Mercurial > babel > old > mirror
comparison scripts/import_cldr.py @ 392:34c0a25b1ed7
Preliminary support for timedelta formatting (see #126), and import/expose the locale plural rules from the CLDR.
| author | cmlenz |
|---|---|
| date | Mon, 14 Jul 2008 22:13:44 +0000 |
| parents | 88e3589ca8df |
| children | 17515f1efee0 |
comparison
equal
deleted
inserted
replaced
| 390:d21f92883d82 | 392:34c0a25b1ed7 |
|---|---|
| 25 | 25 |
| 26 # Make sure we're using Babel source, and not some previously installed version | 26 # Make sure we're using Babel source, and not some previously installed version |
| 27 sys.path.insert(0, os.path.join(os.path.dirname(sys.argv[0]), '..')) | 27 sys.path.insert(0, os.path.join(os.path.dirname(sys.argv[0]), '..')) |
| 28 | 28 |
| 29 from babel import dates, numbers | 29 from babel import dates, numbers |
| 30 from babel.plural import PluralRule | |
| 30 from babel.localedata import Alias | 31 from babel.localedata import Alias |
| 31 | 32 |
| 32 weekdays = {'mon': 0, 'tue': 1, 'wed': 2, 'thu': 3, 'fri': 4, 'sat': 5, | 33 weekdays = {'mon': 0, 'tue': 1, 'wed': 2, 'thu': 3, 'fri': 4, 'sat': 5, |
| 33 'sun': 6} | 34 'sun': 6} |
| 34 | 35 |
| 129 containers = territory_containment.setdefault(territory, set([])) | 130 containers = territory_containment.setdefault(territory, set([])) |
| 130 if group in territory_containment: | 131 if group in territory_containment: |
| 131 containers |= territory_containment[group] | 132 containers |= territory_containment[group] |
| 132 containers.add(group) | 133 containers.add(group) |
| 133 | 134 |
| 135 # prepare the per-locale plural rules definitions | |
| 136 plural_rules = {} | |
| 137 prsup = parse(os.path.join(srcdir, 'supplemental', 'plurals.xml')) | |
| 138 for elem in prsup.findall('//plurals/pluralRules'): | |
| 139 rules = [] | |
| 140 for rule in elem.findall('pluralRule'): | |
| 141 rules.append((rule.attrib['count'], unicode(rule.text))) | |
| 142 pr = PluralRule(rules) | |
| 143 for locale in elem.attrib['locales'].split(): | |
| 144 plural_rules[locale] = pr | |
| 145 | |
| 134 filenames = os.listdir(os.path.join(srcdir, 'main')) | 146 filenames = os.listdir(os.path.join(srcdir, 'main')) |
| 135 filenames.remove('root.xml') | 147 filenames.remove('root.xml') |
| 136 filenames.sort(lambda a,b: len(a)-len(b)) | 148 filenames.sort(lambda a,b: len(a)-len(b)) |
| 137 filenames.insert(0, 'root.xml') | 149 filenames.insert(0, 'root.xml') |
| 138 | 150 |
| 158 else: | 170 else: |
| 159 territory = '001' # world | 171 territory = '001' # world |
| 160 print>>sys.stderr, ' Territory: %r' % territory | 172 print>>sys.stderr, ' Territory: %r' % territory |
| 161 regions = territory_containment.get(territory, []) | 173 regions = territory_containment.get(territory, []) |
| 162 print>>sys.stderr, ' Regions: %r' % regions | 174 print>>sys.stderr, ' Regions: %r' % regions |
| 175 | |
| 176 # plural rules | |
| 177 locale_id = '_'.join(filter(None, [ | |
| 178 language, | |
| 179 territory != '001' and territory or None | |
| 180 ])) | |
| 181 if locale_id in plural_rules: | |
| 182 data['plural_form'] = plural_rules[locale_id] | |
| 163 | 183 |
| 164 # <localeDisplayNames> | 184 # <localeDisplayNames> |
| 165 | 185 |
| 166 territories = data.setdefault('territories', {}) | 186 territories = data.setdefault('territories', {}) |
| 167 for elem in tree.findall('//territories/territory'): | 187 for elem in tree.findall('//territories/territory'): |
| 451 symbol = elem.find('symbol') | 471 symbol = elem.find('symbol') |
| 452 if symbol is not None and 'draft' not in symbol.attrib \ | 472 if symbol is not None and 'draft' not in symbol.attrib \ |
| 453 and 'choice' not in symbol.attrib: | 473 and 'choice' not in symbol.attrib: |
| 454 currency_symbols[code] = unicode(symbol.text) | 474 currency_symbols[code] = unicode(symbol.text) |
| 455 | 475 |
| 476 # <units> | |
| 477 | |
| 478 unit_patterns = data.setdefault('unit_patterns', {}) | |
| 479 for elem in tree.findall('//units/unit'): | |
| 480 unit_type = elem.attrib['type'] | |
| 481 unit_pattern = unit_patterns.setdefault(unit_type, {}) | |
| 482 for pattern in elem.findall('unitPattern'): | |
| 483 unit_patterns[unit_type][pattern.attrib['count']] = \ | |
| 484 unicode(pattern.text) | |
| 485 | |
| 456 outfile = open(os.path.join(destdir, 'localedata', stem + '.dat'), 'wb') | 486 outfile = open(os.path.join(destdir, 'localedata', stem + '.dat'), 'wb') |
| 457 try: | 487 try: |
| 458 pickle.dump(data, outfile, 2) | 488 pickle.dump(data, outfile, 2) |
| 459 finally: | 489 finally: |
| 460 outfile.close() | 490 outfile.close() |