blob: dea0ab770526d69a9b937868d119c3ad50ceb893 [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
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000015
Fredrik Lundh6c86b992000-07-09 17:12:58 +000016# Try importing the _locale module.
17#
18# If this fails, fall back on a basic 'C' locale emulation.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +000019
Tim Peters1baf8292001-01-24 10:13:46 +000020# Yuck: LC_MESSAGES is non-standard: can't tell whether it exists before
21# trying the import. So __all__ is also fiddled at the end of the file.
Guido van Rossum360e4b82007-05-14 22:51:27 +000022__all__ = ["getlocale", "getdefaultlocale", "getpreferredencoding", "Error",
23 "setlocale", "resetlocale", "localeconv", "strcoll", "strxfrm",
24 "str", "atof", "atoi", "format", "format_string", "currency",
25 "normalize", "LC_CTYPE", "LC_COLLATE", "LC_TIME", "LC_MONETARY",
26 "LC_NUMERIC", "LC_ALL", "CHAR_MAX"]
Skip Montanaro17ab1232001-01-24 06:27:27 +000027
Marc-André Lemburg23481142000-06-08 17:49:41 +000028try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +000029
Marc-André Lemburg23481142000-06-08 17:49:41 +000030 from _locale import *
31
32except ImportError:
33
Fredrik Lundh6c86b992000-07-09 17:12:58 +000034 # Locale emulation
35
Marc-André Lemburg23481142000-06-08 17:49:41 +000036 CHAR_MAX = 127
37 LC_ALL = 6
38 LC_COLLATE = 3
39 LC_CTYPE = 0
40 LC_MESSAGES = 5
41 LC_MONETARY = 4
42 LC_NUMERIC = 1
43 LC_TIME = 2
44 Error = ValueError
45
46 def localeconv():
Fredrik Lundh6c86b992000-07-09 17:12:58 +000047 """ localeconv() -> dict.
Marc-André Lemburg23481142000-06-08 17:49:41 +000048 Returns numeric and monetary locale-specific parameters.
49 """
50 # 'C' locale default values
51 return {'grouping': [127],
52 'currency_symbol': '',
53 'n_sign_posn': 127,
Fredrik Lundh6c86b992000-07-09 17:12:58 +000054 'p_cs_precedes': 127,
55 'n_cs_precedes': 127,
56 'mon_grouping': [],
Marc-André Lemburg23481142000-06-08 17:49:41 +000057 'n_sep_by_space': 127,
58 'decimal_point': '.',
59 'negative_sign': '',
60 'positive_sign': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000061 'p_sep_by_space': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000062 'int_curr_symbol': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000063 'p_sign_posn': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000064 'thousands_sep': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000065 'mon_thousands_sep': '',
66 'frac_digits': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000067 'mon_decimal_point': '',
68 'int_frac_digits': 127}
Fredrik Lundh6c86b992000-07-09 17:12:58 +000069
Marc-André Lemburg23481142000-06-08 17:49:41 +000070 def setlocale(category, value=None):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000071 """ setlocale(integer,string=None) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000072 Activates/queries locale processing.
73 """
Martin v. Löwis103d6e72003-03-30 15:42:13 +000074 if value not in (None, '', 'C'):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000075 raise Error, '_locale emulation only supports "C" locale'
Marc-André Lemburg23481142000-06-08 17:49:41 +000076 return 'C'
77
78 def strcoll(a,b):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000079 """ strcoll(string,string) -> int.
Marc-André Lemburg23481142000-06-08 17:49:41 +000080 Compares two strings according to the locale.
81 """
82 return cmp(a,b)
83
84 def strxfrm(s):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000085 """ strxfrm(string) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000086 Returns a string that behaves for cmp locale-aware.
87 """
88 return s
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000089
90### Number formatting APIs
91
92# Author: Martin von Loewis
Thomas Wouters477c8d52006-05-27 19:21:47 +000093# improved by Georg Brandl
Guido van Rossumeef1d4e1997-11-19 19:01:43 +000094
95#perform the grouping from right to left
Thomas Wouters477c8d52006-05-27 19:21:47 +000096def _group(s, monetary=False):
97 conv = localeconv()
98 thousands_sep = conv[monetary and 'mon_thousands_sep' or 'thousands_sep']
99 grouping = conv[monetary and 'mon_grouping' or 'grouping']
100 if not grouping:
101 return (s, 0)
102 result = ""
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000103 seps = 0
104 spaces = ""
105 if s[-1] == ' ':
106 sp = s.find(' ')
107 spaces = s[sp:]
108 s = s[:sp]
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000109 while s and grouping:
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000110 # if grouping is -1, we are done
Thomas Wouters477c8d52006-05-27 19:21:47 +0000111 if grouping[0] == CHAR_MAX:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000112 break
113 # 0: re-use last group ad infinitum
Thomas Wouters477c8d52006-05-27 19:21:47 +0000114 elif grouping[0] != 0:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000115 #process last group
Thomas Wouters477c8d52006-05-27 19:21:47 +0000116 group = grouping[0]
117 grouping = grouping[1:]
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000118 if result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000119 result = s[-group:] + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000120 seps += 1
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000121 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000122 result = s[-group:]
123 s = s[:-group]
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000124 if s and s[-1] not in "0123456789":
125 # the leading string is only spaces and signs
Thomas Wouters477c8d52006-05-27 19:21:47 +0000126 return s + result + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000127 if not result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000128 return s + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000129 if s:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000130 result = s + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000131 seps += 1
Thomas Wouters477c8d52006-05-27 19:21:47 +0000132 return result + spaces, seps
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000133
Thomas Wouters477c8d52006-05-27 19:21:47 +0000134def format(percent, value, grouping=False, monetary=False, *additional):
135 """Returns the locale-aware substitution of a %? specifier
136 (percent).
137
138 additional is for format strings which contain one or more
139 '*' modifiers."""
140 # this is only for one-percent-specifier strings and this should be checked
141 if percent[0] != '%':
142 raise ValueError("format() must be given exactly one %char "
143 "format specifier")
144 if additional:
145 formatted = percent % ((value,) + additional)
146 else:
147 formatted = percent % value
148 # floats and decimal ints need special action!
149 if percent[-1] in 'eEfFgG':
150 seps = 0
151 parts = formatted.split('.')
152 if grouping:
153 parts[0], seps = _group(parts[0], monetary=monetary)
154 decimal_point = localeconv()[monetary and 'mon_decimal_point'
155 or 'decimal_point']
156 formatted = decimal_point.join(parts)
157 while seps:
158 sp = formatted.find(' ')
159 if sp == -1: break
160 formatted = formatted[:sp] + formatted[sp+1:]
161 seps -= 1
162 elif percent[-1] in 'diu':
163 if grouping:
164 formatted = _group(formatted, monetary=monetary)[0]
165 return formatted
166
167import re, operator
168_percent_re = re.compile(r'%(?:\((?P<key>.*?)\))?'
169 r'(?P<modifiers>[-#0-9 +*.hlL]*?)[eEfFgGdiouxXcrs%]')
170
171def format_string(f, val, grouping=False):
172 """Formats a string in the same way that the % formatting would use,
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000173 but takes the current locale into account.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000174 Grouping is applied if the third parameter is true."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000175 percents = list(_percent_re.finditer(f))
176 new_f = _percent_re.sub('%s', f)
177
178 if isinstance(val, tuple):
179 new_val = list(val)
180 i = 0
181 for perc in percents:
182 starcount = perc.group('modifiers').count('*')
183 new_val[i] = format(perc.group(), new_val[i], grouping, False, *new_val[i+1:i+1+starcount])
184 del new_val[i+1:i+1+starcount]
185 i += (1 + starcount)
186 val = tuple(new_val)
187 elif operator.isMappingType(val):
188 for perc in percents:
189 key = perc.group("key")
190 val[key] = format(perc.group(), val[key], grouping)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000191 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000192 # val is a single value
193 val = format(percents[0].group(), val, grouping)
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000194
Thomas Wouters477c8d52006-05-27 19:21:47 +0000195 return new_f % val
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000196
Thomas Wouters477c8d52006-05-27 19:21:47 +0000197def currency(val, symbol=True, grouping=False, international=False):
198 """Formats val according to the currency settings
199 in the current locale."""
200 conv = localeconv()
201
202 # check for illegal values
203 digits = conv[international and 'int_frac_digits' or 'frac_digits']
204 if digits == 127:
205 raise ValueError("Currency formatting is not possible using "
206 "the 'C' locale.")
207
208 s = format('%%.%if' % digits, abs(val), grouping, monetary=True)
209 # '<' and '>' are markers if the sign must be inserted between symbol and value
210 s = '<' + s + '>'
211
212 if symbol:
213 smb = conv[international and 'int_curr_symbol' or 'currency_symbol']
214 precedes = conv[val<0 and 'n_cs_precedes' or 'p_cs_precedes']
215 separated = conv[val<0 and 'n_sep_by_space' or 'p_sep_by_space']
216
217 if precedes:
218 s = smb + (separated and ' ' or '') + s
219 else:
220 s = s + (separated and ' ' or '') + smb
221
222 sign_pos = conv[val<0 and 'n_sign_posn' or 'p_sign_posn']
223 sign = conv[val<0 and 'negative_sign' or 'positive_sign']
224
225 if sign_pos == 0:
226 s = '(' + s + ')'
227 elif sign_pos == 1:
228 s = sign + s
229 elif sign_pos == 2:
230 s = s + sign
231 elif sign_pos == 3:
232 s = s.replace('<', sign)
233 elif sign_pos == 4:
234 s = s.replace('>', sign)
235 else:
236 # the default if nothing specified;
237 # this should be the most fitting sign position
238 s = sign + s
239
240 return s.replace('<', '').replace('>', '')
Martin v. Löwisdb786872001-01-21 18:52:33 +0000241
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000242def str(val):
243 """Convert float to integer, taking the locale into account."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000244 return format("%.12g", val)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000245
Thomas Wouters477c8d52006-05-27 19:21:47 +0000246def atof(string, func=float):
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000247 "Parses a string as a float according to the locale settings."
248 #First, get rid of the grouping
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000249 ts = localeconv()['thousands_sep']
250 if ts:
Skip Montanaro249369c2004-04-10 16:39:32 +0000251 string = string.replace(ts, '')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000252 #next, replace the decimal point with a dot
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000253 dd = localeconv()['decimal_point']
254 if dd:
Skip Montanaro249369c2004-04-10 16:39:32 +0000255 string = string.replace(dd, '.')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000256 #finally, parse the string
Skip Montanaro249369c2004-04-10 16:39:32 +0000257 return func(string)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000258
259def atoi(str):
260 "Converts a string to an integer according to the locale settings."
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000261 return atof(str, int)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000262
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000263def _test():
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000264 setlocale(LC_ALL, "")
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000265 #do grouping
Thomas Wouters477c8d52006-05-27 19:21:47 +0000266 s1 = format("%d", 123456789,1)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000267 print(s1, "is", atoi(s1))
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000268 #standard formatting
Thomas Wouters477c8d52006-05-27 19:21:47 +0000269 s1 = str(3.14)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000270 print(s1, "is", atof(s1))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000271
272### Locale name aliasing engine
273
274# Author: Marc-Andre Lemburg, mal@lemburg.com
Fredrik Lundh37a09822002-10-19 20:19:10 +0000275# Various tweaks by Fredrik Lundh <fredrik@pythonware.com>
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000276
277# store away the low-level version of setlocale (it's
278# overridden below)
279_setlocale = setlocale
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000280
281def normalize(localename):
282
283 """ Returns a normalized locale code for the given locale
284 name.
285
286 The returned locale code is formatted for use with
287 setlocale().
288
289 If normalization fails, the original name is returned
290 unchanged.
291
292 If the given encoding is not known, the function defaults to
293 the default encoding for the locale code just like setlocale()
294 does.
295
296 """
297 # Normalize the locale name and extract the encoding
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000298 fullname = localename.lower()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000299 if ':' in fullname:
300 # ':' is sometimes used as encoding delimiter.
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000301 fullname = fullname.replace(':', '.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000302 if '.' in fullname:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000303 langname, encoding = fullname.split('.')[:2]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000304 fullname = langname + '.' + encoding
305 else:
306 langname = fullname
307 encoding = ''
308
309 # First lookup: fullname (possibly with encoding)
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000310 norm_encoding = encoding.replace('-', '')
311 norm_encoding = norm_encoding.replace('_', '')
312 lookup_name = langname + '.' + encoding
313 code = locale_alias.get(lookup_name, None)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000314 if code is not None:
315 return code
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000316 #print 'first lookup failed'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000317
318 # Second try: langname (without encoding)
319 code = locale_alias.get(langname, None)
320 if code is not None:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000321 #print 'langname lookup succeeded'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000322 if '.' in code:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000323 langname, defenc = code.split('.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000324 else:
325 langname = code
326 defenc = ''
327 if encoding:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000328 # Convert the encoding to a C lib compatible encoding string
329 norm_encoding = encodings.normalize_encoding(encoding)
330 #print 'norm encoding: %r' % norm_encoding
331 norm_encoding = encodings.aliases.aliases.get(norm_encoding,
332 norm_encoding)
333 #print 'aliased encoding: %r' % norm_encoding
334 encoding = locale_encoding_alias.get(norm_encoding,
335 norm_encoding)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000336 else:
337 encoding = defenc
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000338 #print 'found encoding %r' % encoding
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000339 if encoding:
340 return langname + '.' + encoding
341 else:
342 return langname
343
344 else:
345 return localename
346
347def _parse_localename(localename):
348
349 """ Parses the locale code for localename and returns the
350 result as tuple (language code, encoding).
351
352 The localename is normalized and passed through the locale
353 alias engine. A ValueError is raised in case the locale name
354 cannot be parsed.
355
356 The language code corresponds to RFC 1766. code and encoding
357 can be None in case the values cannot be determined or are
Jeremy Hyltona05e2932000-06-28 14:48:01 +0000358 unknown to this implementation.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000359
360 """
361 code = normalize(localename)
Georg Brandlb709c2c2006-01-20 09:07:35 +0000362 if '@' in code:
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000363 # Deal with locale modifiers
364 code, modifier = code.split('@')
365 if modifier == 'euro' and '.' not in code:
366 # Assume Latin-9 for @euro locales. This is bogus,
367 # since some systems may use other encodings for these
368 # locales. Also, we ignore other modifiers.
369 return code, 'iso-8859-15'
Tim Peters230a60c2002-11-09 05:08:07 +0000370
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000371 if '.' in code:
Raymond Hettinger346e67f2005-01-01 06:10:26 +0000372 return tuple(code.split('.')[:2])
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000373 elif code == 'C':
374 return None, None
Andrew M. Kuchling1f877ef2001-08-13 14:50:44 +0000375 raise ValueError, 'unknown locale: %s' % localename
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000376
377def _build_localename(localetuple):
378
379 """ Builds a locale code from the given tuple (language code,
380 encoding).
381
382 No aliasing or normalizing takes place.
383
384 """
385 language, encoding = localetuple
386 if language is None:
387 language = 'C'
388 if encoding is None:
389 return language
390 else:
391 return language + '.' + encoding
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000392
Matthias Klosef3f231f2005-09-20 07:02:49 +0000393def getdefaultlocale(envvars=('LC_ALL', 'LC_CTYPE', 'LANG', 'LANGUAGE')):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000394
395 """ Tries to determine the default locale settings and returns
396 them as tuple (language code, encoding).
397
398 According to POSIX, a program which has not called
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000399 setlocale(LC_ALL, "") runs using the portable 'C' locale.
400 Calling setlocale(LC_ALL, "") lets it use the default locale as
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000401 defined by the LANG variable. Since we don't want to interfere
Thomas Wouters7e474022000-07-16 12:04:32 +0000402 with the current locale setting we thus emulate the behavior
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000403 in the way described above.
404
405 To maintain compatibility with other platforms, not only the
406 LANG variable is tested, but a list of variables given as
407 envvars parameter. The first found to be defined will be
408 used. envvars defaults to the search path used in GNU gettext;
409 it must always contain the variable name 'LANG'.
410
411 Except for the code 'C', the language code corresponds to RFC
412 1766. code and encoding can be None in case the values cannot
413 be determined.
414
415 """
Fredrik Lundh04661322000-07-09 23:16:10 +0000416
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000417 try:
418 # check if it's supported by the _locale module
419 import _locale
420 code, encoding = _locale._getdefaultlocale()
Fredrik Lundh04661322000-07-09 23:16:10 +0000421 except (ImportError, AttributeError):
422 pass
423 else:
Fredrik Lundh663809e2000-07-10 19:32:19 +0000424 # make sure the code/encoding values are valid
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000425 if sys.platform == "win32" and code and code[:2] == "0x":
426 # map windows language identifier to language name
427 code = windows_locale.get(int(code, 0))
Fredrik Lundh663809e2000-07-10 19:32:19 +0000428 # ...add other platform-specific processing here, if
429 # necessary...
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000430 return code, encoding
Fredrik Lundh04661322000-07-09 23:16:10 +0000431
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000432 # fall back on POSIX behaviour
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000433 import os
434 lookup = os.environ.get
435 for variable in envvars:
436 localename = lookup(variable,None)
Martin v. Löwisc8ae31d2004-07-26 12:45:18 +0000437 if localename:
Matthias Klosef3f231f2005-09-20 07:02:49 +0000438 if variable == 'LANGUAGE':
439 localename = localename.split(':')[0]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000440 break
441 else:
442 localename = 'C'
443 return _parse_localename(localename)
444
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000445
446def getlocale(category=LC_CTYPE):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000447
448 """ Returns the current setting for the given locale category as
449 tuple (language code, encoding).
450
451 category may be one of the LC_* value except LC_ALL. It
452 defaults to LC_CTYPE.
453
454 Except for the code 'C', the language code corresponds to RFC
455 1766. code and encoding can be None in case the values cannot
456 be determined.
457
458 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000459 localename = _setlocale(category)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000460 if category == LC_ALL and ';' in localename:
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000461 raise TypeError, 'category LC_ALL is not supported'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000462 return _parse_localename(localename)
463
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000464def setlocale(category, locale=None):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000465
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000466 """ Set the locale for the given category. The locale can be
467 a string, a locale tuple (language code, encoding), or None.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000468
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000469 Locale tuples are converted to strings the locale aliasing
470 engine. Locale strings are passed directly to the C lib.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000471
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000472 category may be given as one of the LC_* values.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000473
474 """
Guido van Rossum572dbf82007-04-27 23:53:51 +0000475 if locale and not isinstance(locale, basestring):
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000476 # convert to string
477 locale = normalize(_build_localename(locale))
478 return _setlocale(category, locale)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000479
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000480def resetlocale(category=LC_ALL):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000481
482 """ Sets the locale for category to the default setting.
483
484 The default setting is determined by calling
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000485 getdefaultlocale(). category defaults to LC_ALL.
486
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000487 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000488 _setlocale(category, _build_localename(getdefaultlocale()))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000489
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000490if sys.platform in ('win32', 'darwin', 'mac'):
491 # On Win32, this will return the ANSI code page
492 # On the Mac, it should return the system encoding;
493 # it might return "ascii" instead
494 def getpreferredencoding(do_setlocale = True):
495 """Return the charset that the user is likely using."""
496 import _locale
Tim Petersa326f472002-11-05 03:49:09 +0000497 return _locale._getdefaultlocale()[1]
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000498else:
499 # On Unix, if CODESET is available, use that.
500 try:
501 CODESET
502 except NameError:
503 # Fall back to parsing environment variables :-(
504 def getpreferredencoding(do_setlocale = True):
505 """Return the charset that the user is likely using,
506 by looking at environment variables."""
507 return getdefaultlocale()[1]
508 else:
509 def getpreferredencoding(do_setlocale = True):
510 """Return the charset that the user is likely using,
511 according to the system configuration."""
512 if do_setlocale:
513 oldloc = setlocale(LC_CTYPE)
514 setlocale(LC_CTYPE, "")
515 result = nl_langinfo(CODESET)
516 setlocale(LC_CTYPE, oldloc)
517 return result
518 else:
519 return nl_langinfo(CODESET)
Tim Peters230a60c2002-11-09 05:08:07 +0000520
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000521
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000522### Database
523#
524# The following data was extracted from the locale.alias file which
525# comes with X11 and then hand edited removing the explicit encoding
526# definitions and adding some more aliases. The file is usually
527# available as /usr/lib/X11/locale/locale.alias.
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000528#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000529
530#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000531# The local_encoding_alias table maps lowercase encoding alias names
532# to C locale encoding names (case-sensitive). Note that normalize()
533# first looks up the encoding in the encodings.aliases dictionary and
534# then applies this mapping to find the correct C lib name for the
535# encoding.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000536#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000537locale_encoding_alias = {
538
539 # Mappings for non-standard encoding names used in locale names
540 '437': 'C',
541 'c': 'C',
542 'en': 'ISO8859-1',
543 'jis': 'JIS7',
544 'jis7': 'JIS7',
545 'ajec': 'eucJP',
546
547 # Mappings from Python codec names to C lib encoding names
548 'ascii': 'ISO8859-1',
549 'latin_1': 'ISO8859-1',
550 'iso8859_1': 'ISO8859-1',
551 'iso8859_10': 'ISO8859-10',
552 'iso8859_11': 'ISO8859-11',
553 'iso8859_13': 'ISO8859-13',
554 'iso8859_14': 'ISO8859-14',
555 'iso8859_15': 'ISO8859-15',
556 'iso8859_2': 'ISO8859-2',
557 'iso8859_3': 'ISO8859-3',
558 'iso8859_4': 'ISO8859-4',
559 'iso8859_5': 'ISO8859-5',
560 'iso8859_6': 'ISO8859-6',
561 'iso8859_7': 'ISO8859-7',
562 'iso8859_8': 'ISO8859-8',
563 'iso8859_9': 'ISO8859-9',
564 'iso2022_jp': 'JIS7',
565 'shift_jis': 'SJIS',
566 'tactis': 'TACTIS',
567 'euc_jp': 'eucJP',
568 'euc_kr': 'eucKR',
Marc-André Lemburgb4cebd42004-12-13 19:56:01 +0000569 'utf_8': 'UTF8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000570 'koi8_r': 'KOI8-R',
571 'koi8_u': 'KOI8-U',
572 # XXX This list is still incomplete. If you know more
573 # mappings, please file a bug report. Thanks.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000574}
575
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000576#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000577# The locale_alias table maps lowercase alias names to C locale names
578# (case-sensitive). Encodings are always separated from the locale
579# name using a dot ('.'); they should only be given in case the
580# language name is needed to interpret the given encoding alias
581# correctly (CJK codes often have this need).
582#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000583# Note that the normalize() function which uses this tables
584# removes '_' and '-' characters from the encoding part of the
585# locale name before doing the lookup. This saves a lot of
586# space in the table.
587#
588# MAL 2004-12-10:
589# Updated alias mapping to most recent locale.alias file
590# from X.org distribution using makelocalealias.py.
591#
592# These are the differences compared to the old mapping (Python 2.4
593# and older):
594#
595# updated 'bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
596# updated 'bg_bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
597# updated 'bulgarian' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
598# updated 'cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
599# updated 'cz_cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
600# updated 'czech' -> 'cs_CS.ISO8859-2' to 'cs_CZ.ISO8859-2'
601# updated 'dutch' -> 'nl_BE.ISO8859-1' to 'nl_NL.ISO8859-1'
602# updated 'et' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
603# updated 'et_ee' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
604# updated 'fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
605# updated 'fi_fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
606# updated 'iw' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
607# updated 'iw_il' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
608# updated 'japanese' -> 'ja_JP.SJIS' to 'ja_JP.eucJP'
609# updated 'lt' -> 'lt_LT.ISO8859-4' to 'lt_LT.ISO8859-13'
610# updated 'lv' -> 'lv_LV.ISO8859-4' to 'lv_LV.ISO8859-13'
611# updated 'sl' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
612# updated 'slovene' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
613# updated 'th_th' -> 'th_TH.TACTIS' to 'th_TH.ISO8859-11'
614# updated 'zh_cn' -> 'zh_CN.eucCN' to 'zh_CN.gb2312'
615# updated 'zh_cn.big5' -> 'zh_TW.eucTW' to 'zh_TW.big5'
616# updated 'zh_tw' -> 'zh_TW.eucTW' to 'zh_TW.big5'
617#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000618locale_alias = {
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000619 'a3': 'a3_AZ.KOI8-C',
620 'a3_az': 'a3_AZ.KOI8-C',
621 'a3_az.koi8c': 'a3_AZ.KOI8-C',
622 'af': 'af_ZA.ISO8859-1',
623 'af_za': 'af_ZA.ISO8859-1',
624 'af_za.iso88591': 'af_ZA.ISO8859-1',
625 'am': 'am_ET.UTF-8',
626 'american': 'en_US.ISO8859-1',
627 'american.iso88591': 'en_US.ISO8859-1',
628 'ar': 'ar_AA.ISO8859-6',
629 'ar_aa': 'ar_AA.ISO8859-6',
630 'ar_aa.iso88596': 'ar_AA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000631 'ar_ae': 'ar_AE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000632 'ar_bh': 'ar_BH.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000633 'ar_dz': 'ar_DZ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000634 'ar_eg': 'ar_EG.ISO8859-6',
635 'ar_eg.iso88596': 'ar_EG.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000636 'ar_iq': 'ar_IQ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000637 'ar_jo': 'ar_JO.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000638 'ar_kw': 'ar_KW.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000639 'ar_lb': 'ar_LB.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000640 'ar_ly': 'ar_LY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000641 'ar_ma': 'ar_MA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000642 'ar_om': 'ar_OM.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000643 'ar_qa': 'ar_QA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000644 'ar_sa': 'ar_SA.ISO8859-6',
645 'ar_sa.iso88596': 'ar_SA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000646 'ar_sd': 'ar_SD.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000647 'ar_sy': 'ar_SY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000648 'ar_tn': 'ar_TN.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000649 'ar_ye': 'ar_YE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000650 'arabic': 'ar_AA.ISO8859-6',
651 'arabic.iso88596': 'ar_AA.ISO8859-6',
652 'az': 'az_AZ.ISO8859-9E',
653 'az_az': 'az_AZ.ISO8859-9E',
654 'az_az.iso88599e': 'az_AZ.ISO8859-9E',
655 'be': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000656 'be_by': 'be_BY.CP1251',
657 'be_by.cp1251': 'be_BY.CP1251',
658 'be_by.microsoftcp1251': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000659 'bg': 'bg_BG.CP1251',
660 'bg_bg': 'bg_BG.CP1251',
661 'bg_bg.cp1251': 'bg_BG.CP1251',
662 'bg_bg.iso88595': 'bg_BG.ISO8859-5',
663 'bg_bg.koi8r': 'bg_BG.KOI8-R',
664 'bg_bg.microsoftcp1251': 'bg_BG.CP1251',
665 'bokmal': 'nb_NO.ISO8859-1',
666 'bokm\xe5l': 'nb_NO.ISO8859-1',
667 'br': 'br_FR.ISO8859-1',
668 'br_fr': 'br_FR.ISO8859-1',
669 'br_fr.iso88591': 'br_FR.ISO8859-1',
670 'br_fr.iso885914': 'br_FR.ISO8859-14',
671 'br_fr.iso885915': 'br_FR.ISO8859-15',
672 'br_fr@euro': 'br_FR.ISO8859-15',
673 'bulgarian': 'bg_BG.CP1251',
674 'c': 'C',
675 'c-french': 'fr_CA.ISO8859-1',
676 'c-french.iso88591': 'fr_CA.ISO8859-1',
677 'c.en': 'C',
678 'c.iso88591': 'en_US.ISO8859-1',
679 'c_c': 'C',
680 'c_c.c': 'C',
681 'ca': 'ca_ES.ISO8859-1',
682 'ca_es': 'ca_ES.ISO8859-1',
683 'ca_es.iso88591': 'ca_ES.ISO8859-1',
684 'ca_es.iso885915': 'ca_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000685 'ca_es@euro': 'ca_ES.ISO8859-15',
686 'catalan': 'ca_ES.ISO8859-1',
687 'cextend': 'en_US.ISO8859-1',
688 'cextend.en': 'en_US.ISO8859-1',
689 'chinese-s': 'zh_CN.eucCN',
690 'chinese-t': 'zh_TW.eucTW',
691 'croatian': 'hr_HR.ISO8859-2',
692 'cs': 'cs_CZ.ISO8859-2',
693 'cs_cs': 'cs_CZ.ISO8859-2',
694 'cs_cs.iso88592': 'cs_CZ.ISO8859-2',
695 'cs_cz': 'cs_CZ.ISO8859-2',
696 'cs_cz.iso88592': 'cs_CZ.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000697 'cy': 'cy_GB.ISO8859-1',
698 'cy_gb': 'cy_GB.ISO8859-1',
699 'cy_gb.iso88591': 'cy_GB.ISO8859-1',
700 'cy_gb.iso885914': 'cy_GB.ISO8859-14',
701 'cy_gb.iso885915': 'cy_GB.ISO8859-15',
702 'cy_gb@euro': 'cy_GB.ISO8859-15',
703 'cz': 'cs_CZ.ISO8859-2',
704 'cz_cz': 'cs_CZ.ISO8859-2',
705 'czech': 'cs_CZ.ISO8859-2',
706 'da': 'da_DK.ISO8859-1',
707 'da_dk': 'da_DK.ISO8859-1',
708 'da_dk.88591': 'da_DK.ISO8859-1',
709 'da_dk.885915': 'da_DK.ISO8859-15',
710 'da_dk.iso88591': 'da_DK.ISO8859-1',
711 'da_dk.iso885915': 'da_DK.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000712 'da_dk@euro': 'da_DK.ISO8859-15',
713 'danish': 'da_DK.ISO8859-1',
714 'danish.iso88591': 'da_DK.ISO8859-1',
715 'dansk': 'da_DK.ISO8859-1',
716 'de': 'de_DE.ISO8859-1',
717 'de_at': 'de_AT.ISO8859-1',
718 'de_at.iso88591': 'de_AT.ISO8859-1',
719 'de_at.iso885915': 'de_AT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000720 'de_at@euro': 'de_AT.ISO8859-15',
721 'de_be': 'de_BE.ISO8859-1',
722 'de_be.iso88591': 'de_BE.ISO8859-1',
723 'de_be.iso885915': 'de_BE.ISO8859-15',
724 'de_be@euro': 'de_BE.ISO8859-15',
725 'de_ch': 'de_CH.ISO8859-1',
726 'de_ch.iso88591': 'de_CH.ISO8859-1',
727 'de_ch.iso885915': 'de_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000728 'de_ch@euro': 'de_CH.ISO8859-15',
729 'de_de': 'de_DE.ISO8859-1',
730 'de_de.88591': 'de_DE.ISO8859-1',
731 'de_de.885915': 'de_DE.ISO8859-15',
732 'de_de.885915@euro': 'de_DE.ISO8859-15',
733 'de_de.iso88591': 'de_DE.ISO8859-1',
734 'de_de.iso885915': 'de_DE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000735 'de_de@euro': 'de_DE.ISO8859-15',
736 'de_lu': 'de_LU.ISO8859-1',
737 'de_lu.iso88591': 'de_LU.ISO8859-1',
738 'de_lu.iso885915': 'de_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000739 'de_lu@euro': 'de_LU.ISO8859-15',
740 'deutsch': 'de_DE.ISO8859-1',
741 'dutch': 'nl_NL.ISO8859-1',
742 'dutch.iso88591': 'nl_BE.ISO8859-1',
743 'ee': 'ee_EE.ISO8859-4',
744 'ee_ee': 'ee_EE.ISO8859-4',
745 'ee_ee.iso88594': 'ee_EE.ISO8859-4',
746 'eesti': 'et_EE.ISO8859-1',
747 'el': 'el_GR.ISO8859-7',
748 'el_gr': 'el_GR.ISO8859-7',
749 'el_gr.iso88597': 'el_GR.ISO8859-7',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000750 'el_gr@euro': 'el_GR.ISO8859-15',
751 'en': 'en_US.ISO8859-1',
752 'en.iso88591': 'en_US.ISO8859-1',
753 'en_au': 'en_AU.ISO8859-1',
754 'en_au.iso88591': 'en_AU.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000755 'en_be': 'en_BE.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000756 'en_be@euro': 'en_BE.ISO8859-15',
757 'en_bw': 'en_BW.ISO8859-1',
758 'en_ca': 'en_CA.ISO8859-1',
759 'en_ca.iso88591': 'en_CA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000760 'en_gb': 'en_GB.ISO8859-1',
761 'en_gb.88591': 'en_GB.ISO8859-1',
762 'en_gb.iso88591': 'en_GB.ISO8859-1',
763 'en_gb.iso885915': 'en_GB.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000764 'en_gb@euro': 'en_GB.ISO8859-15',
765 'en_hk': 'en_HK.ISO8859-1',
766 'en_ie': 'en_IE.ISO8859-1',
767 'en_ie.iso88591': 'en_IE.ISO8859-1',
768 'en_ie.iso885915': 'en_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000769 'en_ie@euro': 'en_IE.ISO8859-15',
770 'en_in': 'en_IN.ISO8859-1',
771 'en_nz': 'en_NZ.ISO8859-1',
772 'en_nz.iso88591': 'en_NZ.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000773 'en_ph': 'en_PH.ISO8859-1',
774 'en_sg': 'en_SG.ISO8859-1',
775 'en_uk': 'en_GB.ISO8859-1',
776 'en_us': 'en_US.ISO8859-1',
777 'en_us.88591': 'en_US.ISO8859-1',
778 'en_us.885915': 'en_US.ISO8859-15',
779 'en_us.iso88591': 'en_US.ISO8859-1',
780 'en_us.iso885915': 'en_US.ISO8859-15',
781 'en_us.iso885915@euro': 'en_US.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000782 'en_us@euro': 'en_US.ISO8859-15',
783 'en_us@euro@euro': 'en_US.ISO8859-15',
784 'en_za': 'en_ZA.ISO8859-1',
785 'en_za.88591': 'en_ZA.ISO8859-1',
786 'en_za.iso88591': 'en_ZA.ISO8859-1',
787 'en_za.iso885915': 'en_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000788 'en_za@euro': 'en_ZA.ISO8859-15',
789 'en_zw': 'en_ZW.ISO8859-1',
790 'eng_gb': 'en_GB.ISO8859-1',
791 'eng_gb.8859': 'en_GB.ISO8859-1',
792 'english': 'en_EN.ISO8859-1',
793 'english.iso88591': 'en_EN.ISO8859-1',
794 'english_uk': 'en_GB.ISO8859-1',
795 'english_uk.8859': 'en_GB.ISO8859-1',
796 'english_united-states': 'en_US.ISO8859-1',
797 'english_united-states.437': 'C',
798 'english_us': 'en_US.ISO8859-1',
799 'english_us.8859': 'en_US.ISO8859-1',
800 'english_us.ascii': 'en_US.ISO8859-1',
801 'eo': 'eo_XX.ISO8859-3',
802 'eo_eo': 'eo_EO.ISO8859-3',
803 'eo_eo.iso88593': 'eo_EO.ISO8859-3',
804 'eo_xx': 'eo_XX.ISO8859-3',
805 'eo_xx.iso88593': 'eo_XX.ISO8859-3',
806 'es': 'es_ES.ISO8859-1',
807 'es_ar': 'es_AR.ISO8859-1',
808 'es_ar.iso88591': 'es_AR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000809 'es_bo': 'es_BO.ISO8859-1',
810 'es_bo.iso88591': 'es_BO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000811 'es_cl': 'es_CL.ISO8859-1',
812 'es_cl.iso88591': 'es_CL.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000813 'es_co': 'es_CO.ISO8859-1',
814 'es_co.iso88591': 'es_CO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000815 'es_cr': 'es_CR.ISO8859-1',
816 'es_cr.iso88591': 'es_CR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000817 'es_do': 'es_DO.ISO8859-1',
818 'es_do.iso88591': 'es_DO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000819 'es_ec': 'es_EC.ISO8859-1',
820 'es_ec.iso88591': 'es_EC.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000821 'es_es': 'es_ES.ISO8859-1',
822 'es_es.88591': 'es_ES.ISO8859-1',
823 'es_es.iso88591': 'es_ES.ISO8859-1',
824 'es_es.iso885915': 'es_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000825 'es_es@euro': 'es_ES.ISO8859-15',
826 'es_gt': 'es_GT.ISO8859-1',
827 'es_gt.iso88591': 'es_GT.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000828 'es_hn': 'es_HN.ISO8859-1',
829 'es_hn.iso88591': 'es_HN.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000830 'es_mx': 'es_MX.ISO8859-1',
831 'es_mx.iso88591': 'es_MX.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000832 'es_ni': 'es_NI.ISO8859-1',
833 'es_ni.iso88591': 'es_NI.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000834 'es_pa': 'es_PA.ISO8859-1',
835 'es_pa.iso88591': 'es_PA.ISO8859-1',
836 'es_pa.iso885915': 'es_PA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000837 'es_pa@euro': 'es_PA.ISO8859-15',
838 'es_pe': 'es_PE.ISO8859-1',
839 'es_pe.iso88591': 'es_PE.ISO8859-1',
840 'es_pe.iso885915': 'es_PE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000841 'es_pe@euro': 'es_PE.ISO8859-15',
842 'es_pr': 'es_PR.ISO8859-1',
843 'es_pr.iso88591': 'es_PR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000844 'es_py': 'es_PY.ISO8859-1',
845 'es_py.iso88591': 'es_PY.ISO8859-1',
846 'es_py.iso885915': 'es_PY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000847 'es_py@euro': 'es_PY.ISO8859-15',
848 'es_sv': 'es_SV.ISO8859-1',
849 'es_sv.iso88591': 'es_SV.ISO8859-1',
850 'es_sv.iso885915': 'es_SV.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000851 'es_sv@euro': 'es_SV.ISO8859-15',
852 'es_us': 'es_US.ISO8859-1',
853 'es_uy': 'es_UY.ISO8859-1',
854 'es_uy.iso88591': 'es_UY.ISO8859-1',
855 'es_uy.iso885915': 'es_UY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000856 'es_uy@euro': 'es_UY.ISO8859-15',
857 'es_ve': 'es_VE.ISO8859-1',
858 'es_ve.iso88591': 'es_VE.ISO8859-1',
859 'es_ve.iso885915': 'es_VE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000860 'es_ve@euro': 'es_VE.ISO8859-15',
861 'estonian': 'et_EE.ISO8859-1',
862 'et': 'et_EE.ISO8859-15',
863 'et_ee': 'et_EE.ISO8859-15',
864 'et_ee.iso88591': 'et_EE.ISO8859-1',
865 'et_ee.iso885913': 'et_EE.ISO8859-13',
866 'et_ee.iso885915': 'et_EE.ISO8859-15',
867 'et_ee.iso88594': 'et_EE.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000868 'et_ee@euro': 'et_EE.ISO8859-15',
869 'eu': 'eu_ES.ISO8859-1',
870 'eu_es': 'eu_ES.ISO8859-1',
871 'eu_es.iso88591': 'eu_ES.ISO8859-1',
872 'eu_es.iso885915': 'eu_ES.ISO8859-15',
873 'eu_es@euro': 'eu_ES.ISO8859-15',
874 'fa': 'fa_IR.UTF-8',
875 'fa_ir': 'fa_IR.UTF-8',
876 'fa_ir.isiri3342': 'fa_IR.ISIRI-3342',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000877 'fi': 'fi_FI.ISO8859-15',
878 'fi_fi': 'fi_FI.ISO8859-15',
879 'fi_fi.88591': 'fi_FI.ISO8859-1',
880 'fi_fi.iso88591': 'fi_FI.ISO8859-1',
881 'fi_fi.iso885915': 'fi_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000882 'fi_fi.utf8@euro': 'fi_FI.UTF-8',
883 'fi_fi@euro': 'fi_FI.ISO8859-15',
884 'finnish': 'fi_FI.ISO8859-1',
885 'finnish.iso88591': 'fi_FI.ISO8859-1',
886 'fo': 'fo_FO.ISO8859-1',
887 'fo_fo': 'fo_FO.ISO8859-1',
888 'fo_fo.iso88591': 'fo_FO.ISO8859-1',
889 'fo_fo.iso885915': 'fo_FO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000890 'fo_fo@euro': 'fo_FO.ISO8859-15',
891 'fr': 'fr_FR.ISO8859-1',
892 'fr_be': 'fr_BE.ISO8859-1',
893 'fr_be.88591': 'fr_BE.ISO8859-1',
894 'fr_be.iso88591': 'fr_BE.ISO8859-1',
895 'fr_be.iso885915': 'fr_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000896 'fr_be@euro': 'fr_BE.ISO8859-15',
897 'fr_ca': 'fr_CA.ISO8859-1',
898 'fr_ca.88591': 'fr_CA.ISO8859-1',
899 'fr_ca.iso88591': 'fr_CA.ISO8859-1',
900 'fr_ca.iso885915': 'fr_CA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000901 'fr_ca@euro': 'fr_CA.ISO8859-15',
902 'fr_ch': 'fr_CH.ISO8859-1',
903 'fr_ch.88591': 'fr_CH.ISO8859-1',
904 'fr_ch.iso88591': 'fr_CH.ISO8859-1',
905 'fr_ch.iso885915': 'fr_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000906 'fr_ch@euro': 'fr_CH.ISO8859-15',
907 'fr_fr': 'fr_FR.ISO8859-1',
908 'fr_fr.88591': 'fr_FR.ISO8859-1',
909 'fr_fr.iso88591': 'fr_FR.ISO8859-1',
910 'fr_fr.iso885915': 'fr_FR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000911 'fr_fr@euro': 'fr_FR.ISO8859-15',
912 'fr_lu': 'fr_LU.ISO8859-1',
913 'fr_lu.88591': 'fr_LU.ISO8859-1',
914 'fr_lu.iso88591': 'fr_LU.ISO8859-1',
915 'fr_lu.iso885915': 'fr_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000916 'fr_lu@euro': 'fr_LU.ISO8859-15',
917 'fran\xe7ais': 'fr_FR.ISO8859-1',
918 'fre_fr': 'fr_FR.ISO8859-1',
919 'fre_fr.8859': 'fr_FR.ISO8859-1',
920 'french': 'fr_FR.ISO8859-1',
921 'french.iso88591': 'fr_CH.ISO8859-1',
922 'french_france': 'fr_FR.ISO8859-1',
923 'french_france.8859': 'fr_FR.ISO8859-1',
924 'ga': 'ga_IE.ISO8859-1',
925 'ga_ie': 'ga_IE.ISO8859-1',
926 'ga_ie.iso88591': 'ga_IE.ISO8859-1',
927 'ga_ie.iso885914': 'ga_IE.ISO8859-14',
928 'ga_ie.iso885915': 'ga_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000929 'ga_ie@euro': 'ga_IE.ISO8859-15',
930 'galego': 'gl_ES.ISO8859-1',
931 'galician': 'gl_ES.ISO8859-1',
932 'gd': 'gd_GB.ISO8859-1',
933 'gd_gb': 'gd_GB.ISO8859-1',
934 'gd_gb.iso88591': 'gd_GB.ISO8859-1',
935 'gd_gb.iso885914': 'gd_GB.ISO8859-14',
936 'gd_gb.iso885915': 'gd_GB.ISO8859-15',
937 'gd_gb@euro': 'gd_GB.ISO8859-15',
938 'ger_de': 'de_DE.ISO8859-1',
939 'ger_de.8859': 'de_DE.ISO8859-1',
940 'german': 'de_DE.ISO8859-1',
941 'german.iso88591': 'de_CH.ISO8859-1',
942 'german_germany': 'de_DE.ISO8859-1',
943 'german_germany.8859': 'de_DE.ISO8859-1',
944 'gl': 'gl_ES.ISO8859-1',
945 'gl_es': 'gl_ES.ISO8859-1',
946 'gl_es.iso88591': 'gl_ES.ISO8859-1',
947 'gl_es.iso885915': 'gl_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000948 'gl_es@euro': 'gl_ES.ISO8859-15',
949 'greek': 'el_GR.ISO8859-7',
950 'greek.iso88597': 'el_GR.ISO8859-7',
951 'gv': 'gv_GB.ISO8859-1',
952 'gv_gb': 'gv_GB.ISO8859-1',
953 'gv_gb.iso88591': 'gv_GB.ISO8859-1',
954 'gv_gb.iso885914': 'gv_GB.ISO8859-14',
955 'gv_gb.iso885915': 'gv_GB.ISO8859-15',
956 'gv_gb@euro': 'gv_GB.ISO8859-15',
957 'he': 'he_IL.ISO8859-8',
958 'he_il': 'he_IL.ISO8859-8',
959 'he_il.cp1255': 'he_IL.CP1255',
960 'he_il.iso88598': 'he_IL.ISO8859-8',
961 'he_il.microsoftcp1255': 'he_IL.CP1255',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000962 'hebrew': 'iw_IL.ISO8859-8',
963 'hebrew.iso88598': 'iw_IL.ISO8859-8',
964 'hi': 'hi_IN.ISCII-DEV',
965 'hi_in': 'hi_IN.ISCII-DEV',
966 'hi_in.isciidev': 'hi_IN.ISCII-DEV',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000967 'hr': 'hr_HR.ISO8859-2',
968 'hr_hr': 'hr_HR.ISO8859-2',
969 'hr_hr.iso88592': 'hr_HR.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000970 'hrvatski': 'hr_HR.ISO8859-2',
971 'hu': 'hu_HU.ISO8859-2',
972 'hu_hu': 'hu_HU.ISO8859-2',
973 'hu_hu.iso88592': 'hu_HU.ISO8859-2',
974 'hungarian': 'hu_HU.ISO8859-2',
975 'icelandic': 'is_IS.ISO8859-1',
976 'icelandic.iso88591': 'is_IS.ISO8859-1',
977 'id': 'id_ID.ISO8859-1',
978 'id_id': 'id_ID.ISO8859-1',
979 'in': 'id_ID.ISO8859-1',
980 'in_id': 'id_ID.ISO8859-1',
981 'is': 'is_IS.ISO8859-1',
982 'is_is': 'is_IS.ISO8859-1',
983 'is_is.iso88591': 'is_IS.ISO8859-1',
984 'is_is.iso885915': 'is_IS.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000985 'is_is@euro': 'is_IS.ISO8859-15',
986 'iso-8859-1': 'en_US.ISO8859-1',
987 'iso-8859-15': 'en_US.ISO8859-15',
988 'iso8859-1': 'en_US.ISO8859-1',
989 'iso8859-15': 'en_US.ISO8859-15',
990 'iso_8859_1': 'en_US.ISO8859-1',
991 'iso_8859_15': 'en_US.ISO8859-15',
992 'it': 'it_IT.ISO8859-1',
993 'it_ch': 'it_CH.ISO8859-1',
994 'it_ch.iso88591': 'it_CH.ISO8859-1',
995 'it_ch.iso885915': 'it_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000996 'it_ch@euro': 'it_CH.ISO8859-15',
997 'it_it': 'it_IT.ISO8859-1',
998 'it_it.88591': 'it_IT.ISO8859-1',
999 'it_it.iso88591': 'it_IT.ISO8859-1',
1000 'it_it.iso885915': 'it_IT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001001 'it_it@euro': 'it_IT.ISO8859-15',
1002 'italian': 'it_IT.ISO8859-1',
1003 'italian.iso88591': 'it_IT.ISO8859-1',
1004 'iu': 'iu_CA.NUNACOM-8',
1005 'iu_ca': 'iu_CA.NUNACOM-8',
1006 'iu_ca.nunacom8': 'iu_CA.NUNACOM-8',
1007 'iw': 'he_IL.ISO8859-8',
1008 'iw_il': 'he_IL.ISO8859-8',
1009 'iw_il.iso88598': 'he_IL.ISO8859-8',
1010 'ja': 'ja_JP.eucJP',
1011 'ja.jis': 'ja_JP.JIS7',
1012 'ja.sjis': 'ja_JP.SJIS',
1013 'ja_jp': 'ja_JP.eucJP',
1014 'ja_jp.ajec': 'ja_JP.eucJP',
1015 'ja_jp.euc': 'ja_JP.eucJP',
1016 'ja_jp.eucjp': 'ja_JP.eucJP',
1017 'ja_jp.iso-2022-jp': 'ja_JP.JIS7',
1018 'ja_jp.iso2022jp': 'ja_JP.JIS7',
1019 'ja_jp.jis': 'ja_JP.JIS7',
1020 'ja_jp.jis7': 'ja_JP.JIS7',
1021 'ja_jp.mscode': 'ja_JP.SJIS',
1022 'ja_jp.sjis': 'ja_JP.SJIS',
1023 'ja_jp.ujis': 'ja_JP.eucJP',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001024 'japan': 'ja_JP.eucJP',
1025 'japanese': 'ja_JP.eucJP',
1026 'japanese-euc': 'ja_JP.eucJP',
1027 'japanese.euc': 'ja_JP.eucJP',
1028 'japanese.sjis': 'ja_JP.SJIS',
1029 'jp_jp': 'ja_JP.eucJP',
1030 'ka': 'ka_GE.GEORGIAN-ACADEMY',
1031 'ka_ge': 'ka_GE.GEORGIAN-ACADEMY',
1032 'ka_ge.georgianacademy': 'ka_GE.GEORGIAN-ACADEMY',
1033 'ka_ge.georgianps': 'ka_GE.GEORGIAN-PS',
1034 'ka_ge.georgianrs': 'ka_GE.GEORGIAN-ACADEMY',
1035 'kl': 'kl_GL.ISO8859-1',
1036 'kl_gl': 'kl_GL.ISO8859-1',
1037 'kl_gl.iso88591': 'kl_GL.ISO8859-1',
1038 'kl_gl.iso885915': 'kl_GL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001039 'kl_gl@euro': 'kl_GL.ISO8859-15',
1040 'ko': 'ko_KR.eucKR',
1041 'ko_kr': 'ko_KR.eucKR',
1042 'ko_kr.euc': 'ko_KR.eucKR',
1043 'ko_kr.euckr': 'ko_KR.eucKR',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001044 'korean': 'ko_KR.eucKR',
1045 'korean.euc': 'ko_KR.eucKR',
1046 'kw': 'kw_GB.ISO8859-1',
1047 'kw_gb': 'kw_GB.ISO8859-1',
1048 'kw_gb.iso88591': 'kw_GB.ISO8859-1',
1049 'kw_gb.iso885914': 'kw_GB.ISO8859-14',
1050 'kw_gb.iso885915': 'kw_GB.ISO8859-15',
1051 'kw_gb@euro': 'kw_GB.ISO8859-15',
1052 'lithuanian': 'lt_LT.ISO8859-13',
1053 'lo': 'lo_LA.MULELAO-1',
1054 'lo_la': 'lo_LA.MULELAO-1',
1055 'lo_la.cp1133': 'lo_LA.IBM-CP1133',
1056 'lo_la.ibmcp1133': 'lo_LA.IBM-CP1133',
1057 'lo_la.mulelao1': 'lo_LA.MULELAO-1',
1058 'lt': 'lt_LT.ISO8859-13',
1059 'lt_lt': 'lt_LT.ISO8859-13',
1060 'lt_lt.iso885913': 'lt_LT.ISO8859-13',
1061 'lt_lt.iso88594': 'lt_LT.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001062 'lv': 'lv_LV.ISO8859-13',
1063 'lv_lv': 'lv_LV.ISO8859-13',
1064 'lv_lv.iso885913': 'lv_LV.ISO8859-13',
1065 'lv_lv.iso88594': 'lv_LV.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001066 'mi': 'mi_NZ.ISO8859-1',
1067 'mi_nz': 'mi_NZ.ISO8859-1',
1068 'mi_nz.iso88591': 'mi_NZ.ISO8859-1',
1069 'mk': 'mk_MK.ISO8859-5',
1070 'mk_mk': 'mk_MK.ISO8859-5',
1071 'mk_mk.cp1251': 'mk_MK.CP1251',
1072 'mk_mk.iso88595': 'mk_MK.ISO8859-5',
1073 'mk_mk.microsoftcp1251': 'mk_MK.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001074 'ms': 'ms_MY.ISO8859-1',
1075 'ms_my': 'ms_MY.ISO8859-1',
1076 'ms_my.iso88591': 'ms_MY.ISO8859-1',
1077 'mt': 'mt_MT.ISO8859-3',
1078 'mt_mt': 'mt_MT.ISO8859-3',
1079 'mt_mt.iso88593': 'mt_MT.ISO8859-3',
1080 'nb': 'nb_NO.ISO8859-1',
1081 'nb_no': 'nb_NO.ISO8859-1',
1082 'nb_no.88591': 'nb_NO.ISO8859-1',
1083 'nb_no.iso88591': 'nb_NO.ISO8859-1',
1084 'nb_no.iso885915': 'nb_NO.ISO8859-15',
1085 'nb_no@euro': 'nb_NO.ISO8859-15',
1086 'nl': 'nl_NL.ISO8859-1',
1087 'nl_be': 'nl_BE.ISO8859-1',
1088 'nl_be.88591': 'nl_BE.ISO8859-1',
1089 'nl_be.iso88591': 'nl_BE.ISO8859-1',
1090 'nl_be.iso885915': 'nl_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001091 'nl_be@euro': 'nl_BE.ISO8859-15',
1092 'nl_nl': 'nl_NL.ISO8859-1',
1093 'nl_nl.88591': 'nl_NL.ISO8859-1',
1094 'nl_nl.iso88591': 'nl_NL.ISO8859-1',
1095 'nl_nl.iso885915': 'nl_NL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001096 'nl_nl@euro': 'nl_NL.ISO8859-15',
1097 'nn': 'nn_NO.ISO8859-1',
1098 'nn_no': 'nn_NO.ISO8859-1',
1099 'nn_no.88591': 'nn_NO.ISO8859-1',
1100 'nn_no.iso88591': 'nn_NO.ISO8859-1',
1101 'nn_no.iso885915': 'nn_NO.ISO8859-15',
1102 'nn_no@euro': 'nn_NO.ISO8859-15',
1103 'no': 'no_NO.ISO8859-1',
1104 'no@nynorsk': 'ny_NO.ISO8859-1',
1105 'no_no': 'no_NO.ISO8859-1',
1106 'no_no.88591': 'no_NO.ISO8859-1',
1107 'no_no.iso88591': 'no_NO.ISO8859-1',
1108 'no_no.iso885915': 'no_NO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001109 'no_no@euro': 'no_NO.ISO8859-15',
1110 'norwegian': 'no_NO.ISO8859-1',
1111 'norwegian.iso88591': 'no_NO.ISO8859-1',
1112 'ny': 'ny_NO.ISO8859-1',
1113 'ny_no': 'ny_NO.ISO8859-1',
1114 'ny_no.88591': 'ny_NO.ISO8859-1',
1115 'ny_no.iso88591': 'ny_NO.ISO8859-1',
1116 'ny_no.iso885915': 'ny_NO.ISO8859-15',
1117 'ny_no@euro': 'ny_NO.ISO8859-15',
1118 'nynorsk': 'nn_NO.ISO8859-1',
1119 'oc': 'oc_FR.ISO8859-1',
1120 'oc_fr': 'oc_FR.ISO8859-1',
1121 'oc_fr.iso88591': 'oc_FR.ISO8859-1',
1122 'oc_fr.iso885915': 'oc_FR.ISO8859-15',
1123 'oc_fr@euro': 'oc_FR.ISO8859-15',
1124 'pd': 'pd_US.ISO8859-1',
1125 'pd_de': 'pd_DE.ISO8859-1',
1126 'pd_de.iso88591': 'pd_DE.ISO8859-1',
1127 'pd_de.iso885915': 'pd_DE.ISO8859-15',
1128 'pd_de@euro': 'pd_DE.ISO8859-15',
1129 'pd_us': 'pd_US.ISO8859-1',
1130 'pd_us.iso88591': 'pd_US.ISO8859-1',
1131 'pd_us.iso885915': 'pd_US.ISO8859-15',
1132 'pd_us@euro': 'pd_US.ISO8859-15',
1133 'ph': 'ph_PH.ISO8859-1',
1134 'ph_ph': 'ph_PH.ISO8859-1',
1135 'ph_ph.iso88591': 'ph_PH.ISO8859-1',
1136 'pl': 'pl_PL.ISO8859-2',
1137 'pl_pl': 'pl_PL.ISO8859-2',
1138 'pl_pl.iso88592': 'pl_PL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001139 'polish': 'pl_PL.ISO8859-2',
1140 'portuguese': 'pt_PT.ISO8859-1',
1141 'portuguese.iso88591': 'pt_PT.ISO8859-1',
1142 'portuguese_brazil': 'pt_BR.ISO8859-1',
1143 'portuguese_brazil.8859': 'pt_BR.ISO8859-1',
1144 'posix': 'C',
1145 'posix-utf2': 'C',
1146 'pp': 'pp_AN.ISO8859-1',
1147 'pp_an': 'pp_AN.ISO8859-1',
1148 'pp_an.iso88591': 'pp_AN.ISO8859-1',
1149 'pt': 'pt_PT.ISO8859-1',
1150 'pt_br': 'pt_BR.ISO8859-1',
1151 'pt_br.88591': 'pt_BR.ISO8859-1',
1152 'pt_br.iso88591': 'pt_BR.ISO8859-1',
1153 'pt_br.iso885915': 'pt_BR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001154 'pt_br@euro': 'pt_BR.ISO8859-15',
1155 'pt_pt': 'pt_PT.ISO8859-1',
1156 'pt_pt.88591': 'pt_PT.ISO8859-1',
1157 'pt_pt.iso88591': 'pt_PT.ISO8859-1',
1158 'pt_pt.iso885915': 'pt_PT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001159 'pt_pt.utf8@euro': 'pt_PT.UTF-8',
1160 'pt_pt@euro': 'pt_PT.ISO8859-15',
1161 'ro': 'ro_RO.ISO8859-2',
1162 'ro_ro': 'ro_RO.ISO8859-2',
1163 'ro_ro.iso88592': 'ro_RO.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001164 'romanian': 'ro_RO.ISO8859-2',
1165 'ru': 'ru_RU.ISO8859-5',
1166 'ru_ru': 'ru_RU.ISO8859-5',
1167 'ru_ru.cp1251': 'ru_RU.CP1251',
1168 'ru_ru.iso88595': 'ru_RU.ISO8859-5',
1169 'ru_ru.koi8r': 'ru_RU.KOI8-R',
1170 'ru_ru.microsoftcp1251': 'ru_RU.CP1251',
1171 'ru_ua': 'ru_UA.KOI8-U',
1172 'ru_ua.cp1251': 'ru_UA.CP1251',
1173 'ru_ua.koi8u': 'ru_UA.KOI8-U',
1174 'ru_ua.microsoftcp1251': 'ru_UA.CP1251',
1175 'rumanian': 'ro_RO.ISO8859-2',
1176 'russian': 'ru_RU.ISO8859-5',
1177 'se_no': 'se_NO.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001178 'serbocroatian': 'sh_YU.ISO8859-2',
1179 'sh': 'sh_YU.ISO8859-2',
1180 'sh_hr': 'sh_HR.ISO8859-2',
1181 'sh_hr.iso88592': 'sh_HR.ISO8859-2',
1182 'sh_sp': 'sh_YU.ISO8859-2',
1183 'sh_yu': 'sh_YU.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001184 'sk': 'sk_SK.ISO8859-2',
1185 'sk_sk': 'sk_SK.ISO8859-2',
1186 'sk_sk.iso88592': 'sk_SK.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001187 'sl': 'sl_SI.ISO8859-2',
1188 'sl_cs': 'sl_CS.ISO8859-2',
1189 'sl_si': 'sl_SI.ISO8859-2',
1190 'sl_si.iso88592': 'sl_SI.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001191 'slovak': 'sk_SK.ISO8859-2',
1192 'slovene': 'sl_SI.ISO8859-2',
1193 'slovenian': 'sl_SI.ISO8859-2',
1194 'sp': 'sp_YU.ISO8859-5',
1195 'sp_yu': 'sp_YU.ISO8859-5',
1196 'spanish': 'es_ES.ISO8859-1',
1197 'spanish.iso88591': 'es_ES.ISO8859-1',
1198 'spanish_spain': 'es_ES.ISO8859-1',
1199 'spanish_spain.8859': 'es_ES.ISO8859-1',
1200 'sq': 'sq_AL.ISO8859-2',
1201 'sq_al': 'sq_AL.ISO8859-2',
1202 'sq_al.iso88592': 'sq_AL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001203 'sr': 'sr_YU.ISO8859-5',
1204 'sr@cyrillic': 'sr_YU.ISO8859-5',
1205 'sr_sp': 'sr_SP.ISO8859-2',
1206 'sr_yu': 'sr_YU.ISO8859-5',
1207 'sr_yu.cp1251@cyrillic': 'sr_YU.CP1251',
1208 'sr_yu.iso88592': 'sr_YU.ISO8859-2',
1209 'sr_yu.iso88595': 'sr_YU.ISO8859-5',
1210 'sr_yu.iso88595@cyrillic': 'sr_YU.ISO8859-5',
1211 'sr_yu.microsoftcp1251@cyrillic': 'sr_YU.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001212 'sr_yu.utf8@cyrillic': 'sr_YU.UTF-8',
1213 'sr_yu@cyrillic': 'sr_YU.ISO8859-5',
1214 'sv': 'sv_SE.ISO8859-1',
1215 'sv_fi': 'sv_FI.ISO8859-1',
1216 'sv_fi.iso88591': 'sv_FI.ISO8859-1',
1217 'sv_fi.iso885915': 'sv_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001218 'sv_fi@euro': 'sv_FI.ISO8859-15',
1219 'sv_se': 'sv_SE.ISO8859-1',
1220 'sv_se.88591': 'sv_SE.ISO8859-1',
1221 'sv_se.iso88591': 'sv_SE.ISO8859-1',
1222 'sv_se.iso885915': 'sv_SE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001223 'sv_se@euro': 'sv_SE.ISO8859-15',
1224 'swedish': 'sv_SE.ISO8859-1',
1225 'swedish.iso88591': 'sv_SE.ISO8859-1',
1226 'ta': 'ta_IN.TSCII-0',
1227 'ta_in': 'ta_IN.TSCII-0',
1228 'ta_in.tscii': 'ta_IN.TSCII-0',
1229 'ta_in.tscii0': 'ta_IN.TSCII-0',
1230 'tg': 'tg_TJ.KOI8-C',
1231 'tg_tj': 'tg_TJ.KOI8-C',
1232 'tg_tj.koi8c': 'tg_TJ.KOI8-C',
1233 'th': 'th_TH.ISO8859-11',
1234 'th_th': 'th_TH.ISO8859-11',
1235 'th_th.iso885911': 'th_TH.ISO8859-11',
1236 'th_th.tactis': 'th_TH.TIS620',
1237 'th_th.tis620': 'th_TH.TIS620',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001238 'thai': 'th_TH.ISO8859-11',
1239 'tl': 'tl_PH.ISO8859-1',
1240 'tl_ph': 'tl_PH.ISO8859-1',
1241 'tl_ph.iso88591': 'tl_PH.ISO8859-1',
1242 'tr': 'tr_TR.ISO8859-9',
1243 'tr_tr': 'tr_TR.ISO8859-9',
1244 'tr_tr.iso88599': 'tr_TR.ISO8859-9',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001245 'tt': 'tt_RU.TATAR-CYR',
1246 'tt_ru': 'tt_RU.TATAR-CYR',
1247 'tt_ru.koi8c': 'tt_RU.KOI8-C',
1248 'tt_ru.tatarcyr': 'tt_RU.TATAR-CYR',
1249 'turkish': 'tr_TR.ISO8859-9',
1250 'turkish.iso88599': 'tr_TR.ISO8859-9',
1251 'uk': 'uk_UA.KOI8-U',
1252 'uk_ua': 'uk_UA.KOI8-U',
1253 'uk_ua.cp1251': 'uk_UA.CP1251',
1254 'uk_ua.iso88595': 'uk_UA.ISO8859-5',
1255 'uk_ua.koi8u': 'uk_UA.KOI8-U',
1256 'uk_ua.microsoftcp1251': 'uk_UA.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001257 'univ': 'en_US.utf',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001258 'universal': 'en_US.utf',
1259 'universal.utf8@ucs4': 'en_US.UTF-8',
1260 'ur': 'ur_PK.CP1256',
1261 'ur_pk': 'ur_PK.CP1256',
1262 'ur_pk.cp1256': 'ur_PK.CP1256',
1263 'ur_pk.microsoftcp1256': 'ur_PK.CP1256',
1264 'uz': 'uz_UZ.UTF-8',
1265 'uz_uz': 'uz_UZ.UTF-8',
1266 'vi': 'vi_VN.TCVN',
1267 'vi_vn': 'vi_VN.TCVN',
1268 'vi_vn.tcvn': 'vi_VN.TCVN',
1269 'vi_vn.tcvn5712': 'vi_VN.TCVN',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001270 'vi_vn.viscii': 'vi_VN.VISCII',
1271 'vi_vn.viscii111': 'vi_VN.VISCII',
1272 'wa': 'wa_BE.ISO8859-1',
1273 'wa_be': 'wa_BE.ISO8859-1',
1274 'wa_be.iso88591': 'wa_BE.ISO8859-1',
1275 'wa_be.iso885915': 'wa_BE.ISO8859-15',
1276 'wa_be@euro': 'wa_BE.ISO8859-15',
1277 'yi': 'yi_US.CP1255',
1278 'yi_us': 'yi_US.CP1255',
1279 'yi_us.cp1255': 'yi_US.CP1255',
1280 'yi_us.microsoftcp1255': 'yi_US.CP1255',
1281 'zh': 'zh_CN.eucCN',
1282 'zh_cn': 'zh_CN.gb2312',
1283 'zh_cn.big5': 'zh_TW.big5',
1284 'zh_cn.euc': 'zh_CN.eucCN',
1285 'zh_cn.gb18030': 'zh_CN.gb18030',
1286 'zh_cn.gb2312': 'zh_CN.gb2312',
1287 'zh_cn.gbk': 'zh_CN.gbk',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001288 'zh_hk': 'zh_HK.big5hkscs',
1289 'zh_hk.big5': 'zh_HK.big5',
1290 'zh_hk.big5hkscs': 'zh_HK.big5hkscs',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001291 'zh_tw': 'zh_TW.big5',
1292 'zh_tw.big5': 'zh_TW.big5',
1293 'zh_tw.euc': 'zh_TW.eucTW',
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001294}
1295
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001296#
Georg Brandlb709c2c2006-01-20 09:07:35 +00001297# This maps Windows language identifiers to locale strings.
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001298#
Tim Peters777f1082006-01-20 20:03:24 +00001299# This list has been updated from
Georg Brandlb709c2c2006-01-20 09:07:35 +00001300# http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/nls_238z.asp
1301# to include every locale up to Windows XP.
Fredrik Lundh37a09822002-10-19 20:19:10 +00001302#
Georg Brandl5035c1c2006-01-20 13:38:26 +00001303# NOTE: this mapping is incomplete. If your language is missing, please
1304# submit a bug report to Python bug manager, which you can find via:
1305# http://www.python.org/dev/
1306# Make sure you include the missing language identifier and the suggested
1307# locale code.
1308#
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001309
1310windows_locale = {
Georg Brandlb709c2c2006-01-20 09:07:35 +00001311 0x0436: "af_ZA", # Afrikaans
1312 0x041c: "sq_AL", # Albanian
1313 0x0401: "ar_SA", # Arabic - Saudi Arabia
1314 0x0801: "ar_IQ", # Arabic - Iraq
1315 0x0c01: "ar_EG", # Arabic - Egypt
1316 0x1001: "ar_LY", # Arabic - Libya
1317 0x1401: "ar_DZ", # Arabic - Algeria
1318 0x1801: "ar_MA", # Arabic - Morocco
1319 0x1c01: "ar_TN", # Arabic - Tunisia
1320 0x2001: "ar_OM", # Arabic - Oman
1321 0x2401: "ar_YE", # Arabic - Yemen
1322 0x2801: "ar_SY", # Arabic - Syria
1323 0x2c01: "ar_JO", # Arabic - Jordan
1324 0x3001: "ar_LB", # Arabic - Lebanon
1325 0x3401: "ar_KW", # Arabic - Kuwait
1326 0x3801: "ar_AE", # Arabic - United Arab Emirates
1327 0x3c01: "ar_BH", # Arabic - Bahrain
1328 0x4001: "ar_QA", # Arabic - Qatar
1329 0x042b: "hy_AM", # Armenian
1330 0x042c: "az_AZ", # Azeri Latin
1331 0x082c: "az_AZ", # Azeri - Cyrillic
1332 0x042d: "eu_ES", # Basque
1333 0x0423: "be_BY", # Belarusian
1334 0x0445: "bn_IN", # Begali
1335 0x201a: "bs_BA", # Bosnian
1336 0x141a: "bs_BA", # Bosnian - Cyrillic
1337 0x047e: "br_FR", # Breton - France
1338 0x0402: "bg_BG", # Bulgarian
1339 0x0403: "ca_ES", # Catalan
1340 0x0004: "zh_CHS",# Chinese - Simplified
1341 0x0404: "zh_TW", # Chinese - Taiwan
1342 0x0804: "zh_CN", # Chinese - PRC
1343 0x0c04: "zh_HK", # Chinese - Hong Kong S.A.R.
1344 0x1004: "zh_SG", # Chinese - Singapore
1345 0x1404: "zh_MO", # Chinese - Macao S.A.R.
1346 0x7c04: "zh_CHT",# Chinese - Traditional
1347 0x041a: "hr_HR", # Croatian
1348 0x101a: "hr_BA", # Croatian - Bosnia
1349 0x0405: "cs_CZ", # Czech
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001350 0x0406: "da_DK", # Danish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001351 0x048c: "gbz_AF",# Dari - Afghanistan
1352 0x0465: "div_MV",# Divehi - Maldives
1353 0x0413: "nl_NL", # Dutch - The Netherlands
1354 0x0813: "nl_BE", # Dutch - Belgium
1355 0x0409: "en_US", # English - United States
1356 0x0809: "en_GB", # English - United Kingdom
1357 0x0c09: "en_AU", # English - Australia
1358 0x1009: "en_CA", # English - Canada
1359 0x1409: "en_NZ", # English - New Zealand
1360 0x1809: "en_IE", # English - Ireland
1361 0x1c09: "en_ZA", # English - South Africa
1362 0x2009: "en_JA", # English - Jamaica
1363 0x2409: "en_CB", # English - Carribbean
1364 0x2809: "en_BZ", # English - Belize
1365 0x2c09: "en_TT", # English - Trinidad
1366 0x3009: "en_ZW", # English - Zimbabwe
1367 0x3409: "en_PH", # English - Phillippines
1368 0x0425: "et_EE", # Estonian
1369 0x0438: "fo_FO", # Faroese
1370 0x0464: "fil_PH",# Filipino
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001371 0x040b: "fi_FI", # Finnish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001372 0x040c: "fr_FR", # French - France
1373 0x080c: "fr_BE", # French - Belgium
1374 0x0c0c: "fr_CA", # French - Canada
1375 0x100c: "fr_CH", # French - Switzerland
1376 0x140c: "fr_LU", # French - Luxembourg
1377 0x180c: "fr_MC", # French - Monaco
1378 0x0462: "fy_NL", # Frisian - Netherlands
1379 0x0456: "gl_ES", # Galician
1380 0x0437: "ka_GE", # Georgian
1381 0x0407: "de_DE", # German - Germany
1382 0x0807: "de_CH", # German - Switzerland
1383 0x0c07: "de_AT", # German - Austria
1384 0x1007: "de_LU", # German - Luxembourg
1385 0x1407: "de_LI", # German - Liechtenstein
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001386 0x0408: "el_GR", # Greek
Georg Brandlb709c2c2006-01-20 09:07:35 +00001387 0x0447: "gu_IN", # Gujarati
1388 0x040d: "he_IL", # Hebrew
1389 0x0439: "hi_IN", # Hindi
1390 0x040e: "hu_HU", # Hungarian
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001391 0x040f: "is_IS", # Icelandic
Georg Brandlb709c2c2006-01-20 09:07:35 +00001392 0x0421: "id_ID", # Indonesian
1393 0x045d: "iu_CA", # Inuktitut
1394 0x085d: "iu_CA", # Inuktitut - Latin
1395 0x083c: "ga_IE", # Irish - Ireland
1396 0x0434: "xh_ZA", # Xhosa - South Africa
1397 0x0435: "zu_ZA", # Zulu
1398 0x0410: "it_IT", # Italian - Italy
1399 0x0810: "it_CH", # Italian - Switzerland
1400 0x0411: "ja_JP", # Japanese
1401 0x044b: "kn_IN", # Kannada - India
1402 0x043f: "kk_KZ", # Kazakh
1403 0x0457: "kok_IN",# Konkani
1404 0x0412: "ko_KR", # Korean
1405 0x0440: "ky_KG", # Kyrgyz
1406 0x0426: "lv_LV", # Latvian
1407 0x0427: "lt_LT", # Lithuanian
1408 0x046e: "lb_LU", # Luxembourgish
1409 0x042f: "mk_MK", # FYRO Macedonian
1410 0x043e: "ms_MY", # Malay - Malaysia
1411 0x083e: "ms_BN", # Malay - Brunei
1412 0x044c: "ml_IN", # Malayalam - India
1413 0x043a: "mt_MT", # Maltese
1414 0x0481: "mi_NZ", # Maori
1415 0x047a: "arn_CL",# Mapudungun
1416 0x044e: "mr_IN", # Marathi
1417 0x047c: "moh_CA",# Mohawk - Canada
1418 0x0450: "mn_MN", # Mongolian
1419 0x0461: "ne_NP", # Nepali
1420 0x0414: "nb_NO", # Norwegian - Bokmal
1421 0x0814: "nn_NO", # Norwegian - Nynorsk
1422 0x0482: "oc_FR", # Occitan - France
1423 0x0448: "or_IN", # Oriya - India
1424 0x0463: "ps_AF", # Pashto - Afghanistan
1425 0x0429: "fa_IR", # Persian
1426 0x0415: "pl_PL", # Polish
1427 0x0416: "pt_BR", # Portuguese - Brazil
1428 0x0816: "pt_PT", # Portuguese - Portugal
1429 0x0446: "pa_IN", # Punjabi
1430 0x046b: "quz_BO",# Quechua (Bolivia)
1431 0x086b: "quz_EC",# Quechua (Ecuador)
1432 0x0c6b: "quz_PE",# Quechua (Peru)
1433 0x0418: "ro_RO", # Romanian - Romania
1434 0x0417: "rm_CH", # Raeto-Romanese
1435 0x0419: "ru_RU", # Russian
1436 0x243b: "smn_FI",# Sami Finland
1437 0x103b: "smj_NO",# Sami Norway
1438 0x143b: "smj_SE",# Sami Sweden
1439 0x043b: "se_NO", # Sami Northern Norway
1440 0x083b: "se_SE", # Sami Northern Sweden
1441 0x0c3b: "se_FI", # Sami Northern Finland
1442 0x203b: "sms_FI",# Sami Skolt
1443 0x183b: "sma_NO",# Sami Southern Norway
1444 0x1c3b: "sma_SE",# Sami Southern Sweden
1445 0x044f: "sa_IN", # Sanskrit
1446 0x0c1a: "sr_SP", # Serbian - Cyrillic
1447 0x1c1a: "sr_BA", # Serbian - Bosnia Cyrillic
1448 0x081a: "sr_SP", # Serbian - Latin
1449 0x181a: "sr_BA", # Serbian - Bosnia Latin
1450 0x046c: "ns_ZA", # Northern Sotho
1451 0x0432: "tn_ZA", # Setswana - Southern Africa
1452 0x041b: "sk_SK", # Slovak
1453 0x0424: "sl_SI", # Slovenian
1454 0x040a: "es_ES", # Spanish - Spain
1455 0x080a: "es_MX", # Spanish - Mexico
1456 0x0c0a: "es_ES", # Spanish - Spain (Modern)
1457 0x100a: "es_GT", # Spanish - Guatemala
1458 0x140a: "es_CR", # Spanish - Costa Rica
1459 0x180a: "es_PA", # Spanish - Panama
1460 0x1c0a: "es_DO", # Spanish - Dominican Republic
1461 0x200a: "es_VE", # Spanish - Venezuela
1462 0x240a: "es_CO", # Spanish - Colombia
1463 0x280a: "es_PE", # Spanish - Peru
1464 0x2c0a: "es_AR", # Spanish - Argentina
1465 0x300a: "es_EC", # Spanish - Ecuador
1466 0x340a: "es_CL", # Spanish - Chile
1467 0x380a: "es_UR", # Spanish - Uruguay
1468 0x3c0a: "es_PY", # Spanish - Paraguay
1469 0x400a: "es_BO", # Spanish - Bolivia
1470 0x440a: "es_SV", # Spanish - El Salvador
1471 0x480a: "es_HN", # Spanish - Honduras
1472 0x4c0a: "es_NI", # Spanish - Nicaragua
1473 0x500a: "es_PR", # Spanish - Puerto Rico
1474 0x0441: "sw_KE", # Swahili
1475 0x041d: "sv_SE", # Swedish - Sweden
1476 0x081d: "sv_FI", # Swedish - Finland
1477 0x045a: "syr_SY",# Syriac
1478 0x0449: "ta_IN", # Tamil
1479 0x0444: "tt_RU", # Tatar
1480 0x044a: "te_IN", # Telugu
1481 0x041e: "th_TH", # Thai
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001482 0x041f: "tr_TR", # Turkish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001483 0x0422: "uk_UA", # Ukrainian
1484 0x0420: "ur_PK", # Urdu
1485 0x0820: "ur_IN", # Urdu - India
1486 0x0443: "uz_UZ", # Uzbek - Latin
1487 0x0843: "uz_UZ", # Uzbek - Cyrillic
1488 0x042a: "vi_VN", # Vietnamese
1489 0x0452: "cy_GB", # Welsh
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001490}
1491
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001492def _print_locale():
1493
1494 """ Test function.
1495 """
1496 categories = {}
1497 def _init_categories(categories=categories):
1498 for k,v in globals().items():
1499 if k[:3] == 'LC_':
1500 categories[k] = v
1501 _init_categories()
1502 del categories['LC_ALL']
1503
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001504 print('Locale defaults as determined by getdefaultlocale():')
1505 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001506 lang, enc = getdefaultlocale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001507 print('Language: ', lang or '(undefined)')
1508 print('Encoding: ', enc or '(undefined)')
1509 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001510
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001511 print('Locale settings on startup:')
1512 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001513 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001514 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001515 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001516 print(' Language: ', lang or '(undefined)')
1517 print(' Encoding: ', enc or '(undefined)')
1518 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001519
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001520 print()
1521 print('Locale settings after calling resetlocale():')
1522 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001523 resetlocale()
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()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001530
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001531 try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001532 setlocale(LC_ALL, "")
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001533 except:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001534 print('NOTE:')
1535 print('setlocale(LC_ALL, "") does not support the default locale')
1536 print('given in the OS environment variables.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001537 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001538 print()
1539 print('Locale settings after calling setlocale(LC_ALL, ""):')
1540 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001541 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001542 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001543 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001544 print(' Language: ', lang or '(undefined)')
1545 print(' Encoding: ', enc or '(undefined)')
1546 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001547
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001548###
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001549
Tim Peters1baf8292001-01-24 10:13:46 +00001550try:
1551 LC_MESSAGES
Skip Montanaro0897f0c2002-03-25 21:40:36 +00001552except NameError:
Tim Peters1baf8292001-01-24 10:13:46 +00001553 pass
1554else:
1555 __all__.append("LC_MESSAGES")
1556
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001557if __name__=='__main__':
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001558 print('Locale aliasing:')
1559 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001560 _print_locale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001561 print()
1562 print('Number formatting:')
1563 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001564 _test()