val<0 and 'n_sep_by_space' or 'p_sep_by_space']
223
224 if precedes:
225 s = smb + (separated and ' ' or '') + s
226 else:
227 s = s + (separated and ' ' or '') + smb
228
229 sign_pos = conv[val<0 and 'n_sign_posn' or 'p_sign_posn']
230 sign = conv[val<0 and 'negative_sign' or 'positive_sign']
231
232 if sign_pos == 0:
233 s = '(' + s + ')'
234 elif sign_pos == 1:
235 s = sign + s
236 elif sign_pos == 2:
237 s = s + sign
238 elif sign_pos == 3:
239 s = s.replace('<', sign)
240 elif sign_pos == 4:
241 s = s.replace('>', sign)
242 else:
243 # the default if nothing specified;
244 # this should be the most fitting sign position
245 s = sign + s
246
247 return s.replace('<', '').replace('>', '')
Martin v. Löwisdb786872001-01-21 18:52:33 +0000248
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000249def str(val):
250 """Convert float to integer, taking the locale into account."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000251 return format("%.12g", val)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000252
Thomas Wouters477c8d52006-05-27 19:21:47 +0000253def atof(string, func=float):
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000254 "Parses a string as a float according to the locale settings."
255 #First, get rid of the grouping
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000256 ts = localeconv()['thousands_sep']
257 if ts:
Skip Montanaro249369c2004-04-10 16:39:32 +0000258 string = string.replace(ts, '')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000259 #next, replace the decimal point with a dot
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000260 dd = localeconv()['decimal_point']
261 if dd:
Skip Montanaro249369c2004-04-10 16:39:32 +0000262 string = string.replace(dd, '.')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000263 #finally, parse the string
Skip Montanaro249369c2004-04-10 16:39:32 +0000264 return func(string)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000265
266def atoi(str):
267 "Converts a string to an integer according to the locale settings."
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000268 return atof(str, int)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000269
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000270def _test():
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000271 setlocale(LC_ALL, "")
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000272 #do grouping
Thomas Wouters477c8d52006-05-27 19:21:47 +0000273 s1 = format("%d", 123456789,1)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000274 print(s1, "is", atoi(s1))
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000275 #standard formatting
Thomas Wouters477c8d52006-05-27 19:21:47 +0000276 s1 = str(3.14)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000277 print(s1, "is", atof(s1))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000278
279### Locale name aliasing engine
280
281# Author: Marc-Andre Lemburg, mal@lemburg.com
Fredrik Lundh37a09822002-10-19 20:19:10 +0000282# Various tweaks by Fredrik Lundh <fredrik@pythonware.com>
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000283
284# store away the low-level version of setlocale (it's
285# overridden below)
286_setlocale = setlocale
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000287
288def normalize(localename):
289
290 """ Returns a normalized locale code for the given locale
291 name.
292
293 The returned locale code is formatted for use with
294 setlocale().
295
296 If normalization fails, the original name is returned
297 unchanged.
298
299 If the given encoding is not known, the function defaults to
300 the default encoding for the locale code just like setlocale()
301 does.
302
303 """
304 # Normalize the locale name and extract the encoding
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000305 fullname = localename.lower()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000306 if ':' in fullname:
307 # ':' is sometimes used as encoding delimiter.
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000308 fullname = fullname.replace(':', '.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000309 if '.' in fullname:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000310 langname, encoding = fullname.split('.')[:2]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000311 fullname = langname + '.' + encoding
312 else:
313 langname = fullname
314 encoding = ''
315
316 # First lookup: fullname (possibly with encoding)
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000317 norm_encoding = encoding.replace('-', '')
318 norm_encoding = norm_encoding.replace('_', '')
319 lookup_name = langname + '.' + encoding
320 code = locale_alias.get(lookup_name, None)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000321 if code is not None:
322 return code
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000323 #print 'first lookup failed'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000324
325 # Second try: langname (without encoding)
326 code = locale_alias.get(langname, None)
327 if code is not None:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000328 #print 'langname lookup succeeded'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000329 if '.' in code:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000330 langname, defenc = code.split('.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000331 else:
332 langname = code
333 defenc = ''
334 if encoding:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000335 # Convert the encoding to a C lib compatible encoding string
336 norm_encoding = encodings.normalize_encoding(encoding)
337 #print 'norm encoding: %r' % norm_encoding
338 norm_encoding = encodings.aliases.aliases.get(norm_encoding,
339 norm_encoding)
340 #print 'aliased encoding: %r' % norm_encoding
341 encoding = locale_encoding_alias.get(norm_encoding,
342 norm_encoding)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000343 else:
344 encoding = defenc
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000345 #print 'found encoding %r' % encoding
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000346 if encoding:
347 return langname + '.' + encoding
348 else:
349 return langname
350
351 else:
352 return localename
353
354def _parse_localename(localename):
355
356 """ Parses the locale code for localename and returns the
357 result as tuple (language code, encoding).
358
359 The localename is normalized and passed through the locale
360 alias engine. A ValueError is raised in case the locale name
361 cannot be parsed.
362
363 The language code corresponds to RFC 1766. code and encoding
364 can be None in case the values cannot be determined or are
Jeremy Hyltona05e2932000-06-28 14:48:01 +0000365 unknown to this implementation.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000366
367 """
368 code = normalize(localename)
Georg Brandlb709c2c2006-01-20 09:07:35 +0000369 if '@' in code:
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000370 # Deal with locale modifiers
371 code, modifier = code.split('@')
372 if modifier == 'euro' and '.' not in code:
373 # Assume Latin-9 for @euro locales. This is bogus,
374 # since some systems may use other encodings for these
375 # locales. Also, we ignore other modifiers.
376 return code, 'iso-8859-15'
Tim Peters230a60c2002-11-09 05:08:07 +0000377
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000378 if '.' in code:
Raymond Hettinger346e67f2005-01-01 06:10:26 +0000379 return tuple(code.split('.')[:2])
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000380 elif code == 'C':
381 return None, None
Collin Winterce36ad82007-08-30 01:19:48 +0000382 raise ValueError('unknown locale: %s' % localename)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000383
384def _build_localename(localetuple):
385
386 """ Builds a locale code from the given tuple (language code,
387 encoding).
388
389 No aliasing or normalizing takes place.
390
391 """
392 language, encoding = localetuple
393 if language is None:
394 language = 'C'
395 if encoding is None:
396 return language
397 else:
398 return language + '.' + encoding
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000399
Matthias Klosef3f231f2005-09-20 07:02:49 +0000400def getdefaultlocale(envvars=('LC_ALL', 'LC_CTYPE', 'LANG', 'LANGUAGE')):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000401
402 """ Tries to determine the default locale settings and returns
403 them as tuple (language code, encoding).
404
405 According to POSIX, a program which has not called
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000406 setlocale(LC_ALL, "") runs using the portable 'C' locale.
407 Calling setlocale(LC_ALL, "") lets it use the default locale as
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000408 defined by the LANG variable. Since we don't want to interfere
Thomas Wouters7e474022000-07-16 12:04:32 +0000409 with the current locale setting we thus emulate the behavior
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000410 in the way described above.
411
412 To maintain compatibility with other platforms, not only the
413 LANG variable is tested, but a list of variables given as
414 envvars parameter. The first found to be defined will be
415 used. envvars defaults to the search path used in GNU gettext;
416 it must always contain the variable name 'LANG'.
417
418 Except for the code 'C', the language code corresponds to RFC
419 1766. code and encoding can be None in case the values cannot
420 be determined.
421
422 """
Fredrik Lundh04661322000-07-09 23:16:10 +0000423
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000424 try:
425 # check if it's supported by the _locale module
426 import _locale
427 code, encoding = _locale._getdefaultlocale()
Fredrik Lundh04661322000-07-09 23:16:10 +0000428 except (ImportError, AttributeError):
429 pass
430 else:
Fredrik Lundh663809e2000-07-10 19:32:19 +0000431 # make sure the code/encoding values are valid
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000432 if sys.platform == "win32" and code and code[:2] == "0x":
433 # map windows language identifier to language name
434 code = windows_locale.get(int(code, 0))
Fredrik Lundh663809e2000-07-10 19:32:19 +0000435 # ...add other platform-specific processing here, if
436 # necessary...
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000437 return code, encoding
Fredrik Lundh04661322000-07-09 23:16:10 +0000438
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000439 # fall back on POSIX behaviour
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000440 import os
441 lookup = os.environ.get
442 for variable in envvars:
443 localename = lookup(variable,None)
Martin v. Löwisc8ae31d2004-07-26 12:45:18 +0000444 if localename:
Matthias Klosef3f231f2005-09-20 07:02:49 +0000445 if variable == 'LANGUAGE':
446 localename = localename.split(':')[0]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000447 break
448 else:
449 localename = 'C'
450 return _parse_localename(localename)
451
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000452
453def getlocale(category=LC_CTYPE):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000454
455 """ Returns the current setting for the given locale category as
456 tuple (language code, encoding).
457
458 category may be one of the LC_* value except LC_ALL. It
459 defaults to LC_CTYPE.
460
461 Except for the code 'C', the language code corresponds to RFC
462 1766. code and encoding can be None in case the values cannot
463 be determined.
464
465 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000466 localename = _setlocale(category)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000467 if category == LC_ALL and ';' in localename:
Collin Winterce36ad82007-08-30 01:19:48 +0000468 raise TypeError('category LC_ALL is not supported')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000469 return _parse_localename(localename)
470
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000471def setlocale(category, locale=None):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000472
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000473 """ Set the locale for the given category. The locale can be
474 a string, a locale tuple (language code, encoding), or None.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000475
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000476 Locale tuples are converted to strings the locale aliasing
477 engine. Locale strings are passed directly to the C lib.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000478
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000479 category may be given as one of the LC_* values.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000480
481 """
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000482 if locale and not isinstance(locale, _builtin_str):
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000483 # convert to string
484 locale = normalize(_build_localename(locale))
485 return _setlocale(category, locale)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000486
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000487def resetlocale(category=LC_ALL):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000488
489 """ Sets the locale for category to the default setting.
490
491 The default setting is determined by calling
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000492 getdefaultlocale(). category defaults to LC_ALL.
493
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000494 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000495 _setlocale(category, _build_localename(getdefaultlocale()))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000496
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000497if sys.platform in ('win32', 'darwin', 'mac'):
498 # On Win32, this will return the ANSI code page
499 # On the Mac, it should return the system encoding;
500 # it might return "ascii" instead
501 def getpreferredencoding(do_setlocale = True):
502 """Return the charset that the user is likely using."""
503 import _locale
Tim Petersa326f472002-11-05 03:49:09 +0000504 return _locale._getdefaultlocale()[1]
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000505else:
506 # On Unix, if CODESET is available, use that.
507 try:
508 CODESET
509 except NameError:
510 # Fall back to parsing environment variables :-(
511 def getpreferredencoding(do_setlocale = True):
512 """Return the charset that the user is likely using,
513 by looking at environment variables."""
Martin v. Löwis071ef772008-03-08 11:24:24 +0000514 res = getdefaultlocale()[1]
515 if res is None:
516 # LANG not set, default conservatively to ASCII
517 res = 'ascii'
518 return res
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000519 else:
520 def getpreferredencoding(do_setlocale = True):
521 """Return the charset that the user is likely using,
522 according to the system configuration."""
523 if do_setlocale:
524 oldloc = setlocale(LC_CTYPE)
525 setlocale(LC_CTYPE, "")
526 result = nl_langinfo(CODESET)
527 setlocale(LC_CTYPE, oldloc)
528 return result
529 else:
530 return nl_langinfo(CODESET)
Tim Peters230a60c2002-11-09 05:08:07 +0000531
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000532
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000533### Database
534#
535# The following data was extracted from the locale.alias file which
536# comes with X11 and then hand edited removing the explicit encoding
537# definitions and adding some more aliases. The file is usually
538# available as /usr/lib/X11/locale/locale.alias.
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000539#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000540
541#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000542# The local_encoding_alias table maps lowercase encoding alias names
543# to C locale encoding names (case-sensitive). Note that normalize()
544# first looks up the encoding in the encodings.aliases dictionary and
545# then applies this mapping to find the correct C lib name for the
546# encoding.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000547#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000548locale_encoding_alias = {
549
550 # Mappings for non-standard encoding names used in locale names
551 '437': 'C',
552 'c': 'C',
553 'en': 'ISO8859-1',
554 'jis': 'JIS7',
555 'jis7': 'JIS7',
556 'ajec': 'eucJP',
557
558 # Mappings from Python codec names to C lib encoding names
559 'ascii': 'ISO8859-1',
560 'latin_1': 'ISO8859-1',
561 'iso8859_1': 'ISO8859-1',
562 'iso8859_10': 'ISO8859-10',
563 'iso8859_11': 'ISO8859-11',
564 'iso8859_13': 'ISO8859-13',
565 'iso8859_14': 'ISO8859-14',
566 'iso8859_15': 'ISO8859-15',
567 'iso8859_2': 'ISO8859-2',
568 'iso8859_3': 'ISO8859-3',
569 'iso8859_4': 'ISO8859-4',
570 'iso8859_5': 'ISO8859-5',
571 'iso8859_6': 'ISO8859-6',
572 'iso8859_7': 'ISO8859-7',
573 'iso8859_8': 'ISO8859-8',
574 'iso8859_9': 'ISO8859-9',
575 'iso2022_jp': 'JIS7',
576 'shift_jis': 'SJIS',
577 'tactis': 'TACTIS',
578 'euc_jp': 'eucJP',
579 'euc_kr': 'eucKR',
Marc-André Lemburgb4cebd42004-12-13 19:56:01 +0000580 'utf_8': 'UTF8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000581 'koi8_r': 'KOI8-R',
582 'koi8_u': 'KOI8-U',
583 # XXX This list is still incomplete. If you know more
584 # mappings, please file a bug report. Thanks.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000585}
586
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000587#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000588# The locale_alias table maps lowercase alias names to C locale names
589# (case-sensitive). Encodings are always separated from the locale
590# name using a dot ('.'); they should only be given in case the
591# language name is needed to interpret the given encoding alias
592# correctly (CJK codes often have this need).
593#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000594# Note that the normalize() function which uses this tables
595# removes '_' and '-' characters from the encoding part of the
596# locale name before doing the lookup. This saves a lot of
597# space in the table.
598#
599# MAL 2004-12-10:
600# Updated alias mapping to most recent locale.alias file
601# from X.org distribution using makelocalealias.py.
602#
603# These are the differences compared to the old mapping (Python 2.4
604# and older):
605#
606# updated 'bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
607# updated 'bg_bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
608# updated 'bulgarian' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
609# updated 'cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
610# updated 'cz_cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
611# updated 'czech' -> 'cs_CS.ISO8859-2' to 'cs_CZ.ISO8859-2'
612# updated 'dutch' -> 'nl_BE.ISO8859-1' to 'nl_NL.ISO8859-1'
613# updated 'et' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
614# updated 'et_ee' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
615# updated 'fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
616# updated 'fi_fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
617# updated 'iw' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
618# updated 'iw_il' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
619# updated 'japanese' -> 'ja_JP.SJIS' to 'ja_JP.eucJP'
620# updated 'lt' -> 'lt_LT.ISO8859-4' to 'lt_LT.ISO8859-13'
621# updated 'lv' -> 'lv_LV.ISO8859-4' to 'lv_LV.ISO8859-13'
622# updated 'sl' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
623# updated 'slovene' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
624# updated 'th_th' -> 'th_TH.TACTIS' to 'th_TH.ISO8859-11'
625# updated 'zh_cn' -> 'zh_CN.eucCN' to 'zh_CN.gb2312'
626# updated 'zh_cn.big5' -> 'zh_TW.eucTW' to 'zh_TW.big5'
627# updated 'zh_tw' -> 'zh_TW.eucTW' to 'zh_TW.big5'
628#
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000629# MAL 2008-05-30:
630# Updated alias mapping to most recent locale.alias file
631# from X.org distribution using makelocalealias.py.
632#
633# These are the differences compared to the old mapping (Python 2.5
634# and older):
635#
636# updated 'cs_cs.iso88592' -> 'cs_CZ.ISO8859-2' to 'cs_CS.ISO8859-2'
637# updated 'serbocroatian' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
638# updated 'sh' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
639# updated 'sh_hr.iso88592' -> 'sh_HR.ISO8859-2' to 'hr_HR.ISO8859-2'
640# updated 'sh_sp' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
641# updated 'sh_yu' -> 'sh_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
642# updated 'sp' -> 'sp_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
643# updated 'sp_yu' -> 'sp_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
644# updated 'sr' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
645# updated 'sr@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
646# updated 'sr_sp' -> 'sr_SP.ISO8859-2' to 'sr_CS.ISO8859-2'
647# updated 'sr_yu' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
648# updated 'sr_yu.cp1251@cyrillic' -> 'sr_YU.CP1251' to 'sr_CS.CP1251'
649# updated 'sr_yu.iso88592' -> 'sr_YU.ISO8859-2' to 'sr_CS.ISO8859-2'
650# updated 'sr_yu.iso88595' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
651# updated 'sr_yu.iso88595@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
652# updated 'sr_yu.microsoftcp1251@cyrillic' -> 'sr_YU.CP1251' to 'sr_CS.CP1251'
653# updated 'sr_yu.utf8@cyrillic' -> 'sr_YU.UTF-8' to 'sr_CS.UTF-8'
654# updated 'sr_yu@cyrillic' -> 'sr_YU.ISO8859-5' to 'sr_CS.ISO8859-5'
655
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000656locale_alias = {
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000657 'a3': 'a3_AZ.KOI8-C',
658 'a3_az': 'a3_AZ.KOI8-C',
659 'a3_az.koi8c': 'a3_AZ.KOI8-C',
660 'af': 'af_ZA.ISO8859-1',
661 'af_za': 'af_ZA.ISO8859-1',
662 'af_za.iso88591': 'af_ZA.ISO8859-1',
663 'am': 'am_ET.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000664 'am_et': 'am_ET.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000665 'american': 'en_US.ISO8859-1',
666 'american.iso88591': 'en_US.ISO8859-1',
667 'ar': 'ar_AA.ISO8859-6',
668 'ar_aa': 'ar_AA.ISO8859-6',
669 'ar_aa.iso88596': 'ar_AA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000670 'ar_ae': 'ar_AE.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000671 'ar_ae.iso88596': 'ar_AE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000672 'ar_bh': 'ar_BH.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000673 'ar_bh.iso88596': 'ar_BH.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000674 'ar_dz': 'ar_DZ.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000675 'ar_dz.iso88596': 'ar_DZ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000676 'ar_eg': 'ar_EG.ISO8859-6',
677 'ar_eg.iso88596': 'ar_EG.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000678 'ar_iq': 'ar_IQ.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000679 'ar_iq.iso88596': 'ar_IQ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000680 'ar_jo': 'ar_JO.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000681 'ar_jo.iso88596': 'ar_JO.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000682 'ar_kw': 'ar_KW.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000683 'ar_kw.iso88596': 'ar_KW.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000684 'ar_lb': 'ar_LB.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000685 'ar_lb.iso88596': 'ar_LB.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000686 'ar_ly': 'ar_LY.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000687 'ar_ly.iso88596': 'ar_LY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000688 'ar_ma': 'ar_MA.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000689 'ar_ma.iso88596': 'ar_MA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000690 'ar_om': 'ar_OM.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000691 'ar_om.iso88596': 'ar_OM.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000692 'ar_qa': 'ar_QA.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000693 'ar_qa.iso88596': 'ar_QA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000694 'ar_sa': 'ar_SA.ISO8859-6',
695 'ar_sa.iso88596': 'ar_SA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000696 'ar_sd': 'ar_SD.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000697 'ar_sd.iso88596': 'ar_SD.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000698 'ar_sy': 'ar_SY.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000699 'ar_sy.iso88596': 'ar_SY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000700 'ar_tn': 'ar_TN.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000701 'ar_tn.iso88596': 'ar_TN.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000702 'ar_ye': 'ar_YE.ISO8859-6',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000703 'ar_ye.iso88596': 'ar_YE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000704 'arabic': 'ar_AA.ISO8859-6',
705 'arabic.iso88596': 'ar_AA.ISO8859-6',
706 'az': 'az_AZ.ISO8859-9E',
707 'az_az': 'az_AZ.ISO8859-9E',
708 'az_az.iso88599e': 'az_AZ.ISO8859-9E',
709 'be': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000710 'be_by': 'be_BY.CP1251',
711 'be_by.cp1251': 'be_BY.CP1251',
712 'be_by.microsoftcp1251': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000713 'bg': 'bg_BG.CP1251',
714 'bg_bg': 'bg_BG.CP1251',
715 'bg_bg.cp1251': 'bg_BG.CP1251',
716 'bg_bg.iso88595': 'bg_BG.ISO8859-5',
717 'bg_bg.koi8r': 'bg_BG.KOI8-R',
718 'bg_bg.microsoftcp1251': 'bg_BG.CP1251',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000719 'bn_in': 'bn_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000720 'bokmal': 'nb_NO.ISO8859-1',
721 'bokm\xe5l': 'nb_NO.ISO8859-1',
722 'br': 'br_FR.ISO8859-1',
723 'br_fr': 'br_FR.ISO8859-1',
724 'br_fr.iso88591': 'br_FR.ISO8859-1',
725 'br_fr.iso885914': 'br_FR.ISO8859-14',
726 'br_fr.iso885915': 'br_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000727 'br_fr.iso885915@euro': 'br_FR.ISO8859-15',
728 'br_fr.utf8@euro': 'br_FR.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000729 'br_fr@euro': 'br_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000730 'bs': 'bs_BA.ISO8859-2',
731 'bs_ba': 'bs_BA.ISO8859-2',
732 'bs_ba.iso88592': 'bs_BA.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000733 'bulgarian': 'bg_BG.CP1251',
734 'c': 'C',
735 'c-french': 'fr_CA.ISO8859-1',
736 'c-french.iso88591': 'fr_CA.ISO8859-1',
737 'c.en': 'C',
738 'c.iso88591': 'en_US.ISO8859-1',
739 'c_c': 'C',
740 'c_c.c': 'C',
741 'ca': 'ca_ES.ISO8859-1',
742 'ca_es': 'ca_ES.ISO8859-1',
743 'ca_es.iso88591': 'ca_ES.ISO8859-1',
744 'ca_es.iso885915': 'ca_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000745 'ca_es.iso885915@euro': 'ca_ES.ISO8859-15',
746 'ca_es.utf8@euro': 'ca_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000747 'ca_es@euro': 'ca_ES.ISO8859-15',
748 'catalan': 'ca_ES.ISO8859-1',
749 'cextend': 'en_US.ISO8859-1',
750 'cextend.en': 'en_US.ISO8859-1',
751 'chinese-s': 'zh_CN.eucCN',
752 'chinese-t': 'zh_TW.eucTW',
753 'croatian': 'hr_HR.ISO8859-2',
754 'cs': 'cs_CZ.ISO8859-2',
755 'cs_cs': 'cs_CZ.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000756 'cs_cs.iso88592': 'cs_CS.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000757 'cs_cz': 'cs_CZ.ISO8859-2',
758 'cs_cz.iso88592': 'cs_CZ.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000759 'cy': 'cy_GB.ISO8859-1',
760 'cy_gb': 'cy_GB.ISO8859-1',
761 'cy_gb.iso88591': 'cy_GB.ISO8859-1',
762 'cy_gb.iso885914': 'cy_GB.ISO8859-14',
763 'cy_gb.iso885915': 'cy_GB.ISO8859-15',
764 'cy_gb@euro': 'cy_GB.ISO8859-15',
765 'cz': 'cs_CZ.ISO8859-2',
766 'cz_cz': 'cs_CZ.ISO8859-2',
767 'czech': 'cs_CZ.ISO8859-2',
768 'da': 'da_DK.ISO8859-1',
769 'da_dk': 'da_DK.ISO8859-1',
770 'da_dk.88591': 'da_DK.ISO8859-1',
771 'da_dk.885915': 'da_DK.ISO8859-15',
772 'da_dk.iso88591': 'da_DK.ISO8859-1',
773 'da_dk.iso885915': 'da_DK.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000774 'da_dk@euro': 'da_DK.ISO8859-15',
775 'danish': 'da_DK.ISO8859-1',
776 'danish.iso88591': 'da_DK.ISO8859-1',
777 'dansk': 'da_DK.ISO8859-1',
778 'de': 'de_DE.ISO8859-1',
779 'de_at': 'de_AT.ISO8859-1',
780 'de_at.iso88591': 'de_AT.ISO8859-1',
781 'de_at.iso885915': 'de_AT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000782 'de_at.iso885915@euro': 'de_AT.ISO8859-15',
783 'de_at.utf8@euro': 'de_AT.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000784 'de_at@euro': 'de_AT.ISO8859-15',
785 'de_be': 'de_BE.ISO8859-1',
786 'de_be.iso88591': 'de_BE.ISO8859-1',
787 'de_be.iso885915': 'de_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000788 'de_be.iso885915@euro': 'de_BE.ISO8859-15',
789 'de_be.utf8@euro': 'de_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000790 'de_be@euro': 'de_BE.ISO8859-15',
791 'de_ch': 'de_CH.ISO8859-1',
792 'de_ch.iso88591': 'de_CH.ISO8859-1',
793 'de_ch.iso885915': 'de_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000794 'de_ch@euro': 'de_CH.ISO8859-15',
795 'de_de': 'de_DE.ISO8859-1',
796 'de_de.88591': 'de_DE.ISO8859-1',
797 'de_de.885915': 'de_DE.ISO8859-15',
798 'de_de.885915@euro': 'de_DE.ISO8859-15',
799 'de_de.iso88591': 'de_DE.ISO8859-1',
800 'de_de.iso885915': 'de_DE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000801 'de_de.iso885915@euro': 'de_DE.ISO8859-15',
802 'de_de.utf8@euro': 'de_DE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000803 'de_de@euro': 'de_DE.ISO8859-15',
804 'de_lu': 'de_LU.ISO8859-1',
805 'de_lu.iso88591': 'de_LU.ISO8859-1',
806 'de_lu.iso885915': 'de_LU.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000807 'de_lu.iso885915@euro': 'de_LU.ISO8859-15',
808 'de_lu.utf8@euro': 'de_LU.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000809 'de_lu@euro': 'de_LU.ISO8859-15',
810 'deutsch': 'de_DE.ISO8859-1',
811 'dutch': 'nl_NL.ISO8859-1',
812 'dutch.iso88591': 'nl_BE.ISO8859-1',
813 'ee': 'ee_EE.ISO8859-4',
814 'ee_ee': 'ee_EE.ISO8859-4',
815 'ee_ee.iso88594': 'ee_EE.ISO8859-4',
816 'eesti': 'et_EE.ISO8859-1',
817 'el': 'el_GR.ISO8859-7',
818 'el_gr': 'el_GR.ISO8859-7',
819 'el_gr.iso88597': 'el_GR.ISO8859-7',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000820 'el_gr@euro': 'el_GR.ISO8859-15',
821 'en': 'en_US.ISO8859-1',
822 'en.iso88591': 'en_US.ISO8859-1',
823 'en_au': 'en_AU.ISO8859-1',
824 'en_au.iso88591': 'en_AU.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000825 'en_be': 'en_BE.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000826 'en_be@euro': 'en_BE.ISO8859-15',
827 'en_bw': 'en_BW.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000828 'en_bw.iso88591': 'en_BW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000829 'en_ca': 'en_CA.ISO8859-1',
830 'en_ca.iso88591': 'en_CA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000831 'en_gb': 'en_GB.ISO8859-1',
832 'en_gb.88591': 'en_GB.ISO8859-1',
833 'en_gb.iso88591': 'en_GB.ISO8859-1',
834 'en_gb.iso885915': 'en_GB.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000835 'en_gb@euro': 'en_GB.ISO8859-15',
836 'en_hk': 'en_HK.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000837 'en_hk.iso88591': 'en_HK.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000838 'en_ie': 'en_IE.ISO8859-1',
839 'en_ie.iso88591': 'en_IE.ISO8859-1',
840 'en_ie.iso885915': 'en_IE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000841 'en_ie.iso885915@euro': 'en_IE.ISO8859-15',
842 'en_ie.utf8@euro': 'en_IE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000843 'en_ie@euro': 'en_IE.ISO8859-15',
844 'en_in': 'en_IN.ISO8859-1',
845 'en_nz': 'en_NZ.ISO8859-1',
846 'en_nz.iso88591': 'en_NZ.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000847 'en_ph': 'en_PH.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000848 'en_ph.iso88591': 'en_PH.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000849 'en_sg': 'en_SG.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000850 'en_sg.iso88591': 'en_SG.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000851 'en_uk': 'en_GB.ISO8859-1',
852 'en_us': 'en_US.ISO8859-1',
853 'en_us.88591': 'en_US.ISO8859-1',
854 'en_us.885915': 'en_US.ISO8859-15',
855 'en_us.iso88591': 'en_US.ISO8859-1',
856 'en_us.iso885915': 'en_US.ISO8859-15',
857 'en_us.iso885915@euro': 'en_US.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000858 'en_us@euro': 'en_US.ISO8859-15',
859 'en_us@euro@euro': 'en_US.ISO8859-15',
860 'en_za': 'en_ZA.ISO8859-1',
861 'en_za.88591': 'en_ZA.ISO8859-1',
862 'en_za.iso88591': 'en_ZA.ISO8859-1',
863 'en_za.iso885915': 'en_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000864 'en_za@euro': 'en_ZA.ISO8859-15',
865 'en_zw': 'en_ZW.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000866 'en_zw.iso88591': 'en_ZW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000867 'eng_gb': 'en_GB.ISO8859-1',
868 'eng_gb.8859': 'en_GB.ISO8859-1',
869 'english': 'en_EN.ISO8859-1',
870 'english.iso88591': 'en_EN.ISO8859-1',
871 'english_uk': 'en_GB.ISO8859-1',
872 'english_uk.8859': 'en_GB.ISO8859-1',
873 'english_united-states': 'en_US.ISO8859-1',
874 'english_united-states.437': 'C',
875 'english_us': 'en_US.ISO8859-1',
876 'english_us.8859': 'en_US.ISO8859-1',
877 'english_us.ascii': 'en_US.ISO8859-1',
878 'eo': 'eo_XX.ISO8859-3',
879 'eo_eo': 'eo_EO.ISO8859-3',
880 'eo_eo.iso88593': 'eo_EO.ISO8859-3',
881 'eo_xx': 'eo_XX.ISO8859-3',
882 'eo_xx.iso88593': 'eo_XX.ISO8859-3',
883 'es': 'es_ES.ISO8859-1',
884 'es_ar': 'es_AR.ISO8859-1',
885 'es_ar.iso88591': 'es_AR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000886 'es_bo': 'es_BO.ISO8859-1',
887 'es_bo.iso88591': 'es_BO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000888 'es_cl': 'es_CL.ISO8859-1',
889 'es_cl.iso88591': 'es_CL.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000890 'es_co': 'es_CO.ISO8859-1',
891 'es_co.iso88591': 'es_CO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000892 'es_cr': 'es_CR.ISO8859-1',
893 'es_cr.iso88591': 'es_CR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000894 'es_do': 'es_DO.ISO8859-1',
895 'es_do.iso88591': 'es_DO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000896 'es_ec': 'es_EC.ISO8859-1',
897 'es_ec.iso88591': 'es_EC.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000898 'es_es': 'es_ES.ISO8859-1',
899 'es_es.88591': 'es_ES.ISO8859-1',
900 'es_es.iso88591': 'es_ES.ISO8859-1',
901 'es_es.iso885915': 'es_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000902 'es_es.iso885915@euro': 'es_ES.ISO8859-15',
903 'es_es.utf8@euro': 'es_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000904 'es_es@euro': 'es_ES.ISO8859-15',
905 'es_gt': 'es_GT.ISO8859-1',
906 'es_gt.iso88591': 'es_GT.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000907 'es_hn': 'es_HN.ISO8859-1',
908 'es_hn.iso88591': 'es_HN.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000909 'es_mx': 'es_MX.ISO8859-1',
910 'es_mx.iso88591': 'es_MX.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000911 'es_ni': 'es_NI.ISO8859-1',
912 'es_ni.iso88591': 'es_NI.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000913 'es_pa': 'es_PA.ISO8859-1',
914 'es_pa.iso88591': 'es_PA.ISO8859-1',
915 'es_pa.iso885915': 'es_PA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000916 'es_pa@euro': 'es_PA.ISO8859-15',
917 'es_pe': 'es_PE.ISO8859-1',
918 'es_pe.iso88591': 'es_PE.ISO8859-1',
919 'es_pe.iso885915': 'es_PE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000920 'es_pe@euro': 'es_PE.ISO8859-15',
921 'es_pr': 'es_PR.ISO8859-1',
922 'es_pr.iso88591': 'es_PR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000923 'es_py': 'es_PY.ISO8859-1',
924 'es_py.iso88591': 'es_PY.ISO8859-1',
925 'es_py.iso885915': 'es_PY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000926 'es_py@euro': 'es_PY.ISO8859-15',
927 'es_sv': 'es_SV.ISO8859-1',
928 'es_sv.iso88591': 'es_SV.ISO8859-1',
929 'es_sv.iso885915': 'es_SV.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000930 'es_sv@euro': 'es_SV.ISO8859-15',
931 'es_us': 'es_US.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000932 'es_us.iso88591': 'es_US.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000933 'es_uy': 'es_UY.ISO8859-1',
934 'es_uy.iso88591': 'es_UY.ISO8859-1',
935 'es_uy.iso885915': 'es_UY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000936 'es_uy@euro': 'es_UY.ISO8859-15',
937 'es_ve': 'es_VE.ISO8859-1',
938 'es_ve.iso88591': 'es_VE.ISO8859-1',
939 'es_ve.iso885915': 'es_VE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000940 'es_ve@euro': 'es_VE.ISO8859-15',
941 'estonian': 'et_EE.ISO8859-1',
942 'et': 'et_EE.ISO8859-15',
943 'et_ee': 'et_EE.ISO8859-15',
944 'et_ee.iso88591': 'et_EE.ISO8859-1',
945 'et_ee.iso885913': 'et_EE.ISO8859-13',
946 'et_ee.iso885915': 'et_EE.ISO8859-15',
947 'et_ee.iso88594': 'et_EE.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000948 'et_ee@euro': 'et_EE.ISO8859-15',
949 'eu': 'eu_ES.ISO8859-1',
950 'eu_es': 'eu_ES.ISO8859-1',
951 'eu_es.iso88591': 'eu_ES.ISO8859-1',
952 'eu_es.iso885915': 'eu_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000953 'eu_es.iso885915@euro': 'eu_ES.ISO8859-15',
954 'eu_es.utf8@euro': 'eu_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000955 'eu_es@euro': 'eu_ES.ISO8859-15',
956 'fa': 'fa_IR.UTF-8',
957 'fa_ir': 'fa_IR.UTF-8',
958 'fa_ir.isiri3342': 'fa_IR.ISIRI-3342',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000959 'fi': 'fi_FI.ISO8859-15',
960 'fi_fi': 'fi_FI.ISO8859-15',
961 'fi_fi.88591': 'fi_FI.ISO8859-1',
962 'fi_fi.iso88591': 'fi_FI.ISO8859-1',
963 'fi_fi.iso885915': 'fi_FI.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000964 'fi_fi.iso885915@euro': 'fi_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000965 'fi_fi.utf8@euro': 'fi_FI.UTF-8',
966 'fi_fi@euro': 'fi_FI.ISO8859-15',
967 'finnish': 'fi_FI.ISO8859-1',
968 'finnish.iso88591': 'fi_FI.ISO8859-1',
969 'fo': 'fo_FO.ISO8859-1',
970 'fo_fo': 'fo_FO.ISO8859-1',
971 'fo_fo.iso88591': 'fo_FO.ISO8859-1',
972 'fo_fo.iso885915': 'fo_FO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000973 'fo_fo@euro': 'fo_FO.ISO8859-15',
974 'fr': 'fr_FR.ISO8859-1',
975 'fr_be': 'fr_BE.ISO8859-1',
976 'fr_be.88591': 'fr_BE.ISO8859-1',
977 'fr_be.iso88591': 'fr_BE.ISO8859-1',
978 'fr_be.iso885915': 'fr_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000979 'fr_be.iso885915@euro': 'fr_BE.ISO8859-15',
980 'fr_be.utf8@euro': 'fr_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000981 'fr_be@euro': 'fr_BE.ISO8859-15',
982 'fr_ca': 'fr_CA.ISO8859-1',
983 'fr_ca.88591': 'fr_CA.ISO8859-1',
984 'fr_ca.iso88591': 'fr_CA.ISO8859-1',
985 'fr_ca.iso885915': 'fr_CA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000986 'fr_ca@euro': 'fr_CA.ISO8859-15',
987 'fr_ch': 'fr_CH.ISO8859-1',
988 'fr_ch.88591': 'fr_CH.ISO8859-1',
989 'fr_ch.iso88591': 'fr_CH.ISO8859-1',
990 'fr_ch.iso885915': 'fr_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000991 'fr_ch@euro': 'fr_CH.ISO8859-15',
992 'fr_fr': 'fr_FR.ISO8859-1',
993 'fr_fr.88591': 'fr_FR.ISO8859-1',
994 'fr_fr.iso88591': 'fr_FR.ISO8859-1',
995 'fr_fr.iso885915': 'fr_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +0000996 'fr_fr.iso885915@euro': 'fr_FR.ISO8859-15',
997 'fr_fr.utf8@euro': 'fr_FR.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000998 'fr_fr@euro': 'fr_FR.ISO8859-15',
999 'fr_lu': 'fr_LU.ISO8859-1',
1000 'fr_lu.88591': 'fr_LU.ISO8859-1',
1001 'fr_lu.iso88591': 'fr_LU.ISO8859-1',
1002 'fr_lu.iso885915': 'fr_LU.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001003 'fr_lu.iso885915@euro': 'fr_LU.ISO8859-15',
1004 'fr_lu.utf8@euro': 'fr_LU.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001005 'fr_lu@euro': 'fr_LU.ISO8859-15',
1006 'fran\xe7ais': 'fr_FR.ISO8859-1',
1007 'fre_fr': 'fr_FR.ISO8859-1',
1008 'fre_fr.8859': 'fr_FR.ISO8859-1',
1009 'french': 'fr_FR.ISO8859-1',
1010 'french.iso88591': 'fr_CH.ISO8859-1',
1011 'french_france': 'fr_FR.ISO8859-1',
1012 'french_france.8859': 'fr_FR.ISO8859-1',
1013 'ga': 'ga_IE.ISO8859-1',
1014 'ga_ie': 'ga_IE.ISO8859-1',
1015 'ga_ie.iso88591': 'ga_IE.ISO8859-1',
1016 'ga_ie.iso885914': 'ga_IE.ISO8859-14',
1017 'ga_ie.iso885915': 'ga_IE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001018 'ga_ie.iso885915@euro': 'ga_IE.ISO8859-15',
1019 'ga_ie.utf8@euro': 'ga_IE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001020 'ga_ie@euro': 'ga_IE.ISO8859-15',
1021 'galego': 'gl_ES.ISO8859-1',
1022 'galician': 'gl_ES.ISO8859-1',
1023 'gd': 'gd_GB.ISO8859-1',
1024 'gd_gb': 'gd_GB.ISO8859-1',
1025 'gd_gb.iso88591': 'gd_GB.ISO8859-1',
1026 'gd_gb.iso885914': 'gd_GB.ISO8859-14',
1027 'gd_gb.iso885915': 'gd_GB.ISO8859-15',
1028 'gd_gb@euro': 'gd_GB.ISO8859-15',
1029 'ger_de': 'de_DE.ISO8859-1',
1030 'ger_de.8859': 'de_DE.ISO8859-1',
1031 'german': 'de_DE.ISO8859-1',
1032 'german.iso88591': 'de_CH.ISO8859-1',
1033 'german_germany': 'de_DE.ISO8859-1',
1034 'german_germany.8859': 'de_DE.ISO8859-1',
1035 'gl': 'gl_ES.ISO8859-1',
1036 'gl_es': 'gl_ES.ISO8859-1',
1037 'gl_es.iso88591': 'gl_ES.ISO8859-1',
1038 'gl_es.iso885915': 'gl_ES.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001039 'gl_es.iso885915@euro': 'gl_ES.ISO8859-15',
1040 'gl_es.utf8@euro': 'gl_ES.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001041 'gl_es@euro': 'gl_ES.ISO8859-15',
1042 'greek': 'el_GR.ISO8859-7',
1043 'greek.iso88597': 'el_GR.ISO8859-7',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001044 'gu_in': 'gu_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001045 'gv': 'gv_GB.ISO8859-1',
1046 'gv_gb': 'gv_GB.ISO8859-1',
1047 'gv_gb.iso88591': 'gv_GB.ISO8859-1',
1048 'gv_gb.iso885914': 'gv_GB.ISO8859-14',
1049 'gv_gb.iso885915': 'gv_GB.ISO8859-15',
1050 'gv_gb@euro': 'gv_GB.ISO8859-15',
1051 'he': 'he_IL.ISO8859-8',
1052 'he_il': 'he_IL.ISO8859-8',
1053 'he_il.cp1255': 'he_IL.CP1255',
1054 'he_il.iso88598': 'he_IL.ISO8859-8',
1055 'he_il.microsoftcp1255': 'he_IL.CP1255',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001056 'hebrew': 'iw_IL.ISO8859-8',
1057 'hebrew.iso88598': 'iw_IL.ISO8859-8',
1058 'hi': 'hi_IN.ISCII-DEV',
1059 'hi_in': 'hi_IN.ISCII-DEV',
1060 'hi_in.isciidev': 'hi_IN.ISCII-DEV',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001061 'hr': 'hr_HR.ISO8859-2',
1062 'hr_hr': 'hr_HR.ISO8859-2',
1063 'hr_hr.iso88592': 'hr_HR.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001064 'hrvatski': 'hr_HR.ISO8859-2',
1065 'hu': 'hu_HU.ISO8859-2',
1066 'hu_hu': 'hu_HU.ISO8859-2',
1067 'hu_hu.iso88592': 'hu_HU.ISO8859-2',
1068 'hungarian': 'hu_HU.ISO8859-2',
1069 'icelandic': 'is_IS.ISO8859-1',
1070 'icelandic.iso88591': 'is_IS.ISO8859-1',
1071 'id': 'id_ID.ISO8859-1',
1072 'id_id': 'id_ID.ISO8859-1',
1073 'in': 'id_ID.ISO8859-1',
1074 'in_id': 'id_ID.ISO8859-1',
1075 'is': 'is_IS.ISO8859-1',
1076 'is_is': 'is_IS.ISO8859-1',
1077 'is_is.iso88591': 'is_IS.ISO8859-1',
1078 'is_is.iso885915': 'is_IS.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001079 'is_is@euro': 'is_IS.ISO8859-15',
1080 'iso-8859-1': 'en_US.ISO8859-1',
1081 'iso-8859-15': 'en_US.ISO8859-15',
1082 'iso8859-1': 'en_US.ISO8859-1',
1083 'iso8859-15': 'en_US.ISO8859-15',
1084 'iso_8859_1': 'en_US.ISO8859-1',
1085 'iso_8859_15': 'en_US.ISO8859-15',
1086 'it': 'it_IT.ISO8859-1',
1087 'it_ch': 'it_CH.ISO8859-1',
1088 'it_ch.iso88591': 'it_CH.ISO8859-1',
1089 'it_ch.iso885915': 'it_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001090 'it_ch@euro': 'it_CH.ISO8859-15',
1091 'it_it': 'it_IT.ISO8859-1',
1092 'it_it.88591': 'it_IT.ISO8859-1',
1093 'it_it.iso88591': 'it_IT.ISO8859-1',
1094 'it_it.iso885915': 'it_IT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001095 'it_it.iso885915@euro': 'it_IT.ISO8859-15',
1096 'it_it.utf8@euro': 'it_IT.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001097 'it_it@euro': 'it_IT.ISO8859-15',
1098 'italian': 'it_IT.ISO8859-1',
1099 'italian.iso88591': 'it_IT.ISO8859-1',
1100 'iu': 'iu_CA.NUNACOM-8',
1101 'iu_ca': 'iu_CA.NUNACOM-8',
1102 'iu_ca.nunacom8': 'iu_CA.NUNACOM-8',
1103 'iw': 'he_IL.ISO8859-8',
1104 'iw_il': 'he_IL.ISO8859-8',
1105 'iw_il.iso88598': 'he_IL.ISO8859-8',
1106 'ja': 'ja_JP.eucJP',
1107 'ja.jis': 'ja_JP.JIS7',
1108 'ja.sjis': 'ja_JP.SJIS',
1109 'ja_jp': 'ja_JP.eucJP',
1110 'ja_jp.ajec': 'ja_JP.eucJP',
1111 'ja_jp.euc': 'ja_JP.eucJP',
1112 'ja_jp.eucjp': 'ja_JP.eucJP',
1113 'ja_jp.iso-2022-jp': 'ja_JP.JIS7',
1114 'ja_jp.iso2022jp': 'ja_JP.JIS7',
1115 'ja_jp.jis': 'ja_JP.JIS7',
1116 'ja_jp.jis7': 'ja_JP.JIS7',
1117 'ja_jp.mscode': 'ja_JP.SJIS',
1118 'ja_jp.sjis': 'ja_JP.SJIS',
1119 'ja_jp.ujis': 'ja_JP.eucJP',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001120 'japan': 'ja_JP.eucJP',
1121 'japanese': 'ja_JP.eucJP',
1122 'japanese-euc': 'ja_JP.eucJP',
1123 'japanese.euc': 'ja_JP.eucJP',
1124 'japanese.sjis': 'ja_JP.SJIS',
1125 'jp_jp': 'ja_JP.eucJP',
1126 'ka': 'ka_GE.GEORGIAN-ACADEMY',
1127 'ka_ge': 'ka_GE.GEORGIAN-ACADEMY',
1128 'ka_ge.georgianacademy': 'ka_GE.GEORGIAN-ACADEMY',
1129 'ka_ge.georgianps': 'ka_GE.GEORGIAN-PS',
1130 'ka_ge.georgianrs': 'ka_GE.GEORGIAN-ACADEMY',
1131 'kl': 'kl_GL.ISO8859-1',
1132 'kl_gl': 'kl_GL.ISO8859-1',
1133 'kl_gl.iso88591': 'kl_GL.ISO8859-1',
1134 'kl_gl.iso885915': 'kl_GL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001135 'kl_gl@euro': 'kl_GL.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001136 'km_kh': 'km_KH.UTF-8',
1137 'kn_in': 'kn_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001138 'ko': 'ko_KR.eucKR',
1139 'ko_kr': 'ko_KR.eucKR',
1140 'ko_kr.euc': 'ko_KR.eucKR',
1141 'ko_kr.euckr': 'ko_KR.eucKR',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001142 'korean': 'ko_KR.eucKR',
1143 'korean.euc': 'ko_KR.eucKR',
1144 'kw': 'kw_GB.ISO8859-1',
1145 'kw_gb': 'kw_GB.ISO8859-1',
1146 'kw_gb.iso88591': 'kw_GB.ISO8859-1',
1147 'kw_gb.iso885914': 'kw_GB.ISO8859-14',
1148 'kw_gb.iso885915': 'kw_GB.ISO8859-15',
1149 'kw_gb@euro': 'kw_GB.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001150 'ky': 'ky_KG.UTF-8',
1151 'ky_kg': 'ky_KG.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001152 'lithuanian': 'lt_LT.ISO8859-13',
1153 'lo': 'lo_LA.MULELAO-1',
1154 'lo_la': 'lo_LA.MULELAO-1',
1155 'lo_la.cp1133': 'lo_LA.IBM-CP1133',
1156 'lo_la.ibmcp1133': 'lo_LA.IBM-CP1133',
1157 'lo_la.mulelao1': 'lo_LA.MULELAO-1',
1158 'lt': 'lt_LT.ISO8859-13',
1159 'lt_lt': 'lt_LT.ISO8859-13',
1160 'lt_lt.iso885913': 'lt_LT.ISO8859-13',
1161 'lt_lt.iso88594': 'lt_LT.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001162 'lv': 'lv_LV.ISO8859-13',
1163 'lv_lv': 'lv_LV.ISO8859-13',
1164 'lv_lv.iso885913': 'lv_LV.ISO8859-13',
1165 'lv_lv.iso88594': 'lv_LV.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001166 'mi': 'mi_NZ.ISO8859-1',
1167 'mi_nz': 'mi_NZ.ISO8859-1',
1168 'mi_nz.iso88591': 'mi_NZ.ISO8859-1',
1169 'mk': 'mk_MK.ISO8859-5',
1170 'mk_mk': 'mk_MK.ISO8859-5',
1171 'mk_mk.cp1251': 'mk_MK.CP1251',
1172 'mk_mk.iso88595': 'mk_MK.ISO8859-5',
1173 'mk_mk.microsoftcp1251': 'mk_MK.CP1251',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001174 'mr_in': 'mr_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001175 'ms': 'ms_MY.ISO8859-1',
1176 'ms_my': 'ms_MY.ISO8859-1',
1177 'ms_my.iso88591': 'ms_MY.ISO8859-1',
1178 'mt': 'mt_MT.ISO8859-3',
1179 'mt_mt': 'mt_MT.ISO8859-3',
1180 'mt_mt.iso88593': 'mt_MT.ISO8859-3',
1181 'nb': 'nb_NO.ISO8859-1',
1182 'nb_no': 'nb_NO.ISO8859-1',
1183 'nb_no.88591': 'nb_NO.ISO8859-1',
1184 'nb_no.iso88591': 'nb_NO.ISO8859-1',
1185 'nb_no.iso885915': 'nb_NO.ISO8859-15',
1186 'nb_no@euro': 'nb_NO.ISO8859-15',
1187 'nl': 'nl_NL.ISO8859-1',
1188 'nl_be': 'nl_BE.ISO8859-1',
1189 'nl_be.88591': 'nl_BE.ISO8859-1',
1190 'nl_be.iso88591': 'nl_BE.ISO8859-1',
1191 'nl_be.iso885915': 'nl_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001192 'nl_be.iso885915@euro': 'nl_BE.ISO8859-15',
1193 'nl_be.utf8@euro': 'nl_BE.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001194 'nl_be@euro': 'nl_BE.ISO8859-15',
1195 'nl_nl': 'nl_NL.ISO8859-1',
1196 'nl_nl.88591': 'nl_NL.ISO8859-1',
1197 'nl_nl.iso88591': 'nl_NL.ISO8859-1',
1198 'nl_nl.iso885915': 'nl_NL.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001199 'nl_nl.iso885915@euro': 'nl_NL.ISO8859-15',
1200 'nl_nl.utf8@euro': 'nl_NL.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001201 'nl_nl@euro': 'nl_NL.ISO8859-15',
1202 'nn': 'nn_NO.ISO8859-1',
1203 'nn_no': 'nn_NO.ISO8859-1',
1204 'nn_no.88591': 'nn_NO.ISO8859-1',
1205 'nn_no.iso88591': 'nn_NO.ISO8859-1',
1206 'nn_no.iso885915': 'nn_NO.ISO8859-15',
1207 'nn_no@euro': 'nn_NO.ISO8859-15',
1208 'no': 'no_NO.ISO8859-1',
1209 'no@nynorsk': 'ny_NO.ISO8859-1',
1210 'no_no': 'no_NO.ISO8859-1',
1211 'no_no.88591': 'no_NO.ISO8859-1',
1212 'no_no.iso88591': 'no_NO.ISO8859-1',
1213 'no_no.iso885915': 'no_NO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001214 'no_no@euro': 'no_NO.ISO8859-15',
1215 'norwegian': 'no_NO.ISO8859-1',
1216 'norwegian.iso88591': 'no_NO.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001217 'nr': 'nr_ZA.ISO8859-1',
1218 'nr_za': 'nr_ZA.ISO8859-1',
1219 'nr_za.iso88591': 'nr_ZA.ISO8859-1',
1220 'nso': 'nso_ZA.ISO8859-15',
1221 'nso_za': 'nso_ZA.ISO8859-15',
1222 'nso_za.iso885915': 'nso_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001223 'ny': 'ny_NO.ISO8859-1',
1224 'ny_no': 'ny_NO.ISO8859-1',
1225 'ny_no.88591': 'ny_NO.ISO8859-1',
1226 'ny_no.iso88591': 'ny_NO.ISO8859-1',
1227 'ny_no.iso885915': 'ny_NO.ISO8859-15',
1228 'ny_no@euro': 'ny_NO.ISO8859-15',
1229 'nynorsk': 'nn_NO.ISO8859-1',
1230 'oc': 'oc_FR.ISO8859-1',
1231 'oc_fr': 'oc_FR.ISO8859-1',
1232 'oc_fr.iso88591': 'oc_FR.ISO8859-1',
1233 'oc_fr.iso885915': 'oc_FR.ISO8859-15',
1234 'oc_fr@euro': 'oc_FR.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001235 'pa_in': 'pa_IN.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001236 'pd': 'pd_US.ISO8859-1',
1237 'pd_de': 'pd_DE.ISO8859-1',
1238 'pd_de.iso88591': 'pd_DE.ISO8859-1',
1239 'pd_de.iso885915': 'pd_DE.ISO8859-15',
1240 'pd_de@euro': 'pd_DE.ISO8859-15',
1241 'pd_us': 'pd_US.ISO8859-1',
1242 'pd_us.iso88591': 'pd_US.ISO8859-1',
1243 'pd_us.iso885915': 'pd_US.ISO8859-15',
1244 'pd_us@euro': 'pd_US.ISO8859-15',
1245 'ph': 'ph_PH.ISO8859-1',
1246 'ph_ph': 'ph_PH.ISO8859-1',
1247 'ph_ph.iso88591': 'ph_PH.ISO8859-1',
1248 'pl': 'pl_PL.ISO8859-2',
1249 'pl_pl': 'pl_PL.ISO8859-2',
1250 'pl_pl.iso88592': 'pl_PL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001251 'polish': 'pl_PL.ISO8859-2',
1252 'portuguese': 'pt_PT.ISO8859-1',
1253 'portuguese.iso88591': 'pt_PT.ISO8859-1',
1254 'portuguese_brazil': 'pt_BR.ISO8859-1',
1255 'portuguese_brazil.8859': 'pt_BR.ISO8859-1',
1256 'posix': 'C',
1257 'posix-utf2': 'C',
1258 'pp': 'pp_AN.ISO8859-1',
1259 'pp_an': 'pp_AN.ISO8859-1',
1260 'pp_an.iso88591': 'pp_AN.ISO8859-1',
1261 'pt': 'pt_PT.ISO8859-1',
1262 'pt_br': 'pt_BR.ISO8859-1',
1263 'pt_br.88591': 'pt_BR.ISO8859-1',
1264 'pt_br.iso88591': 'pt_BR.ISO8859-1',
1265 'pt_br.iso885915': 'pt_BR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001266 'pt_br@euro': 'pt_BR.ISO8859-15',
1267 'pt_pt': 'pt_PT.ISO8859-1',
1268 'pt_pt.88591': 'pt_PT.ISO8859-1',
1269 'pt_pt.iso88591': 'pt_PT.ISO8859-1',
1270 'pt_pt.iso885915': 'pt_PT.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001271 'pt_pt.iso885915@euro': 'pt_PT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001272 'pt_pt.utf8@euro': 'pt_PT.UTF-8',
1273 'pt_pt@euro': 'pt_PT.ISO8859-15',
1274 'ro': 'ro_RO.ISO8859-2',
1275 'ro_ro': 'ro_RO.ISO8859-2',
1276 'ro_ro.iso88592': 'ro_RO.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001277 'romanian': 'ro_RO.ISO8859-2',
1278 'ru': 'ru_RU.ISO8859-5',
1279 'ru_ru': 'ru_RU.ISO8859-5',
1280 'ru_ru.cp1251': 'ru_RU.CP1251',
1281 'ru_ru.iso88595': 'ru_RU.ISO8859-5',
1282 'ru_ru.koi8r': 'ru_RU.KOI8-R',
1283 'ru_ru.microsoftcp1251': 'ru_RU.CP1251',
1284 'ru_ua': 'ru_UA.KOI8-U',
1285 'ru_ua.cp1251': 'ru_UA.CP1251',
1286 'ru_ua.koi8u': 'ru_UA.KOI8-U',
1287 'ru_ua.microsoftcp1251': 'ru_UA.CP1251',
1288 'rumanian': 'ro_RO.ISO8859-2',
1289 'russian': 'ru_RU.ISO8859-5',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001290 'rw': 'rw_RW.ISO8859-1',
1291 'rw_rw': 'rw_RW.ISO8859-1',
1292 'rw_rw.iso88591': 'rw_RW.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001293 'se_no': 'se_NO.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001294 'serbocroatian': 'sr_CS.ISO8859-2',
1295 'sh': 'sr_CS.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001296 'sh_hr': 'sh_HR.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001297 'sh_hr.iso88592': 'hr_HR.ISO8859-2',
1298 'sh_sp': 'sr_CS.ISO8859-2',
1299 'sh_yu': 'sr_CS.ISO8859-2',
1300 'si': 'si_LK.UTF-8',
1301 'si_lk': 'si_LK.UTF-8',
1302 'sinhala': 'si_LK.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001303 'sk': 'sk_SK.ISO8859-2',
1304 'sk_sk': 'sk_SK.ISO8859-2',
1305 'sk_sk.iso88592': 'sk_SK.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001306 'sl': 'sl_SI.ISO8859-2',
1307 'sl_cs': 'sl_CS.ISO8859-2',
1308 'sl_si': 'sl_SI.ISO8859-2',
1309 'sl_si.iso88592': 'sl_SI.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001310 'slovak': 'sk_SK.ISO8859-2',
1311 'slovene': 'sl_SI.ISO8859-2',
1312 'slovenian': 'sl_SI.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001313 'sp': 'sr_CS.ISO8859-5',
1314 'sp_yu': 'sr_CS.ISO8859-5',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001315 'spanish': 'es_ES.ISO8859-1',
1316 'spanish.iso88591': 'es_ES.ISO8859-1',
1317 'spanish_spain': 'es_ES.ISO8859-1',
1318 'spanish_spain.8859': 'es_ES.ISO8859-1',
1319 'sq': 'sq_AL.ISO8859-2',
1320 'sq_al': 'sq_AL.ISO8859-2',
1321 'sq_al.iso88592': 'sq_AL.ISO8859-2',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001322 'sr': 'sr_CS.ISO8859-5',
1323 'sr@cyrillic': 'sr_CS.ISO8859-5',
1324 'sr@latn': 'sr_CS.ISO8859-2',
1325 'sr_cs.iso88592': 'sr_CS.ISO8859-2',
1326 'sr_cs.iso88592@latn': 'sr_CS.ISO8859-2',
1327 'sr_cs.iso88595': 'sr_CS.ISO8859-5',
1328 'sr_cs.utf8@latn': 'sr_CS.UTF-8',
1329 'sr_cs@latn': 'sr_CS.ISO8859-2',
1330 'sr_sp': 'sr_CS.ISO8859-2',
1331 'sr_yu': 'sr_CS.ISO8859-5',
1332 'sr_yu.cp1251@cyrillic': 'sr_CS.CP1251',
1333 'sr_yu.iso88592': 'sr_CS.ISO8859-2',
1334 'sr_yu.iso88595': 'sr_CS.ISO8859-5',
1335 'sr_yu.iso88595@cyrillic': 'sr_CS.ISO8859-5',
1336 'sr_yu.microsoftcp1251@cyrillic': 'sr_CS.CP1251',
1337 'sr_yu.utf8@cyrillic': 'sr_CS.UTF-8',
1338 'sr_yu@cyrillic': 'sr_CS.ISO8859-5',
1339 'ss': 'ss_ZA.ISO8859-1',
1340 'ss_za': 'ss_ZA.ISO8859-1',
1341 'ss_za.iso88591': 'ss_ZA.ISO8859-1',
1342 'st': 'st_ZA.ISO8859-1',
1343 'st_za': 'st_ZA.ISO8859-1',
1344 'st_za.iso88591': 'st_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001345 'sv': 'sv_SE.ISO8859-1',
1346 'sv_fi': 'sv_FI.ISO8859-1',
1347 'sv_fi.iso88591': 'sv_FI.ISO8859-1',
1348 'sv_fi.iso885915': 'sv_FI.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001349 'sv_fi.iso885915@euro': 'sv_FI.ISO8859-15',
1350 'sv_fi.utf8@euro': 'sv_FI.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001351 'sv_fi@euro': 'sv_FI.ISO8859-15',
1352 'sv_se': 'sv_SE.ISO8859-1',
1353 'sv_se.88591': 'sv_SE.ISO8859-1',
1354 'sv_se.iso88591': 'sv_SE.ISO8859-1',
1355 'sv_se.iso885915': 'sv_SE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001356 'sv_se@euro': 'sv_SE.ISO8859-15',
1357 'swedish': 'sv_SE.ISO8859-1',
1358 'swedish.iso88591': 'sv_SE.ISO8859-1',
1359 'ta': 'ta_IN.TSCII-0',
1360 'ta_in': 'ta_IN.TSCII-0',
1361 'ta_in.tscii': 'ta_IN.TSCII-0',
1362 'ta_in.tscii0': 'ta_IN.TSCII-0',
1363 'tg': 'tg_TJ.KOI8-C',
1364 'tg_tj': 'tg_TJ.KOI8-C',
1365 'tg_tj.koi8c': 'tg_TJ.KOI8-C',
1366 'th': 'th_TH.ISO8859-11',
1367 'th_th': 'th_TH.ISO8859-11',
1368 'th_th.iso885911': 'th_TH.ISO8859-11',
1369 'th_th.tactis': 'th_TH.TIS620',
1370 'th_th.tis620': 'th_TH.TIS620',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001371 'thai': 'th_TH.ISO8859-11',
1372 'tl': 'tl_PH.ISO8859-1',
1373 'tl_ph': 'tl_PH.ISO8859-1',
1374 'tl_ph.iso88591': 'tl_PH.ISO8859-1',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001375 'tn': 'tn_ZA.ISO8859-15',
1376 'tn_za': 'tn_ZA.ISO8859-15',
1377 'tn_za.iso885915': 'tn_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001378 'tr': 'tr_TR.ISO8859-9',
1379 'tr_tr': 'tr_TR.ISO8859-9',
1380 'tr_tr.iso88599': 'tr_TR.ISO8859-9',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001381 'ts': 'ts_ZA.ISO8859-1',
1382 'ts_za': 'ts_ZA.ISO8859-1',
1383 'ts_za.iso88591': 'ts_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001384 'tt': 'tt_RU.TATAR-CYR',
1385 'tt_ru': 'tt_RU.TATAR-CYR',
1386 'tt_ru.koi8c': 'tt_RU.KOI8-C',
1387 'tt_ru.tatarcyr': 'tt_RU.TATAR-CYR',
1388 'turkish': 'tr_TR.ISO8859-9',
1389 'turkish.iso88599': 'tr_TR.ISO8859-9',
1390 'uk': 'uk_UA.KOI8-U',
1391 'uk_ua': 'uk_UA.KOI8-U',
1392 'uk_ua.cp1251': 'uk_UA.CP1251',
1393 'uk_ua.iso88595': 'uk_UA.ISO8859-5',
1394 'uk_ua.koi8u': 'uk_UA.KOI8-U',
1395 'uk_ua.microsoftcp1251': 'uk_UA.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001396 'univ': 'en_US.utf',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001397 'universal': 'en_US.utf',
1398 'universal.utf8@ucs4': 'en_US.UTF-8',
1399 'ur': 'ur_PK.CP1256',
1400 'ur_pk': 'ur_PK.CP1256',
1401 'ur_pk.cp1256': 'ur_PK.CP1256',
1402 'ur_pk.microsoftcp1256': 'ur_PK.CP1256',
1403 'uz': 'uz_UZ.UTF-8',
1404 'uz_uz': 'uz_UZ.UTF-8',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001405 'uz_uz.iso88591': 'uz_UZ.ISO8859-1',
1406 'uz_uz.utf8@cyrillic': 'uz_UZ.UTF-8',
1407 'uz_uz@cyrillic': 'uz_UZ.UTF-8',
1408 've': 've_ZA.UTF-8',
1409 've_za': 've_ZA.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001410 'vi': 'vi_VN.TCVN',
1411 'vi_vn': 'vi_VN.TCVN',
1412 'vi_vn.tcvn': 'vi_VN.TCVN',
1413 'vi_vn.tcvn5712': 'vi_VN.TCVN',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001414 'vi_vn.viscii': 'vi_VN.VISCII',
1415 'vi_vn.viscii111': 'vi_VN.VISCII',
1416 'wa': 'wa_BE.ISO8859-1',
1417 'wa_be': 'wa_BE.ISO8859-1',
1418 'wa_be.iso88591': 'wa_BE.ISO8859-1',
1419 'wa_be.iso885915': 'wa_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001420 'wa_be.iso885915@euro': 'wa_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001421 'wa_be@euro': 'wa_BE.ISO8859-15',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001422 'xh': 'xh_ZA.ISO8859-1',
1423 'xh_za': 'xh_ZA.ISO8859-1',
1424 'xh_za.iso88591': 'xh_ZA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001425 'yi': 'yi_US.CP1255',
1426 'yi_us': 'yi_US.CP1255',
1427 'yi_us.cp1255': 'yi_US.CP1255',
1428 'yi_us.microsoftcp1255': 'yi_US.CP1255',
1429 'zh': 'zh_CN.eucCN',
1430 'zh_cn': 'zh_CN.gb2312',
1431 'zh_cn.big5': 'zh_TW.big5',
1432 'zh_cn.euc': 'zh_CN.eucCN',
1433 'zh_cn.gb18030': 'zh_CN.gb18030',
1434 'zh_cn.gb2312': 'zh_CN.gb2312',
1435 'zh_cn.gbk': 'zh_CN.gbk',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001436 'zh_hk': 'zh_HK.big5hkscs',
1437 'zh_hk.big5': 'zh_HK.big5',
1438 'zh_hk.big5hkscs': 'zh_HK.big5hkscs',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001439 'zh_tw': 'zh_TW.big5',
1440 'zh_tw.big5': 'zh_TW.big5',
1441 'zh_tw.euc': 'zh_TW.eucTW',
Georg Brandlf08a9dd2008-06-10 16:57:31 +00001442 'zh_tw.euctw': 'zh_TW.eucTW',
1443 'zu': 'zu_ZA.ISO8859-1',
1444 'zu_za': 'zu_ZA.ISO8859-1',
1445 'zu_za.iso88591': 'zu_ZA.ISO8859-1',
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001446}
1447
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001448#
Georg Brandlb709c2c2006-01-20 09:07:35 +00001449# This maps Windows language identifiers to locale strings.
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001450#
Tim Peters777f1082006-01-20 20:03:24 +00001451# This list has been updated from
Georg Brandlb709c2c2006-01-20 09:07:35 +00001452# http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/nls_238z.asp
1453# to include every locale up to Windows XP.
Fredrik Lundh37a09822002-10-19 20:19:10 +00001454#
Georg Brandl5035c1c2006-01-20 13:38:26 +00001455# NOTE: this mapping is incomplete. If your language is missing, please
1456# submit a bug report to Python bug manager, which you can find via:
1457# http://www.python.org/dev/
1458# Make sure you include the missing language identifier and the suggested
1459# locale code.
1460#
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001461
1462windows_locale = {
Georg Brandlb709c2c2006-01-20 09:07:35 +00001463 0x0436: "af_ZA", # Afrikaans
1464 0x041c: "sq_AL", # Albanian
1465 0x0401: "ar_SA", # Arabic - Saudi Arabia
1466 0x0801: "ar_IQ", # Arabic - Iraq
1467 0x0c01: "ar_EG", # Arabic - Egypt
1468 0x1001: "ar_LY", # Arabic - Libya
1469 0x1401: "ar_DZ", # Arabic - Algeria
1470 0x1801: "ar_MA", # Arabic - Morocco
1471 0x1c01: "ar_TN", # Arabic - Tunisia
1472 0x2001: "ar_OM", # Arabic - Oman
1473 0x2401: "ar_YE", # Arabic - Yemen
1474 0x2801: "ar_SY", # Arabic - Syria
1475 0x2c01: "ar_JO", # Arabic - Jordan
1476 0x3001: "ar_LB", # Arabic - Lebanon
1477 0x3401: "ar_KW", # Arabic - Kuwait
1478 0x3801: "ar_AE", # Arabic - United Arab Emirates
1479 0x3c01: "ar_BH", # Arabic - Bahrain
1480 0x4001: "ar_QA", # Arabic - Qatar
1481 0x042b: "hy_AM", # Armenian
1482 0x042c: "az_AZ", # Azeri Latin
1483 0x082c: "az_AZ", # Azeri - Cyrillic
1484 0x042d: "eu_ES", # Basque
1485 0x0423: "be_BY", # Belarusian
1486 0x0445: "bn_IN", # Begali
1487 0x201a: "bs_BA", # Bosnian
1488 0x141a: "bs_BA", # Bosnian - Cyrillic
1489 0x047e: "br_FR", # Breton - France
1490 0x0402: "bg_BG", # Bulgarian
1491 0x0403: "ca_ES", # Catalan
1492 0x0004: "zh_CHS",# Chinese - Simplified
1493 0x0404: "zh_TW", # Chinese - Taiwan
1494 0x0804: "zh_CN", # Chinese - PRC
1495 0x0c04: "zh_HK", # Chinese - Hong Kong S.A.R.
1496 0x1004: "zh_SG", # Chinese - Singapore
1497 0x1404: "zh_MO", # Chinese - Macao S.A.R.
1498 0x7c04: "zh_CHT",# Chinese - Traditional
1499 0x041a: "hr_HR", # Croatian
1500 0x101a: "hr_BA", # Croatian - Bosnia
1501 0x0405: "cs_CZ", # Czech
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001502 0x0406: "da_DK", # Danish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001503 0x048c: "gbz_AF",# Dari - Afghanistan
1504 0x0465: "div_MV",# Divehi - Maldives
1505 0x0413: "nl_NL", # Dutch - The Netherlands
1506 0x0813: "nl_BE", # Dutch - Belgium
1507 0x0409: "en_US", # English - United States
1508 0x0809: "en_GB", # English - United Kingdom
1509 0x0c09: "en_AU", # English - Australia
1510 0x1009: "en_CA", # English - Canada
1511 0x1409: "en_NZ", # English - New Zealand
1512 0x1809: "en_IE", # English - Ireland
1513 0x1c09: "en_ZA", # English - South Africa
1514 0x2009: "en_JA", # English - Jamaica
1515 0x2409: "en_CB", # English - Carribbean
1516 0x2809: "en_BZ", # English - Belize
1517 0x2c09: "en_TT", # English - Trinidad
1518 0x3009: "en_ZW", # English - Zimbabwe
1519 0x3409: "en_PH", # English - Phillippines
1520 0x0425: "et_EE", # Estonian
1521 0x0438: "fo_FO", # Faroese
1522 0x0464: "fil_PH",# Filipino
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001523 0x040b: "fi_FI", # Finnish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001524 0x040c: "fr_FR", # French - France
1525 0x080c: "fr_BE", # French - Belgium
1526 0x0c0c: "fr_CA", # French - Canada
1527 0x100c: "fr_CH", # French - Switzerland
1528 0x140c: "fr_LU", # French - Luxembourg
1529 0x180c: "fr_MC", # French - Monaco
1530 0x0462: "fy_NL", # Frisian - Netherlands
1531 0x0456: "gl_ES", # Galician
1532 0x0437: "ka_GE", # Georgian
1533 0x0407: "de_DE", # German - Germany
1534 0x0807: "de_CH", # German - Switzerland
1535 0x0c07: "de_AT", # German - Austria
1536 0x1007: "de_LU", # German - Luxembourg
1537 0x1407: "de_LI", # German - Liechtenstein
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001538 0x0408: "el_GR", # Greek
Georg Brandlb709c2c2006-01-20 09:07:35 +00001539 0x0447: "gu_IN", # Gujarati
1540 0x040d: "he_IL", # Hebrew
1541 0x0439: "hi_IN", # Hindi
1542 0x040e: "hu_HU", # Hungarian
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001543 0x040f: "is_IS", # Icelandic
Georg Brandlb709c2c2006-01-20 09:07:35 +00001544 0x0421: "id_ID", # Indonesian
1545 0x045d: "iu_CA", # Inuktitut
1546 0x085d: "iu_CA", # Inuktitut - Latin
1547 0x083c: "ga_IE", # Irish - Ireland
1548 0x0434: "xh_ZA", # Xhosa - South Africa
1549 0x0435: "zu_ZA", # Zulu
1550 0x0410: "it_IT", # Italian - Italy
1551 0x0810: "it_CH", # Italian - Switzerland
1552 0x0411: "ja_JP", # Japanese
1553 0x044b: "kn_IN", # Kannada - India
1554 0x043f: "kk_KZ", # Kazakh
1555 0x0457: "kok_IN",# Konkani
1556 0x0412: "ko_KR", # Korean
1557 0x0440: "ky_KG", # Kyrgyz
1558 0x0426: "lv_LV", # Latvian
1559 0x0427: "lt_LT", # Lithuanian
1560 0x046e: "lb_LU", # Luxembourgish
1561 0x042f: "mk_MK", # FYRO Macedonian
1562 0x043e: "ms_MY", # Malay - Malaysia
1563 0x083e: "ms_BN", # Malay - Brunei
1564 0x044c: "ml_IN", # Malayalam - India
1565 0x043a: "mt_MT", # Maltese
1566 0x0481: "mi_NZ", # Maori
1567 0x047a: "arn_CL",# Mapudungun
1568 0x044e: "mr_IN", # Marathi
1569 0x047c: "moh_CA",# Mohawk - Canada
1570 0x0450: "mn_MN", # Mongolian
1571 0x0461: "ne_NP", # Nepali
1572 0x0414: "nb_NO", # Norwegian - Bokmal
1573 0x0814: "nn_NO", # Norwegian - Nynorsk
1574 0x0482: "oc_FR", # Occitan - France
1575 0x0448: "or_IN", # Oriya - India
1576 0x0463: "ps_AF", # Pashto - Afghanistan
1577 0x0429: "fa_IR", # Persian
1578 0x0415: "pl_PL", # Polish
1579 0x0416: "pt_BR", # Portuguese - Brazil
1580 0x0816: "pt_PT", # Portuguese - Portugal
1581 0x0446: "pa_IN", # Punjabi
1582 0x046b: "quz_BO",# Quechua (Bolivia)
1583 0x086b: "quz_EC",# Quechua (Ecuador)
1584 0x0c6b: "quz_PE",# Quechua (Peru)
1585 0x0418: "ro_RO", # Romanian - Romania
1586 0x0417: "rm_CH", # Raeto-Romanese
1587 0x0419: "ru_RU", # Russian
1588 0x243b: "smn_FI",# Sami Finland
1589 0x103b: "smj_NO",# Sami Norway
1590 0x143b: "smj_SE",# Sami Sweden
1591 0x043b: "se_NO", # Sami Northern Norway
1592 0x083b: "se_SE", # Sami Northern Sweden
1593 0x0c3b: "se_FI", # Sami Northern Finland
1594 0x203b: "sms_FI",# Sami Skolt
1595 0x183b: "sma_NO",# Sami Southern Norway
1596 0x1c3b: "sma_SE",# Sami Southern Sweden
1597 0x044f: "sa_IN", # Sanskrit
1598 0x0c1a: "sr_SP", # Serbian - Cyrillic
1599 0x1c1a: "sr_BA", # Serbian - Bosnia Cyrillic
1600 0x081a: "sr_SP", # Serbian - Latin
1601 0x181a: "sr_BA", # Serbian - Bosnia Latin
1602 0x046c: "ns_ZA", # Northern Sotho
1603 0x0432: "tn_ZA", # Setswana - Southern Africa
1604 0x041b: "sk_SK", # Slovak
1605 0x0424: "sl_SI", # Slovenian
1606 0x040a: "es_ES", # Spanish - Spain
1607 0x080a: "es_MX", # Spanish - Mexico
1608 0x0c0a: "es_ES", # Spanish - Spain (Modern)
1609 0x100a: "es_GT", # Spanish - Guatemala
1610 0x140a: "es_CR", # Spanish - Costa Rica
1611 0x180a: "es_PA", # Spanish - Panama
1612 0x1c0a: "es_DO", # Spanish - Dominican Republic
1613 0x200a: "es_VE", # Spanish - Venezuela
1614 0x240a: "es_CO", # Spanish - Colombia
1615 0x280a: "es_PE", # Spanish - Peru
1616 0x2c0a: "es_AR", # Spanish - Argentina
1617 0x300a: "es_EC", # Spanish - Ecuador
1618 0x340a: "es_CL", # Spanish - Chile
1619 0x380a: "es_UR", # Spanish - Uruguay
1620 0x3c0a: "es_PY", # Spanish - Paraguay
1621 0x400a: "es_BO", # Spanish - Bolivia
1622 0x440a: "es_SV", # Spanish - El Salvador
1623 0x480a: "es_HN", # Spanish - Honduras
1624 0x4c0a: "es_NI", # Spanish - Nicaragua
1625 0x500a: "es_PR", # Spanish - Puerto Rico
1626 0x0441: "sw_KE", # Swahili
1627 0x041d: "sv_SE", # Swedish - Sweden
1628 0x081d: "sv_FI", # Swedish - Finland
1629 0x045a: "syr_SY",# Syriac
1630 0x0449: "ta_IN", # Tamil
1631 0x0444: "tt_RU", # Tatar
1632 0x044a: "te_IN", # Telugu
1633 0x041e: "th_TH", # Thai
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001634 0x041f: "tr_TR", # Turkish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001635 0x0422: "uk_UA", # Ukrainian
1636 0x0420: "ur_PK", # Urdu
1637 0x0820: "ur_IN", # Urdu - India
1638 0x0443: "uz_UZ", # Uzbek - Latin
1639 0x0843: "uz_UZ", # Uzbek - Cyrillic
1640 0x042a: "vi_VN", # Vietnamese
1641 0x0452: "cy_GB", # Welsh
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001642}
1643
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001644def _print_locale():
1645
1646 """ Test function.
1647 """
1648 categories = {}
1649 def _init_categories(categories=categories):
1650 for k,v in globals().items():
1651 if k[:3] == 'LC_':
1652 categories[k] = v
1653 _init_categories()
1654 del categories['LC_ALL']
1655
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001656 print('Locale defaults as determined by getdefaultlocale():')
1657 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001658 lang, enc = getdefaultlocale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001659 print('Language: ', lang or '(undefined)')
1660 print('Encoding: ', enc or '(undefined)')
1661 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001662
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001663 print('Locale settings on startup:')
1664 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001665 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001666 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001667 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001668 print(' Language: ', lang or '(undefined)')
1669 print(' Encoding: ', enc or '(undefined)')
1670 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001671
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001672 print()
1673 print('Locale settings after calling resetlocale():')
1674 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001675 resetlocale()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001676 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001677 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001678 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001679 print(' Language: ', lang or '(undefined)')
1680 print(' Encoding: ', enc or '(undefined)')
1681 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001682
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001683 try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001684 setlocale(LC_ALL, "")
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001685 except:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001686 print('NOTE:')
1687 print('setlocale(LC_ALL, "") does not support the default locale')
1688 print('given in the OS environment variables.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001689 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001690 print()
1691 print('Locale settings after calling setlocale(LC_ALL, ""):')
1692 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001693 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001694 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001695 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001696 print(' Language: ', lang or '(undefined)')
1697 print(' Encoding: ', enc or '(undefined)')
1698 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001699
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001700###
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001701
Tim Peters1baf8292001-01-24 10:13:46 +00001702try:
1703 LC_MESSAGES
Skip Montanaro0897f0c2002-03-25 21:40:36 +00001704except NameError:
Tim Peters1baf8292001-01-24 10:13:46 +00001705 pass
1706else:
1707 __all__.append("LC_MESSAGES")
1708
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001709if __name__=='__main__':
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001710 print('Locale aliasing:')
1711 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001712 _print_locale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001713 print()
1714 print('Number formatting:')
1715 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001716 _test()