babel/old/babel-test: scripts/import

annotate scripts/import_cldr.py @ 21:ddfac856c34f

Change pot header's first line, "Translations Template for %%(project)s." instead of "SOME DESCRIPTIVE TITLE.". '''`project`''' and '''`version`''' now default to '''PROJECT''' and '''VERSION''' respectively. Fixed a bug regarding '''Content-Transfer-Encoding''', it shouldn't be the charset, and we're defaulting to `8bit` untill someone complains.

author	palgarvio
date	Thu, 31 May 2007 18:04:41 +0000
parents	76985c08a339
children	7d37639a7411

rev	line source
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	1 #!/usr/bin/env python
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	2 # -- coding: utf-8 --
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	3 #
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	4 # Copyright (C) 2007 Edgewall Software
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	5 # All rights reserved.
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	6 #
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	7 # This software is licensed as described in the file COPYING, which
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	8 # you should have received as part of this distribution. The terms
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	9 # are also available at http://babel.edgewall.org/wiki/License.
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	10 #
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	11 # This software consists of voluntary contributions made by many
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	12 # individuals. For the exact contribution history, see the revision
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	13 # history and logs, available at http://babel.edgewall.org/log/.
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	14
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	15 import copy
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	16 from optparse import OptionParser
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	17 import os
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	18 import pickle
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	19 import sys
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	20 try:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	21 from xml.etree.ElementTree import parse
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	22 except ImportError:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	23 from elementtree.ElementTree import parse
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	24
9 3be73c6f01f1 Add basic support for number format patterns. jonas parents: 8 diff changeset	25 from babel import dates, numbers
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	26
15 76985c08a339 Minor date formatting improvements. cmlenz parents: 13 diff changeset	27 weekdays = {'mon': 0, 'tue': 1, 'wed': 2, 'thu': 3, 'fri': 4, 'sat': 5,
76985c08a339 Minor date formatting improvements. cmlenz parents: 13 diff changeset	28 'sun': 6}
8 9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	29
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	30 try:
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	31 any
9132c9218745 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:
9132c9218745 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):
9132c9218745 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))
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	35
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	36 def _parent(locale):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	37 parts = locale.split('_')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	38 if len(parts) == 1:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	39 return 'root'
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	40 else:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	41 return '_'.join(parts[:-1])
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	42
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	43 def _text(elem):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	44 buf = [elem.text or '']
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	45 for child in elem:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	46 buf.append(_text(child))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	47 buf.append(elem.tail or '')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	48 return u''.join(filter(None, buf)).strip()
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	49
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	50 def main():
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	51 parser = OptionParser(usage='%prog path/to/cldr')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	52 options, args = parser.parse_args()
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	53 if len(args) != 1:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	54 parser.error('incorrect number of arguments')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	55
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	56 srcdir = args[0]
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	57 destdir = os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])),
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	58 '..', 'babel', 'localedata')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	59
8 9132c9218745 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'))
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	61
9132c9218745 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
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	63 regions = {}
9132c9218745 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'):
9132c9218745 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()
9132c9218745 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
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	67
9132c9218745 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
9132c9218745 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 = {}
9132c9218745 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()
9132c9218745 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()
9132c9218745 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:
9132c9218745 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:
9132c9218745 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([]))
9132c9218745 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:
9132c9218745 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]
9132c9218745 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)
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	78
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	79 filenames = os.listdir(os.path.join(srcdir, 'main'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	80 filenames.remove('root.xml')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	81 filenames.sort(lambda a,b: len(a)-len(b))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	82 filenames.insert(0, 'root.xml')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	83
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	84 dicts = {}
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	85
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	86 for filename in filenames:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	87 print>>sys.stderr, 'Processing input file %r' % filename
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	88 stem, ext = os.path.splitext(filename)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	89 if ext != '.xml':
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	90 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	91
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	92 data = {}
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	93 if stem != 'root':
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	94 data.update(copy.deepcopy(dicts[_parent(stem)]))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	95 tree = parse(os.path.join(srcdir, 'main', filename))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	96
8 9132c9218745 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
9132c9218745 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')
9132c9218745 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:
9132c9218745 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']
9132c9218745 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
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	102
9132c9218745 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
9132c9218745 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')
9132c9218745 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:
9132c9218745 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']
13 b6c0de43fa40 Extended and documented `LazyProxy`. cmlenz parents: 9 diff changeset	107 else:
b6c0de43fa40 Extended and documented `LazyProxy`. cmlenz parents: 9 diff changeset	108 territory = '001' # world
8 9132c9218745 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, ' Territory: %r' % territory
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	110 regions = territory_containment.get(territory, [])
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	111 print>>sys.stderr, ' Regions: %r' % regions
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	112
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	113 # <localeDisplayNames>
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	114
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	115 territories = data.setdefault('territories', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	116 for elem in tree.findall('//territories/territory'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	117 if 'draft' in elem.attrib and elem.attrib['type'] in territories:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	118 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	119 territories[elem.attrib['type']] = _text(elem)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	120
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	121 languages = data.setdefault('languages', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	122 for elem in tree.findall('//languages/language'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	123 if 'draft' in elem.attrib and elem.attrib['type'] in languages:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	124 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	125 languages[elem.attrib['type']] = _text(elem)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	126
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	127 variants = data.setdefault('variants', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	128 for elem in tree.findall('//variants/variant'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	129 if 'draft' in elem.attrib and elem.attrib['type'] in variants:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	130 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	131 variants[elem.attrib['type']] = _text(elem)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	132
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	133 scripts = data.setdefault('scripts', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	134 for elem in tree.findall('//scripts/script'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	135 if 'draft' in elem.attrib and elem.attrib['type'] in scripts:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	136 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	137 scripts[elem.attrib['type']] = _text(elem)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	138
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	139 # <dates>
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	140
8 9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	141 week_data = data.setdefault('week_data', {})
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	142 supelem = sup.find('//weekData')
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	143
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	144 for elem in supelem.findall('minDays'):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	145 territories = elem.attrib['territories'].split()
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	146 if territory in territories or any([r in territories for r in regions]):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	147 week_data['min_days'] = int(elem.attrib['count'])
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	148
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	149 for elem in supelem.findall('firstDay'):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	150 territories = elem.attrib['territories'].split()
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	151 if territory in territories or any([r in territories for r in regions]):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	152 week_data['first_day'] = weekdays[elem.attrib['day']]
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	153
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	154 for elem in supelem.findall('weekendStart'):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	155 territories = elem.attrib['territories'].split()
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	156 if territory in territories or any([r in territories for r in regions]):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	157 week_data['weekend_start'] = weekdays[elem.attrib['day']]
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	158
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	159 for elem in supelem.findall('weekendEnd'):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	160 territories = elem.attrib['territories'].split()
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	161 if territory in territories or any([r in territories for r in regions]):
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	162 week_data['weekend_end'] = weekdays[elem.attrib['day']]
9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	163
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	164 time_zones = data.setdefault('time_zones', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	165 for elem in tree.findall('//timeZoneNames/zone'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	166 time_zones[elem.tag] = unicode(elem.findtext('displayName'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	167
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	168 for calendar in tree.findall('//calendars/calendar'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	169 if calendar.attrib['type'] != 'gregorian':
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	170 # TODO: support other calendar types
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	171 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	172
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	173 months = data.setdefault('months', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	174 for ctxt in calendar.findall('months/monthContext'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	175 ctxts = months.setdefault(ctxt.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	176 for width in ctxt.findall('monthWidth'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	177 widths = ctxts.setdefault(width.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	178 for elem in width.findall('month'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	179 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	180 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	181 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	182
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	183 days = data.setdefault('days', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	184 for ctxt in calendar.findall('days/dayContext'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	185 ctxts = days.setdefault(ctxt.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	186 for width in ctxt.findall('dayWidth'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	187 widths = ctxts.setdefault(width.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	188 for elem in width.findall('day'):
8 9132c9218745 Pull in some supplemental data from the CLDR, for things like the first day of the week. cmlenz parents: 1 diff changeset	189 dtype = weekdays[elem.attrib['type']]
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	190 if 'draft' in elem.attrib and dtype in widths:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	191 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	192 widths[dtype] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	193
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	194 quarters = data.setdefault('quarters', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	195 for ctxt in calendar.findall('quarters/quarterContext'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	196 ctxts = quarters.setdefault(ctxt.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	197 for width in ctxt.findall('quarterWidth'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	198 widths = ctxts.setdefault(width.attrib['type'], {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	199 for elem in width.findall('quarter'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	200 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	201 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	202 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	203
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	204 eras = data.setdefault('eras', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	205 for width in calendar.findall('eras/*'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	206 ewidth = {'eraNames': 'wide', 'eraAbbr': 'abbreviated'}[width.tag]
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	207 widths = eras.setdefault(ewidth, {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	208 for elem in width.findall('era'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	209 if 'draft' in elem.attrib and int(elem.attrib['type']) in widths:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	210 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	211 widths[int(elem.attrib.get('type'))] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	212
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	213 # AM/PM
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	214 periods = data.setdefault('periods', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	215 for elem in calendar.findall('am'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	216 if 'draft' in elem.attrib and elem.tag in periods:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	217 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	218 periods[elem.tag] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	219 for elem in calendar.findall('pm'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	220 if 'draft' in elem.attrib and elem.tag in periods:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	221 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	222 periods[elem.tag] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	223
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	224 date_formats = data.setdefault('date_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	225 for elem in calendar.findall('dateFormats/dateFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	226 if 'draft' in elem.attrib and elem.attrib.get('type') in date_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	227 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	228 try:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	229 date_formats[elem.attrib.get('type')] = \
9 3be73c6f01f1 Add basic support for number format patterns. jonas parents: 8 diff changeset	230 dates.parse_pattern(unicode(elem.findtext('dateFormat/pattern')))
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	231 except ValueError, e:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	232 print e
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	233
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	234 time_formats = data.setdefault('time_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	235 for elem in calendar.findall('timeFormats/timeFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	236 if 'draft' in elem.attrib and elem.attrib.get('type') in time_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	237 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	238 try:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	239 time_formats[elem.attrib.get('type')] = \
9 3be73c6f01f1 Add basic support for number format patterns. jonas parents: 8 diff changeset	240 dates.parse_pattern(unicode(elem.findtext('timeFormat/pattern')))
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	241 except ValueError, e:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	242 print e
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	243
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	244 # <numbers>
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	245
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	246 number_symbols = data.setdefault('number_symbols', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	247 for elem in tree.findall('//numbers/symbols/*'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	248 number_symbols[elem.tag] = unicode(elem.text)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	249
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	250 decimal_formats = data.setdefault('decimal_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	251 for elem in tree.findall('//decimalFormats/decimalFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	252 if 'draft' in elem.attrib and elem.attrib.get('type') in decimal_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	253 continue
9 3be73c6f01f1 Add basic support for number format patterns. jonas parents: 8 diff changeset	254 decimal_formats[elem.attrib.get('type')] = numbers.parse_pattern(unicode(elem.findtext('decimalFormat/pattern')))
1 f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	255
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	256 scientific_formats = data.setdefault('scientific_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	257 for elem in tree.findall('//scientificFormats/scientificFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	258 if 'draft' in elem.attrib and elem.attrib.get('type') in scientific_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	259 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	260 scientific_formats[elem.attrib.get('type')] = unicode(elem.findtext('scientificFormat/pattern'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	261
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	262 currency_formats = data.setdefault('currency_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	263 for elem in tree.findall('//currencyFormats/currencyFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	264 if 'draft' in elem.attrib and elem.attrib.get('type') in currency_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	265 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	266 currency_formats[elem.attrib.get('type')] = unicode(elem.findtext('currencyFormat/pattern'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	267
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	268 percent_formats = data.setdefault('percent_formats', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	269 for elem in tree.findall('//percentFormats/percentFormatLength'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	270 if 'draft' in elem.attrib and elem.attrib.get('type') in percent_formats:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	271 continue
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	272 percent_formats[elem.attrib.get('type')] = unicode(elem.findtext('percentFormat/pattern'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	273
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	274 currencies = data.setdefault('currencies', {})
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	275 for elem in tree.findall('//currencies/currency'):
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	276 currencies[elem.attrib['type']] = {
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	277 'display_name': unicode(elem.findtext('displayName')),
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	278 'symbol': unicode(elem.findtext('symbol'))
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	279 }
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	280
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	281 dicts[stem] = data
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	282 outfile = open(os.path.join(destdir, stem + '.dat'), 'wb')
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	283 try:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	284 pickle.dump(data, outfile, 2)
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	285 finally:
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	286 outfile.close()
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	287
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	288 if __name__ == '__main__':
f71ca60f2a4a Import of initial code base. cmlenz parents: diff changeset	289 main()

Mercurial > babel > old > babel-test

annotate scripts/import_cldr.py @ 21:ddfac856c34f