Mercurial > babel > mirror
annotate babel/numbers.py @ 26:6041782ea677 trunk
* Reduce size of locale data pickles by only storing the data provided by each locale itself, and merging inherited data at runtime.
* Move locale data loading from `babel.core` into a separate `babel.localedata` module.
* Add curency names and symbols to locale data.
| author | cmlenz |
|---|---|
| date | Sun, 03 Jun 2007 15:27:27 +0000 |
| parents | d1e6944f2ff0 |
| children | ac0957a8fd8b |
| 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 | |
| 27 from babel.core import Locale | |
| 28 from babel.util import default_locale | |
| 29 | |
| 30 __all__ = ['format_number', 'format_decimal', 'format_currency', | |
| 31 'format_percent', 'format_scientific', 'parse_number', | |
| 32 'parse_decimal'] | |
| 33 __docformat__ = 'restructuredtext en' | |
| 34 | |
| 35 LC_NUMERIC = default_locale('LC_NUMERIC') | |
| 36 | |
| 37 def get_decimal_symbol(locale=LC_NUMERIC): | |
| 38 """Return the symbol used by the locale to separate decimal fractions. | |
| 39 | |
| 40 >>> get_decimal_symbol('en_US') | |
| 41 u'.' | |
| 42 | |
| 43 :param locale: the `Locale` object or locale identifier | |
| 44 :return: the decimal symbol | |
| 45 :rtype: `unicode` | |
| 46 """ | |
| 47 return Locale.parse(locale).number_symbols.get('decimal', u'.') | |
| 48 | |
| 49 def get_group_symbol(locale=LC_NUMERIC): | |
| 50 """Return the symbol used by the locale to separate groups of thousands. | |
| 51 | |
| 52 >>> get_group_symbol('en_US') | |
| 53 u',' | |
| 54 | |
| 55 :param locale: the `Locale` object or locale identifier | |
| 56 :return: the group symbol | |
| 57 :rtype: `unicode` | |
| 58 """ | |
| 9 | 59 return Locale.parse(locale).number_symbols.get('group', u',') |
| 1 | 60 |
| 61 def format_number(number, locale=LC_NUMERIC): | |
| 62 """Returns the given number formatted for a specific locale. | |
| 63 | |
| 64 >>> format_number(1099, locale='en_US') | |
| 65 u'1,099' | |
| 66 | |
| 67 :param number: the number to format | |
| 68 :param locale: the `Locale` object or locale identifier | |
| 69 :return: the formatted number | |
| 70 :rtype: `unicode` | |
| 71 """ | |
| 9 | 72 # Do we really need this one? |
| 73 return format_decimal(number, locale=locale) | |
| 1 | 74 |
| 9 | 75 def format_decimal(number, format=None, locale=LC_NUMERIC): |
| 1 | 76 """Returns the given decimal number formatted for a specific locale. |
| 77 | |
| 9 | 78 >>> format_decimal(1, locale='en_US') |
| 79 u'1' | |
| 80 >>> format_decimal(1.2345, locale='en_US') | |
| 81 u'1.234' | |
| 82 >>> format_decimal(1.2345, locale='sv_SE') | |
| 83 u'1,234' | |
| 84 >>> format_decimal(12345, locale='de_DE') | |
| 85 u'12.345' | |
| 86 >>> format_decimal(-1.2345, format='#,##0.##;-#', locale='sv_SE') | |
| 87 u'-1,23' | |
| 88 >>> format_decimal(-1.2345, format='#,##0.##;(#)', locale='sv_SE') | |
| 89 u'(1,23)' | |
| 90 | |
| 1 | 91 The appropriate thousands grouping and the decimal separator are used for |
| 92 each locale: | |
| 93 | |
| 9 | 94 >>> format_decimal(12345, locale='en_US') |
| 95 u'12,345' | |
| 96 | |
| 1 | 97 :param number: the number to format |
| 9 | 98 :param format: |
| 1 | 99 :param locale: the `Locale` object or locale identifier |
| 100 :return: the formatted decimal number | |
| 101 :rtype: `unicode` | |
| 102 """ | |
| 103 locale = Locale.parse(locale) | |
| 9 | 104 pattern = locale.decimal_formats.get(format) |
| 105 if not pattern: | |
| 106 pattern = parse_pattern(format) | |
| 107 return pattern.apply(number, locale) | |
| 1 | 108 |
|
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
|
109 def format_currency(number, locale=LC_NUMERIC): |
| 1 | 110 """Returns formatted currency value. |
| 111 | |
| 112 >>> format_currency(1099.98, locale='en_US') | |
| 113 u'1,099.98' | |
| 114 | |
|
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
|
115 :param number: the number to format |
| 1 | 116 :param locale: the `Locale` object or locale identifier |
| 117 :return: the formatted currency value | |
| 118 :rtype: `unicode` | |
| 119 """ | |
|
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
|
120 return format_decimal(number, locale=locale) |
| 1 | 121 |
|
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
|
122 def format_percent(number, format=None, locale=LC_NUMERIC): |
| 22 | 123 """Returns formatted percent value for a specific locale. |
| 124 | |
| 125 >>> format_percent(0.34, locale='en_US') | |
| 126 u'34%' | |
| 127 >>> format_percent(25.1234, locale='en_US') | |
| 128 u'2,512%' | |
| 129 >>> format_percent(25.1234, locale='sv_SE') | |
| 130 u'2\\xa0512 %' | |
| 131 | |
| 132 :param number: the percent number to format | |
| 133 :param format: | |
| 134 :param locale: the `Locale` object or locale identifier | |
| 135 :return: the formatted percent number | |
| 136 :rtype: `unicode` | |
| 137 """ | |
| 138 locale = Locale.parse(locale) | |
| 139 pattern = locale.percent_formats.get(format) | |
| 140 if not pattern: | |
| 141 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
|
142 return pattern.apply(number, locale) |
| 1 | 143 |
|
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
|
144 def format_scientific(number, locale=LC_NUMERIC): |
| 1 | 145 raise NotImplementedError |
| 146 | |
| 147 def parse_number(string, locale=LC_NUMERIC): | |
| 148 """Parse localized number string into a long integer. | |
| 149 | |
| 150 >>> parse_number('1,099', locale='en_US') | |
| 151 1099L | |
| 152 >>> parse_number('1.099', locale='de_DE') | |
| 153 1099L | |
| 154 | |
| 155 :param string: the string to parse | |
| 156 :param locale: the `Locale` object or locale identifier | |
| 157 :return: the parsed number | |
| 158 :rtype: `long` | |
| 159 :raise `ValueError`: if the string can not be converted to a number | |
| 160 """ | |
| 161 return long(string.replace(get_group_symbol(locale), '')) | |
| 162 | |
| 163 def parse_decimal(string, locale=LC_NUMERIC): | |
| 164 """Parse localized decimal string into a float. | |
| 165 | |
| 166 >>> parse_decimal('1,099.98', locale='en_US') | |
| 167 1099.98 | |
| 168 >>> parse_decimal('1.099,98', locale='de_DE') | |
| 169 1099.98 | |
| 170 | |
| 171 :param string: the string to parse | |
| 172 :param locale: the `Locale` object or locale identifier | |
| 173 :return: the parsed decimal number | |
| 174 :rtype: `float` | |
| 175 :raise `ValueError`: if the string can not be converted to a decimal number | |
| 176 """ | |
| 177 locale = Locale.parse(locale) | |
| 178 string = string.replace(get_group_symbol(locale), '') \ | |
| 179 .replace(get_decimal_symbol(locale), '.') | |
| 180 return float(string) | |
| 9 | 181 |
| 182 | |
| 183 PREFIX_END = r'[^0-9@#.,]' | |
| 184 NUMBER_TOKEN = r'[0-9@#.\-,E]' | |
| 185 | |
| 186 PREFIX_PATTERN = r"(?P<prefix>(?:'[^']*'|%s)*)" % PREFIX_END | |
| 187 NUMBER_PATTERN = r"(?P<number>%s+)" % NUMBER_TOKEN | |
| 188 SUFFIX_PATTERN = r"(?P<suffix>.*)" | |
| 189 | |
| 190 number_re = re.compile(r"%s%s%s" % (PREFIX_PATTERN, NUMBER_PATTERN, | |
| 191 SUFFIX_PATTERN)) | |
| 192 | |
| 193 # TODO: | |
| 194 # Filling | |
| 195 # Rounding | |
| 196 # Scientific notation | |
| 197 # Significant Digits | |
| 198 def parse_pattern(pattern): | |
| 199 """Parse number format patterns""" | |
| 200 if isinstance(pattern, NumberPattern): | |
| 201 return pattern | |
| 202 | |
| 203 # Do we have a negative subpattern? | |
| 204 if ';' in pattern: | |
| 205 pattern, neg_pattern = pattern.split(';', 1) | |
| 206 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 207 neg_prefix, _, neg_suffix = number_re.search(neg_pattern).groups() | |
| 208 else: | |
| 209 pos_prefix, number, pos_suffix = number_re.search(pattern).groups() | |
| 210 neg_prefix = '-' + pos_prefix | |
| 211 neg_suffix = pos_suffix | |
| 22 | 212 if '.' in number: |
| 213 integer, fraction = number.rsplit('.', 1) | |
| 214 else: | |
| 215 integer = number | |
| 216 fraction = '' | |
| 9 | 217 min_frac = max_frac = 0 |
| 218 | |
| 219 def parse_precision(p): | |
| 220 """Calculate the min and max allowed digits""" | |
| 221 min = max = 0 | |
| 222 for c in p: | |
| 223 if c == '0': | |
| 224 min += 1 | |
| 225 max += 1 | |
| 226 elif c == '#': | |
| 227 max += 1 | |
| 228 else: | |
| 229 break | |
| 230 return min, max | |
| 231 | |
| 232 def parse_grouping(p): | |
| 233 """Parse primary and secondary digit grouping | |
| 234 | |
| 235 >>> parse_grouping('##') | |
| 236 0, 0 | |
| 237 >>> parse_grouping('#,###') | |
| 238 3, 3 | |
| 239 >>> parse_grouping('#,####,###') | |
| 240 3, 4 | |
| 241 """ | |
| 242 width = len(p) | |
| 243 g1 = p.rfind(',') | |
| 244 if g1 == -1: | |
| 245 return 1000, 1000 | |
| 246 g1 = width - g1 - 1 | |
| 247 g2 = p[:-g1 - 1].rfind(',') | |
| 248 if g2 == -1: | |
| 249 return g1, g1 | |
| 250 g2 = width - g1 - g2 - 2 | |
| 251 return g1, g2 | |
| 252 | |
| 253 int_precision = parse_precision(integer) | |
| 254 frac_precision = parse_precision(fraction) | |
| 255 grouping = parse_grouping(integer) | |
| 256 int_precision = (int_precision[0], 1000) # Unlimited | |
| 257 return NumberPattern(pattern, (pos_prefix, neg_prefix), | |
| 258 (pos_suffix, neg_suffix), grouping, | |
| 259 int_precision, frac_precision) | |
| 260 | |
| 261 | |
| 262 class NumberPattern(object): | |
| 22 | 263 |
| 9 | 264 def __init__(self, pattern, prefix, suffix, grouping, |
| 265 int_precision, frac_precision): | |
| 266 self.pattern = pattern | |
| 267 self.prefix = prefix | |
| 268 self.suffix = suffix | |
| 269 self.grouping = grouping | |
| 270 self.int_precision = int_precision | |
| 271 self.frac_precision = frac_precision | |
| 22 | 272 if '%' in ''.join(self.prefix + self.suffix): |
| 273 self.scale = 100.0 | |
| 274 elif u'‰' in ''.join(self.prefix + self.suffix): | |
| 275 self.scale = 1000.0 | |
| 276 else: | |
| 277 self.scale = 1.0 | |
| 9 | 278 |
| 279 def __repr__(self): | |
| 280 return '<%s %r>' % (type(self).__name__, self.pattern) | |
| 281 | |
| 282 def apply(self, value, locale): | |
| 22 | 283 value *= self.scale |
| 9 | 284 negative = int(value < 0) |
| 22 | 285 a, b = str(value).split('.') |
| 9 | 286 a = a.lstrip('-') |
| 287 return '%s%s%s%s' % (self.prefix[negative], | |
| 288 self._format_int(a, locale), | |
| 289 self._format_frac(b, locale), | |
| 290 self.suffix[negative]) | |
| 291 | |
| 292 def _format_int(self, value, locale): | |
| 293 min, max = self.int_precision | |
| 294 width = len(value) | |
| 295 if width < min: | |
| 296 value += '0' * (min - width) | |
| 297 gsize = self.grouping[0] | |
| 298 ret = '' | |
| 299 symbol = get_group_symbol(locale) | |
| 300 while len(value) > gsize: | |
| 301 ret = symbol + value[-gsize:] + ret | |
| 302 value = value[:-gsize] | |
| 303 gsize = self.grouping[1] | |
| 304 return value + ret | |
| 305 | |
| 306 def _format_frac(self, value, locale): | |
| 307 min, max = self.frac_precision | |
| 22 | 308 if max == 0 or (min == 0 and int(value) == 0): |
| 9 | 309 return '' |
| 310 width = len(value) | |
| 311 if width < min: | |
| 312 value += '0' * (min - width) | |
| 313 if width > max: | |
| 314 value = value[:max] # FIXME: Rounding?!? | |
| 315 return get_decimal_symbol(locale) + value |