Mercurial > babel > old > babel-test
annotate babel/numbers.py @ 101:1312ad6b624d
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
| author | cmlenz |
|---|---|
| date | Wed, 13 Jun 2007 08:26:23 +0000 |
| parents | 4dcdb1d367ec |
| children | b75ae5def3b1 |
| 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
4dcdb1d367ec
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
4dcdb1d367ec
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 |
| 36 def get_decimal_symbol(locale=LC_NUMERIC): | |
| 37 """Return the symbol used by the locale to separate decimal fractions. | |
| 38 | |
| 39 >>> get_decimal_symbol('en_US') | |
| 40 u'.' | |
| 41 | |
| 42 :param locale: the `Locale` object or locale identifier | |
| 43 :return: the decimal symbol | |
| 44 :rtype: `unicode` | |
| 45 """ | |
| 46 return Locale.parse(locale).number_symbols.get('decimal', u'.') | |
| 47 | |
| 48 def get_group_symbol(locale=LC_NUMERIC): | |
| 49 """Return the symbol used by the locale to separate groups of thousands. | |
| 50 | |
| 51 >>> get_group_symbol('en_US') | |
| 52 u',' | |
| 53 | |
| 54 :param locale: the `Locale` object or locale identifier | |
| 55 :return: the group symbol | |
| 56 :rtype: `unicode` | |
| 57 """ | |
| 9 | 58 return Locale.parse(locale).number_symbols.get('group', u',') |
| 1 | 59 |
| 60 def format_number(number, locale=LC_NUMERIC): | |
|
101
1312ad6b624d
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
61 """Return the given number formatted for a specific locale. |
| 1 | 62 |
| 63 >>> format_number(1099, locale='en_US') | |
| 64 u'1,099' | |
| 65 | |
| 66 :param number: the number to format | |
| 67 :param locale: the `Locale` object or locale identifier | |
| 68 :return: the formatted number | |
| 69 :rtype: `unicode` | |
| 70 """ | |
| 9 | 71 # Do we really need this one? |
| 72 return format_decimal(number, locale=locale) | |
| 1 | 73 |
| 9 | 74 def format_decimal(number, format=None, locale=LC_NUMERIC): |
|
101
1312ad6b624d
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
75 """Return the given decimal number formatted for a specific locale. |
| 1 | 76 |
| 9 | 77 >>> format_decimal(1.2345, locale='en_US') |
| 78 u'1.234' | |
| 50 | 79 >>> format_decimal(1.2346, locale='en_US') |
| 80 u'1.235' | |
| 81 >>> format_decimal(-1.2346, locale='en_US') | |
| 82 u'-1.235' | |
| 9 | 83 >>> format_decimal(1.2345, locale='sv_SE') |
| 84 u'1,234' | |
| 50 | 85 >>> format_decimal(12345, locale='de') |
| 9 | 86 u'12.345' |
| 87 | |
| 1 | 88 The appropriate thousands grouping and the decimal separator are used for |
| 89 each locale: | |
| 90 | |
| 9 | 91 >>> format_decimal(12345, locale='en_US') |
| 92 u'12,345' | |
| 93 | |
| 1 | 94 :param number: the number to format |
| 9 | 95 :param format: |
| 1 | 96 :param locale: the `Locale` object or locale identifier |
| 97 :return: the formatted decimal number | |
| 98 :rtype: `unicode` | |
| 99 """ | |
| 100 locale = Locale.parse(locale) | |
| 9 | 101 pattern = locale.decimal_formats.get(format) |
| 102 if not pattern: | |
| 103 pattern = parse_pattern(format) | |
| 104 return pattern.apply(number, locale) | |
| 1 | 105 |
| 32 | 106 def format_currency(number, currency, locale=LC_NUMERIC): |
|
101
1312ad6b624d
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
107 """Return formatted currency value. |
| 1 | 108 |
| 32 | 109 >>> format_currency(1099.98, 'USD', locale='en_US') |
| 1 | 110 u'1,099.98' |
| 111 | |
|
26
710090104678
* 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
|
112 :param number: the number to format |
| 32 | 113 :param currency: the currency code |
| 1 | 114 :param locale: the `Locale` object or locale identifier |
| 115 :return: the formatted currency value | |
| 116 :rtype: `unicode` | |
| 117 """ | |
|
26
710090104678
* 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
|
118 return format_decimal(number, locale=locale) |
| 1 | 119 |
|
26
710090104678
* 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
|
120 def format_percent(number, format=None, locale=LC_NUMERIC): |
|
101
1312ad6b624d
Add wrapper class bundling the various formatting functions bound to a specific locale and time-zone.
cmlenz
parents:
72
diff
changeset
|
121 """Return formatted percent value for a specific locale. |
| 22 | 122 |
| 123 >>> format_percent(0.34, locale='en_US') | |
| 124 u'34%' | |
| 125 >>> format_percent(25.1234, locale='en_US') | |
| 126 u'2,512%' | |
| 127 >>> format_percent(25.1234, locale='sv_SE') | |
| 128 u'2\\xa0512 %' | |
| 129 | |
| 130 :param number: the percent number to format | |
| 131 :param format: | |
| 132 :param locale: the `Locale` object or locale identifier | |
| 133 :return: the formatted percent number | |
| 134 :rtype: `unicode` | |
| 135 """ | |
| 136 locale = Locale.parse(locale) | |
| 137 pattern = locale.percent_formats.get(format) | |
| 138 if not pattern: | |
| 139 pattern = parse_pattern(format) | |
|
26
710090104678
* 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
|
140 return pattern.apply(number, locale) |
| 1 | 141 |
|
26
710090104678
* 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
|
142 def format_scientific(number, locale=LC_NUMERIC): |
| 1 | 143 raise NotImplementedError |
| 144 | |
| 32 | 145 |
| 146 | |
| 147 class NumberFormatError(ValueError): | |
| 148 """Exception raised when a string cannot be parsed into a number.""" | |
| 149 | |
| 150 | |
| 1 | 151 def parse_number(string, locale=LC_NUMERIC): |
| 152 """Parse localized number string into a long integer. | |
| 153 | |
| 154 >>> parse_number('1,099', locale='en_US') | |
| 155 1099L | |
| 156 >>> parse_number('1.099', locale='de_DE') | |
| 157 1099L | |
| 158 | |
| 32 | 159 When the given string cannot be parsed, an exception is raised: |
| 160 | |
| 161 >>> parse_number('1.099,98', locale='de') | |
| 162 Traceback (most recent call last): | |
| 163 ... | |
| 164 NumberFormatError: '1.099,98' is not a valid number | |
| 165 | |
| 1 | 166 :param string: the string to parse |
| 167 :param locale: the `Locale` object or locale identifier | |
| 168 :return: the parsed number | |
| 169 :rtype: `long` | |
| 32 | 170 :raise `NumberFormatError`: if the string can not be converted to a number |
| 1 | 171 """ |
| 32 | 172 try: |
| 173 return long(string.replace(get_group_symbol(locale), '')) | |
| 174 except ValueError: | |
| 175 raise NumberFormatError('%r is not a valid number' % string) | |
| 1 | 176 |
| 177 def parse_decimal(string, locale=LC_NUMERIC): | |
| 178 """Parse localized decimal string into a float. | |
| 179 | |
| 180 >>> parse_decimal('1,099.98', locale='en_US') | |
| 181 1099.98 | |
| 32 | 182 >>> parse_decimal('1.099,98', locale='de') |
| 1 | 183 1099.98 |
| 184 | |
| 32 | 185 When the given string cannot be parsed, an exception is raised: |
| 186 | |
| 187 >>> parse_decimal('2,109,998', locale='de') | |
| 188 Traceback (most recent call last): | |
| 189 ... | |
| 190 NumberFormatError: '2,109,998' is not a valid decimal number | |
| 191 | |
| 1 | 192 :param string: the string to parse |
| 193 :param locale: the `Locale` object or locale identifier | |
| 194 :return: the parsed decimal number | |
| 195 :rtype: `float` | |
| 32 | 196 :raise `NumberFormatError`: if the string can not be converted to a |
| 197 decimal number | |
| 1 | 198 """ |
| 199 locale = Locale.parse(locale) | |
| 32 | 200 try: |
| 201 return float(string.replace(get_group_symbol(locale), '') | |
| 202 .replace(get_decimal_symbol(locale), '.')) | |
| 203 except ValueError: | |
| 204 raise NumberFormatError('%r is not a valid decimal number' % string) | |
| 9 | 205 |
| 206 | |
| 207 PREFIX_END = r'[^0-9@#.,]' | |
| 208 NUMBER_TOKEN = r'[0-9@#.\-,E]' | |
| 209 | |
| 210 PREFIX_PATTERN = r"(?P<prefix>(?:'[^']*'|%s)*)" % PREFIX_END | |
| 211 NUMBER_PATTERN = r"(?P<number>%s+)" % NUMBER_TOKEN | |
| 212 SUFFIX_PATTERN = r"(?P<suffix>.*)" | |
| 213 | |
| 214 number_re = re.compile(r"%s%s%s" % (PREFIX_PATTERN, NUMBER_PATTERN, | |
| 215 SUFFIX_PATTERN)) | |
| 216 | |
| 217 # TODO: | |
| 218 # Filling | |
| 50 | 219 # Rounding increment in pattern |
| 9 | 220 # Scientific notation |
| 221 # Significant Digits | |
| 222 def parse_pattern(pattern): | |
| 223 """Parse number format patterns""" | |
| 224 if isinstance(pattern, NumberPattern): | |
| 225 return pattern | |
| 226 | |
| 227 # Do we have a negative subpattern? | |
| 228 if ';' in pattern: | |
| 229 pattern, neg_pattern = pattern.split(';', 1) | |
| 230 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 231 neg_prefix, _, neg_suffix = number_re.search(neg_pattern).groups() | |
| 232 else: | |
| 233 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 234 neg_prefix = '-' + pos_prefix | |
| 235 neg_suffix = pos_suffix | |
| 22 | 236 if '.' in number: |
| 237 integer, fraction = number.rsplit('.', 1) | |
| 238 else: | |
| 239 integer = number | |
| 240 fraction = '' | |
| 9 | 241 min_frac = max_frac = 0 |
| 242 | |
| 243 def parse_precision(p): | |
| 244 """Calculate the min and max allowed digits""" | |
| 245 min = max = 0 | |
| 246 for c in p: | |
| 247 if c == '0': | |
| 248 min += 1 | |
| 249 max += 1 | |
| 250 elif c == '#': | |
| 251 max += 1 | |
| 252 else: | |
| 253 break | |
| 254 return min, max | |
| 255 | |
| 256 def parse_grouping(p): | |
| 257 """Parse primary and secondary digit grouping | |
| 258 | |
| 259 >>> parse_grouping('##') | |
| 260 0, 0 | |
| 261 >>> parse_grouping('#,###') | |
| 262 3, 3 | |
| 263 >>> parse_grouping('#,####,###') | |
| 264 3, 4 | |
| 265 """ | |
| 266 width = len(p) | |
| 267 g1 = p.rfind(',') | |
| 268 if g1 == -1: | |
| 269 return 1000, 1000 | |
| 270 g1 = width - g1 - 1 | |
| 271 g2 = p[:-g1 - 1].rfind(',') | |
| 272 if g2 == -1: | |
| 273 return g1, g1 | |
| 274 g2 = width - g1 - g2 - 2 | |
| 275 return g1, g2 | |
| 276 | |
| 277 int_precision = parse_precision(integer) | |
| 278 frac_precision = parse_precision(fraction) | |
| 279 grouping = parse_grouping(integer) | |
| 280 int_precision = (int_precision[0], 1000) # Unlimited | |
| 281 return NumberPattern(pattern, (pos_prefix, neg_prefix), | |
| 282 (pos_suffix, neg_suffix), grouping, | |
| 283 int_precision, frac_precision) | |
| 284 | |
| 285 | |
| 286 class NumberPattern(object): | |
| 22 | 287 |
| 9 | 288 def __init__(self, pattern, prefix, suffix, grouping, |
| 289 int_precision, frac_precision): | |
| 290 self.pattern = pattern | |
| 291 self.prefix = prefix | |
| 292 self.suffix = suffix | |
| 293 self.grouping = grouping | |
| 294 self.int_precision = int_precision | |
| 295 self.frac_precision = frac_precision | |
| 50 | 296 self.format = '%%.%df' % self.frac_precision[1] |
| 22 | 297 if '%' in ''.join(self.prefix + self.suffix): |
| 298 self.scale = 100.0 | |
| 299 elif u'‰' in ''.join(self.prefix + self.suffix): | |
| 300 self.scale = 1000.0 | |
| 301 else: | |
| 302 self.scale = 1.0 | |
| 9 | 303 |
| 304 def __repr__(self): | |
| 305 return '<%s %r>' % (type(self).__name__, self.pattern) | |
| 306 | |
| 307 def apply(self, value, locale): | |
| 22 | 308 value *= self.scale |
| 9 | 309 negative = int(value < 0) |
| 50 | 310 a = self.format % value |
| 311 if self.frac_precision[1] > 0: | |
| 312 a, b = a.split('.') | |
| 313 else: | |
| 314 b = '' | |
| 9 | 315 a = a.lstrip('-') |
| 316 return '%s%s%s%s' % (self.prefix[negative], | |
| 317 self._format_int(a, locale), | |
| 318 self._format_frac(b, locale), | |
| 319 self.suffix[negative]) | |
| 320 | |
| 321 def _format_int(self, value, locale): | |
| 322 min, max = self.int_precision | |
| 323 width = len(value) | |
| 324 if width < min: | |
| 325 value += '0' * (min - width) | |
| 326 gsize = self.grouping[0] | |
| 327 ret = '' | |
| 328 symbol = get_group_symbol(locale) | |
| 329 while len(value) > gsize: | |
| 330 ret = symbol + value[-gsize:] + ret | |
| 331 value = value[:-gsize] | |
| 332 gsize = self.grouping[1] | |
| 333 return value + ret | |
| 334 | |
| 335 def _format_frac(self, value, locale): | |
| 336 min, max = self.frac_precision | |
| 22 | 337 if max == 0 or (min == 0 and int(value) == 0): |
| 9 | 338 return '' |
| 339 width = len(value) | |
| 50 | 340 while len(value) > min and value[-1] == '0': |
| 341 value = value[:-1] | |
| 9 | 342 return get_decimal_symbol(locale) + value |