babel/mirror: scripts/import

annotate scripts/import_cldr.py @ 9:9ed6cf5975a1 trunk

Add basic support for number format patterns.

author	jonas
date	Wed, 30 May 2007 21:30:14 +0000
parents	29f6f9a90f14
children	368650dc3423

rev	line source
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	1 #!/usr/bin/env python
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	2 # -- coding: utf-8 --
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	3 #
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	4 # Copyright (C) 2007 Edgewall Software
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	5 # All rights reserved.
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	6 #
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	7 # This software is licensed as described in the file COPYING, which
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	8 # you should have received as part of this distribution. The terms
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	9 # are also available at http://babel.edgewall.org/wiki/License.
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	10 #
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	11 # This software consists of voluntary contributions made by many
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	12 # individuals. For the exact contribution history, see the revision
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	13 # history and logs, available at http://babel.edgewall.org/log/.
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	14
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	15 import copy
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	16 from optparse import OptionParser
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	17 import os
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	18 import pickle
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	19 import sys
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	20 try:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	21 from xml.etree.ElementTree import parse
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	22 except ImportError:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	23 from elementtree.ElementTree import parse
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	24
9 9ed6cf5975a1 Add basic support for number format patterns. jonas parents: 8 diff changeset	25 from babel import dates, numbers
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	26
8 29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	27 weekdays = {'mon': 1, 'tue': 2, 'wed': 3, 'thu': 4, 'fri': 5, 'sat': 6,
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	28 'sun': 7}
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	29
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	30 try:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	31 any
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	32 except NameError:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	33 def any(iterable):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	34 return filter(None, list(iterable))
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	35
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	36 def _parent(locale):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	37 parts = locale.split('_')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	38 if len(parts) == 1:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	39 return 'root'
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	40 else:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	41 return '_'.join(parts[:-1])
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	42
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	43 def _text(elem):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	44 buf = [elem.text or '']
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	45 for child in elem:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	46 buf.append(_text(child))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	47 buf.append(elem.tail or '')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	48 return u''.join(filter(None, buf)).strip()
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	49
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	50 def main():
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	51 parser = OptionParser(usage='%prog path/to/cldr')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	52 options, args = parser.parse_args()
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	53 if len(args) != 1:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	54 parser.error('incorrect number of arguments')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	55
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	56 srcdir = args[0]
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	57 destdir = os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])),
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	58 '..', 'babel', 'localedata')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	59
8 29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	60 sup = parse(os.path.join(srcdir, 'supplemental', 'supplementalData.xml'))
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	61
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	62 # build a territory containment mapping for inheritance
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	63 regions = {}
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	64 for elem in sup.findall('//territoryContainment/group'):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	65 regions[elem.attrib['type']] = elem.attrib['contains'].split()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	66 from pprint import pprint
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	67
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	68 # Resolve territory containment
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	69 territory_containment = {}
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	70 region_items = regions.items()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	71 region_items.sort()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	72 for group, territory_list in region_items:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	73 for territory in territory_list:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	74 containers = territory_containment.setdefault(territory, set([]))
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	75 if group in territory_containment:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	76 containers \|= territory_containment[group]
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	77 containers.add(group)
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	78
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	79 filenames = os.listdir(os.path.join(srcdir, 'main'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	80 filenames.remove('root.xml')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	81 filenames.sort(lambda a,b: len(a)-len(b))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	82 filenames.insert(0, 'root.xml')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	83
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	84 dicts = {}
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	85
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	86 for filename in filenames:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	87 print>>sys.stderr, 'Processing input file %r' % filename
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	88 stem, ext = os.path.splitext(filename)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	89 if ext != '.xml':
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	90 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	91
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	92 data = {}
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	93 if stem != 'root':
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	94 data.update(copy.deepcopy(dicts[_parent(stem)]))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	95 tree = parse(os.path.join(srcdir, 'main', filename))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	96
8 29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	97 language = None
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	98 elem = tree.find('//identity/language')
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	99 if elem is not None:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	100 language = elem.attrib['type']
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	101 print>>sys.stderr, ' Language: %r' % language
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	102
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	103 territory = None
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	104 elem = tree.find('//identity/territory')
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	105 if elem is not None:
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	106 territory = elem.attrib['type']
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	107 print>>sys.stderr, ' Territory: %r' % territory
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	108 regions = territory_containment.get(territory, [])
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	109 print>>sys.stderr, ' Regions: %r' % regions
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	110
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	111 # <localeDisplayNames>
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	112
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	113 territories = data.setdefault('territories', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	114 for elem in tree.findall('//territories/territory'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	115 if 'draft' in elem.attrib and elem.attrib['type'] in territories:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	116 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	117 territories[elem.attrib['type']] = _text(elem)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	118
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	119 languages = data.setdefault('languages', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	120 for elem in tree.findall('//languages/language'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	121 if 'draft' in elem.attrib and elem.attrib['type'] in languages:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	122 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	123 languages[elem.attrib['type']] = _text(elem)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	124
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	125 variants = data.setdefault('variants', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	126 for elem in tree.findall('//variants/variant'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	127 if 'draft' in elem.attrib and elem.attrib['type'] in variants:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	128 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	129 variants[elem.attrib['type']] = _text(elem)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	130
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	131 scripts = data.setdefault('scripts', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	132 for elem in tree.findall('//scripts/script'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	133 if 'draft' in elem.attrib and elem.attrib['type'] in scripts:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	134 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	135 scripts[elem.attrib['type']] = _text(elem)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	136
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	137 # <dates>
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	138
8 29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	139 week_data = data.setdefault('week_data', {})
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	140 supelem = sup.find('//weekData')
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	141
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	142 for elem in supelem.findall('minDays'):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	143 territories = elem.attrib['territories'].split()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	144 if territory in territories or any([r in territories for r in regions]):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	145 week_data['min_days'] = int(elem.attrib['count'])
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	146
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	147 for elem in supelem.findall('firstDay'):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	148 territories = elem.attrib['territories'].split()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	149 if territory in territories or any([r in territories for r in regions]):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	150 week_data['first_day'] = weekdays[elem.attrib['day']]
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	151
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	152 for elem in supelem.findall('weekendStart'):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	153 territories = elem.attrib['territories'].split()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	154 if territory in territories or any([r in territories for r in regions]):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	155 week_data['weekend_start'] = weekdays[elem.attrib['day']]
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	156
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	157 for elem in supelem.findall('weekendEnd'):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	158 territories = elem.attrib['territories'].split()
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	159 if territory in territories or any([r in territories for r in regions]):
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	160 week_data['weekend_end'] = weekdays[elem.attrib['day']]
29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	161
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	162 time_zones = data.setdefault('time_zones', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	163 for elem in tree.findall('//timeZoneNames/zone'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	164 time_zones[elem.tag] = unicode(elem.findtext('displayName'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	165
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	166 for calendar in tree.findall('//calendars/calendar'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	167 if calendar.attrib['type'] != 'gregorian':
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	168 # TODO: support other calendar types
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	169 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	170
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	171 months = data.setdefault('months', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	172 for ctxt in calendar.findall('months/monthContext'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	173 ctxts = months.setdefault(ctxt.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	174 for width in ctxt.findall('monthWidth'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	175 widths = ctxts.setdefault(width.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	176 for elem in width.findall('month'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	177 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	178 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	179 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	180
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	181 days = data.setdefault('days', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	182 for ctxt in calendar.findall('days/dayContext'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	183 ctxts = days.setdefault(ctxt.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	184 for width in ctxt.findall('dayWidth'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	185 widths = ctxts.setdefault(width.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	186 for elem in width.findall('day'):
8 29f6f9a90f14 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	187 dtype = weekdays[elem.attrib['type']]
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	188 if 'draft' in elem.attrib and dtype in widths:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	189 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	190 widths[dtype] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	191
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	192 quarters = data.setdefault('quarters', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	193 for ctxt in calendar.findall('quarters/quarterContext'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	194 ctxts = quarters.setdefault(ctxt.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	195 for width in ctxt.findall('quarterWidth'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	196 widths = ctxts.setdefault(width.attrib['type'], {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	197 for elem in width.findall('quarter'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	198 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	199 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	200 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	201
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	202 eras = data.setdefault('eras', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	203 for width in calendar.findall('eras/*'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	204 ewidth = {'eraNames': 'wide', 'eraAbbr': 'abbreviated'}[width.tag]
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	205 widths = eras.setdefault(ewidth, {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	206 for elem in width.findall('era'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	207 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	208 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	209 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	210
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	211 # AM/PM
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	212 periods = data.setdefault('periods', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	213 for elem in calendar.findall('am'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	214 if 'draft' in elem.attrib and elem.tag in periods:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	215 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	216 periods[elem.tag] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	217 for elem in calendar.findall('pm'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	218 if 'draft' in elem.attrib and elem.tag in periods:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	219 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	220 periods[elem.tag] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	221
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	222 date_formats = data.setdefault('date_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	223 for elem in calendar.findall('dateFormats/dateFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	224 if 'draft' in elem.attrib and elem.attrib.get('type') in date_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	225 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	226 try:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	227 date_formats[elem.attrib.get('type')] = \
9 9ed6cf5975a1 Add basic support for number format patterns. jonas parents: 8 diff changeset	228 dates.parse_pattern(unicode(elem.findtext('dateFormat/pattern')))
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	229 except ValueError, e:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	230 print e
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	231
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	232 time_formats = data.setdefault('time_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	233 for elem in calendar.findall('timeFormats/timeFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	234 if 'draft' in elem.attrib and elem.attrib.get('type') in time_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	235 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	236 try:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	237 time_formats[elem.attrib.get('type')] = \
9 9ed6cf5975a1 Add basic support for number format patterns. jonas parents: 8 diff changeset	238 dates.parse_pattern(unicode(elem.findtext('timeFormat/pattern')))
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	239 except ValueError, e:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	240 print e
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	241
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	242 # <numbers>
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	243
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	244 number_symbols = data.setdefault('number_symbols', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	245 for elem in tree.findall('//numbers/symbols/*'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	246 number_symbols[elem.tag] = unicode(elem.text)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	247
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	248 decimal_formats = data.setdefault('decimal_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	249 for elem in tree.findall('//decimalFormats/decimalFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	250 if 'draft' in elem.attrib and elem.attrib.get('type') in decimal_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	251 continue
9 9ed6cf5975a1 Add basic support for number format patterns. jonas parents: 8 diff changeset	252 decimal_formats[elem.attrib.get('type')] = numbers.parse_pattern(unicode(elem.findtext('decimalFormat/pattern')))
1 7870274479f5 Import of initial code base. cmlenz parents: diff changeset	253
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	254 scientific_formats = data.setdefault('scientific_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	255 for elem in tree.findall('//scientificFormats/scientificFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	256 if 'draft' in elem.attrib and elem.attrib.get('type') in scientific_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	257 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	258 scientific_formats[elem.attrib.get('type')] = unicode(elem.findtext('scientificFormat/pattern'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	259
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	260 currency_formats = data.setdefault('currency_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	261 for elem in tree.findall('//currencyFormats/currencyFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	262 if 'draft' in elem.attrib and elem.attrib.get('type') in currency_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	263 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	264 currency_formats[elem.attrib.get('type')] = unicode(elem.findtext('currencyFormat/pattern'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	265
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	266 percent_formats = data.setdefault('percent_formats', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	267 for elem in tree.findall('//percentFormats/percentFormatLength'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	268 if 'draft' in elem.attrib and elem.attrib.get('type') in percent_formats:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	269 continue
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	270 percent_formats[elem.attrib.get('type')] = unicode(elem.findtext('percentFormat/pattern'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	271
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	272 currencies = data.setdefault('currencies', {})
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	273 for elem in tree.findall('//currencies/currency'):
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	274 currencies[elem.attrib['type']] = {
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	275 'display_name': unicode(elem.findtext('displayName')),
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	276 'symbol': unicode(elem.findtext('symbol'))
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	277 }
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	278
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	279 dicts[stem] = data
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	280 outfile = open(os.path.join(destdir, stem + '.dat'), 'wb')
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	281 try:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	282 pickle.dump(data, outfile, 2)
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	283 finally:
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	284 outfile.close()
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	285
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	286 if __name__ == '__main__':
7870274479f5 Import of initial code base. cmlenz parents: diff changeset	287 main()

Mercurial > babel > mirror

annotate scripts/import_cldr.py @ 9:9ed6cf5975a1 trunk