blob: eb8b33d5ec01b45cdd9850cb3dbfa370eb0323ac [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#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000629locale_alias = {
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000630 'a3': 'a3_AZ.KOI8-C',
631 'a3_az': 'a3_AZ.KOI8-C',
632 'a3_az.koi8c': 'a3_AZ.KOI8-C',
633 'af': 'af_ZA.ISO8859-1',
634 'af_za': 'af_ZA.ISO8859-1',
635 'af_za.iso88591': 'af_ZA.ISO8859-1',
636 'am': 'am_ET.UTF-8',
637 'american': 'en_US.ISO8859-1',
638 'american.iso88591': 'en_US.ISO8859-1',
639 'ar': 'ar_AA.ISO8859-6',
640 'ar_aa': 'ar_AA.ISO8859-6',
641 'ar_aa.iso88596': 'ar_AA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000642 'ar_ae': 'ar_AE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000643 'ar_bh': 'ar_BH.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000644 'ar_dz': 'ar_DZ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000645 'ar_eg': 'ar_EG.ISO8859-6',
646 'ar_eg.iso88596': 'ar_EG.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000647 'ar_iq': 'ar_IQ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000648 'ar_jo': 'ar_JO.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000649 'ar_kw': 'ar_KW.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000650 'ar_lb': 'ar_LB.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000651 'ar_ly': 'ar_LY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000652 'ar_ma': 'ar_MA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000653 'ar_om': 'ar_OM.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000654 'ar_qa': 'ar_QA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000655 'ar_sa': 'ar_SA.ISO8859-6',
656 'ar_sa.iso88596': 'ar_SA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000657 'ar_sd': 'ar_SD.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000658 'ar_sy': 'ar_SY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000659 'ar_tn': 'ar_TN.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000660 'ar_ye': 'ar_YE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000661 'arabic': 'ar_AA.ISO8859-6',
662 'arabic.iso88596': 'ar_AA.ISO8859-6',
663 'az': 'az_AZ.ISO8859-9E',
664 'az_az': 'az_AZ.ISO8859-9E',
665 'az_az.iso88599e': 'az_AZ.ISO8859-9E',
666 'be': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000667 'be_by': 'be_BY.CP1251',
668 'be_by.cp1251': 'be_BY.CP1251',
669 'be_by.microsoftcp1251': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000670 'bg': 'bg_BG.CP1251',
671 'bg_bg': 'bg_BG.CP1251',
672 'bg_bg.cp1251': 'bg_BG.CP1251',
673 'bg_bg.iso88595': 'bg_BG.ISO8859-5',
674 'bg_bg.koi8r': 'bg_BG.KOI8-R',
675 'bg_bg.microsoftcp1251': 'bg_BG.CP1251',
676 'bokmal': 'nb_NO.ISO8859-1',
677 'bokm\xe5l': 'nb_NO.ISO8859-1',
678 'br': 'br_FR.ISO8859-1',
679 'br_fr': 'br_FR.ISO8859-1',
680 'br_fr.iso88591': 'br_FR.ISO8859-1',
681 'br_fr.iso885914': 'br_FR.ISO8859-14',
682 'br_fr.iso885915': 'br_FR.ISO8859-15',
683 'br_fr@euro': 'br_FR.ISO8859-15',
684 'bulgarian': 'bg_BG.CP1251',
685 'c': 'C',
686 'c-french': 'fr_CA.ISO8859-1',
687 'c-french.iso88591': 'fr_CA.ISO8859-1',
688 'c.en': 'C',
689 'c.iso88591': 'en_US.ISO8859-1',
690 'c_c': 'C',
691 'c_c.c': 'C',
692 'ca': 'ca_ES.ISO8859-1',
693 'ca_es': 'ca_ES.ISO8859-1',
694 'ca_es.iso88591': 'ca_ES.ISO8859-1',
695 'ca_es.iso885915': 'ca_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000696 'ca_es@euro': 'ca_ES.ISO8859-15',
697 'catalan': 'ca_ES.ISO8859-1',
698 'cextend': 'en_US.ISO8859-1',
699 'cextend.en': 'en_US.ISO8859-1',
700 'chinese-s': 'zh_CN.eucCN',
701 'chinese-t': 'zh_TW.eucTW',
702 'croatian': 'hr_HR.ISO8859-2',
703 'cs': 'cs_CZ.ISO8859-2',
704 'cs_cs': 'cs_CZ.ISO8859-2',
705 'cs_cs.iso88592': 'cs_CZ.ISO8859-2',
706 'cs_cz': 'cs_CZ.ISO8859-2',
707 'cs_cz.iso88592': 'cs_CZ.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000708 'cy': 'cy_GB.ISO8859-1',
709 'cy_gb': 'cy_GB.ISO8859-1',
710 'cy_gb.iso88591': 'cy_GB.ISO8859-1',
711 'cy_gb.iso885914': 'cy_GB.ISO8859-14',
712 'cy_gb.iso885915': 'cy_GB.ISO8859-15',
713 'cy_gb@euro': 'cy_GB.ISO8859-15',
714 'cz': 'cs_CZ.ISO8859-2',
715 'cz_cz': 'cs_CZ.ISO8859-2',
716 'czech': 'cs_CZ.ISO8859-2',
717 'da': 'da_DK.ISO8859-1',
718 'da_dk': 'da_DK.ISO8859-1',
719 'da_dk.88591': 'da_DK.ISO8859-1',
720 'da_dk.885915': 'da_DK.ISO8859-15',
721 'da_dk.iso88591': 'da_DK.ISO8859-1',
722 'da_dk.iso885915': 'da_DK.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000723 'da_dk@euro': 'da_DK.ISO8859-15',
724 'danish': 'da_DK.ISO8859-1',
725 'danish.iso88591': 'da_DK.ISO8859-1',
726 'dansk': 'da_DK.ISO8859-1',
727 'de': 'de_DE.ISO8859-1',
728 'de_at': 'de_AT.ISO8859-1',
729 'de_at.iso88591': 'de_AT.ISO8859-1',
730 'de_at.iso885915': 'de_AT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000731 'de_at@euro': 'de_AT.ISO8859-15',
732 'de_be': 'de_BE.ISO8859-1',
733 'de_be.iso88591': 'de_BE.ISO8859-1',
734 'de_be.iso885915': 'de_BE.ISO8859-15',
735 'de_be@euro': 'de_BE.ISO8859-15',
736 'de_ch': 'de_CH.ISO8859-1',
737 'de_ch.iso88591': 'de_CH.ISO8859-1',
738 'de_ch.iso885915': 'de_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000739 'de_ch@euro': 'de_CH.ISO8859-15',
740 'de_de': 'de_DE.ISO8859-1',
741 'de_de.88591': 'de_DE.ISO8859-1',
742 'de_de.885915': 'de_DE.ISO8859-15',
743 'de_de.885915@euro': 'de_DE.ISO8859-15',
744 'de_de.iso88591': 'de_DE.ISO8859-1',
745 'de_de.iso885915': 'de_DE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000746 'de_de@euro': 'de_DE.ISO8859-15',
747 'de_lu': 'de_LU.ISO8859-1',
748 'de_lu.iso88591': 'de_LU.ISO8859-1',
749 'de_lu.iso885915': 'de_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000750 'de_lu@euro': 'de_LU.ISO8859-15',
751 'deutsch': 'de_DE.ISO8859-1',
752 'dutch': 'nl_NL.ISO8859-1',
753 'dutch.iso88591': 'nl_BE.ISO8859-1',
754 'ee': 'ee_EE.ISO8859-4',
755 'ee_ee': 'ee_EE.ISO8859-4',
756 'ee_ee.iso88594': 'ee_EE.ISO8859-4',
757 'eesti': 'et_EE.ISO8859-1',
758 'el': 'el_GR.ISO8859-7',
759 'el_gr': 'el_GR.ISO8859-7',
760 'el_gr.iso88597': 'el_GR.ISO8859-7',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000761 'el_gr@euro': 'el_GR.ISO8859-15',
762 'en': 'en_US.ISO8859-1',
763 'en.iso88591': 'en_US.ISO8859-1',
764 'en_au': 'en_AU.ISO8859-1',
765 'en_au.iso88591': 'en_AU.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000766 'en_be': 'en_BE.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000767 'en_be@euro': 'en_BE.ISO8859-15',
768 'en_bw': 'en_BW.ISO8859-1',
769 'en_ca': 'en_CA.ISO8859-1',
770 'en_ca.iso88591': 'en_CA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000771 'en_gb': 'en_GB.ISO8859-1',
772 'en_gb.88591': 'en_GB.ISO8859-1',
773 'en_gb.iso88591': 'en_GB.ISO8859-1',
774 'en_gb.iso885915': 'en_GB.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000775 'en_gb@euro': 'en_GB.ISO8859-15',
776 'en_hk': 'en_HK.ISO8859-1',
777 'en_ie': 'en_IE.ISO8859-1',
778 'en_ie.iso88591': 'en_IE.ISO8859-1',
779 'en_ie.iso885915': 'en_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000780 'en_ie@euro': 'en_IE.ISO8859-15',
781 'en_in': 'en_IN.ISO8859-1',
782 'en_nz': 'en_NZ.ISO8859-1',
783 'en_nz.iso88591': 'en_NZ.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000784 'en_ph': 'en_PH.ISO8859-1',
785 'en_sg': 'en_SG.ISO8859-1',
786 'en_uk': 'en_GB.ISO8859-1',
787 'en_us': 'en_US.ISO8859-1',
788 'en_us.88591': 'en_US.ISO8859-1',
789 'en_us.885915': 'en_US.ISO8859-15',
790 'en_us.iso88591': 'en_US.ISO8859-1',
791 'en_us.iso885915': 'en_US.ISO8859-15',
792 'en_us.iso885915@euro': 'en_US.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000793 'en_us@euro': 'en_US.ISO8859-15',
794 'en_us@euro@euro': 'en_US.ISO8859-15',
795 'en_za': 'en_ZA.ISO8859-1',
796 'en_za.88591': 'en_ZA.ISO8859-1',
797 'en_za.iso88591': 'en_ZA.ISO8859-1',
798 'en_za.iso885915': 'en_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000799 'en_za@euro': 'en_ZA.ISO8859-15',
800 'en_zw': 'en_ZW.ISO8859-1',
801 'eng_gb': 'en_GB.ISO8859-1',
802 'eng_gb.8859': 'en_GB.ISO8859-1',
803 'english': 'en_EN.ISO8859-1',
804 'english.iso88591': 'en_EN.ISO8859-1',
805 'english_uk': 'en_GB.ISO8859-1',
806 'english_uk.8859': 'en_GB.ISO8859-1',
807 'english_united-states': 'en_US.ISO8859-1',
808 'english_united-states.437': 'C',
809 'english_us': 'en_US.ISO8859-1',
810 'english_us.8859': 'en_US.ISO8859-1',
811 'english_us.ascii': 'en_US.ISO8859-1',
812 'eo': 'eo_XX.ISO8859-3',
813 'eo_eo': 'eo_EO.ISO8859-3',
814 'eo_eo.iso88593': 'eo_EO.ISO8859-3',
815 'eo_xx': 'eo_XX.ISO8859-3',
816 'eo_xx.iso88593': 'eo_XX.ISO8859-3',
817 'es': 'es_ES.ISO8859-1',
818 'es_ar': 'es_AR.ISO8859-1',
819 'es_ar.iso88591': 'es_AR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000820 'es_bo': 'es_BO.ISO8859-1',
821 'es_bo.iso88591': 'es_BO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000822 'es_cl': 'es_CL.ISO8859-1',
823 'es_cl.iso88591': 'es_CL.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000824 'es_co': 'es_CO.ISO8859-1',
825 'es_co.iso88591': 'es_CO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000826 'es_cr': 'es_CR.ISO8859-1',
827 'es_cr.iso88591': 'es_CR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000828 'es_do': 'es_DO.ISO8859-1',
829 'es_do.iso88591': 'es_DO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000830 'es_ec': 'es_EC.ISO8859-1',
831 'es_ec.iso88591': 'es_EC.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000832 'es_es': 'es_ES.ISO8859-1',
833 'es_es.88591': 'es_ES.ISO8859-1',
834 'es_es.iso88591': 'es_ES.ISO8859-1',
835 'es_es.iso885915': 'es_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000836 'es_es@euro': 'es_ES.ISO8859-15',
837 'es_gt': 'es_GT.ISO8859-1',
838 'es_gt.iso88591': 'es_GT.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000839 'es_hn': 'es_HN.ISO8859-1',
840 'es_hn.iso88591': 'es_HN.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000841 'es_mx': 'es_MX.ISO8859-1',
842 'es_mx.iso88591': 'es_MX.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000843 'es_ni': 'es_NI.ISO8859-1',
844 'es_ni.iso88591': 'es_NI.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000845 'es_pa': 'es_PA.ISO8859-1',
846 'es_pa.iso88591': 'es_PA.ISO8859-1',
847 'es_pa.iso885915': 'es_PA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000848 'es_pa@euro': 'es_PA.ISO8859-15',
849 'es_pe': 'es_PE.ISO8859-1',
850 'es_pe.iso88591': 'es_PE.ISO8859-1',
851 'es_pe.iso885915': 'es_PE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000852 'es_pe@euro': 'es_PE.ISO8859-15',
853 'es_pr': 'es_PR.ISO8859-1',
854 'es_pr.iso88591': 'es_PR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000855 'es_py': 'es_PY.ISO8859-1',
856 'es_py.iso88591': 'es_PY.ISO8859-1',
857 'es_py.iso885915': 'es_PY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000858 'es_py@euro': 'es_PY.ISO8859-15',
859 'es_sv': 'es_SV.ISO8859-1',
860 'es_sv.iso88591': 'es_SV.ISO8859-1',
861 'es_sv.iso885915': 'es_SV.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000862 'es_sv@euro': 'es_SV.ISO8859-15',
863 'es_us': 'es_US.ISO8859-1',
864 'es_uy': 'es_UY.ISO8859-1',
865 'es_uy.iso88591': 'es_UY.ISO8859-1',
866 'es_uy.iso885915': 'es_UY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000867 'es_uy@euro': 'es_UY.ISO8859-15',
868 'es_ve': 'es_VE.ISO8859-1',
869 'es_ve.iso88591': 'es_VE.ISO8859-1',
870 'es_ve.iso885915': 'es_VE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000871 'es_ve@euro': 'es_VE.ISO8859-15',
872 'estonian': 'et_EE.ISO8859-1',
873 'et': 'et_EE.ISO8859-15',
874 'et_ee': 'et_EE.ISO8859-15',
875 'et_ee.iso88591': 'et_EE.ISO8859-1',
876 'et_ee.iso885913': 'et_EE.ISO8859-13',
877 'et_ee.iso885915': 'et_EE.ISO8859-15',
878 'et_ee.iso88594': 'et_EE.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000879 'et_ee@euro': 'et_EE.ISO8859-15',
880 'eu': 'eu_ES.ISO8859-1',
881 'eu_es': 'eu_ES.ISO8859-1',
882 'eu_es.iso88591': 'eu_ES.ISO8859-1',
883 'eu_es.iso885915': 'eu_ES.ISO8859-15',
884 'eu_es@euro': 'eu_ES.ISO8859-15',
885 'fa': 'fa_IR.UTF-8',
886 'fa_ir': 'fa_IR.UTF-8',
887 'fa_ir.isiri3342': 'fa_IR.ISIRI-3342',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000888 'fi': 'fi_FI.ISO8859-15',
889 'fi_fi': 'fi_FI.ISO8859-15',
890 'fi_fi.88591': 'fi_FI.ISO8859-1',
891 'fi_fi.iso88591': 'fi_FI.ISO8859-1',
892 'fi_fi.iso885915': 'fi_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000893 'fi_fi.utf8@euro': 'fi_FI.UTF-8',
894 'fi_fi@euro': 'fi_FI.ISO8859-15',
895 'finnish': 'fi_FI.ISO8859-1',
896 'finnish.iso88591': 'fi_FI.ISO8859-1',
897 'fo': 'fo_FO.ISO8859-1',
898 'fo_fo': 'fo_FO.ISO8859-1',
899 'fo_fo.iso88591': 'fo_FO.ISO8859-1',
900 'fo_fo.iso885915': 'fo_FO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000901 'fo_fo@euro': 'fo_FO.ISO8859-15',
902 'fr': 'fr_FR.ISO8859-1',
903 'fr_be': 'fr_BE.ISO8859-1',
904 'fr_be.88591': 'fr_BE.ISO8859-1',
905 'fr_be.iso88591': 'fr_BE.ISO8859-1',
906 'fr_be.iso885915': 'fr_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000907 'fr_be@euro': 'fr_BE.ISO8859-15',
908 'fr_ca': 'fr_CA.ISO8859-1',
909 'fr_ca.88591': 'fr_CA.ISO8859-1',
910 'fr_ca.iso88591': 'fr_CA.ISO8859-1',
911 'fr_ca.iso885915': 'fr_CA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000912 'fr_ca@euro': 'fr_CA.ISO8859-15',
913 'fr_ch': 'fr_CH.ISO8859-1',
914 'fr_ch.88591': 'fr_CH.ISO8859-1',
915 'fr_ch.iso88591': 'fr_CH.ISO8859-1',
916 'fr_ch.iso885915': 'fr_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000917 'fr_ch@euro': 'fr_CH.ISO8859-15',
918 'fr_fr': 'fr_FR.ISO8859-1',
919 'fr_fr.88591': 'fr_FR.ISO8859-1',
920 'fr_fr.iso88591': 'fr_FR.ISO8859-1',
921 'fr_fr.iso885915': 'fr_FR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000922 'fr_fr@euro': 'fr_FR.ISO8859-15',
923 'fr_lu': 'fr_LU.ISO8859-1',
924 'fr_lu.88591': 'fr_LU.ISO8859-1',
925 'fr_lu.iso88591': 'fr_LU.ISO8859-1',
926 'fr_lu.iso885915': 'fr_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000927 'fr_lu@euro': 'fr_LU.ISO8859-15',
928 'fran\xe7ais': 'fr_FR.ISO8859-1',
929 'fre_fr': 'fr_FR.ISO8859-1',
930 'fre_fr.8859': 'fr_FR.ISO8859-1',
931 'french': 'fr_FR.ISO8859-1',
932 'french.iso88591': 'fr_CH.ISO8859-1',
933 'french_france': 'fr_FR.ISO8859-1',
934 'french_france.8859': 'fr_FR.ISO8859-1',
935 'ga': 'ga_IE.ISO8859-1',
936 'ga_ie': 'ga_IE.ISO8859-1',
937 'ga_ie.iso88591': 'ga_IE.ISO8859-1',
938 'ga_ie.iso885914': 'ga_IE.ISO8859-14',
939 'ga_ie.iso885915': 'ga_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000940 'ga_ie@euro': 'ga_IE.ISO8859-15',
941 'galego': 'gl_ES.ISO8859-1',
942 'galician': 'gl_ES.ISO8859-1',
943 'gd': 'gd_GB.ISO8859-1',
944 'gd_gb': 'gd_GB.ISO8859-1',
945 'gd_gb.iso88591': 'gd_GB.ISO8859-1',
946 'gd_gb.iso885914': 'gd_GB.ISO8859-14',
947 'gd_gb.iso885915': 'gd_GB.ISO8859-15',
948 'gd_gb@euro': 'gd_GB.ISO8859-15',
949 'ger_de': 'de_DE.ISO8859-1',
950 'ger_de.8859': 'de_DE.ISO8859-1',
951 'german': 'de_DE.ISO8859-1',
952 'german.iso88591': 'de_CH.ISO8859-1',
953 'german_germany': 'de_DE.ISO8859-1',
954 'german_germany.8859': 'de_DE.ISO8859-1',
955 'gl': 'gl_ES.ISO8859-1',
956 'gl_es': 'gl_ES.ISO8859-1',
957 'gl_es.iso88591': 'gl_ES.ISO8859-1',
958 'gl_es.iso885915': 'gl_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000959 'gl_es@euro': 'gl_ES.ISO8859-15',
960 'greek': 'el_GR.ISO8859-7',
961 'greek.iso88597': 'el_GR.ISO8859-7',
962 'gv': 'gv_GB.ISO8859-1',
963 'gv_gb': 'gv_GB.ISO8859-1',
964 'gv_gb.iso88591': 'gv_GB.ISO8859-1',
965 'gv_gb.iso885914': 'gv_GB.ISO8859-14',
966 'gv_gb.iso885915': 'gv_GB.ISO8859-15',
967 'gv_gb@euro': 'gv_GB.ISO8859-15',
968 'he': 'he_IL.ISO8859-8',
969 'he_il': 'he_IL.ISO8859-8',
970 'he_il.cp1255': 'he_IL.CP1255',
971 'he_il.iso88598': 'he_IL.ISO8859-8',
972 'he_il.microsoftcp1255': 'he_IL.CP1255',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000973 'hebrew': 'iw_IL.ISO8859-8',
974 'hebrew.iso88598': 'iw_IL.ISO8859-8',
975 'hi': 'hi_IN.ISCII-DEV',
976 'hi_in': 'hi_IN.ISCII-DEV',
977 'hi_in.isciidev': 'hi_IN.ISCII-DEV',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000978 'hr': 'hr_HR.ISO8859-2',
979 'hr_hr': 'hr_HR.ISO8859-2',
980 'hr_hr.iso88592': 'hr_HR.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000981 'hrvatski': 'hr_HR.ISO8859-2',
982 'hu': 'hu_HU.ISO8859-2',
983 'hu_hu': 'hu_HU.ISO8859-2',
984 'hu_hu.iso88592': 'hu_HU.ISO8859-2',
985 'hungarian': 'hu_HU.ISO8859-2',
986 'icelandic': 'is_IS.ISO8859-1',
987 'icelandic.iso88591': 'is_IS.ISO8859-1',
988 'id': 'id_ID.ISO8859-1',
989 'id_id': 'id_ID.ISO8859-1',
990 'in': 'id_ID.ISO8859-1',
991 'in_id': 'id_ID.ISO8859-1',
992 'is': 'is_IS.ISO8859-1',
993 'is_is': 'is_IS.ISO8859-1',
994 'is_is.iso88591': 'is_IS.ISO8859-1',
995 'is_is.iso885915': 'is_IS.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000996 'is_is@euro': 'is_IS.ISO8859-15',
997 'iso-8859-1': 'en_US.ISO8859-1',
998 'iso-8859-15': 'en_US.ISO8859-15',
999 'iso8859-1': 'en_US.ISO8859-1',
1000 'iso8859-15': 'en_US.ISO8859-15',
1001 'iso_8859_1': 'en_US.ISO8859-1',
1002 'iso_8859_15': 'en_US.ISO8859-15',
1003 'it': 'it_IT.ISO8859-1',
1004 'it_ch': 'it_CH.ISO8859-1',
1005 'it_ch.iso88591': 'it_CH.ISO8859-1',
1006 'it_ch.iso885915': 'it_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001007 'it_ch@euro': 'it_CH.ISO8859-15',
1008 'it_it': 'it_IT.ISO8859-1',
1009 'it_it.88591': 'it_IT.ISO8859-1',
1010 'it_it.iso88591': 'it_IT.ISO8859-1',
1011 'it_it.iso885915': 'it_IT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001012 'it_it@euro': 'it_IT.ISO8859-15',
1013 'italian': 'it_IT.ISO8859-1',
1014 'italian.iso88591': 'it_IT.ISO8859-1',
1015 'iu': 'iu_CA.NUNACOM-8',
1016 'iu_ca': 'iu_CA.NUNACOM-8',
1017 'iu_ca.nunacom8': 'iu_CA.NUNACOM-8',
1018 'iw': 'he_IL.ISO8859-8',
1019 'iw_il': 'he_IL.ISO8859-8',
1020 'iw_il.iso88598': 'he_IL.ISO8859-8',
1021 'ja': 'ja_JP.eucJP',
1022 'ja.jis': 'ja_JP.JIS7',
1023 'ja.sjis': 'ja_JP.SJIS',
1024 'ja_jp': 'ja_JP.eucJP',
1025 'ja_jp.ajec': 'ja_JP.eucJP',
1026 'ja_jp.euc': 'ja_JP.eucJP',
1027 'ja_jp.eucjp': 'ja_JP.eucJP',
1028 'ja_jp.iso-2022-jp': 'ja_JP.JIS7',
1029 'ja_jp.iso2022jp': 'ja_JP.JIS7',
1030 'ja_jp.jis': 'ja_JP.JIS7',
1031 'ja_jp.jis7': 'ja_JP.JIS7',
1032 'ja_jp.mscode': 'ja_JP.SJIS',
1033 'ja_jp.sjis': 'ja_JP.SJIS',
1034 'ja_jp.ujis': 'ja_JP.eucJP',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001035 'japan': 'ja_JP.eucJP',
1036 'japanese': 'ja_JP.eucJP',
1037 'japanese-euc': 'ja_JP.eucJP',
1038 'japanese.euc': 'ja_JP.eucJP',
1039 'japanese.sjis': 'ja_JP.SJIS',
1040 'jp_jp': 'ja_JP.eucJP',
1041 'ka': 'ka_GE.GEORGIAN-ACADEMY',
1042 'ka_ge': 'ka_GE.GEORGIAN-ACADEMY',
1043 'ka_ge.georgianacademy': 'ka_GE.GEORGIAN-ACADEMY',
1044 'ka_ge.georgianps': 'ka_GE.GEORGIAN-PS',
1045 'ka_ge.georgianrs': 'ka_GE.GEORGIAN-ACADEMY',
1046 'kl': 'kl_GL.ISO8859-1',
1047 'kl_gl': 'kl_GL.ISO8859-1',
1048 'kl_gl.iso88591': 'kl_GL.ISO8859-1',
1049 'kl_gl.iso885915': 'kl_GL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001050 'kl_gl@euro': 'kl_GL.ISO8859-15',
1051 'ko': 'ko_KR.eucKR',
1052 'ko_kr': 'ko_KR.eucKR',
1053 'ko_kr.euc': 'ko_KR.eucKR',
1054 'ko_kr.euckr': 'ko_KR.eucKR',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001055 'korean': 'ko_KR.eucKR',
1056 'korean.euc': 'ko_KR.eucKR',
1057 'kw': 'kw_GB.ISO8859-1',
1058 'kw_gb': 'kw_GB.ISO8859-1',
1059 'kw_gb.iso88591': 'kw_GB.ISO8859-1',
1060 'kw_gb.iso885914': 'kw_GB.ISO8859-14',
1061 'kw_gb.iso885915': 'kw_GB.ISO8859-15',
1062 'kw_gb@euro': 'kw_GB.ISO8859-15',
1063 'lithuanian': 'lt_LT.ISO8859-13',
1064 'lo': 'lo_LA.MULELAO-1',
1065 'lo_la': 'lo_LA.MULELAO-1',
1066 'lo_la.cp1133': 'lo_LA.IBM-CP1133',
1067 'lo_la.ibmcp1133': 'lo_LA.IBM-CP1133',
1068 'lo_la.mulelao1': 'lo_LA.MULELAO-1',
1069 'lt': 'lt_LT.ISO8859-13',
1070 'lt_lt': 'lt_LT.ISO8859-13',
1071 'lt_lt.iso885913': 'lt_LT.ISO8859-13',
1072 'lt_lt.iso88594': 'lt_LT.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001073 'lv': 'lv_LV.ISO8859-13',
1074 'lv_lv': 'lv_LV.ISO8859-13',
1075 'lv_lv.iso885913': 'lv_LV.ISO8859-13',
1076 'lv_lv.iso88594': 'lv_LV.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001077 'mi': 'mi_NZ.ISO8859-1',
1078 'mi_nz': 'mi_NZ.ISO8859-1',
1079 'mi_nz.iso88591': 'mi_NZ.ISO8859-1',
1080 'mk': 'mk_MK.ISO8859-5',
1081 'mk_mk': 'mk_MK.ISO8859-5',
1082 'mk_mk.cp1251': 'mk_MK.CP1251',
1083 'mk_mk.iso88595': 'mk_MK.ISO8859-5',
1084 'mk_mk.microsoftcp1251': 'mk_MK.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001085 'ms': 'ms_MY.ISO8859-1',
1086 'ms_my': 'ms_MY.ISO8859-1',
1087 'ms_my.iso88591': 'ms_MY.ISO8859-1',
1088 'mt': 'mt_MT.ISO8859-3',
1089 'mt_mt': 'mt_MT.ISO8859-3',
1090 'mt_mt.iso88593': 'mt_MT.ISO8859-3',
1091 'nb': 'nb_NO.ISO8859-1',
1092 'nb_no': 'nb_NO.ISO8859-1',
1093 'nb_no.88591': 'nb_NO.ISO8859-1',
1094 'nb_no.iso88591': 'nb_NO.ISO8859-1',
1095 'nb_no.iso885915': 'nb_NO.ISO8859-15',
1096 'nb_no@euro': 'nb_NO.ISO8859-15',
1097 'nl': 'nl_NL.ISO8859-1',
1098 'nl_be': 'nl_BE.ISO8859-1',
1099 'nl_be.88591': 'nl_BE.ISO8859-1',
1100 'nl_be.iso88591': 'nl_BE.ISO8859-1',
1101 'nl_be.iso885915': 'nl_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001102 'nl_be@euro': 'nl_BE.ISO8859-15',
1103 'nl_nl': 'nl_NL.ISO8859-1',
1104 'nl_nl.88591': 'nl_NL.ISO8859-1',
1105 'nl_nl.iso88591': 'nl_NL.ISO8859-1',
1106 'nl_nl.iso885915': 'nl_NL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001107 'nl_nl@euro': 'nl_NL.ISO8859-15',
1108 'nn': 'nn_NO.ISO8859-1',
1109 'nn_no': 'nn_NO.ISO8859-1',
1110 'nn_no.88591': 'nn_NO.ISO8859-1',
1111 'nn_no.iso88591': 'nn_NO.ISO8859-1',
1112 'nn_no.iso885915': 'nn_NO.ISO8859-15',
1113 'nn_no@euro': 'nn_NO.ISO8859-15',
1114 'no': 'no_NO.ISO8859-1',
1115 'no@nynorsk': 'ny_NO.ISO8859-1',
1116 'no_no': 'no_NO.ISO8859-1',
1117 'no_no.88591': 'no_NO.ISO8859-1',
1118 'no_no.iso88591': 'no_NO.ISO8859-1',
1119 'no_no.iso885915': 'no_NO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001120 'no_no@euro': 'no_NO.ISO8859-15',
1121 'norwegian': 'no_NO.ISO8859-1',
1122 'norwegian.iso88591': 'no_NO.ISO8859-1',
1123 'ny': 'ny_NO.ISO8859-1',
1124 'ny_no': 'ny_NO.ISO8859-1',
1125 'ny_no.88591': 'ny_NO.ISO8859-1',
1126 'ny_no.iso88591': 'ny_NO.ISO8859-1',
1127 'ny_no.iso885915': 'ny_NO.ISO8859-15',
1128 'ny_no@euro': 'ny_NO.ISO8859-15',
1129 'nynorsk': 'nn_NO.ISO8859-1',
1130 'oc': 'oc_FR.ISO8859-1',
1131 'oc_fr': 'oc_FR.ISO8859-1',
1132 'oc_fr.iso88591': 'oc_FR.ISO8859-1',
1133 'oc_fr.iso885915': 'oc_FR.ISO8859-15',
1134 'oc_fr@euro': 'oc_FR.ISO8859-15',
1135 'pd': 'pd_US.ISO8859-1',
1136 'pd_de': 'pd_DE.ISO8859-1',
1137 'pd_de.iso88591': 'pd_DE.ISO8859-1',
1138 'pd_de.iso885915': 'pd_DE.ISO8859-15',
1139 'pd_de@euro': 'pd_DE.ISO8859-15',
1140 'pd_us': 'pd_US.ISO8859-1',
1141 'pd_us.iso88591': 'pd_US.ISO8859-1',
1142 'pd_us.iso885915': 'pd_US.ISO8859-15',
1143 'pd_us@euro': 'pd_US.ISO8859-15',
1144 'ph': 'ph_PH.ISO8859-1',
1145 'ph_ph': 'ph_PH.ISO8859-1',
1146 'ph_ph.iso88591': 'ph_PH.ISO8859-1',
1147 'pl': 'pl_PL.ISO8859-2',
1148 'pl_pl': 'pl_PL.ISO8859-2',
1149 'pl_pl.iso88592': 'pl_PL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001150 'polish': 'pl_PL.ISO8859-2',
1151 'portuguese': 'pt_PT.ISO8859-1',
1152 'portuguese.iso88591': 'pt_PT.ISO8859-1',
1153 'portuguese_brazil': 'pt_BR.ISO8859-1',
1154 'portuguese_brazil.8859': 'pt_BR.ISO8859-1',
1155 'posix': 'C',
1156 'posix-utf2': 'C',
1157 'pp': 'pp_AN.ISO8859-1',
1158 'pp_an': 'pp_AN.ISO8859-1',
1159 'pp_an.iso88591': 'pp_AN.ISO8859-1',
1160 'pt': 'pt_PT.ISO8859-1',
1161 'pt_br': 'pt_BR.ISO8859-1',
1162 'pt_br.88591': 'pt_BR.ISO8859-1',
1163 'pt_br.iso88591': 'pt_BR.ISO8859-1',
1164 'pt_br.iso885915': 'pt_BR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001165 'pt_br@euro': 'pt_BR.ISO8859-15',
1166 'pt_pt': 'pt_PT.ISO8859-1',
1167 'pt_pt.88591': 'pt_PT.ISO8859-1',
1168 'pt_pt.iso88591': 'pt_PT.ISO8859-1',
1169 'pt_pt.iso885915': 'pt_PT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001170 'pt_pt.utf8@euro': 'pt_PT.UTF-8',
1171 'pt_pt@euro': 'pt_PT.ISO8859-15',
1172 'ro': 'ro_RO.ISO8859-2',
1173 'ro_ro': 'ro_RO.ISO8859-2',
1174 'ro_ro.iso88592': 'ro_RO.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001175 'romanian': 'ro_RO.ISO8859-2',
1176 'ru': 'ru_RU.ISO8859-5',
1177 'ru_ru': 'ru_RU.ISO8859-5',
1178 'ru_ru.cp1251': 'ru_RU.CP1251',
1179 'ru_ru.iso88595': 'ru_RU.ISO8859-5',
1180 'ru_ru.koi8r': 'ru_RU.KOI8-R',
1181 'ru_ru.microsoftcp1251': 'ru_RU.CP1251',
1182 'ru_ua': 'ru_UA.KOI8-U',
1183 'ru_ua.cp1251': 'ru_UA.CP1251',
1184 'ru_ua.koi8u': 'ru_UA.KOI8-U',
1185 'ru_ua.microsoftcp1251': 'ru_UA.CP1251',
1186 'rumanian': 'ro_RO.ISO8859-2',
1187 'russian': 'ru_RU.ISO8859-5',
1188 'se_no': 'se_NO.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001189 'serbocroatian': 'sh_YU.ISO8859-2',
1190 'sh': 'sh_YU.ISO8859-2',
1191 'sh_hr': 'sh_HR.ISO8859-2',
1192 'sh_hr.iso88592': 'sh_HR.ISO8859-2',
1193 'sh_sp': 'sh_YU.ISO8859-2',
1194 'sh_yu': 'sh_YU.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001195 'sk': 'sk_SK.ISO8859-2',
1196 'sk_sk': 'sk_SK.ISO8859-2',
1197 'sk_sk.iso88592': 'sk_SK.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001198 'sl': 'sl_SI.ISO8859-2',
1199 'sl_cs': 'sl_CS.ISO8859-2',
1200 'sl_si': 'sl_SI.ISO8859-2',
1201 'sl_si.iso88592': 'sl_SI.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001202 'slovak': 'sk_SK.ISO8859-2',
1203 'slovene': 'sl_SI.ISO8859-2',
1204 'slovenian': 'sl_SI.ISO8859-2',
1205 'sp': 'sp_YU.ISO8859-5',
1206 'sp_yu': 'sp_YU.ISO8859-5',
1207 'spanish': 'es_ES.ISO8859-1',
1208 'spanish.iso88591': 'es_ES.ISO8859-1',
1209 'spanish_spain': 'es_ES.ISO8859-1',
1210 'spanish_spain.8859': 'es_ES.ISO8859-1',
1211 'sq': 'sq_AL.ISO8859-2',
1212 'sq_al': 'sq_AL.ISO8859-2',
1213 'sq_al.iso88592': 'sq_AL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001214 'sr': 'sr_YU.ISO8859-5',
1215 'sr@cyrillic': 'sr_YU.ISO8859-5',
1216 'sr_sp': 'sr_SP.ISO8859-2',
1217 'sr_yu': 'sr_YU.ISO8859-5',
1218 'sr_yu.cp1251@cyrillic': 'sr_YU.CP1251',
1219 'sr_yu.iso88592': 'sr_YU.ISO8859-2',
1220 'sr_yu.iso88595': 'sr_YU.ISO8859-5',
1221 'sr_yu.iso88595@cyrillic': 'sr_YU.ISO8859-5',
1222 'sr_yu.microsoftcp1251@cyrillic': 'sr_YU.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001223 'sr_yu.utf8@cyrillic': 'sr_YU.UTF-8',
1224 'sr_yu@cyrillic': 'sr_YU.ISO8859-5',
1225 'sv': 'sv_SE.ISO8859-1',
1226 'sv_fi': 'sv_FI.ISO8859-1',
1227 'sv_fi.iso88591': 'sv_FI.ISO8859-1',
1228 'sv_fi.iso885915': 'sv_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001229 'sv_fi@euro': 'sv_FI.ISO8859-15',
1230 'sv_se': 'sv_SE.ISO8859-1',
1231 'sv_se.88591': 'sv_SE.ISO8859-1',
1232 'sv_se.iso88591': 'sv_SE.ISO8859-1',
1233 'sv_se.iso885915': 'sv_SE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001234 'sv_se@euro': 'sv_SE.ISO8859-15',
1235 'swedish': 'sv_SE.ISO8859-1',
1236 'swedish.iso88591': 'sv_SE.ISO8859-1',
1237 'ta': 'ta_IN.TSCII-0',
1238 'ta_in': 'ta_IN.TSCII-0',
1239 'ta_in.tscii': 'ta_IN.TSCII-0',
1240 'ta_in.tscii0': 'ta_IN.TSCII-0',
1241 'tg': 'tg_TJ.KOI8-C',
1242 'tg_tj': 'tg_TJ.KOI8-C',
1243 'tg_tj.koi8c': 'tg_TJ.KOI8-C',
1244 'th': 'th_TH.ISO8859-11',
1245 'th_th': 'th_TH.ISO8859-11',
1246 'th_th.iso885911': 'th_TH.ISO8859-11',
1247 'th_th.tactis': 'th_TH.TIS620',
1248 'th_th.tis620': 'th_TH.TIS620',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001249 'thai': 'th_TH.ISO8859-11',
1250 'tl': 'tl_PH.ISO8859-1',
1251 'tl_ph': 'tl_PH.ISO8859-1',
1252 'tl_ph.iso88591': 'tl_PH.ISO8859-1',
1253 'tr': 'tr_TR.ISO8859-9',
1254 'tr_tr': 'tr_TR.ISO8859-9',
1255 'tr_tr.iso88599': 'tr_TR.ISO8859-9',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001256 'tt': 'tt_RU.TATAR-CYR',
1257 'tt_ru': 'tt_RU.TATAR-CYR',
1258 'tt_ru.koi8c': 'tt_RU.KOI8-C',
1259 'tt_ru.tatarcyr': 'tt_RU.TATAR-CYR',
1260 'turkish': 'tr_TR.ISO8859-9',
1261 'turkish.iso88599': 'tr_TR.ISO8859-9',
1262 'uk': 'uk_UA.KOI8-U',
1263 'uk_ua': 'uk_UA.KOI8-U',
1264 'uk_ua.cp1251': 'uk_UA.CP1251',
1265 'uk_ua.iso88595': 'uk_UA.ISO8859-5',
1266 'uk_ua.koi8u': 'uk_UA.KOI8-U',
1267 'uk_ua.microsoftcp1251': 'uk_UA.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001268 'univ': 'en_US.utf',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001269 'universal': 'en_US.utf',
1270 'universal.utf8@ucs4': 'en_US.UTF-8',
1271 'ur': 'ur_PK.CP1256',
1272 'ur_pk': 'ur_PK.CP1256',
1273 'ur_pk.cp1256': 'ur_PK.CP1256',
1274 'ur_pk.microsoftcp1256': 'ur_PK.CP1256',
1275 'uz': 'uz_UZ.UTF-8',
1276 'uz_uz': 'uz_UZ.UTF-8',
1277 'vi': 'vi_VN.TCVN',
1278 'vi_vn': 'vi_VN.TCVN',
1279 'vi_vn.tcvn': 'vi_VN.TCVN',
1280 'vi_vn.tcvn5712': 'vi_VN.TCVN',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001281 'vi_vn.viscii': 'vi_VN.VISCII',
1282 'vi_vn.viscii111': 'vi_VN.VISCII',
1283 'wa': 'wa_BE.ISO8859-1',
1284 'wa_be': 'wa_BE.ISO8859-1',
1285 'wa_be.iso88591': 'wa_BE.ISO8859-1',
1286 'wa_be.iso885915': 'wa_BE.ISO8859-15',
1287 'wa_be@euro': 'wa_BE.ISO8859-15',
1288 'yi': 'yi_US.CP1255',
1289 'yi_us': 'yi_US.CP1255',
1290 'yi_us.cp1255': 'yi_US.CP1255',
1291 'yi_us.microsoftcp1255': 'yi_US.CP1255',
1292 'zh': 'zh_CN.eucCN',
1293 'zh_cn': 'zh_CN.gb2312',
1294 'zh_cn.big5': 'zh_TW.big5',
1295 'zh_cn.euc': 'zh_CN.eucCN',
1296 'zh_cn.gb18030': 'zh_CN.gb18030',
1297 'zh_cn.gb2312': 'zh_CN.gb2312',
1298 'zh_cn.gbk': 'zh_CN.gbk',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001299 'zh_hk': 'zh_HK.big5hkscs',
1300 'zh_hk.big5': 'zh_HK.big5',
1301 'zh_hk.big5hkscs': 'zh_HK.big5hkscs',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001302 'zh_tw': 'zh_TW.big5',
1303 'zh_tw.big5': 'zh_TW.big5',
1304 'zh_tw.euc': 'zh_TW.eucTW',
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001305}
1306
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001307#
Georg Brandlb709c2c2006-01-20 09:07:35 +00001308# This maps Windows language identifiers to locale strings.
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001309#
Tim Peters777f1082006-01-20 20:03:24 +00001310# This list has been updated from
Georg Brandlb709c2c2006-01-20 09:07:35 +00001311# http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/nls_238z.asp
1312# to include every locale up to Windows XP.
Fredrik Lundh37a09822002-10-19 20:19:10 +00001313#
Georg Brandl5035c1c2006-01-20 13:38:26 +00001314# NOTE: this mapping is incomplete. If your language is missing, please
1315# submit a bug report to Python bug manager, which you can find via:
1316# http://www.python.org/dev/
1317# Make sure you include the missing language identifier and the suggested
1318# locale code.
1319#
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001320
1321windows_locale = {
Georg Brandlb709c2c2006-01-20 09:07:35 +00001322 0x0436: "af_ZA", # Afrikaans
1323 0x041c: "sq_AL", # Albanian
1324 0x0401: "ar_SA", # Arabic - Saudi Arabia
1325 0x0801: "ar_IQ", # Arabic - Iraq
1326 0x0c01: "ar_EG", # Arabic - Egypt
1327 0x1001: "ar_LY", # Arabic - Libya
1328 0x1401: "ar_DZ", # Arabic - Algeria
1329 0x1801: "ar_MA", # Arabic - Morocco
1330 0x1c01: "ar_TN", # Arabic - Tunisia
1331 0x2001: "ar_OM", # Arabic - Oman
1332 0x2401: "ar_YE", # Arabic - Yemen
1333 0x2801: "ar_SY", # Arabic - Syria
1334 0x2c01: "ar_JO", # Arabic - Jordan
1335 0x3001: "ar_LB", # Arabic - Lebanon
1336 0x3401: "ar_KW", # Arabic - Kuwait
1337 0x3801: "ar_AE", # Arabic - United Arab Emirates
1338 0x3c01: "ar_BH", # Arabic - Bahrain
1339 0x4001: "ar_QA", # Arabic - Qatar
1340 0x042b: "hy_AM", # Armenian
1341 0x042c: "az_AZ", # Azeri Latin
1342 0x082c: "az_AZ", # Azeri - Cyrillic
1343 0x042d: "eu_ES", # Basque
1344 0x0423: "be_BY", # Belarusian
1345 0x0445: "bn_IN", # Begali
1346 0x201a: "bs_BA", # Bosnian
1347 0x141a: "bs_BA", # Bosnian - Cyrillic
1348 0x047e: "br_FR", # Breton - France
1349 0x0402: "bg_BG", # Bulgarian
1350 0x0403: "ca_ES", # Catalan
1351 0x0004: "zh_CHS",# Chinese - Simplified
1352 0x0404: "zh_TW", # Chinese - Taiwan
1353 0x0804: "zh_CN", # Chinese - PRC
1354 0x0c04: "zh_HK", # Chinese - Hong Kong S.A.R.
1355 0x1004: "zh_SG", # Chinese - Singapore
1356 0x1404: "zh_MO", # Chinese - Macao S.A.R.
1357 0x7c04: "zh_CHT",# Chinese - Traditional
1358 0x041a: "hr_HR", # Croatian
1359 0x101a: "hr_BA", # Croatian - Bosnia
1360 0x0405: "cs_CZ", # Czech
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001361 0x0406: "da_DK", # Danish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001362 0x048c: "gbz_AF",# Dari - Afghanistan
1363 0x0465: "div_MV",# Divehi - Maldives
1364 0x0413: "nl_NL", # Dutch - The Netherlands
1365 0x0813: "nl_BE", # Dutch - Belgium
1366 0x0409: "en_US", # English - United States
1367 0x0809: "en_GB", # English - United Kingdom
1368 0x0c09: "en_AU", # English - Australia
1369 0x1009: "en_CA", # English - Canada
1370 0x1409: "en_NZ", # English - New Zealand
1371 0x1809: "en_IE", # English - Ireland
1372 0x1c09: "en_ZA", # English - South Africa
1373 0x2009: "en_JA", # English - Jamaica
1374 0x2409: "en_CB", # English - Carribbean
1375 0x2809: "en_BZ", # English - Belize
1376 0x2c09: "en_TT", # English - Trinidad
1377 0x3009: "en_ZW", # English - Zimbabwe
1378 0x3409: "en_PH", # English - Phillippines
1379 0x0425: "et_EE", # Estonian
1380 0x0438: "fo_FO", # Faroese
1381 0x0464: "fil_PH",# Filipino
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001382 0x040b: "fi_FI", # Finnish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001383 0x040c: "fr_FR", # French - France
1384 0x080c: "fr_BE", # French - Belgium
1385 0x0c0c: "fr_CA", # French - Canada
1386 0x100c: "fr_CH", # French - Switzerland
1387 0x140c: "fr_LU", # French - Luxembourg
1388 0x180c: "fr_MC", # French - Monaco
1389 0x0462: "fy_NL", # Frisian - Netherlands
1390 0x0456: "gl_ES", # Galician
1391 0x0437: "ka_GE", # Georgian
1392 0x0407: "de_DE", # German - Germany
1393 0x0807: "de_CH", # German - Switzerland
1394 0x0c07: "de_AT", # German - Austria
1395 0x1007: "de_LU", # German - Luxembourg
1396 0x1407: "de_LI", # German - Liechtenstein
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001397 0x0408: "el_GR", # Greek
Georg Brandlb709c2c2006-01-20 09:07:35 +00001398 0x0447: "gu_IN", # Gujarati
1399 0x040d: "he_IL", # Hebrew
1400 0x0439: "hi_IN", # Hindi
1401 0x040e: "hu_HU", # Hungarian
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001402 0x040f: "is_IS", # Icelandic
Georg Brandlb709c2c2006-01-20 09:07:35 +00001403 0x0421: "id_ID", # Indonesian
1404 0x045d: "iu_CA", # Inuktitut
1405 0x085d: "iu_CA", # Inuktitut - Latin
1406 0x083c: "ga_IE", # Irish - Ireland
1407 0x0434: "xh_ZA", # Xhosa - South Africa
1408 0x0435: "zu_ZA", # Zulu
1409 0x0410: "it_IT", # Italian - Italy
1410 0x0810: "it_CH", # Italian - Switzerland
1411 0x0411: "ja_JP", # Japanese
1412 0x044b: "kn_IN", # Kannada - India
1413 0x043f: "kk_KZ", # Kazakh
1414 0x0457: "kok_IN",# Konkani
1415 0x0412: "ko_KR", # Korean
1416 0x0440: "ky_KG", # Kyrgyz
1417 0x0426: "lv_LV", # Latvian
1418 0x0427: "lt_LT", # Lithuanian
1419 0x046e: "lb_LU", # Luxembourgish
1420 0x042f: "mk_MK", # FYRO Macedonian
1421 0x043e: "ms_MY", # Malay - Malaysia
1422 0x083e: "ms_BN", # Malay - Brunei
1423 0x044c: "ml_IN", # Malayalam - India
1424 0x043a: "mt_MT", # Maltese
1425 0x0481: "mi_NZ", # Maori
1426 0x047a: "arn_CL",# Mapudungun
1427 0x044e: "mr_IN", # Marathi
1428 0x047c: "moh_CA",# Mohawk - Canada
1429 0x0450: "mn_MN", # Mongolian
1430 0x0461: "ne_NP", # Nepali
1431 0x0414: "nb_NO", # Norwegian - Bokmal
1432 0x0814: "nn_NO", # Norwegian - Nynorsk
1433 0x0482: "oc_FR", # Occitan - France
1434 0x0448: "or_IN", # Oriya - India
1435 0x0463: "ps_AF", # Pashto - Afghanistan
1436 0x0429: "fa_IR", # Persian
1437 0x0415: "pl_PL", # Polish
1438 0x0416: "pt_BR", # Portuguese - Brazil
1439 0x0816: "pt_PT", # Portuguese - Portugal
1440 0x0446: "pa_IN", # Punjabi
1441 0x046b: "quz_BO",# Quechua (Bolivia)
1442 0x086b: "quz_EC",# Quechua (Ecuador)
1443 0x0c6b: "quz_PE",# Quechua (Peru)
1444 0x0418: "ro_RO", # Romanian - Romania
1445 0x0417: "rm_CH", # Raeto-Romanese
1446 0x0419: "ru_RU", # Russian
1447 0x243b: "smn_FI",# Sami Finland
1448 0x103b: "smj_NO",# Sami Norway
1449 0x143b: "smj_SE",# Sami Sweden
1450 0x043b: "se_NO", # Sami Northern Norway
1451 0x083b: "se_SE", # Sami Northern Sweden
1452 0x0c3b: "se_FI", # Sami Northern Finland
1453 0x203b: "sms_FI",# Sami Skolt
1454 0x183b: "sma_NO",# Sami Southern Norway
1455 0x1c3b: "sma_SE",# Sami Southern Sweden
1456 0x044f: "sa_IN", # Sanskrit
1457 0x0c1a: "sr_SP", # Serbian - Cyrillic
1458 0x1c1a: "sr_BA", # Serbian - Bosnia Cyrillic
1459 0x081a: "sr_SP", # Serbian - Latin
1460 0x181a: "sr_BA", # Serbian - Bosnia Latin
1461 0x046c: "ns_ZA", # Northern Sotho
1462 0x0432: "tn_ZA", # Setswana - Southern Africa
1463 0x041b: "sk_SK", # Slovak
1464 0x0424: "sl_SI", # Slovenian
1465 0x040a: "es_ES", # Spanish - Spain
1466 0x080a: "es_MX", # Spanish - Mexico
1467 0x0c0a: "es_ES", # Spanish - Spain (Modern)
1468 0x100a: "es_GT", # Spanish - Guatemala
1469 0x140a: "es_CR", # Spanish - Costa Rica
1470 0x180a: "es_PA", # Spanish - Panama
1471 0x1c0a: "es_DO", # Spanish - Dominican Republic
1472 0x200a: "es_VE", # Spanish - Venezuela
1473 0x240a: "es_CO", # Spanish - Colombia
1474 0x280a: "es_PE", # Spanish - Peru
1475 0x2c0a: "es_AR", # Spanish - Argentina
1476 0x300a: "es_EC", # Spanish - Ecuador
1477 0x340a: "es_CL", # Spanish - Chile
1478 0x380a: "es_UR", # Spanish - Uruguay
1479 0x3c0a: "es_PY", # Spanish - Paraguay
1480 0x400a: "es_BO", # Spanish - Bolivia
1481 0x440a: "es_SV", # Spanish - El Salvador
1482 0x480a: "es_HN", # Spanish - Honduras
1483 0x4c0a: "es_NI", # Spanish - Nicaragua
1484 0x500a: "es_PR", # Spanish - Puerto Rico
1485 0x0441: "sw_KE", # Swahili
1486 0x041d: "sv_SE", # Swedish - Sweden
1487 0x081d: "sv_FI", # Swedish - Finland
1488 0x045a: "syr_SY",# Syriac
1489 0x0449: "ta_IN", # Tamil
1490 0x0444: "tt_RU", # Tatar
1491 0x044a: "te_IN", # Telugu
1492 0x041e: "th_TH", # Thai
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001493 0x041f: "tr_TR", # Turkish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001494 0x0422: "uk_UA", # Ukrainian
1495 0x0420: "ur_PK", # Urdu
1496 0x0820: "ur_IN", # Urdu - India
1497 0x0443: "uz_UZ", # Uzbek - Latin
1498 0x0843: "uz_UZ", # Uzbek - Cyrillic
1499 0x042a: "vi_VN", # Vietnamese
1500 0x0452: "cy_GB", # Welsh
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001501}
1502
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001503def _print_locale():
1504
1505 """ Test function.
1506 """
1507 categories = {}
1508 def _init_categories(categories=categories):
1509 for k,v in globals().items():
1510 if k[:3] == 'LC_':
1511 categories[k] = v
1512 _init_categories()
1513 del categories['LC_ALL']
1514
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001515 print('Locale defaults as determined by getdefaultlocale():')
1516 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001517 lang, enc = getdefaultlocale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001518 print('Language: ', lang or '(undefined)')
1519 print('Encoding: ', enc or '(undefined)')
1520 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001521
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001522 print('Locale settings on startup:')
1523 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001524 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001525 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001526 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001527 print(' Language: ', lang or '(undefined)')
1528 print(' Encoding: ', enc or '(undefined)')
1529 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001530
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001531 print()
1532 print('Locale settings after calling resetlocale():')
1533 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001534 resetlocale()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001535 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001536 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001537 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001538 print(' Language: ', lang or '(undefined)')
1539 print(' Encoding: ', enc or '(undefined)')
1540 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001541
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001542 try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001543 setlocale(LC_ALL, "")
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001544 except:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001545 print('NOTE:')
1546 print('setlocale(LC_ALL, "") does not support the default locale')
1547 print('given in the OS environment variables.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001548 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001549 print()
1550 print('Locale settings after calling setlocale(LC_ALL, ""):')
1551 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001552 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001553 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001554 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001555 print(' Language: ', lang or '(undefined)')
1556 print(' Encoding: ', enc or '(undefined)')
1557 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001558
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001559###
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001560
Tim Peters1baf8292001-01-24 10:13:46 +00001561try:
1562 LC_MESSAGES
Skip Montanaro0897f0c2002-03-25 21:40:36 +00001563except NameError:
Tim Peters1baf8292001-01-24 10:13:46 +00001564 pass
1565else:
1566 __all__.append("LC_MESSAGES")
1567
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001568if __name__=='__main__':
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001569 print('Locale aliasing:')
1570 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001571 _print_locale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001572 print()
1573 print('Number formatting:')
1574 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001575 _test()