Mercurial > babel > mirror
annotate babel/numbers.py @ 137:289002e555b1 trunk
Some code cleanup.
| author | jonas |
|---|---|
| date | Tue, 19 Jun 2007 19:44:02 +0000 |
| parents | 32b28673cfd5 |
| children | 5125934e5f27 |
| rev | line source |
|---|---|
| 1 | 1 # -*- coding: utf-8 -*- |
| 2 # | |
| 3 # Copyright (C) 2007 Edgewall Software | |
| 4 # All rights reserved. | |
| 5 # | |
| 6 # This software is licensed as described in the file COPYING, which | |
| 7 # you should have received as part of this distribution. The terms | |
| 8 # are also available at http://babel.edgewall.org/wiki/License. | |
| 9 # | |
| 10 # This software consists of voluntary contributions made by many | |
| 11 # individuals. For the exact contribution history, see the revision | |
| 12 # history and logs, available at http://babel.edgewall.org/log/. | |
| 13 | |
| 14 """Locale dependent formatting and parsing of numeric data. | |
| 15 | |
| 16 The default locale for the functions in this module is determined by the | |
| 17 following environment variables, in that order: | |
| 18 | |
| 19 * ``LC_NUMERIC``, | |
| 20 * ``LC_ALL``, and | |
| 21 * ``LANG`` | |
| 22 """ | |
| 23 # TODO: percent and scientific formatting | |
| 24 | |
| 25 import re | |
| 26 | |
|
72
e0bb7dce49ea
More explicit module-level function names in `babel.core`. Added `Locale.negotiate` class method.
cmlenz
parents:
50
diff
changeset
|
27 from babel.core import default_locale, Locale |
| 1 | 28 |
| 29 __all__ = ['format_number', 'format_decimal', 'format_currency', | |
| 30 'format_percent', 'format_scientific', 'parse_number', | |
| 32 | 31 'parse_decimal', 'NumberFormatError'] |
| 1 | 32 __docformat__ = 'restructuredtext en' |
| 33 | |
|
72
e0bb7dce49ea
More explicit module-level function names in `babel.core`. Added `Locale.negotiate` class method.
cmlenz
parents:
50
diff
changeset
|
34 LC_NUMERIC = default_locale('LC_NUMERIC') |
| 1 | 35 |
| 125 | 36 def get_currency_symbol(currency, locale=LC_NUMERIC): |
| 37 """Return the symbol used by the locale for the specified currency. | |
| 38 | |
| 39 >>> get_currency_symbol('USD', 'en_US') | |
| 40 u'$' | |
| 41 | |
| 42 :param currency: the currency code | |
| 43 :param locale: the `Locale` object or locale identifier | |
| 44 :return: the currency symbol | |
| 45 :rtype: `unicode` | |
| 46 """ | |
| 47 return Locale.parse(locale).currency_symbols.get(currency, currency) | |
| 48 | |
| 1 | 49 def get_decimal_symbol(locale=LC_NUMERIC): |
| 50 """Return the symbol used by the locale to separate decimal fractions. | |
| 51 | |
| 52 >>> get_decimal_symbol('en_US') | |
| 53 u'.' | |
| 54 | |
| 55 :param locale: the `Locale` object or locale identifier | |
| 56 :return: the decimal symbol | |
| 57 :rtype: `unicode` | |
| 58 """ | |
| 59 return Locale.parse(locale).number_symbols.get('decimal', u'.') | |
| 60 | |
| 61 def get_group_symbol(locale=LC_NUMERIC): | |
| 62 """Return the symbol used by the locale to separate groups of thousands. | |
| 63 | |
| 64 >>> get_group_symbol('en_US') | |
| 65 u',' | |
| 66 | |
| 67 :param locale: the `Locale` object or locale identifier | |
| 68 :return: the group symbol | |
| 69 :rtype: `unicode` | |
| 70 """ | |
| 9 | 71 return Locale.parse(locale).number_symbols.get('group', u',') |
| 1 | 72 |
| 73 def format_number(number, locale=LC_NUMERIC): | |
|
101
0f641136aa6b
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
74 """Return the given number formatted for a specific locale. |
| 1 | 75 |
| 76 >>> format_number(1099, locale='en_US') | |
| 77 u'1,099' | |
| 78 | |
| 79 :param number: the number to format | |
| 80 :param locale: the `Locale` object or locale identifier | |
| 81 :return: the formatted number | |
| 82 :rtype: `unicode` | |
| 83 """ | |
| 9 | 84 # Do we really need this one? |
| 85 return format_decimal(number, locale=locale) | |
| 1 | 86 |
| 9 | 87 def format_decimal(number, format=None, locale=LC_NUMERIC): |
|
101
0f641136aa6b
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
88 """Return the given decimal number formatted for a specific locale. |
| 1 | 89 |
| 9 | 90 >>> format_decimal(1.2345, locale='en_US') |
| 91 u'1.234' | |
| 50 | 92 >>> format_decimal(1.2346, locale='en_US') |
| 93 u'1.235' | |
| 94 >>> format_decimal(-1.2346, locale='en_US') | |
| 95 u'-1.235' | |
| 9 | 96 >>> format_decimal(1.2345, locale='sv_SE') |
| 97 u'1,234' | |
| 50 | 98 >>> format_decimal(12345, locale='de') |
| 9 | 99 u'12.345' |
| 100 | |
| 1 | 101 The appropriate thousands grouping and the decimal separator are used for |
| 102 each locale: | |
| 103 | |
| 125 | 104 >>> format_decimal(12345.5, locale='en_US') |
| 105 u'12,345.5' | |
| 9 | 106 |
| 1 | 107 :param number: the number to format |
| 9 | 108 :param format: |
| 1 | 109 :param locale: the `Locale` object or locale identifier |
| 110 :return: the formatted decimal number | |
| 111 :rtype: `unicode` | |
| 112 """ | |
| 113 locale = Locale.parse(locale) | |
| 125 | 114 if not format: |
| 115 format = locale.decimal_formats.get(format) | |
| 116 pattern = parse_pattern(format) | |
| 9 | 117 return pattern.apply(number, locale) |
| 1 | 118 |
| 125 | 119 def format_currency(number, currency, format=None, locale=LC_NUMERIC): |
| 133 | 120 u"""Return formatted currency value. |
| 1 | 121 |
| 32 | 122 >>> format_currency(1099.98, 'USD', locale='en_US') |
| 125 | 123 u'$1,099.98' |
| 124 >>> format_currency(1099.98, 'USD', locale='es_CO') | |
| 125 u'US$1.099,98' | |
| 126 >>> format_currency(1099.98, 'EUR', locale='de_DE') | |
| 127 u'1.099,98 \\u20ac' | |
| 128 | |
| 129 The pattern can also be specified explicitly: | |
| 130 | |
| 131 >>> format_currency(1099.98, 'EUR', u'\xa4\xa4 #,##0.00', locale='en_US') | |
| 132 u'EUR 1,099.98' | |
| 1 | 133 |
|
26
6041782ea677
* Reduce size of locale data pickles by only storing the data provided by each locale itself, and merging inherited data at runtime.
cmlenz
parents:
22
diff
changeset
|
134 :param number: the number to format |
| 32 | 135 :param currency: the currency code |
| 1 | 136 :param locale: the `Locale` object or locale identifier |
| 137 :return: the formatted currency value | |
| 138 :rtype: `unicode` | |
| 139 """ | |
| 125 | 140 locale = Locale.parse(locale) |
| 141 if not format: | |
| 142 format = locale.currency_formats.get(format) | |
| 143 pattern = parse_pattern(format) | |
| 144 return pattern.apply(number, locale, currency=currency) | |
| 1 | 145 |
|
26
6041782ea677
* Reduce size of locale data pickles by only storing the data provided by each locale itself, and merging inherited data at runtime.
cmlenz
parents:
22
diff
changeset
|
146 def format_percent(number, format=None, locale=LC_NUMERIC): |
|
101
0f641136aa6b
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
147 """Return formatted percent value for a specific locale. |
| 22 | 148 |
| 149 >>> format_percent(0.34, locale='en_US') | |
| 150 u'34%' | |
| 151 >>> format_percent(25.1234, locale='en_US') | |
| 152 u'2,512%' | |
| 153 >>> format_percent(25.1234, locale='sv_SE') | |
| 154 u'2\\xa0512 %' | |
| 155 | |
| 126 | 156 The format pattern can also be specified explicitly: |
| 157 | |
| 158 >>> format_percent(25.1234, u'#,##0\u2030', locale='en_US') | |
| 159 u'25,123\u2030' | |
| 160 | |
| 22 | 161 :param number: the percent number to format |
| 162 :param format: | |
| 163 :param locale: the `Locale` object or locale identifier | |
| 164 :return: the formatted percent number | |
| 165 :rtype: `unicode` | |
| 166 """ | |
| 167 locale = Locale.parse(locale) | |
| 125 | 168 if not format: |
| 169 format = locale.percent_formats.get(format) | |
| 170 pattern = parse_pattern(format) | |
|
26
6041782ea677
* Reduce size of locale data pickles by only storing the data provided by each locale itself, and merging inherited data at runtime.
cmlenz
parents:
22
diff
changeset
|
171 return pattern.apply(number, locale) |
| 1 | 172 |
|
26
6041782ea677
* Reduce size of locale data pickles by only storing the data provided by each locale itself, and merging inherited data at runtime.
cmlenz
parents:
22
diff
changeset
|
173 def format_scientific(number, locale=LC_NUMERIC): |
| 126 | 174 # TODO: implement |
| 1 | 175 raise NotImplementedError |
| 176 | |
| 32 | 177 |
| 178 class NumberFormatError(ValueError): | |
| 179 """Exception raised when a string cannot be parsed into a number.""" | |
| 180 | |
| 181 | |
| 1 | 182 def parse_number(string, locale=LC_NUMERIC): |
| 183 """Parse localized number string into a long integer. | |
| 184 | |
| 185 >>> parse_number('1,099', locale='en_US') | |
| 186 1099L | |
| 187 >>> parse_number('1.099', locale='de_DE') | |
| 188 1099L | |
| 189 | |
| 32 | 190 When the given string cannot be parsed, an exception is raised: |
| 191 | |
| 192 >>> parse_number('1.099,98', locale='de') | |
| 193 Traceback (most recent call last): | |
| 194 ... | |
| 195 NumberFormatError: '1.099,98' is not a valid number | |
| 196 | |
| 1 | 197 :param string: the string to parse |
| 198 :param locale: the `Locale` object or locale identifier | |
| 199 :return: the parsed number | |
| 200 :rtype: `long` | |
| 32 | 201 :raise `NumberFormatError`: if the string can not be converted to a number |
| 1 | 202 """ |
| 32 | 203 try: |
| 204 return long(string.replace(get_group_symbol(locale), '')) | |
| 205 except ValueError: | |
| 206 raise NumberFormatError('%r is not a valid number' % string) | |
| 1 | 207 |
| 208 def parse_decimal(string, locale=LC_NUMERIC): | |
| 209 """Parse localized decimal string into a float. | |
| 210 | |
| 211 >>> parse_decimal('1,099.98', locale='en_US') | |
| 212 1099.98 | |
| 32 | 213 >>> parse_decimal('1.099,98', locale='de') |
| 1 | 214 1099.98 |
| 215 | |
| 32 | 216 When the given string cannot be parsed, an exception is raised: |
| 217 | |
| 218 >>> parse_decimal('2,109,998', locale='de') | |
| 219 Traceback (most recent call last): | |
| 220 ... | |
| 221 NumberFormatError: '2,109,998' is not a valid decimal number | |
| 222 | |
| 1 | 223 :param string: the string to parse |
| 224 :param locale: the `Locale` object or locale identifier | |
| 225 :return: the parsed decimal number | |
| 226 :rtype: `float` | |
| 32 | 227 :raise `NumberFormatError`: if the string can not be converted to a |
| 228 decimal number | |
| 1 | 229 """ |
| 230 locale = Locale.parse(locale) | |
| 32 | 231 try: |
| 232 return float(string.replace(get_group_symbol(locale), '') | |
| 233 .replace(get_decimal_symbol(locale), '.')) | |
| 234 except ValueError: | |
| 235 raise NumberFormatError('%r is not a valid decimal number' % string) | |
| 9 | 236 |
| 237 | |
| 238 PREFIX_END = r'[^0-9@#.,]' | |
| 239 NUMBER_TOKEN = r'[0-9@#.\-,E]' | |
| 240 | |
| 241 PREFIX_PATTERN = r"(?P<prefix>(?:'[^']*'|%s)*)" % PREFIX_END | |
| 242 NUMBER_PATTERN = r"(?P<number>%s+)" % NUMBER_TOKEN | |
| 243 SUFFIX_PATTERN = r"(?P<suffix>.*)" | |
| 244 | |
| 127 | 245 number_re = re.compile(r"%s%s%s" % (PREFIX_PATTERN, NUMBER_PATTERN, |
| 9 | 246 SUFFIX_PATTERN)) |
| 247 | |
| 248 # TODO: | |
| 249 # Filling | |
| 50 | 250 # Rounding increment in pattern |
| 9 | 251 # Scientific notation |
| 252 # Significant Digits | |
| 253 def parse_pattern(pattern): | |
| 254 """Parse number format patterns""" | |
| 255 if isinstance(pattern, NumberPattern): | |
| 256 return pattern | |
| 257 | |
| 258 # Do we have a negative subpattern? | |
| 259 if ';' in pattern: | |
| 260 pattern, neg_pattern = pattern.split(';', 1) | |
| 261 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 262 neg_prefix, _, neg_suffix = number_re.search(neg_pattern).groups() | |
| 263 else: | |
| 264 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 265 neg_prefix = '-' + pos_prefix | |
| 266 neg_suffix = pos_suffix | |
| 22 | 267 if '.' in number: |
| 268 integer, fraction = number.rsplit('.', 1) | |
| 269 else: | |
| 270 integer = number | |
| 271 fraction = '' | |
| 9 | 272 min_frac = max_frac = 0 |
| 273 | |
| 274 def parse_precision(p): | |
| 275 """Calculate the min and max allowed digits""" | |
| 276 min = max = 0 | |
| 277 for c in p: | |
| 278 if c == '0': | |
| 279 min += 1 | |
| 280 max += 1 | |
| 281 elif c == '#': | |
| 282 max += 1 | |
| 283 else: | |
| 284 break | |
| 285 return min, max | |
| 286 | |
| 287 def parse_grouping(p): | |
| 288 """Parse primary and secondary digit grouping | |
| 289 | |
| 290 >>> parse_grouping('##') | |
| 291 0, 0 | |
| 292 >>> parse_grouping('#,###') | |
| 293 3, 3 | |
| 294 >>> parse_grouping('#,####,###') | |
| 295 3, 4 | |
| 296 """ | |
| 297 width = len(p) | |
| 298 g1 = p.rfind(',') | |
| 299 if g1 == -1: | |
| 300 return 1000, 1000 | |
| 301 g1 = width - g1 - 1 | |
| 302 g2 = p[:-g1 - 1].rfind(',') | |
| 303 if g2 == -1: | |
| 304 return g1, g1 | |
| 305 g2 = width - g1 - g2 - 2 | |
| 306 return g1, g2 | |
| 307 | |
| 308 int_precision = parse_precision(integer) | |
| 309 frac_precision = parse_precision(fraction) | |
| 310 grouping = parse_grouping(integer) | |
| 311 int_precision = (int_precision[0], 1000) # Unlimited | |
| 312 return NumberPattern(pattern, (pos_prefix, neg_prefix), | |
| 313 (pos_suffix, neg_suffix), grouping, | |
| 314 int_precision, frac_precision) | |
| 315 | |
| 316 | |
| 317 class NumberPattern(object): | |
| 22 | 318 |
| 9 | 319 def __init__(self, pattern, prefix, suffix, grouping, |
| 320 int_precision, frac_precision): | |
| 321 self.pattern = pattern | |
| 322 self.prefix = prefix | |
| 323 self.suffix = suffix | |
| 324 self.grouping = grouping | |
| 325 self.int_precision = int_precision | |
| 326 self.frac_precision = frac_precision | |
| 137 | 327 self.format = '%%#.%df' % self.frac_precision[1] |
| 22 | 328 if '%' in ''.join(self.prefix + self.suffix): |
| 329 self.scale = 100.0 | |
| 330 elif u'‰' in ''.join(self.prefix + self.suffix): | |
| 331 self.scale = 1000.0 | |
| 332 else: | |
| 333 self.scale = 1.0 | |
| 9 | 334 |
| 335 def __repr__(self): | |
| 336 return '<%s %r>' % (type(self).__name__, self.pattern) | |
| 337 | |
| 125 | 338 def apply(self, value, locale, currency=None): |
| 22 | 339 value *= self.scale |
| 9 | 340 negative = int(value < 0) |
| 137 | 341 a, b = (self.format % abs(value)).split('.', 1) |
| 126 | 342 retval = u'%s%s%s%s' % (self.prefix[negative], |
| 343 self._format_int(a, locale), | |
| 344 self._format_frac(b, locale), | |
| 345 self.suffix[negative]) | |
| 125 | 346 if u'¤' in retval: |
| 347 retval = retval.replace(u'¤¤', currency.upper()) | |
| 348 retval = retval.replace(u'¤', get_currency_symbol(currency, locale)) | |
| 349 return retval | |
| 9 | 350 |
| 351 def _format_int(self, value, locale): | |
| 352 min, max = self.int_precision | |
| 353 width = len(value) | |
| 354 if width < min: | |
| 355 value += '0' * (min - width) | |
| 356 gsize = self.grouping[0] | |
| 357 ret = '' | |
| 358 symbol = get_group_symbol(locale) | |
| 359 while len(value) > gsize: | |
| 360 ret = symbol + value[-gsize:] + ret | |
| 361 value = value[:-gsize] | |
| 362 gsize = self.grouping[1] | |
| 363 return value + ret | |
| 364 | |
| 365 def _format_frac(self, value, locale): | |
| 366 min, max = self.frac_precision | |
| 22 | 367 if max == 0 or (min == 0 and int(value) == 0): |
| 9 | 368 return '' |
| 369 width = len(value) | |
| 50 | 370 while len(value) > min and value[-1] == '0': |
| 371 value = value[:-1] | |
| 9 | 372 return get_decimal_symbol(locale) + value |