blob: ce0ca817eb138d5fdb43a22fffb2190581268dbf [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
Marc-André Lemburg23481142000-06-08 17:49:41 +000029try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +000030
Marc-André Lemburg23481142000-06-08 17:49:41 +000031 from _locale import *
32
33except ImportError:
34
Fredrik Lundh6c86b992000-07-09 17:12:58 +000035 # Locale emulation
36
Marc-André Lemburg23481142000-06-08 17:49:41 +000037 CHAR_MAX = 127
38 LC_ALL = 6
39 LC_COLLATE = 3
40 LC_CTYPE = 0
41 LC_MESSAGES = 5
42 LC_MONETARY = 4
43 LC_NUMERIC = 1
44 LC_TIME = 2
45 Error = ValueError
46
47 def localeconv():
Fredrik Lundh6c86b992000-07-09 17:12:58 +000048 """ localeconv() -> dict.
Marc-André Lemburg23481142000-06-08 17:49:41 +000049 Returns numeric and monetary locale-specific parameters.
50 """
51 # 'C' locale default values
52 return {'grouping': [127],
53 'currency_symbol': '',
54 'n_sign_posn': 127,
Fredrik Lundh6c86b992000-07-09 17:12:58 +000055 'p_cs_precedes': 127,
56 'n_cs_precedes': 127,
57 'mon_grouping': [],
Marc-André Lemburg23481142000-06-08 17:49:41 +000058 'n_sep_by_space': 127,
59 'decimal_point': '.',
60 'negative_sign': '',
61 'positive_sign': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000062 'p_sep_by_space': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000063 'int_curr_symbol': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000064 'p_sign_posn': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000065 'thousands_sep': '',
Fredrik Lundh6c86b992000-07-09 17:12:58 +000066 'mon_thousands_sep': '',
67 'frac_digits': 127,
Marc-André Lemburg23481142000-06-08 17:49:41 +000068 'mon_decimal_point': '',
69 'int_frac_digits': 127}
Fredrik Lundh6c86b992000-07-09 17:12:58 +000070
Marc-André Lemburg23481142000-06-08 17:49:41 +000071 def setlocale(category, value=None):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000072 """ setlocale(integer,string=None) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000073 Activates/queries locale processing.
74 """
Martin v. Löwis103d6e72003-03-30 15:42:13 +000075 if value not in (None, '', 'C'):
Collin Winterce36ad82007-08-30 01:19:48 +000076 raise Error('_locale emulation only supports "C" locale')
Marc-André Lemburg23481142000-06-08 17:49:41 +000077 return 'C'
78
79 def strcoll(a,b):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000080 """ strcoll(string,string) -> int.
Marc-André Lemburg23481142000-06-08 17:49:41 +000081 Compares two strings according to the locale.
82 """
83 return cmp(a,b)
84
85 def strxfrm(s):
Fredrik Lundh6c86b992000-07-09 17:12:58 +000086 """ strxfrm(string) -> string.
Marc-André Lemburg23481142000-06-08 17:49:41 +000087 Returns a string that behaves for cmp locale-aware.
88 """
89 return s
Marc-André Lemburg5431bc32000-06-07 09:11:40 +000090
91### Number formatting APIs
92
93# Author: Martin von Loewis
Thomas Wouters477c8d52006-05-27 19:21:47 +000094# improved by Georg Brandl
Guido van Rossumeef1d4e1997-11-19 19:01:43 +000095
96#perform the grouping from right to left
Thomas Wouters477c8d52006-05-27 19:21:47 +000097def _group(s, monetary=False):
98 conv = localeconv()
99 thousands_sep = conv[monetary and 'mon_thousands_sep' or 'thousands_sep']
100 grouping = conv[monetary and 'mon_grouping' or 'grouping']
101 if not grouping:
102 return (s, 0)
103 result = ""
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000104 seps = 0
105 spaces = ""
106 if s[-1] == ' ':
107 sp = s.find(' ')
108 spaces = s[sp:]
109 s = s[:sp]
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000110 while s and grouping:
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000111 # if grouping is -1, we are done
Thomas Wouters477c8d52006-05-27 19:21:47 +0000112 if grouping[0] == CHAR_MAX:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000113 break
114 # 0: re-use last group ad infinitum
Thomas Wouters477c8d52006-05-27 19:21:47 +0000115 elif grouping[0] != 0:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000116 #process last group
Thomas Wouters477c8d52006-05-27 19:21:47 +0000117 group = grouping[0]
118 grouping = grouping[1:]
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000119 if result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000120 result = s[-group:] + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000121 seps += 1
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000122 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000123 result = s[-group:]
124 s = s[:-group]
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000125 if s and s[-1] not in "0123456789":
126 # the leading string is only spaces and signs
Thomas Wouters477c8d52006-05-27 19:21:47 +0000127 return s + result + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000128 if not result:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000129 return s + spaces, seps
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000130 if s:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000131 result = s + thousands_sep + result
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000132 seps += 1
Thomas Wouters477c8d52006-05-27 19:21:47 +0000133 return result + spaces, seps
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000134
Thomas Wouters477c8d52006-05-27 19:21:47 +0000135def format(percent, value, grouping=False, monetary=False, *additional):
136 """Returns the locale-aware substitution of a %? specifier
137 (percent).
138
139 additional is for format strings which contain one or more
140 '*' modifiers."""
141 # this is only for one-percent-specifier strings and this should be checked
142 if percent[0] != '%':
143 raise ValueError("format() must be given exactly one %char "
144 "format specifier")
145 if additional:
146 formatted = percent % ((value,) + additional)
147 else:
148 formatted = percent % value
149 # floats and decimal ints need special action!
150 if percent[-1] in 'eEfFgG':
151 seps = 0
152 parts = formatted.split('.')
153 if grouping:
154 parts[0], seps = _group(parts[0], monetary=monetary)
155 decimal_point = localeconv()[monetary and 'mon_decimal_point'
156 or 'decimal_point']
157 formatted = decimal_point.join(parts)
158 while seps:
159 sp = formatted.find(' ')
160 if sp == -1: break
161 formatted = formatted[:sp] + formatted[sp+1:]
162 seps -= 1
163 elif percent[-1] in 'diu':
164 if grouping:
165 formatted = _group(formatted, monetary=monetary)[0]
166 return formatted
167
168import re, operator
169_percent_re = re.compile(r'%(?:\((?P<key>.*?)\))?'
170 r'(?P<modifiers>[-#0-9 +*.hlL]*?)[eEfFgGdiouxXcrs%]')
171
172def format_string(f, val, grouping=False):
173 """Formats a string in the same way that the % formatting would use,
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000174 but takes the current locale into account.
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000175 Grouping is applied if the third parameter is true."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000176 percents = list(_percent_re.finditer(f))
177 new_f = _percent_re.sub('%s', f)
178
179 if isinstance(val, tuple):
180 new_val = list(val)
181 i = 0
182 for perc in percents:
183 starcount = perc.group('modifiers').count('*')
184 new_val[i] = format(perc.group(), new_val[i], grouping, False, *new_val[i+1:i+1+starcount])
185 del new_val[i+1:i+1+starcount]
186 i += (1 + starcount)
187 val = tuple(new_val)
188 elif operator.isMappingType(val):
189 for perc in percents:
190 key = perc.group("key")
191 val[key] = format(perc.group(), val[key], grouping)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000192 else:
Thomas Wouters477c8d52006-05-27 19:21:47 +0000193 # val is a single value
194 val = format(percents[0].group(), val, grouping)
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000195
Thomas Wouters477c8d52006-05-27 19:21:47 +0000196 return new_f % val
Martin v. Löwis88ad12a2001-04-13 08:09:50 +0000197
Thomas Wouters477c8d52006-05-27 19:21:47 +0000198def currency(val, symbol=True, grouping=False, international=False):
199 """Formats val according to the currency settings
200 in the current locale."""
201 conv = localeconv()
202
203 # check for illegal values
204 digits = conv[international and 'int_frac_digits' or 'frac_digits']
205 if digits == 127:
206 raise ValueError("Currency formatting is not possible using "
207 "the 'C' locale.")
208
209 s = format('%%.%if' % digits, abs(val), grouping, monetary=True)
210 # '<' and '>' are markers if the sign must be inserted between symbol and value
211 s = '<' + s + '>'
212
213 if symbol:
214 smb = conv[international and 'int_curr_symbol' or 'currency_symbol']
215 precedes = conv[val<0 and 'n_cs_precedes' or 'p_cs_precedes']
216 separated = conv[val<0 and 'n_sep_by_space' or 'p_sep_by_space']
217
218 if precedes:
219 s = smb + (separated and ' ' or '') + s
220 else:
221 s = s + (separated and ' ' or '') + smb
222
223 sign_pos = conv[val<0 and 'n_sign_posn' or 'p_sign_posn']
224 sign = conv[val<0 and 'negative_sign' or 'positive_sign']
225
226 if sign_pos == 0:
227 s = '(' + s + ')'
228 elif sign_pos == 1:
229 s = sign + s
230 elif sign_pos == 2:
231 s = s + sign
232 elif sign_pos == 3:
233 s = s.replace('<', sign)
234 elif sign_pos == 4:
235 s = s.replace('>', sign)
236 else:
237 # the default if nothing specified;
238 # this should be the most fitting sign position
239 s = sign + s
240
241 return s.replace('<', '').replace('>', '')
Martin v. Löwisdb786872001-01-21 18:52:33 +0000242
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000243def str(val):
244 """Convert float to integer, taking the locale into account."""
Thomas Wouters477c8d52006-05-27 19:21:47 +0000245 return format("%.12g", val)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000246
Thomas Wouters477c8d52006-05-27 19:21:47 +0000247def atof(string, func=float):
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000248 "Parses a string as a float according to the locale settings."
249 #First, get rid of the grouping
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000250 ts = localeconv()['thousands_sep']
251 if ts:
Skip Montanaro249369c2004-04-10 16:39:32 +0000252 string = string.replace(ts, '')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000253 #next, replace the decimal point with a dot
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000254 dd = localeconv()['decimal_point']
255 if dd:
Skip Montanaro249369c2004-04-10 16:39:32 +0000256 string = string.replace(dd, '.')
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000257 #finally, parse the string
Skip Montanaro249369c2004-04-10 16:39:32 +0000258 return func(string)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000259
260def atoi(str):
261 "Converts a string to an integer according to the locale settings."
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000262 return atof(str, int)
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000263
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000264def _test():
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000265 setlocale(LC_ALL, "")
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000266 #do grouping
Thomas Wouters477c8d52006-05-27 19:21:47 +0000267 s1 = format("%d", 123456789,1)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000268 print(s1, "is", atoi(s1))
Guido van Rossumeef1d4e1997-11-19 19:01:43 +0000269 #standard formatting
Thomas Wouters477c8d52006-05-27 19:21:47 +0000270 s1 = str(3.14)
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000271 print(s1, "is", atof(s1))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000272
273### Locale name aliasing engine
274
275# Author: Marc-Andre Lemburg, mal@lemburg.com
Fredrik Lundh37a09822002-10-19 20:19:10 +0000276# Various tweaks by Fredrik Lundh <fredrik@pythonware.com>
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000277
278# store away the low-level version of setlocale (it's
279# overridden below)
280_setlocale = setlocale
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000281
282def normalize(localename):
283
284 """ Returns a normalized locale code for the given locale
285 name.
286
287 The returned locale code is formatted for use with
288 setlocale().
289
290 If normalization fails, the original name is returned
291 unchanged.
292
293 If the given encoding is not known, the function defaults to
294 the default encoding for the locale code just like setlocale()
295 does.
296
297 """
298 # Normalize the locale name and extract the encoding
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000299 fullname = localename.lower()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000300 if ':' in fullname:
301 # ':' is sometimes used as encoding delimiter.
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000302 fullname = fullname.replace(':', '.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000303 if '.' in fullname:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000304 langname, encoding = fullname.split('.')[:2]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000305 fullname = langname + '.' + encoding
306 else:
307 langname = fullname
308 encoding = ''
309
310 # First lookup: fullname (possibly with encoding)
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000311 norm_encoding = encoding.replace('-', '')
312 norm_encoding = norm_encoding.replace('_', '')
313 lookup_name = langname + '.' + encoding
314 code = locale_alias.get(lookup_name, None)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000315 if code is not None:
316 return code
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000317 #print 'first lookup failed'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000318
319 # Second try: langname (without encoding)
320 code = locale_alias.get(langname, None)
321 if code is not None:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000322 #print 'langname lookup succeeded'
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000323 if '.' in code:
Eric S. Raymondbe9b5072001-02-09 10:48:30 +0000324 langname, defenc = code.split('.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000325 else:
326 langname = code
327 defenc = ''
328 if encoding:
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000329 # Convert the encoding to a C lib compatible encoding string
330 norm_encoding = encodings.normalize_encoding(encoding)
331 #print 'norm encoding: %r' % norm_encoding
332 norm_encoding = encodings.aliases.aliases.get(norm_encoding,
333 norm_encoding)
334 #print 'aliased encoding: %r' % norm_encoding
335 encoding = locale_encoding_alias.get(norm_encoding,
336 norm_encoding)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000337 else:
338 encoding = defenc
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000339 #print 'found encoding %r' % encoding
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000340 if encoding:
341 return langname + '.' + encoding
342 else:
343 return langname
344
345 else:
346 return localename
347
348def _parse_localename(localename):
349
350 """ Parses the locale code for localename and returns the
351 result as tuple (language code, encoding).
352
353 The localename is normalized and passed through the locale
354 alias engine. A ValueError is raised in case the locale name
355 cannot be parsed.
356
357 The language code corresponds to RFC 1766. code and encoding
358 can be None in case the values cannot be determined or are
Jeremy Hyltona05e2932000-06-28 14:48:01 +0000359 unknown to this implementation.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000360
361 """
362 code = normalize(localename)
Georg Brandlb709c2c2006-01-20 09:07:35 +0000363 if '@' in code:
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000364 # Deal with locale modifiers
365 code, modifier = code.split('@')
366 if modifier == 'euro' and '.' not in code:
367 # Assume Latin-9 for @euro locales. This is bogus,
368 # since some systems may use other encodings for these
369 # locales. Also, we ignore other modifiers.
370 return code, 'iso-8859-15'
Tim Peters230a60c2002-11-09 05:08:07 +0000371
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000372 if '.' in code:
Raymond Hettinger346e67f2005-01-01 06:10:26 +0000373 return tuple(code.split('.')[:2])
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000374 elif code == 'C':
375 return None, None
Collin Winterce36ad82007-08-30 01:19:48 +0000376 raise ValueError('unknown locale: %s' % localename)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000377
378def _build_localename(localetuple):
379
380 """ Builds a locale code from the given tuple (language code,
381 encoding).
382
383 No aliasing or normalizing takes place.
384
385 """
386 language, encoding = localetuple
387 if language is None:
388 language = 'C'
389 if encoding is None:
390 return language
391 else:
392 return language + '.' + encoding
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000393
Matthias Klosef3f231f2005-09-20 07:02:49 +0000394def getdefaultlocale(envvars=('LC_ALL', 'LC_CTYPE', 'LANG', 'LANGUAGE')):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000395
396 """ Tries to determine the default locale settings and returns
397 them as tuple (language code, encoding).
398
399 According to POSIX, a program which has not called
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000400 setlocale(LC_ALL, "") runs using the portable 'C' locale.
401 Calling setlocale(LC_ALL, "") lets it use the default locale as
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000402 defined by the LANG variable. Since we don't want to interfere
Thomas Wouters7e474022000-07-16 12:04:32 +0000403 with the current locale setting we thus emulate the behavior
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000404 in the way described above.
405
406 To maintain compatibility with other platforms, not only the
407 LANG variable is tested, but a list of variables given as
408 envvars parameter. The first found to be defined will be
409 used. envvars defaults to the search path used in GNU gettext;
410 it must always contain the variable name 'LANG'.
411
412 Except for the code 'C', the language code corresponds to RFC
413 1766. code and encoding can be None in case the values cannot
414 be determined.
415
416 """
Fredrik Lundh04661322000-07-09 23:16:10 +0000417
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000418 try:
419 # check if it's supported by the _locale module
420 import _locale
421 code, encoding = _locale._getdefaultlocale()
Fredrik Lundh04661322000-07-09 23:16:10 +0000422 except (ImportError, AttributeError):
423 pass
424 else:
Fredrik Lundh663809e2000-07-10 19:32:19 +0000425 # make sure the code/encoding values are valid
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000426 if sys.platform == "win32" and code and code[:2] == "0x":
427 # map windows language identifier to language name
428 code = windows_locale.get(int(code, 0))
Fredrik Lundh663809e2000-07-10 19:32:19 +0000429 # ...add other platform-specific processing here, if
430 # necessary...
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000431 return code, encoding
Fredrik Lundh04661322000-07-09 23:16:10 +0000432
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000433 # fall back on POSIX behaviour
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000434 import os
435 lookup = os.environ.get
436 for variable in envvars:
437 localename = lookup(variable,None)
Martin v. Löwisc8ae31d2004-07-26 12:45:18 +0000438 if localename:
Matthias Klosef3f231f2005-09-20 07:02:49 +0000439 if variable == 'LANGUAGE':
440 localename = localename.split(':')[0]
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000441 break
442 else:
443 localename = 'C'
444 return _parse_localename(localename)
445
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000446
447def getlocale(category=LC_CTYPE):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000448
449 """ Returns the current setting for the given locale category as
450 tuple (language code, encoding).
451
452 category may be one of the LC_* value except LC_ALL. It
453 defaults to LC_CTYPE.
454
455 Except for the code 'C', the language code corresponds to RFC
456 1766. code and encoding can be None in case the values cannot
457 be determined.
458
459 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000460 localename = _setlocale(category)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000461 if category == LC_ALL and ';' in localename:
Collin Winterce36ad82007-08-30 01:19:48 +0000462 raise TypeError('category LC_ALL is not supported')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000463 return _parse_localename(localename)
464
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000465def setlocale(category, locale=None):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000466
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000467 """ Set the locale for the given category. The locale can be
468 a string, a locale tuple (language code, encoding), or None.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000469
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000470 Locale tuples are converted to strings the locale aliasing
471 engine. Locale strings are passed directly to the C lib.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000472
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000473 category may be given as one of the LC_* values.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000474
475 """
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000476 if locale and not isinstance(locale, _builtin_str):
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000477 # convert to string
478 locale = normalize(_build_localename(locale))
479 return _setlocale(category, locale)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000480
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000481def resetlocale(category=LC_ALL):
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000482
483 """ Sets the locale for category to the default setting.
484
485 The default setting is determined by calling
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000486 getdefaultlocale(). category defaults to LC_ALL.
487
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000488 """
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000489 _setlocale(category, _build_localename(getdefaultlocale()))
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000490
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000491if sys.platform in ('win32', 'darwin', 'mac'):
492 # On Win32, this will return the ANSI code page
493 # On the Mac, it should return the system encoding;
494 # it might return "ascii" instead
495 def getpreferredencoding(do_setlocale = True):
496 """Return the charset that the user is likely using."""
497 import _locale
Tim Petersa326f472002-11-05 03:49:09 +0000498 return _locale._getdefaultlocale()[1]
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000499else:
500 # On Unix, if CODESET is available, use that.
501 try:
502 CODESET
503 except NameError:
504 # Fall back to parsing environment variables :-(
505 def getpreferredencoding(do_setlocale = True):
506 """Return the charset that the user is likely using,
507 by looking at environment variables."""
508 return getdefaultlocale()[1]
509 else:
510 def getpreferredencoding(do_setlocale = True):
511 """Return the charset that the user is likely using,
512 according to the system configuration."""
513 if do_setlocale:
514 oldloc = setlocale(LC_CTYPE)
515 setlocale(LC_CTYPE, "")
516 result = nl_langinfo(CODESET)
517 setlocale(LC_CTYPE, oldloc)
518 return result
519 else:
520 return nl_langinfo(CODESET)
Tim Peters230a60c2002-11-09 05:08:07 +0000521
Martin v. Löwisf0a46682002-11-03 17:20:12 +0000522
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000523### Database
524#
525# The following data was extracted from the locale.alias file which
526# comes with X11 and then hand edited removing the explicit encoding
527# definitions and adding some more aliases. The file is usually
528# available as /usr/lib/X11/locale/locale.alias.
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000529#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000530
531#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000532# The local_encoding_alias table maps lowercase encoding alias names
533# to C locale encoding names (case-sensitive). Note that normalize()
534# first looks up the encoding in the encodings.aliases dictionary and
535# then applies this mapping to find the correct C lib name for the
536# encoding.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000537#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000538locale_encoding_alias = {
539
540 # Mappings for non-standard encoding names used in locale names
541 '437': 'C',
542 'c': 'C',
543 'en': 'ISO8859-1',
544 'jis': 'JIS7',
545 'jis7': 'JIS7',
546 'ajec': 'eucJP',
547
548 # Mappings from Python codec names to C lib encoding names
549 'ascii': 'ISO8859-1',
550 'latin_1': 'ISO8859-1',
551 'iso8859_1': 'ISO8859-1',
552 'iso8859_10': 'ISO8859-10',
553 'iso8859_11': 'ISO8859-11',
554 'iso8859_13': 'ISO8859-13',
555 'iso8859_14': 'ISO8859-14',
556 'iso8859_15': 'ISO8859-15',
557 'iso8859_2': 'ISO8859-2',
558 'iso8859_3': 'ISO8859-3',
559 'iso8859_4': 'ISO8859-4',
560 'iso8859_5': 'ISO8859-5',
561 'iso8859_6': 'ISO8859-6',
562 'iso8859_7': 'ISO8859-7',
563 'iso8859_8': 'ISO8859-8',
564 'iso8859_9': 'ISO8859-9',
565 'iso2022_jp': 'JIS7',
566 'shift_jis': 'SJIS',
567 'tactis': 'TACTIS',
568 'euc_jp': 'eucJP',
569 'euc_kr': 'eucKR',
Marc-André Lemburgb4cebd42004-12-13 19:56:01 +0000570 'utf_8': 'UTF8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000571 'koi8_r': 'KOI8-R',
572 'koi8_u': 'KOI8-U',
573 # XXX This list is still incomplete. If you know more
574 # mappings, please file a bug report. Thanks.
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000575}
576
Fredrik Lundh6c86b992000-07-09 17:12:58 +0000577#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000578# The locale_alias table maps lowercase alias names to C locale names
579# (case-sensitive). Encodings are always separated from the locale
580# name using a dot ('.'); they should only be given in case the
581# language name is needed to interpret the given encoding alias
582# correctly (CJK codes often have this need).
583#
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000584# Note that the normalize() function which uses this tables
585# removes '_' and '-' characters from the encoding part of the
586# locale name before doing the lookup. This saves a lot of
587# space in the table.
588#
589# MAL 2004-12-10:
590# Updated alias mapping to most recent locale.alias file
591# from X.org distribution using makelocalealias.py.
592#
593# These are the differences compared to the old mapping (Python 2.4
594# and older):
595#
596# updated 'bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
597# updated 'bg_bg' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
598# updated 'bulgarian' -> 'bg_BG.ISO8859-5' to 'bg_BG.CP1251'
599# updated 'cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
600# updated 'cz_cz' -> 'cz_CZ.ISO8859-2' to 'cs_CZ.ISO8859-2'
601# updated 'czech' -> 'cs_CS.ISO8859-2' to 'cs_CZ.ISO8859-2'
602# updated 'dutch' -> 'nl_BE.ISO8859-1' to 'nl_NL.ISO8859-1'
603# updated 'et' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
604# updated 'et_ee' -> 'et_EE.ISO8859-4' to 'et_EE.ISO8859-15'
605# updated 'fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
606# updated 'fi_fi' -> 'fi_FI.ISO8859-1' to 'fi_FI.ISO8859-15'
607# updated 'iw' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
608# updated 'iw_il' -> 'iw_IL.ISO8859-8' to 'he_IL.ISO8859-8'
609# updated 'japanese' -> 'ja_JP.SJIS' to 'ja_JP.eucJP'
610# updated 'lt' -> 'lt_LT.ISO8859-4' to 'lt_LT.ISO8859-13'
611# updated 'lv' -> 'lv_LV.ISO8859-4' to 'lv_LV.ISO8859-13'
612# updated 'sl' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
613# updated 'slovene' -> 'sl_CS.ISO8859-2' to 'sl_SI.ISO8859-2'
614# updated 'th_th' -> 'th_TH.TACTIS' to 'th_TH.ISO8859-11'
615# updated 'zh_cn' -> 'zh_CN.eucCN' to 'zh_CN.gb2312'
616# updated 'zh_cn.big5' -> 'zh_TW.eucTW' to 'zh_TW.big5'
617# updated 'zh_tw' -> 'zh_TW.eucTW' to 'zh_TW.big5'
618#
Marc-André Lemburg5431bc32000-06-07 09:11:40 +0000619locale_alias = {
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000620 'a3': 'a3_AZ.KOI8-C',
621 'a3_az': 'a3_AZ.KOI8-C',
622 'a3_az.koi8c': 'a3_AZ.KOI8-C',
623 'af': 'af_ZA.ISO8859-1',
624 'af_za': 'af_ZA.ISO8859-1',
625 'af_za.iso88591': 'af_ZA.ISO8859-1',
626 'am': 'am_ET.UTF-8',
627 'american': 'en_US.ISO8859-1',
628 'american.iso88591': 'en_US.ISO8859-1',
629 'ar': 'ar_AA.ISO8859-6',
630 'ar_aa': 'ar_AA.ISO8859-6',
631 'ar_aa.iso88596': 'ar_AA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000632 'ar_ae': 'ar_AE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000633 'ar_bh': 'ar_BH.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000634 'ar_dz': 'ar_DZ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000635 'ar_eg': 'ar_EG.ISO8859-6',
636 'ar_eg.iso88596': 'ar_EG.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000637 'ar_iq': 'ar_IQ.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000638 'ar_jo': 'ar_JO.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000639 'ar_kw': 'ar_KW.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000640 'ar_lb': 'ar_LB.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000641 'ar_ly': 'ar_LY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000642 'ar_ma': 'ar_MA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000643 'ar_om': 'ar_OM.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000644 'ar_qa': 'ar_QA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000645 'ar_sa': 'ar_SA.ISO8859-6',
646 'ar_sa.iso88596': 'ar_SA.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000647 'ar_sd': 'ar_SD.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000648 'ar_sy': 'ar_SY.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000649 'ar_tn': 'ar_TN.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000650 'ar_ye': 'ar_YE.ISO8859-6',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000651 'arabic': 'ar_AA.ISO8859-6',
652 'arabic.iso88596': 'ar_AA.ISO8859-6',
653 'az': 'az_AZ.ISO8859-9E',
654 'az_az': 'az_AZ.ISO8859-9E',
655 'az_az.iso88599e': 'az_AZ.ISO8859-9E',
656 'be': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000657 'be_by': 'be_BY.CP1251',
658 'be_by.cp1251': 'be_BY.CP1251',
659 'be_by.microsoftcp1251': 'be_BY.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000660 'bg': 'bg_BG.CP1251',
661 'bg_bg': 'bg_BG.CP1251',
662 'bg_bg.cp1251': 'bg_BG.CP1251',
663 'bg_bg.iso88595': 'bg_BG.ISO8859-5',
664 'bg_bg.koi8r': 'bg_BG.KOI8-R',
665 'bg_bg.microsoftcp1251': 'bg_BG.CP1251',
666 'bokmal': 'nb_NO.ISO8859-1',
667 'bokm\xe5l': 'nb_NO.ISO8859-1',
668 'br': 'br_FR.ISO8859-1',
669 'br_fr': 'br_FR.ISO8859-1',
670 'br_fr.iso88591': 'br_FR.ISO8859-1',
671 'br_fr.iso885914': 'br_FR.ISO8859-14',
672 'br_fr.iso885915': 'br_FR.ISO8859-15',
673 'br_fr@euro': 'br_FR.ISO8859-15',
674 'bulgarian': 'bg_BG.CP1251',
675 'c': 'C',
676 'c-french': 'fr_CA.ISO8859-1',
677 'c-french.iso88591': 'fr_CA.ISO8859-1',
678 'c.en': 'C',
679 'c.iso88591': 'en_US.ISO8859-1',
680 'c_c': 'C',
681 'c_c.c': 'C',
682 'ca': 'ca_ES.ISO8859-1',
683 'ca_es': 'ca_ES.ISO8859-1',
684 'ca_es.iso88591': 'ca_ES.ISO8859-1',
685 'ca_es.iso885915': 'ca_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000686 'ca_es@euro': 'ca_ES.ISO8859-15',
687 'catalan': 'ca_ES.ISO8859-1',
688 'cextend': 'en_US.ISO8859-1',
689 'cextend.en': 'en_US.ISO8859-1',
690 'chinese-s': 'zh_CN.eucCN',
691 'chinese-t': 'zh_TW.eucTW',
692 'croatian': 'hr_HR.ISO8859-2',
693 'cs': 'cs_CZ.ISO8859-2',
694 'cs_cs': 'cs_CZ.ISO8859-2',
695 'cs_cs.iso88592': 'cs_CZ.ISO8859-2',
696 'cs_cz': 'cs_CZ.ISO8859-2',
697 'cs_cz.iso88592': 'cs_CZ.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000698 'cy': 'cy_GB.ISO8859-1',
699 'cy_gb': 'cy_GB.ISO8859-1',
700 'cy_gb.iso88591': 'cy_GB.ISO8859-1',
701 'cy_gb.iso885914': 'cy_GB.ISO8859-14',
702 'cy_gb.iso885915': 'cy_GB.ISO8859-15',
703 'cy_gb@euro': 'cy_GB.ISO8859-15',
704 'cz': 'cs_CZ.ISO8859-2',
705 'cz_cz': 'cs_CZ.ISO8859-2',
706 'czech': 'cs_CZ.ISO8859-2',
707 'da': 'da_DK.ISO8859-1',
708 'da_dk': 'da_DK.ISO8859-1',
709 'da_dk.88591': 'da_DK.ISO8859-1',
710 'da_dk.885915': 'da_DK.ISO8859-15',
711 'da_dk.iso88591': 'da_DK.ISO8859-1',
712 'da_dk.iso885915': 'da_DK.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000713 'da_dk@euro': 'da_DK.ISO8859-15',
714 'danish': 'da_DK.ISO8859-1',
715 'danish.iso88591': 'da_DK.ISO8859-1',
716 'dansk': 'da_DK.ISO8859-1',
717 'de': 'de_DE.ISO8859-1',
718 'de_at': 'de_AT.ISO8859-1',
719 'de_at.iso88591': 'de_AT.ISO8859-1',
720 'de_at.iso885915': 'de_AT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000721 'de_at@euro': 'de_AT.ISO8859-15',
722 'de_be': 'de_BE.ISO8859-1',
723 'de_be.iso88591': 'de_BE.ISO8859-1',
724 'de_be.iso885915': 'de_BE.ISO8859-15',
725 'de_be@euro': 'de_BE.ISO8859-15',
726 'de_ch': 'de_CH.ISO8859-1',
727 'de_ch.iso88591': 'de_CH.ISO8859-1',
728 'de_ch.iso885915': 'de_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000729 'de_ch@euro': 'de_CH.ISO8859-15',
730 'de_de': 'de_DE.ISO8859-1',
731 'de_de.88591': 'de_DE.ISO8859-1',
732 'de_de.885915': 'de_DE.ISO8859-15',
733 'de_de.885915@euro': 'de_DE.ISO8859-15',
734 'de_de.iso88591': 'de_DE.ISO8859-1',
735 'de_de.iso885915': 'de_DE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000736 'de_de@euro': 'de_DE.ISO8859-15',
737 'de_lu': 'de_LU.ISO8859-1',
738 'de_lu.iso88591': 'de_LU.ISO8859-1',
739 'de_lu.iso885915': 'de_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000740 'de_lu@euro': 'de_LU.ISO8859-15',
741 'deutsch': 'de_DE.ISO8859-1',
742 'dutch': 'nl_NL.ISO8859-1',
743 'dutch.iso88591': 'nl_BE.ISO8859-1',
744 'ee': 'ee_EE.ISO8859-4',
745 'ee_ee': 'ee_EE.ISO8859-4',
746 'ee_ee.iso88594': 'ee_EE.ISO8859-4',
747 'eesti': 'et_EE.ISO8859-1',
748 'el': 'el_GR.ISO8859-7',
749 'el_gr': 'el_GR.ISO8859-7',
750 'el_gr.iso88597': 'el_GR.ISO8859-7',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000751 'el_gr@euro': 'el_GR.ISO8859-15',
752 'en': 'en_US.ISO8859-1',
753 'en.iso88591': 'en_US.ISO8859-1',
754 'en_au': 'en_AU.ISO8859-1',
755 'en_au.iso88591': 'en_AU.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000756 'en_be': 'en_BE.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000757 'en_be@euro': 'en_BE.ISO8859-15',
758 'en_bw': 'en_BW.ISO8859-1',
759 'en_ca': 'en_CA.ISO8859-1',
760 'en_ca.iso88591': 'en_CA.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000761 'en_gb': 'en_GB.ISO8859-1',
762 'en_gb.88591': 'en_GB.ISO8859-1',
763 'en_gb.iso88591': 'en_GB.ISO8859-1',
764 'en_gb.iso885915': 'en_GB.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000765 'en_gb@euro': 'en_GB.ISO8859-15',
766 'en_hk': 'en_HK.ISO8859-1',
767 'en_ie': 'en_IE.ISO8859-1',
768 'en_ie.iso88591': 'en_IE.ISO8859-1',
769 'en_ie.iso885915': 'en_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000770 'en_ie@euro': 'en_IE.ISO8859-15',
771 'en_in': 'en_IN.ISO8859-1',
772 'en_nz': 'en_NZ.ISO8859-1',
773 'en_nz.iso88591': 'en_NZ.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000774 'en_ph': 'en_PH.ISO8859-1',
775 'en_sg': 'en_SG.ISO8859-1',
776 'en_uk': 'en_GB.ISO8859-1',
777 'en_us': 'en_US.ISO8859-1',
778 'en_us.88591': 'en_US.ISO8859-1',
779 'en_us.885915': 'en_US.ISO8859-15',
780 'en_us.iso88591': 'en_US.ISO8859-1',
781 'en_us.iso885915': 'en_US.ISO8859-15',
782 'en_us.iso885915@euro': 'en_US.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000783 'en_us@euro': 'en_US.ISO8859-15',
784 'en_us@euro@euro': 'en_US.ISO8859-15',
785 'en_za': 'en_ZA.ISO8859-1',
786 'en_za.88591': 'en_ZA.ISO8859-1',
787 'en_za.iso88591': 'en_ZA.ISO8859-1',
788 'en_za.iso885915': 'en_ZA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000789 'en_za@euro': 'en_ZA.ISO8859-15',
790 'en_zw': 'en_ZW.ISO8859-1',
791 'eng_gb': 'en_GB.ISO8859-1',
792 'eng_gb.8859': 'en_GB.ISO8859-1',
793 'english': 'en_EN.ISO8859-1',
794 'english.iso88591': 'en_EN.ISO8859-1',
795 'english_uk': 'en_GB.ISO8859-1',
796 'english_uk.8859': 'en_GB.ISO8859-1',
797 'english_united-states': 'en_US.ISO8859-1',
798 'english_united-states.437': 'C',
799 'english_us': 'en_US.ISO8859-1',
800 'english_us.8859': 'en_US.ISO8859-1',
801 'english_us.ascii': 'en_US.ISO8859-1',
802 'eo': 'eo_XX.ISO8859-3',
803 'eo_eo': 'eo_EO.ISO8859-3',
804 'eo_eo.iso88593': 'eo_EO.ISO8859-3',
805 'eo_xx': 'eo_XX.ISO8859-3',
806 'eo_xx.iso88593': 'eo_XX.ISO8859-3',
807 'es': 'es_ES.ISO8859-1',
808 'es_ar': 'es_AR.ISO8859-1',
809 'es_ar.iso88591': 'es_AR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000810 'es_bo': 'es_BO.ISO8859-1',
811 'es_bo.iso88591': 'es_BO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000812 'es_cl': 'es_CL.ISO8859-1',
813 'es_cl.iso88591': 'es_CL.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000814 'es_co': 'es_CO.ISO8859-1',
815 'es_co.iso88591': 'es_CO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000816 'es_cr': 'es_CR.ISO8859-1',
817 'es_cr.iso88591': 'es_CR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000818 'es_do': 'es_DO.ISO8859-1',
819 'es_do.iso88591': 'es_DO.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000820 'es_ec': 'es_EC.ISO8859-1',
821 'es_ec.iso88591': 'es_EC.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000822 'es_es': 'es_ES.ISO8859-1',
823 'es_es.88591': 'es_ES.ISO8859-1',
824 'es_es.iso88591': 'es_ES.ISO8859-1',
825 'es_es.iso885915': 'es_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000826 'es_es@euro': 'es_ES.ISO8859-15',
827 'es_gt': 'es_GT.ISO8859-1',
828 'es_gt.iso88591': 'es_GT.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000829 'es_hn': 'es_HN.ISO8859-1',
830 'es_hn.iso88591': 'es_HN.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000831 'es_mx': 'es_MX.ISO8859-1',
832 'es_mx.iso88591': 'es_MX.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000833 'es_ni': 'es_NI.ISO8859-1',
834 'es_ni.iso88591': 'es_NI.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000835 'es_pa': 'es_PA.ISO8859-1',
836 'es_pa.iso88591': 'es_PA.ISO8859-1',
837 'es_pa.iso885915': 'es_PA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000838 'es_pa@euro': 'es_PA.ISO8859-15',
839 'es_pe': 'es_PE.ISO8859-1',
840 'es_pe.iso88591': 'es_PE.ISO8859-1',
841 'es_pe.iso885915': 'es_PE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000842 'es_pe@euro': 'es_PE.ISO8859-15',
843 'es_pr': 'es_PR.ISO8859-1',
844 'es_pr.iso88591': 'es_PR.ISO8859-1',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000845 'es_py': 'es_PY.ISO8859-1',
846 'es_py.iso88591': 'es_PY.ISO8859-1',
847 'es_py.iso885915': 'es_PY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000848 'es_py@euro': 'es_PY.ISO8859-15',
849 'es_sv': 'es_SV.ISO8859-1',
850 'es_sv.iso88591': 'es_SV.ISO8859-1',
851 'es_sv.iso885915': 'es_SV.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000852 'es_sv@euro': 'es_SV.ISO8859-15',
853 'es_us': 'es_US.ISO8859-1',
854 'es_uy': 'es_UY.ISO8859-1',
855 'es_uy.iso88591': 'es_UY.ISO8859-1',
856 'es_uy.iso885915': 'es_UY.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000857 'es_uy@euro': 'es_UY.ISO8859-15',
858 'es_ve': 'es_VE.ISO8859-1',
859 'es_ve.iso88591': 'es_VE.ISO8859-1',
860 'es_ve.iso885915': 'es_VE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000861 'es_ve@euro': 'es_VE.ISO8859-15',
862 'estonian': 'et_EE.ISO8859-1',
863 'et': 'et_EE.ISO8859-15',
864 'et_ee': 'et_EE.ISO8859-15',
865 'et_ee.iso88591': 'et_EE.ISO8859-1',
866 'et_ee.iso885913': 'et_EE.ISO8859-13',
867 'et_ee.iso885915': 'et_EE.ISO8859-15',
868 'et_ee.iso88594': 'et_EE.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000869 'et_ee@euro': 'et_EE.ISO8859-15',
870 'eu': 'eu_ES.ISO8859-1',
871 'eu_es': 'eu_ES.ISO8859-1',
872 'eu_es.iso88591': 'eu_ES.ISO8859-1',
873 'eu_es.iso885915': 'eu_ES.ISO8859-15',
874 'eu_es@euro': 'eu_ES.ISO8859-15',
875 'fa': 'fa_IR.UTF-8',
876 'fa_ir': 'fa_IR.UTF-8',
877 'fa_ir.isiri3342': 'fa_IR.ISIRI-3342',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000878 'fi': 'fi_FI.ISO8859-15',
879 'fi_fi': 'fi_FI.ISO8859-15',
880 'fi_fi.88591': 'fi_FI.ISO8859-1',
881 'fi_fi.iso88591': 'fi_FI.ISO8859-1',
882 'fi_fi.iso885915': 'fi_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000883 'fi_fi.utf8@euro': 'fi_FI.UTF-8',
884 'fi_fi@euro': 'fi_FI.ISO8859-15',
885 'finnish': 'fi_FI.ISO8859-1',
886 'finnish.iso88591': 'fi_FI.ISO8859-1',
887 'fo': 'fo_FO.ISO8859-1',
888 'fo_fo': 'fo_FO.ISO8859-1',
889 'fo_fo.iso88591': 'fo_FO.ISO8859-1',
890 'fo_fo.iso885915': 'fo_FO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000891 'fo_fo@euro': 'fo_FO.ISO8859-15',
892 'fr': 'fr_FR.ISO8859-1',
893 'fr_be': 'fr_BE.ISO8859-1',
894 'fr_be.88591': 'fr_BE.ISO8859-1',
895 'fr_be.iso88591': 'fr_BE.ISO8859-1',
896 'fr_be.iso885915': 'fr_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000897 'fr_be@euro': 'fr_BE.ISO8859-15',
898 'fr_ca': 'fr_CA.ISO8859-1',
899 'fr_ca.88591': 'fr_CA.ISO8859-1',
900 'fr_ca.iso88591': 'fr_CA.ISO8859-1',
901 'fr_ca.iso885915': 'fr_CA.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000902 'fr_ca@euro': 'fr_CA.ISO8859-15',
903 'fr_ch': 'fr_CH.ISO8859-1',
904 'fr_ch.88591': 'fr_CH.ISO8859-1',
905 'fr_ch.iso88591': 'fr_CH.ISO8859-1',
906 'fr_ch.iso885915': 'fr_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000907 'fr_ch@euro': 'fr_CH.ISO8859-15',
908 'fr_fr': 'fr_FR.ISO8859-1',
909 'fr_fr.88591': 'fr_FR.ISO8859-1',
910 'fr_fr.iso88591': 'fr_FR.ISO8859-1',
911 'fr_fr.iso885915': 'fr_FR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000912 'fr_fr@euro': 'fr_FR.ISO8859-15',
913 'fr_lu': 'fr_LU.ISO8859-1',
914 'fr_lu.88591': 'fr_LU.ISO8859-1',
915 'fr_lu.iso88591': 'fr_LU.ISO8859-1',
916 'fr_lu.iso885915': 'fr_LU.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000917 'fr_lu@euro': 'fr_LU.ISO8859-15',
918 'fran\xe7ais': 'fr_FR.ISO8859-1',
919 'fre_fr': 'fr_FR.ISO8859-1',
920 'fre_fr.8859': 'fr_FR.ISO8859-1',
921 'french': 'fr_FR.ISO8859-1',
922 'french.iso88591': 'fr_CH.ISO8859-1',
923 'french_france': 'fr_FR.ISO8859-1',
924 'french_france.8859': 'fr_FR.ISO8859-1',
925 'ga': 'ga_IE.ISO8859-1',
926 'ga_ie': 'ga_IE.ISO8859-1',
927 'ga_ie.iso88591': 'ga_IE.ISO8859-1',
928 'ga_ie.iso885914': 'ga_IE.ISO8859-14',
929 'ga_ie.iso885915': 'ga_IE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000930 'ga_ie@euro': 'ga_IE.ISO8859-15',
931 'galego': 'gl_ES.ISO8859-1',
932 'galician': 'gl_ES.ISO8859-1',
933 'gd': 'gd_GB.ISO8859-1',
934 'gd_gb': 'gd_GB.ISO8859-1',
935 'gd_gb.iso88591': 'gd_GB.ISO8859-1',
936 'gd_gb.iso885914': 'gd_GB.ISO8859-14',
937 'gd_gb.iso885915': 'gd_GB.ISO8859-15',
938 'gd_gb@euro': 'gd_GB.ISO8859-15',
939 'ger_de': 'de_DE.ISO8859-1',
940 'ger_de.8859': 'de_DE.ISO8859-1',
941 'german': 'de_DE.ISO8859-1',
942 'german.iso88591': 'de_CH.ISO8859-1',
943 'german_germany': 'de_DE.ISO8859-1',
944 'german_germany.8859': 'de_DE.ISO8859-1',
945 'gl': 'gl_ES.ISO8859-1',
946 'gl_es': 'gl_ES.ISO8859-1',
947 'gl_es.iso88591': 'gl_ES.ISO8859-1',
948 'gl_es.iso885915': 'gl_ES.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000949 'gl_es@euro': 'gl_ES.ISO8859-15',
950 'greek': 'el_GR.ISO8859-7',
951 'greek.iso88597': 'el_GR.ISO8859-7',
952 'gv': 'gv_GB.ISO8859-1',
953 'gv_gb': 'gv_GB.ISO8859-1',
954 'gv_gb.iso88591': 'gv_GB.ISO8859-1',
955 'gv_gb.iso885914': 'gv_GB.ISO8859-14',
956 'gv_gb.iso885915': 'gv_GB.ISO8859-15',
957 'gv_gb@euro': 'gv_GB.ISO8859-15',
958 'he': 'he_IL.ISO8859-8',
959 'he_il': 'he_IL.ISO8859-8',
960 'he_il.cp1255': 'he_IL.CP1255',
961 'he_il.iso88598': 'he_IL.ISO8859-8',
962 'he_il.microsoftcp1255': 'he_IL.CP1255',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000963 'hebrew': 'iw_IL.ISO8859-8',
964 'hebrew.iso88598': 'iw_IL.ISO8859-8',
965 'hi': 'hi_IN.ISCII-DEV',
966 'hi_in': 'hi_IN.ISCII-DEV',
967 'hi_in.isciidev': 'hi_IN.ISCII-DEV',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000968 'hr': 'hr_HR.ISO8859-2',
969 'hr_hr': 'hr_HR.ISO8859-2',
970 'hr_hr.iso88592': 'hr_HR.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000971 'hrvatski': 'hr_HR.ISO8859-2',
972 'hu': 'hu_HU.ISO8859-2',
973 'hu_hu': 'hu_HU.ISO8859-2',
974 'hu_hu.iso88592': 'hu_HU.ISO8859-2',
975 'hungarian': 'hu_HU.ISO8859-2',
976 'icelandic': 'is_IS.ISO8859-1',
977 'icelandic.iso88591': 'is_IS.ISO8859-1',
978 'id': 'id_ID.ISO8859-1',
979 'id_id': 'id_ID.ISO8859-1',
980 'in': 'id_ID.ISO8859-1',
981 'in_id': 'id_ID.ISO8859-1',
982 'is': 'is_IS.ISO8859-1',
983 'is_is': 'is_IS.ISO8859-1',
984 'is_is.iso88591': 'is_IS.ISO8859-1',
985 'is_is.iso885915': 'is_IS.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000986 'is_is@euro': 'is_IS.ISO8859-15',
987 'iso-8859-1': 'en_US.ISO8859-1',
988 'iso-8859-15': 'en_US.ISO8859-15',
989 'iso8859-1': 'en_US.ISO8859-1',
990 'iso8859-15': 'en_US.ISO8859-15',
991 'iso_8859_1': 'en_US.ISO8859-1',
992 'iso_8859_15': 'en_US.ISO8859-15',
993 'it': 'it_IT.ISO8859-1',
994 'it_ch': 'it_CH.ISO8859-1',
995 'it_ch.iso88591': 'it_CH.ISO8859-1',
996 'it_ch.iso885915': 'it_CH.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +0000997 'it_ch@euro': 'it_CH.ISO8859-15',
998 'it_it': 'it_IT.ISO8859-1',
999 'it_it.88591': 'it_IT.ISO8859-1',
1000 'it_it.iso88591': 'it_IT.ISO8859-1',
1001 'it_it.iso885915': 'it_IT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001002 'it_it@euro': 'it_IT.ISO8859-15',
1003 'italian': 'it_IT.ISO8859-1',
1004 'italian.iso88591': 'it_IT.ISO8859-1',
1005 'iu': 'iu_CA.NUNACOM-8',
1006 'iu_ca': 'iu_CA.NUNACOM-8',
1007 'iu_ca.nunacom8': 'iu_CA.NUNACOM-8',
1008 'iw': 'he_IL.ISO8859-8',
1009 'iw_il': 'he_IL.ISO8859-8',
1010 'iw_il.iso88598': 'he_IL.ISO8859-8',
1011 'ja': 'ja_JP.eucJP',
1012 'ja.jis': 'ja_JP.JIS7',
1013 'ja.sjis': 'ja_JP.SJIS',
1014 'ja_jp': 'ja_JP.eucJP',
1015 'ja_jp.ajec': 'ja_JP.eucJP',
1016 'ja_jp.euc': 'ja_JP.eucJP',
1017 'ja_jp.eucjp': 'ja_JP.eucJP',
1018 'ja_jp.iso-2022-jp': 'ja_JP.JIS7',
1019 'ja_jp.iso2022jp': 'ja_JP.JIS7',
1020 'ja_jp.jis': 'ja_JP.JIS7',
1021 'ja_jp.jis7': 'ja_JP.JIS7',
1022 'ja_jp.mscode': 'ja_JP.SJIS',
1023 'ja_jp.sjis': 'ja_JP.SJIS',
1024 'ja_jp.ujis': 'ja_JP.eucJP',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001025 'japan': 'ja_JP.eucJP',
1026 'japanese': 'ja_JP.eucJP',
1027 'japanese-euc': 'ja_JP.eucJP',
1028 'japanese.euc': 'ja_JP.eucJP',
1029 'japanese.sjis': 'ja_JP.SJIS',
1030 'jp_jp': 'ja_JP.eucJP',
1031 'ka': 'ka_GE.GEORGIAN-ACADEMY',
1032 'ka_ge': 'ka_GE.GEORGIAN-ACADEMY',
1033 'ka_ge.georgianacademy': 'ka_GE.GEORGIAN-ACADEMY',
1034 'ka_ge.georgianps': 'ka_GE.GEORGIAN-PS',
1035 'ka_ge.georgianrs': 'ka_GE.GEORGIAN-ACADEMY',
1036 'kl': 'kl_GL.ISO8859-1',
1037 'kl_gl': 'kl_GL.ISO8859-1',
1038 'kl_gl.iso88591': 'kl_GL.ISO8859-1',
1039 'kl_gl.iso885915': 'kl_GL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001040 'kl_gl@euro': 'kl_GL.ISO8859-15',
1041 'ko': 'ko_KR.eucKR',
1042 'ko_kr': 'ko_KR.eucKR',
1043 'ko_kr.euc': 'ko_KR.eucKR',
1044 'ko_kr.euckr': 'ko_KR.eucKR',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001045 'korean': 'ko_KR.eucKR',
1046 'korean.euc': 'ko_KR.eucKR',
1047 'kw': 'kw_GB.ISO8859-1',
1048 'kw_gb': 'kw_GB.ISO8859-1',
1049 'kw_gb.iso88591': 'kw_GB.ISO8859-1',
1050 'kw_gb.iso885914': 'kw_GB.ISO8859-14',
1051 'kw_gb.iso885915': 'kw_GB.ISO8859-15',
1052 'kw_gb@euro': 'kw_GB.ISO8859-15',
1053 'lithuanian': 'lt_LT.ISO8859-13',
1054 'lo': 'lo_LA.MULELAO-1',
1055 'lo_la': 'lo_LA.MULELAO-1',
1056 'lo_la.cp1133': 'lo_LA.IBM-CP1133',
1057 'lo_la.ibmcp1133': 'lo_LA.IBM-CP1133',
1058 'lo_la.mulelao1': 'lo_LA.MULELAO-1',
1059 'lt': 'lt_LT.ISO8859-13',
1060 'lt_lt': 'lt_LT.ISO8859-13',
1061 'lt_lt.iso885913': 'lt_LT.ISO8859-13',
1062 'lt_lt.iso88594': 'lt_LT.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001063 'lv': 'lv_LV.ISO8859-13',
1064 'lv_lv': 'lv_LV.ISO8859-13',
1065 'lv_lv.iso885913': 'lv_LV.ISO8859-13',
1066 'lv_lv.iso88594': 'lv_LV.ISO8859-4',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001067 'mi': 'mi_NZ.ISO8859-1',
1068 'mi_nz': 'mi_NZ.ISO8859-1',
1069 'mi_nz.iso88591': 'mi_NZ.ISO8859-1',
1070 'mk': 'mk_MK.ISO8859-5',
1071 'mk_mk': 'mk_MK.ISO8859-5',
1072 'mk_mk.cp1251': 'mk_MK.CP1251',
1073 'mk_mk.iso88595': 'mk_MK.ISO8859-5',
1074 'mk_mk.microsoftcp1251': 'mk_MK.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001075 'ms': 'ms_MY.ISO8859-1',
1076 'ms_my': 'ms_MY.ISO8859-1',
1077 'ms_my.iso88591': 'ms_MY.ISO8859-1',
1078 'mt': 'mt_MT.ISO8859-3',
1079 'mt_mt': 'mt_MT.ISO8859-3',
1080 'mt_mt.iso88593': 'mt_MT.ISO8859-3',
1081 'nb': 'nb_NO.ISO8859-1',
1082 'nb_no': 'nb_NO.ISO8859-1',
1083 'nb_no.88591': 'nb_NO.ISO8859-1',
1084 'nb_no.iso88591': 'nb_NO.ISO8859-1',
1085 'nb_no.iso885915': 'nb_NO.ISO8859-15',
1086 'nb_no@euro': 'nb_NO.ISO8859-15',
1087 'nl': 'nl_NL.ISO8859-1',
1088 'nl_be': 'nl_BE.ISO8859-1',
1089 'nl_be.88591': 'nl_BE.ISO8859-1',
1090 'nl_be.iso88591': 'nl_BE.ISO8859-1',
1091 'nl_be.iso885915': 'nl_BE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001092 'nl_be@euro': 'nl_BE.ISO8859-15',
1093 'nl_nl': 'nl_NL.ISO8859-1',
1094 'nl_nl.88591': 'nl_NL.ISO8859-1',
1095 'nl_nl.iso88591': 'nl_NL.ISO8859-1',
1096 'nl_nl.iso885915': 'nl_NL.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001097 'nl_nl@euro': 'nl_NL.ISO8859-15',
1098 'nn': 'nn_NO.ISO8859-1',
1099 'nn_no': 'nn_NO.ISO8859-1',
1100 'nn_no.88591': 'nn_NO.ISO8859-1',
1101 'nn_no.iso88591': 'nn_NO.ISO8859-1',
1102 'nn_no.iso885915': 'nn_NO.ISO8859-15',
1103 'nn_no@euro': 'nn_NO.ISO8859-15',
1104 'no': 'no_NO.ISO8859-1',
1105 'no@nynorsk': 'ny_NO.ISO8859-1',
1106 'no_no': 'no_NO.ISO8859-1',
1107 'no_no.88591': 'no_NO.ISO8859-1',
1108 'no_no.iso88591': 'no_NO.ISO8859-1',
1109 'no_no.iso885915': 'no_NO.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001110 'no_no@euro': 'no_NO.ISO8859-15',
1111 'norwegian': 'no_NO.ISO8859-1',
1112 'norwegian.iso88591': 'no_NO.ISO8859-1',
1113 'ny': 'ny_NO.ISO8859-1',
1114 'ny_no': 'ny_NO.ISO8859-1',
1115 'ny_no.88591': 'ny_NO.ISO8859-1',
1116 'ny_no.iso88591': 'ny_NO.ISO8859-1',
1117 'ny_no.iso885915': 'ny_NO.ISO8859-15',
1118 'ny_no@euro': 'ny_NO.ISO8859-15',
1119 'nynorsk': 'nn_NO.ISO8859-1',
1120 'oc': 'oc_FR.ISO8859-1',
1121 'oc_fr': 'oc_FR.ISO8859-1',
1122 'oc_fr.iso88591': 'oc_FR.ISO8859-1',
1123 'oc_fr.iso885915': 'oc_FR.ISO8859-15',
1124 'oc_fr@euro': 'oc_FR.ISO8859-15',
1125 'pd': 'pd_US.ISO8859-1',
1126 'pd_de': 'pd_DE.ISO8859-1',
1127 'pd_de.iso88591': 'pd_DE.ISO8859-1',
1128 'pd_de.iso885915': 'pd_DE.ISO8859-15',
1129 'pd_de@euro': 'pd_DE.ISO8859-15',
1130 'pd_us': 'pd_US.ISO8859-1',
1131 'pd_us.iso88591': 'pd_US.ISO8859-1',
1132 'pd_us.iso885915': 'pd_US.ISO8859-15',
1133 'pd_us@euro': 'pd_US.ISO8859-15',
1134 'ph': 'ph_PH.ISO8859-1',
1135 'ph_ph': 'ph_PH.ISO8859-1',
1136 'ph_ph.iso88591': 'ph_PH.ISO8859-1',
1137 'pl': 'pl_PL.ISO8859-2',
1138 'pl_pl': 'pl_PL.ISO8859-2',
1139 'pl_pl.iso88592': 'pl_PL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001140 'polish': 'pl_PL.ISO8859-2',
1141 'portuguese': 'pt_PT.ISO8859-1',
1142 'portuguese.iso88591': 'pt_PT.ISO8859-1',
1143 'portuguese_brazil': 'pt_BR.ISO8859-1',
1144 'portuguese_brazil.8859': 'pt_BR.ISO8859-1',
1145 'posix': 'C',
1146 'posix-utf2': 'C',
1147 'pp': 'pp_AN.ISO8859-1',
1148 'pp_an': 'pp_AN.ISO8859-1',
1149 'pp_an.iso88591': 'pp_AN.ISO8859-1',
1150 'pt': 'pt_PT.ISO8859-1',
1151 'pt_br': 'pt_BR.ISO8859-1',
1152 'pt_br.88591': 'pt_BR.ISO8859-1',
1153 'pt_br.iso88591': 'pt_BR.ISO8859-1',
1154 'pt_br.iso885915': 'pt_BR.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001155 'pt_br@euro': 'pt_BR.ISO8859-15',
1156 'pt_pt': 'pt_PT.ISO8859-1',
1157 'pt_pt.88591': 'pt_PT.ISO8859-1',
1158 'pt_pt.iso88591': 'pt_PT.ISO8859-1',
1159 'pt_pt.iso885915': 'pt_PT.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001160 'pt_pt.utf8@euro': 'pt_PT.UTF-8',
1161 'pt_pt@euro': 'pt_PT.ISO8859-15',
1162 'ro': 'ro_RO.ISO8859-2',
1163 'ro_ro': 'ro_RO.ISO8859-2',
1164 'ro_ro.iso88592': 'ro_RO.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001165 'romanian': 'ro_RO.ISO8859-2',
1166 'ru': 'ru_RU.ISO8859-5',
1167 'ru_ru': 'ru_RU.ISO8859-5',
1168 'ru_ru.cp1251': 'ru_RU.CP1251',
1169 'ru_ru.iso88595': 'ru_RU.ISO8859-5',
1170 'ru_ru.koi8r': 'ru_RU.KOI8-R',
1171 'ru_ru.microsoftcp1251': 'ru_RU.CP1251',
1172 'ru_ua': 'ru_UA.KOI8-U',
1173 'ru_ua.cp1251': 'ru_UA.CP1251',
1174 'ru_ua.koi8u': 'ru_UA.KOI8-U',
1175 'ru_ua.microsoftcp1251': 'ru_UA.CP1251',
1176 'rumanian': 'ro_RO.ISO8859-2',
1177 'russian': 'ru_RU.ISO8859-5',
1178 'se_no': 'se_NO.UTF-8',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001179 'serbocroatian': 'sh_YU.ISO8859-2',
1180 'sh': 'sh_YU.ISO8859-2',
1181 'sh_hr': 'sh_HR.ISO8859-2',
1182 'sh_hr.iso88592': 'sh_HR.ISO8859-2',
1183 'sh_sp': 'sh_YU.ISO8859-2',
1184 'sh_yu': 'sh_YU.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001185 'sk': 'sk_SK.ISO8859-2',
1186 'sk_sk': 'sk_SK.ISO8859-2',
1187 'sk_sk.iso88592': 'sk_SK.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001188 'sl': 'sl_SI.ISO8859-2',
1189 'sl_cs': 'sl_CS.ISO8859-2',
1190 'sl_si': 'sl_SI.ISO8859-2',
1191 'sl_si.iso88592': 'sl_SI.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001192 'slovak': 'sk_SK.ISO8859-2',
1193 'slovene': 'sl_SI.ISO8859-2',
1194 'slovenian': 'sl_SI.ISO8859-2',
1195 'sp': 'sp_YU.ISO8859-5',
1196 'sp_yu': 'sp_YU.ISO8859-5',
1197 'spanish': 'es_ES.ISO8859-1',
1198 'spanish.iso88591': 'es_ES.ISO8859-1',
1199 'spanish_spain': 'es_ES.ISO8859-1',
1200 'spanish_spain.8859': 'es_ES.ISO8859-1',
1201 'sq': 'sq_AL.ISO8859-2',
1202 'sq_al': 'sq_AL.ISO8859-2',
1203 'sq_al.iso88592': 'sq_AL.ISO8859-2',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001204 'sr': 'sr_YU.ISO8859-5',
1205 'sr@cyrillic': 'sr_YU.ISO8859-5',
1206 'sr_sp': 'sr_SP.ISO8859-2',
1207 'sr_yu': 'sr_YU.ISO8859-5',
1208 'sr_yu.cp1251@cyrillic': 'sr_YU.CP1251',
1209 'sr_yu.iso88592': 'sr_YU.ISO8859-2',
1210 'sr_yu.iso88595': 'sr_YU.ISO8859-5',
1211 'sr_yu.iso88595@cyrillic': 'sr_YU.ISO8859-5',
1212 'sr_yu.microsoftcp1251@cyrillic': 'sr_YU.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001213 'sr_yu.utf8@cyrillic': 'sr_YU.UTF-8',
1214 'sr_yu@cyrillic': 'sr_YU.ISO8859-5',
1215 'sv': 'sv_SE.ISO8859-1',
1216 'sv_fi': 'sv_FI.ISO8859-1',
1217 'sv_fi.iso88591': 'sv_FI.ISO8859-1',
1218 'sv_fi.iso885915': 'sv_FI.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001219 'sv_fi@euro': 'sv_FI.ISO8859-15',
1220 'sv_se': 'sv_SE.ISO8859-1',
1221 'sv_se.88591': 'sv_SE.ISO8859-1',
1222 'sv_se.iso88591': 'sv_SE.ISO8859-1',
1223 'sv_se.iso885915': 'sv_SE.ISO8859-15',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001224 'sv_se@euro': 'sv_SE.ISO8859-15',
1225 'swedish': 'sv_SE.ISO8859-1',
1226 'swedish.iso88591': 'sv_SE.ISO8859-1',
1227 'ta': 'ta_IN.TSCII-0',
1228 'ta_in': 'ta_IN.TSCII-0',
1229 'ta_in.tscii': 'ta_IN.TSCII-0',
1230 'ta_in.tscii0': 'ta_IN.TSCII-0',
1231 'tg': 'tg_TJ.KOI8-C',
1232 'tg_tj': 'tg_TJ.KOI8-C',
1233 'tg_tj.koi8c': 'tg_TJ.KOI8-C',
1234 'th': 'th_TH.ISO8859-11',
1235 'th_th': 'th_TH.ISO8859-11',
1236 'th_th.iso885911': 'th_TH.ISO8859-11',
1237 'th_th.tactis': 'th_TH.TIS620',
1238 'th_th.tis620': 'th_TH.TIS620',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001239 'thai': 'th_TH.ISO8859-11',
1240 'tl': 'tl_PH.ISO8859-1',
1241 'tl_ph': 'tl_PH.ISO8859-1',
1242 'tl_ph.iso88591': 'tl_PH.ISO8859-1',
1243 'tr': 'tr_TR.ISO8859-9',
1244 'tr_tr': 'tr_TR.ISO8859-9',
1245 'tr_tr.iso88599': 'tr_TR.ISO8859-9',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001246 'tt': 'tt_RU.TATAR-CYR',
1247 'tt_ru': 'tt_RU.TATAR-CYR',
1248 'tt_ru.koi8c': 'tt_RU.KOI8-C',
1249 'tt_ru.tatarcyr': 'tt_RU.TATAR-CYR',
1250 'turkish': 'tr_TR.ISO8859-9',
1251 'turkish.iso88599': 'tr_TR.ISO8859-9',
1252 'uk': 'uk_UA.KOI8-U',
1253 'uk_ua': 'uk_UA.KOI8-U',
1254 'uk_ua.cp1251': 'uk_UA.CP1251',
1255 'uk_ua.iso88595': 'uk_UA.ISO8859-5',
1256 'uk_ua.koi8u': 'uk_UA.KOI8-U',
1257 'uk_ua.microsoftcp1251': 'uk_UA.CP1251',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001258 'univ': 'en_US.utf',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001259 'universal': 'en_US.utf',
1260 'universal.utf8@ucs4': 'en_US.UTF-8',
1261 'ur': 'ur_PK.CP1256',
1262 'ur_pk': 'ur_PK.CP1256',
1263 'ur_pk.cp1256': 'ur_PK.CP1256',
1264 'ur_pk.microsoftcp1256': 'ur_PK.CP1256',
1265 'uz': 'uz_UZ.UTF-8',
1266 'uz_uz': 'uz_UZ.UTF-8',
1267 'vi': 'vi_VN.TCVN',
1268 'vi_vn': 'vi_VN.TCVN',
1269 'vi_vn.tcvn': 'vi_VN.TCVN',
1270 'vi_vn.tcvn5712': 'vi_VN.TCVN',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001271 'vi_vn.viscii': 'vi_VN.VISCII',
1272 'vi_vn.viscii111': 'vi_VN.VISCII',
1273 'wa': 'wa_BE.ISO8859-1',
1274 'wa_be': 'wa_BE.ISO8859-1',
1275 'wa_be.iso88591': 'wa_BE.ISO8859-1',
1276 'wa_be.iso885915': 'wa_BE.ISO8859-15',
1277 'wa_be@euro': 'wa_BE.ISO8859-15',
1278 'yi': 'yi_US.CP1255',
1279 'yi_us': 'yi_US.CP1255',
1280 'yi_us.cp1255': 'yi_US.CP1255',
1281 'yi_us.microsoftcp1255': 'yi_US.CP1255',
1282 'zh': 'zh_CN.eucCN',
1283 'zh_cn': 'zh_CN.gb2312',
1284 'zh_cn.big5': 'zh_TW.big5',
1285 'zh_cn.euc': 'zh_CN.eucCN',
1286 'zh_cn.gb18030': 'zh_CN.gb18030',
1287 'zh_cn.gb2312': 'zh_CN.gb2312',
1288 'zh_cn.gbk': 'zh_CN.gbk',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001289 'zh_hk': 'zh_HK.big5hkscs',
1290 'zh_hk.big5': 'zh_HK.big5',
1291 'zh_hk.big5hkscs': 'zh_HK.big5hkscs',
Marc-André Lemburgbb4f1bd2004-12-10 21:58:14 +00001292 'zh_tw': 'zh_TW.big5',
1293 'zh_tw.big5': 'zh_TW.big5',
1294 'zh_tw.euc': 'zh_TW.eucTW',
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001295}
1296
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001297#
Georg Brandlb709c2c2006-01-20 09:07:35 +00001298# This maps Windows language identifiers to locale strings.
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001299#
Tim Peters777f1082006-01-20 20:03:24 +00001300# This list has been updated from
Georg Brandlb709c2c2006-01-20 09:07:35 +00001301# http://msdn.microsoft.com/library/default.asp?url=/library/en-us/intl/nls_238z.asp
1302# to include every locale up to Windows XP.
Fredrik Lundh37a09822002-10-19 20:19:10 +00001303#
Georg Brandl5035c1c2006-01-20 13:38:26 +00001304# NOTE: this mapping is incomplete. If your language is missing, please
1305# submit a bug report to Python bug manager, which you can find via:
1306# http://www.python.org/dev/
1307# Make sure you include the missing language identifier and the suggested
1308# locale code.
1309#
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001310
1311windows_locale = {
Georg Brandlb709c2c2006-01-20 09:07:35 +00001312 0x0436: "af_ZA", # Afrikaans
1313 0x041c: "sq_AL", # Albanian
1314 0x0401: "ar_SA", # Arabic - Saudi Arabia
1315 0x0801: "ar_IQ", # Arabic - Iraq
1316 0x0c01: "ar_EG", # Arabic - Egypt
1317 0x1001: "ar_LY", # Arabic - Libya
1318 0x1401: "ar_DZ", # Arabic - Algeria
1319 0x1801: "ar_MA", # Arabic - Morocco
1320 0x1c01: "ar_TN", # Arabic - Tunisia
1321 0x2001: "ar_OM", # Arabic - Oman
1322 0x2401: "ar_YE", # Arabic - Yemen
1323 0x2801: "ar_SY", # Arabic - Syria
1324 0x2c01: "ar_JO", # Arabic - Jordan
1325 0x3001: "ar_LB", # Arabic - Lebanon
1326 0x3401: "ar_KW", # Arabic - Kuwait
1327 0x3801: "ar_AE", # Arabic - United Arab Emirates
1328 0x3c01: "ar_BH", # Arabic - Bahrain
1329 0x4001: "ar_QA", # Arabic - Qatar
1330 0x042b: "hy_AM", # Armenian
1331 0x042c: "az_AZ", # Azeri Latin
1332 0x082c: "az_AZ", # Azeri - Cyrillic
1333 0x042d: "eu_ES", # Basque
1334 0x0423: "be_BY", # Belarusian
1335 0x0445: "bn_IN", # Begali
1336 0x201a: "bs_BA", # Bosnian
1337 0x141a: "bs_BA", # Bosnian - Cyrillic
1338 0x047e: "br_FR", # Breton - France
1339 0x0402: "bg_BG", # Bulgarian
1340 0x0403: "ca_ES", # Catalan
1341 0x0004: "zh_CHS",# Chinese - Simplified
1342 0x0404: "zh_TW", # Chinese - Taiwan
1343 0x0804: "zh_CN", # Chinese - PRC
1344 0x0c04: "zh_HK", # Chinese - Hong Kong S.A.R.
1345 0x1004: "zh_SG", # Chinese - Singapore
1346 0x1404: "zh_MO", # Chinese - Macao S.A.R.
1347 0x7c04: "zh_CHT",# Chinese - Traditional
1348 0x041a: "hr_HR", # Croatian
1349 0x101a: "hr_BA", # Croatian - Bosnia
1350 0x0405: "cs_CZ", # Czech
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001351 0x0406: "da_DK", # Danish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001352 0x048c: "gbz_AF",# Dari - Afghanistan
1353 0x0465: "div_MV",# Divehi - Maldives
1354 0x0413: "nl_NL", # Dutch - The Netherlands
1355 0x0813: "nl_BE", # Dutch - Belgium
1356 0x0409: "en_US", # English - United States
1357 0x0809: "en_GB", # English - United Kingdom
1358 0x0c09: "en_AU", # English - Australia
1359 0x1009: "en_CA", # English - Canada
1360 0x1409: "en_NZ", # English - New Zealand
1361 0x1809: "en_IE", # English - Ireland
1362 0x1c09: "en_ZA", # English - South Africa
1363 0x2009: "en_JA", # English - Jamaica
1364 0x2409: "en_CB", # English - Carribbean
1365 0x2809: "en_BZ", # English - Belize
1366 0x2c09: "en_TT", # English - Trinidad
1367 0x3009: "en_ZW", # English - Zimbabwe
1368 0x3409: "en_PH", # English - Phillippines
1369 0x0425: "et_EE", # Estonian
1370 0x0438: "fo_FO", # Faroese
1371 0x0464: "fil_PH",# Filipino
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001372 0x040b: "fi_FI", # Finnish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001373 0x040c: "fr_FR", # French - France
1374 0x080c: "fr_BE", # French - Belgium
1375 0x0c0c: "fr_CA", # French - Canada
1376 0x100c: "fr_CH", # French - Switzerland
1377 0x140c: "fr_LU", # French - Luxembourg
1378 0x180c: "fr_MC", # French - Monaco
1379 0x0462: "fy_NL", # Frisian - Netherlands
1380 0x0456: "gl_ES", # Galician
1381 0x0437: "ka_GE", # Georgian
1382 0x0407: "de_DE", # German - Germany
1383 0x0807: "de_CH", # German - Switzerland
1384 0x0c07: "de_AT", # German - Austria
1385 0x1007: "de_LU", # German - Luxembourg
1386 0x1407: "de_LI", # German - Liechtenstein
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001387 0x0408: "el_GR", # Greek
Georg Brandlb709c2c2006-01-20 09:07:35 +00001388 0x0447: "gu_IN", # Gujarati
1389 0x040d: "he_IL", # Hebrew
1390 0x0439: "hi_IN", # Hindi
1391 0x040e: "hu_HU", # Hungarian
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001392 0x040f: "is_IS", # Icelandic
Georg Brandlb709c2c2006-01-20 09:07:35 +00001393 0x0421: "id_ID", # Indonesian
1394 0x045d: "iu_CA", # Inuktitut
1395 0x085d: "iu_CA", # Inuktitut - Latin
1396 0x083c: "ga_IE", # Irish - Ireland
1397 0x0434: "xh_ZA", # Xhosa - South Africa
1398 0x0435: "zu_ZA", # Zulu
1399 0x0410: "it_IT", # Italian - Italy
1400 0x0810: "it_CH", # Italian - Switzerland
1401 0x0411: "ja_JP", # Japanese
1402 0x044b: "kn_IN", # Kannada - India
1403 0x043f: "kk_KZ", # Kazakh
1404 0x0457: "kok_IN",# Konkani
1405 0x0412: "ko_KR", # Korean
1406 0x0440: "ky_KG", # Kyrgyz
1407 0x0426: "lv_LV", # Latvian
1408 0x0427: "lt_LT", # Lithuanian
1409 0x046e: "lb_LU", # Luxembourgish
1410 0x042f: "mk_MK", # FYRO Macedonian
1411 0x043e: "ms_MY", # Malay - Malaysia
1412 0x083e: "ms_BN", # Malay - Brunei
1413 0x044c: "ml_IN", # Malayalam - India
1414 0x043a: "mt_MT", # Maltese
1415 0x0481: "mi_NZ", # Maori
1416 0x047a: "arn_CL",# Mapudungun
1417 0x044e: "mr_IN", # Marathi
1418 0x047c: "moh_CA",# Mohawk - Canada
1419 0x0450: "mn_MN", # Mongolian
1420 0x0461: "ne_NP", # Nepali
1421 0x0414: "nb_NO", # Norwegian - Bokmal
1422 0x0814: "nn_NO", # Norwegian - Nynorsk
1423 0x0482: "oc_FR", # Occitan - France
1424 0x0448: "or_IN", # Oriya - India
1425 0x0463: "ps_AF", # Pashto - Afghanistan
1426 0x0429: "fa_IR", # Persian
1427 0x0415: "pl_PL", # Polish
1428 0x0416: "pt_BR", # Portuguese - Brazil
1429 0x0816: "pt_PT", # Portuguese - Portugal
1430 0x0446: "pa_IN", # Punjabi
1431 0x046b: "quz_BO",# Quechua (Bolivia)
1432 0x086b: "quz_EC",# Quechua (Ecuador)
1433 0x0c6b: "quz_PE",# Quechua (Peru)
1434 0x0418: "ro_RO", # Romanian - Romania
1435 0x0417: "rm_CH", # Raeto-Romanese
1436 0x0419: "ru_RU", # Russian
1437 0x243b: "smn_FI",# Sami Finland
1438 0x103b: "smj_NO",# Sami Norway
1439 0x143b: "smj_SE",# Sami Sweden
1440 0x043b: "se_NO", # Sami Northern Norway
1441 0x083b: "se_SE", # Sami Northern Sweden
1442 0x0c3b: "se_FI", # Sami Northern Finland
1443 0x203b: "sms_FI",# Sami Skolt
1444 0x183b: "sma_NO",# Sami Southern Norway
1445 0x1c3b: "sma_SE",# Sami Southern Sweden
1446 0x044f: "sa_IN", # Sanskrit
1447 0x0c1a: "sr_SP", # Serbian - Cyrillic
1448 0x1c1a: "sr_BA", # Serbian - Bosnia Cyrillic
1449 0x081a: "sr_SP", # Serbian - Latin
1450 0x181a: "sr_BA", # Serbian - Bosnia Latin
1451 0x046c: "ns_ZA", # Northern Sotho
1452 0x0432: "tn_ZA", # Setswana - Southern Africa
1453 0x041b: "sk_SK", # Slovak
1454 0x0424: "sl_SI", # Slovenian
1455 0x040a: "es_ES", # Spanish - Spain
1456 0x080a: "es_MX", # Spanish - Mexico
1457 0x0c0a: "es_ES", # Spanish - Spain (Modern)
1458 0x100a: "es_GT", # Spanish - Guatemala
1459 0x140a: "es_CR", # Spanish - Costa Rica
1460 0x180a: "es_PA", # Spanish - Panama
1461 0x1c0a: "es_DO", # Spanish - Dominican Republic
1462 0x200a: "es_VE", # Spanish - Venezuela
1463 0x240a: "es_CO", # Spanish - Colombia
1464 0x280a: "es_PE", # Spanish - Peru
1465 0x2c0a: "es_AR", # Spanish - Argentina
1466 0x300a: "es_EC", # Spanish - Ecuador
1467 0x340a: "es_CL", # Spanish - Chile
1468 0x380a: "es_UR", # Spanish - Uruguay
1469 0x3c0a: "es_PY", # Spanish - Paraguay
1470 0x400a: "es_BO", # Spanish - Bolivia
1471 0x440a: "es_SV", # Spanish - El Salvador
1472 0x480a: "es_HN", # Spanish - Honduras
1473 0x4c0a: "es_NI", # Spanish - Nicaragua
1474 0x500a: "es_PR", # Spanish - Puerto Rico
1475 0x0441: "sw_KE", # Swahili
1476 0x041d: "sv_SE", # Swedish - Sweden
1477 0x081d: "sv_FI", # Swedish - Finland
1478 0x045a: "syr_SY",# Syriac
1479 0x0449: "ta_IN", # Tamil
1480 0x0444: "tt_RU", # Tatar
1481 0x044a: "te_IN", # Telugu
1482 0x041e: "th_TH", # Thai
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001483 0x041f: "tr_TR", # Turkish
Georg Brandlb709c2c2006-01-20 09:07:35 +00001484 0x0422: "uk_UA", # Ukrainian
1485 0x0420: "ur_PK", # Urdu
1486 0x0820: "ur_IN", # Urdu - India
1487 0x0443: "uz_UZ", # Uzbek - Latin
1488 0x0843: "uz_UZ", # Uzbek - Cyrillic
1489 0x042a: "vi_VN", # Vietnamese
1490 0x0452: "cy_GB", # Welsh
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001491}
1492
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001493def _print_locale():
1494
1495 """ Test function.
1496 """
1497 categories = {}
1498 def _init_categories(categories=categories):
1499 for k,v in globals().items():
1500 if k[:3] == 'LC_':
1501 categories[k] = v
1502 _init_categories()
1503 del categories['LC_ALL']
1504
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001505 print('Locale defaults as determined by getdefaultlocale():')
1506 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001507 lang, enc = getdefaultlocale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001508 print('Language: ', lang or '(undefined)')
1509 print('Encoding: ', enc or '(undefined)')
1510 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001511
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001512 print('Locale settings on startup:')
1513 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001514 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001515 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001516 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001517 print(' Language: ', lang or '(undefined)')
1518 print(' Encoding: ', enc or '(undefined)')
1519 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001520
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001521 print()
1522 print('Locale settings after calling resetlocale():')
1523 print('-'*72)
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001524 resetlocale()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001525 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001526 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001527 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001528 print(' Language: ', lang or '(undefined)')
1529 print(' Encoding: ', enc or '(undefined)')
1530 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001531
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001532 try:
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001533 setlocale(LC_ALL, "")
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001534 except:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001535 print('NOTE:')
1536 print('setlocale(LC_ALL, "") does not support the default locale')
1537 print('given in the OS environment variables.')
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001538 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001539 print()
1540 print('Locale settings after calling setlocale(LC_ALL, ""):')
1541 print('-'*72)
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001542 for name,category in categories.items():
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001543 print(name, '...')
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001544 lang, enc = getlocale(category)
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001545 print(' Language: ', lang or '(undefined)')
1546 print(' Encoding: ', enc or '(undefined)')
1547 print()
Fredrik Lundh6c86b992000-07-09 17:12:58 +00001548
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001549###
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001550
Tim Peters1baf8292001-01-24 10:13:46 +00001551try:
1552 LC_MESSAGES
Skip Montanaro0897f0c2002-03-25 21:40:36 +00001553except NameError:
Tim Peters1baf8292001-01-24 10:13:46 +00001554 pass
1555else:
1556 __all__.append("LC_MESSAGES")
1557
Guido van Rossumeef1d4e1997-11-19 19:01:43 +00001558if __name__=='__main__':
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001559 print('Locale aliasing:')
1560 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001561 _print_locale()
Guido van Rossumbe19ed72007-02-09 05:37:30 +00001562 print()
1563 print('Number formatting:')
1564 print()
Marc-André Lemburg5431bc32000-06-07 09:11:40 +00001565 _test()