def textPatternData(self):
for key in ('quotationStart', 'alternateQuotationEnd',
'quotationEnd', 'alternateQuotationStart'):
yield key, self.find('delimiters/' + key)
for key in ('start', 'middle', 'end'):
yield ('listPatternPart' + key.capitalize(),
self.__fromLdmlListPattern(self.find(
'listPatterns/listPattern/listPatternPart[{}]'.format(key))))
yield ('listPatternPartTwo',
self.__fromLdmlListPattern(self.find(
'listPatterns/listPattern/listPatternPart[2]')))
stem = 'dates/calendars/calendar[gregorian]/'
# TODO: is wide really the right width to use here ?
# abbreviated might be an option ... or try both ?
meridiem = stem + 'dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/'
for key in ('am', 'pm'):
yield key, self.find(meridiem + 'dayPeriod[{}]'.format(key),
draft = Node.draftScore('contributed'))
for pair in (('long', 'full'), ('short', 'short')):
for key in ('time', 'date'):
yield (pair[0] + key.capitalize() + 'Format',
convert_date(self.find(
stem + '{}Formats/{}FormatLength[{}]/{}Format/pattern'.format(
key, key, pair[1], key))))
def _generateLocaleInfo(path, language_code, script_code, country_code, variant_code=""):
if not path.endswith(".xml"):
return {}
if language_code == 'root':
# just skip it
return {}
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
raise xpathlite.Error('we do not support variants ("%s")' % variant_code)
language_id = enumdata.languageCodeToId(language_code)
if language_id <= 0:
raise xpathlite.Error('unknown language code "%s"' % language_code)
script_id = enumdata.scriptCodeToId(script_code)
if script_id == -1:
raise xpathlite.Error('unknown script code "%s"' % script_code)
# we should handle fully qualified names with the territory
if not country_code:
return {}
country_id = enumdata.countryCodeToId(country_code)
if country_id <= 0:
raise xpathlite.Error('unknown country code "%s"' % country_code)
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = dict(
language=enumdata.language_list[language_id][0],
language_code=language_code, language_id=language_id,
script=enumdata.script_list[script_id][0],
script_code=script_code, script_id=script_id,
country=enumdata.country_list[country_id][0],
country_code=country_code, country_id=country_id,
variant_code=variant_code)
(dir_name, file_name) = os.path.split(path)
def from_supplement(tag,
path=os.path.join(dir_name, '..', 'supplemental',
'supplementalData.xml')):
return findTagsInFile(path, tag)
currencies = from_supplement('currencyData/region[iso3166=%s]' % country_code)
result['currencyIsoCode'] = ''
result['currencyDigits'] = 2
result['currencyRounding'] = 1
if currencies:
for e in currencies:
if e[0] == 'currency':
t = [x[1] == 'false' for x in e[1] if x[0] == 'tender']
if t and t[0]:
pass
elif not any(x[0] == 'to' for x in e[1]):
result['currencyIsoCode'] = (x[1] for x in e[1] if x[0] == 'iso4217').next()
break
if result['currencyIsoCode']:
t = from_supplement("currencyData/fractions/info[iso4217=%s]"
% result['currencyIsoCode'])
if t and t[0][0] == 'info':
result['currencyDigits'] = (int(x[1]) for x in t[0][1] if x[0] == 'digits').next()
result['currencyRounding'] = (int(x[1]) for x in t[0][1] if x[0] == 'rounding').next()
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def findEntryDef(path, xpath, value=''):
try:
return findEntry(path, xpath)
except xpathlite.Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
# in CLDR 1.9 number system was refactored for numbers (but not for currency)
# so if previous findEntry doesn't work we should try this:
try:
return findEntry(path, xpath.replace("/symbols/", "/symbols[numberSystem=" + numbering_system + "]/"))
except xpathlite.Error:
# fallback to default
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
result['list'] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result['percent'] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
try:
numbering_systems = {}
for ns in findTagsInFile(os.path.join(cldr_dir, '..', 'supplemental',
'numberingSystems.xml'),
'numberingSystems'):
tmp = {}
id = ""
for data in ns[1:][0]: # ns looks like this: [u'numberingSystem', [(u'digits', u'0123456789'), (u'type', u'numeric'), (u'id', u'latn')]]
tmp[data[0]] = data[1]
if data[0] == u"id":
id = data[1]
numbering_systems[id] = tmp
result['zero'] = numbering_systems[numbering_system][u"digits"][0]
except e:
sys.stderr.write("Native zero detection problem:\n" + str(e) + "\n")
result['zero'] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result['quotationStart'] = findEntry(path, "delimiters/quotationStart")
result['quotationEnd'] = findEntry(path, "delimiters/quotationEnd")
result['alternateQuotationStart'] = findEntry(path, "delimiters/alternateQuotationStart")
result['alternateQuotationEnd'] = findEntry(path, "delimiters/alternateQuotationEnd")
result['listPatternPartStart'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[start]"))
result['listPatternPartMiddle'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[middle]"))
result['listPatternPartEnd'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[end]"))
result['listPatternPartTwo'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[2]"))
result['am'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]", draft)
result['pm'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]", draft)
result['longDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"))
result['shortDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"))
result['longTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"))
result['shortTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"))
endonym = None
if country_code and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s_%s]" % (language_code, script_code, country_code))
if not endonym and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, country_code))
if not endonym:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s]" % (language_code))
result['language_endonym'] = endonym
result['country_endonym'] = findEntryDef(path, "localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(path, "numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern", numbering_system)
currency_format = parse_number_format(currency_format, result)
result['currencyFormat'] = currency_format[0]
result['currencyNegativeFormat'] = ''
if len(currency_format) > 1:
result['currencyNegativeFormat'] = currency_format[1]
result['currencySymbol'] = ''
result['currencyDisplayName'] = ''
if result['currencyIsoCode']:
result['currencySymbol'] = findEntryDef(path, "numbers/currencies/currency[%s]/symbol" % result['currencyIsoCode'])
result['currencyDisplayName'] = ';'.join(
findEntryDef(path, 'numbers/currencies/currency[' + result['currencyIsoCode']
+ ']/displayName' + tail)
for tail in ['',] + [
'[count=%s]' % x for x in ('zero', 'one', 'two', 'few', 'many', 'other')
]) + ';'
def findUnitDef(path, stem, fallback=''):
# The displayName for a quantified unit in en.xml is kByte
# instead of kB (etc.), so prefer any unitPattern provided:
for count in ('many', 'few', 'two', 'other', 'zero', 'one'):
try:
ans = findEntry(path, stem + 'unitPattern[count=%s]' % count)
except xpathlite.Error:
continue
# TODO: epxloit count-handling, instead of discarding placeholders
if ans.startswith('{0}'):
ans = ans[3:].lstrip()
if ans:
return ans
return findEntryDef(path, stem + 'displayName', fallback)
# First without quantifier, then quantified each way:
result['byte_unit'] = findEntryDef(
path, 'units/unitLength[type=long]/unit[type=digital-byte]/displayName',
'bytes')
stem = 'units/unitLength[type=short]/unit[type=digital-%sbyte]/'
known = [] # cases where we *do* have a given version:
result['byte_si_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem, 'B', known))
# IEC 60027-2
# http://physics.nist.gov/cuu/Units/binary.html
result['byte_iec_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem % '%sbi', 'iB', known))
# Used for month and day data:
namings = (
('standaloneLong', 'stand-alone', 'wide'),
('standaloneShort', 'stand-alone', 'abbreviated'),
('standaloneNarrow', 'stand-alone', 'narrow'),
('long', 'format', 'wide'),
('short', 'format', 'abbreviated'),
('narrow', 'format', 'narrow'),
)
# Month data:
for cal in ('gregorian',): # We shall want to add to this
stem = 'dates/calendars/calendar[' + cal + ']/months/'
for (key, mode, size) in namings:
prop = 'monthContext[' + mode + ']/monthWidth[' + size + ']/'
result[key + 'Months'] = ';'.join(
findEntry(path, stem + prop + "month[%d]" % i)
for i in range(1, 13)) + ';'
# Day data (for Gregorian, at least):
stem = 'dates/calendars/calendar[gregorian]/days/'
days = ('sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat')
for (key, mode, size) in namings:
prop = 'dayContext[' + mode + ']/dayWidth[' + size + ']/day'
result[key + 'Days'] = ';'.join(
findEntry(path, stem + prop + '[' + day + ']')
for day in days) + ';'
return Locale(result)
def generateLocaleInfo(path):
(dir_name, file_name) = os.path.split(path)
if not path.endswith(".xml"):
return {}
# skip legacy/compatibility ones
alias = findAlias(path)
if alias:
raise xpathlite.Error('alias to "%s"' % alias)
language_code = findEntryInFile(path, "identity/language", attribute="type")[0]
if language_code == "root":
# just skip it
return {}
country_code = findEntryInFile(path, "identity/territory", attribute="type")[0]
script_code = findEntryInFile(path, "identity/script", attribute="type")[0]
variant_code = findEntryInFile(path, "identity/variant", attribute="type")[0]
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
raise xpathlite.Error('we do not support variants ("%s")' % variant_code)
language_id = enumdata.languageCodeToId(language_code)
if language_id <= 0:
raise xpathlite.Error('unknown language code "%s"' % language_code)
language = enumdata.language_list[language_id][0]
script_id = enumdata.scriptCodeToId(script_code)
if script_id == -1:
raise xpathlite.Error('unknown script code "%s"' % script_code)
script = enumdata.script_list[script_id][0]
# we should handle fully qualified names with the territory
if not country_code:
return {}
country_id = enumdata.countryCodeToId(country_code)
if country_id <= 0:
raise xpathlite.Error('unknown country code "%s"' % country_code)
country = enumdata.country_list[country_id][0]
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = {}
result["language"] = language
result["script"] = script
result["country"] = country
result["language_code"] = language_code
result["country_code"] = country_code
result["script_code"] = script_code
result["variant_code"] = variant_code
result["language_id"] = language_id
result["script_id"] = script_id
result["country_id"] = country_id
supplementalPath = dir_name + "/../supplemental/supplementalData.xml"
currencies = findTagsInFile(supplementalPath, "currencyData/region[iso3166=%s]" % country_code)
result["currencyIsoCode"] = ""
result["currencyDigits"] = 2
result["currencyRounding"] = 1
if currencies:
for e in currencies:
if e[0] == "currency":
tender = True
t = filter(lambda x: x[0] == "tender", e[1])
if t and t[0][1] == "false":
tender = False
if tender and not filter(lambda x: x[0] == "to", e[1]):
result["currencyIsoCode"] = filter(lambda x: x[0] == "iso4217", e[1])[0][1]
break
if result["currencyIsoCode"]:
t = findTagsInFile(supplementalPath, "currencyData/fractions/info[iso4217=%s]" % result["currencyIsoCode"])
if t and t[0][0] == "info":
result["currencyDigits"] = int(filter(lambda x: x[0] == "digits", t[0][1])[0][1])
result["currencyRounding"] = int(filter(lambda x: x[0] == "rounding", t[0][1])[0][1])
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def findEntryDef(path, xpath, value=""):
try:
return findEntry(path, xpath)
except xpathlite.Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
# in CLDR 1.9 number system was refactored for numbers (but not for currency)
# so if previous findEntry doesn't work we should try this:
try:
return findEntry(
path, xpath.replace("/symbols/", "/symbols[numberSystem=" + numbering_system + "]/")
)
except xpathlite.Error:
# fallback to default
pass
return findEntry(path, xpath)
result["decimal"] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result["group"] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
result["list"] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result["percent"] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
try:
numbering_systems = {}
for ns in findTagsInFile(cldr_dir + "/../supplemental/numberingSystems.xml", "numberingSystems"):
tmp = {}
id = ""
for data in ns[1:][
0
]: # ns looks like this: [u'numberingSystem', [(u'digits', u'0123456789'), (u'type', u'numeric'), (u'id', u'latn')]]
tmp[data[0]] = data[1]
if data[0] == u"id":
id = data[1]
numbering_systems[id] = tmp
result["zero"] = numbering_systems[numbering_system][u"digits"][0]
except e:
sys.stderr.write("Native zero detection problem:\n" + str(e) + "\n")
result["zero"] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result["minus"] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result["plus"] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result["exp"] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result["quotationStart"] = findEntry(path, "delimiters/quotationStart")
result["quotationEnd"] = findEntry(path, "delimiters/quotationEnd")
result["alternateQuotationStart"] = findEntry(path, "delimiters/alternateQuotationStart")
result["alternateQuotationEnd"] = findEntry(path, "delimiters/alternateQuotationEnd")
result["listPatternPartStart"] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[start]")
)
result["listPatternPartMiddle"] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[middle]")
)
result["listPatternPartEnd"] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[end]")
)
result["listPatternPartTwo"] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[2]")
)
result["am"] = findEntry(
path,
"dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]",
draft,
)
result["pm"] = findEntry(
path,
"dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]",
draft,
)
result["longDateFormat"] = convert_date(
findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern")
)
result["shortDateFormat"] = convert_date(
findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern")
)
result["longTimeFormat"] = convert_date(
findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern")
)
result["shortTimeFormat"] = convert_date(
findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern")
)
endonym = None
if country_code and script_code:
endonym = findEntryDef(
path, "localeDisplayNames/languages/language[type=%s_%s_%s]" % (language_code, script_code, country_code)
)
if not endonym and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(
path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, country_code)
)
if not endonym:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s]" % (language_code))
result["language_endonym"] = endonym
result["country_endonym"] = findEntryDef(path, "localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(
path, "numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern", numbering_system
)
currency_format = parse_number_format(currency_format, result)
result["currencyFormat"] = currency_format[0]
result["currencyNegativeFormat"] = ""
if len(currency_format) > 1:
result["currencyNegativeFormat"] = currency_format[1]
result["currencySymbol"] = ""
result["currencyDisplayName"] = ""
if result["currencyIsoCode"]:
result["currencySymbol"] = findEntryDef(
path, "numbers/currencies/currency[%s]/symbol" % result["currencyIsoCode"]
)
display_name_path = "numbers/currencies/currency[%s]/displayName" % result["currencyIsoCode"]
result["currencyDisplayName"] = (
findEntryDef(path, display_name_path)
+ ";"
+ findEntryDef(path, display_name_path + "[count=zero]")
+ ";"
+ findEntryDef(path, display_name_path + "[count=one]")
+ ";"
+ findEntryDef(path, display_name_path + "[count=two]")
+ ";"
+ findEntryDef(path, display_name_path + "[count=few]")
+ ";"
+ findEntryDef(path, display_name_path + "[count=many]")
+ ";"
+ findEntryDef(path, display_name_path + "[count=other]")
+ ";"
)
standalone_long_month_path = (
"dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[wide]/month"
)
result["standaloneLongMonths"] = (
findEntry(path, standalone_long_month_path + "[1]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[2]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[3]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[4]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[5]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[6]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[7]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[8]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[9]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[10]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[11]")
+ ";"
+ findEntry(path, standalone_long_month_path + "[12]")
+ ";"
)
standalone_short_month_path = (
"dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[abbreviated]/month"
)
result["standaloneShortMonths"] = (
findEntry(path, standalone_short_month_path + "[1]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[2]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[3]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[4]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[5]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[6]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[7]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[8]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[9]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[10]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[11]")
+ ";"
+ findEntry(path, standalone_short_month_path + "[12]")
+ ";"
)
standalone_narrow_month_path = (
"dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[narrow]/month"
)
result["standaloneNarrowMonths"] = (
findEntry(path, standalone_narrow_month_path + "[1]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[2]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[3]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[4]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[5]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[6]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[7]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[8]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[9]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[10]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[11]")
+ ";"
+ findEntry(path, standalone_narrow_month_path + "[12]")
+ ";"
)
long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[wide]/month"
result["longMonths"] = (
findEntry(path, long_month_path + "[1]")
+ ";"
+ findEntry(path, long_month_path + "[2]")
+ ";"
+ findEntry(path, long_month_path + "[3]")
+ ";"
+ findEntry(path, long_month_path + "[4]")
+ ";"
+ findEntry(path, long_month_path + "[5]")
+ ";"
+ findEntry(path, long_month_path + "[6]")
+ ";"
+ findEntry(path, long_month_path + "[7]")
+ ";"
+ findEntry(path, long_month_path + "[8]")
+ ";"
+ findEntry(path, long_month_path + "[9]")
+ ";"
+ findEntry(path, long_month_path + "[10]")
+ ";"
+ findEntry(path, long_month_path + "[11]")
+ ";"
+ findEntry(path, long_month_path + "[12]")
+ ";"
)
short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[abbreviated]/month"
result["shortMonths"] = (
findEntry(path, short_month_path + "[1]")
+ ";"
+ findEntry(path, short_month_path + "[2]")
+ ";"
+ findEntry(path, short_month_path + "[3]")
+ ";"
+ findEntry(path, short_month_path + "[4]")
+ ";"
+ findEntry(path, short_month_path + "[5]")
+ ";"
+ findEntry(path, short_month_path + "[6]")
+ ";"
+ findEntry(path, short_month_path + "[7]")
+ ";"
+ findEntry(path, short_month_path + "[8]")
+ ";"
+ findEntry(path, short_month_path + "[9]")
+ ";"
+ findEntry(path, short_month_path + "[10]")
+ ";"
+ findEntry(path, short_month_path + "[11]")
+ ";"
+ findEntry(path, short_month_path + "[12]")
+ ";"
)
narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[narrow]/month"
result["narrowMonths"] = (
findEntry(path, narrow_month_path + "[1]")
+ ";"
+ findEntry(path, narrow_month_path + "[2]")
+ ";"
+ findEntry(path, narrow_month_path + "[3]")
+ ";"
+ findEntry(path, narrow_month_path + "[4]")
+ ";"
+ findEntry(path, narrow_month_path + "[5]")
+ ";"
+ findEntry(path, narrow_month_path + "[6]")
+ ";"
+ findEntry(path, narrow_month_path + "[7]")
+ ";"
+ findEntry(path, narrow_month_path + "[8]")
+ ";"
+ findEntry(path, narrow_month_path + "[9]")
+ ";"
+ findEntry(path, narrow_month_path + "[10]")
+ ";"
+ findEntry(path, narrow_month_path + "[11]")
+ ";"
+ findEntry(path, narrow_month_path + "[12]")
+ ";"
)
long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[wide]/day"
result["longDays"] = (
findEntry(path, long_day_path + "[sun]")
+ ";"
+ findEntry(path, long_day_path + "[mon]")
+ ";"
+ findEntry(path, long_day_path + "[tue]")
+ ";"
+ findEntry(path, long_day_path + "[wed]")
+ ";"
+ findEntry(path, long_day_path + "[thu]")
+ ";"
+ findEntry(path, long_day_path + "[fri]")
+ ";"
+ findEntry(path, long_day_path + "[sat]")
+ ";"
)
short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[abbreviated]/day"
result["shortDays"] = (
findEntry(path, short_day_path + "[sun]")
+ ";"
+ findEntry(path, short_day_path + "[mon]")
+ ";"
+ findEntry(path, short_day_path + "[tue]")
+ ";"
+ findEntry(path, short_day_path + "[wed]")
+ ";"
+ findEntry(path, short_day_path + "[thu]")
+ ";"
+ findEntry(path, short_day_path + "[fri]")
+ ";"
+ findEntry(path, short_day_path + "[sat]")
+ ";"
)
narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[narrow]/day"
result["narrowDays"] = (
findEntry(path, narrow_day_path + "[sun]")
+ ";"
+ findEntry(path, narrow_day_path + "[mon]")
+ ";"
+ findEntry(path, narrow_day_path + "[tue]")
+ ";"
+ findEntry(path, narrow_day_path + "[wed]")
+ ";"
+ findEntry(path, narrow_day_path + "[thu]")
+ ";"
+ findEntry(path, narrow_day_path + "[fri]")
+ ";"
+ findEntry(path, narrow_day_path + "[sat]")
+ ";"
)
standalone_long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[wide]/day"
result["standaloneLongDays"] = (
findEntry(path, standalone_long_day_path + "[sun]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[mon]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[tue]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[wed]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[thu]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[fri]")
+ ";"
+ findEntry(path, standalone_long_day_path + "[sat]")
+ ";"
)
standalone_short_day_path = (
"dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[abbreviated]/day"
)
result["standaloneShortDays"] = (
findEntry(path, standalone_short_day_path + "[sun]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[mon]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[tue]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[wed]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[thu]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[fri]")
+ ";"
+ findEntry(path, standalone_short_day_path + "[sat]")
+ ";"
)
standalone_narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[narrow]/day"
result["standaloneNarrowDays"] = (
findEntry(path, standalone_narrow_day_path + "[sun]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[mon]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[tue]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[wed]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[thu]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[fri]")
+ ";"
+ findEntry(path, standalone_narrow_day_path + "[sat]")
+ ";"
)
return result
def generateLocaleInfo(path):
(dir_name, file_name) = os.path.split(path)
if not path.endswith(".xml"):
return {}
language_code = findEntryInFile(path, "identity/language", attribute="type")[0]
if language_code == 'root':
# just skip it
return {}
country_code = findEntryInFile(path, "identity/territory", attribute="type")[0]
script_code = findEntryInFile(path, "identity/script", attribute="type")[0]
variant_code = findEntryInFile(path, "identity/variant", attribute="type")[0]
# we should handle fully qualified names with the territory
if not country_code:
return {}
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
return {}
language_id = enumdata.languageCodeToId(language_code)
if language_id == -1:
sys.stderr.write("unnknown language code \"" + language_code + "\"\n")
return {}
language = enumdata.language_list[language_id][0]
script_id = enumdata.scriptCodeToId(script_code)
if script_code == -1:
sys.stderr.write("unnknown script code \"" + script_code + "\"\n")
return {}
script = "AnyScript"
if script_id != -1:
script = enumdata.script_list[script_id][0]
country_id = enumdata.countryCodeToId(country_code)
country = ""
if country_id != -1:
country = enumdata.country_list[country_id][0]
if country == "":
sys.stderr.write("unnknown country code \"" + country_code + "\"\n")
return {}
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = {}
result['language'] = language
result['script'] = script
result['country'] = country
result['language_code'] = language_code
result['country_code'] = country_code
result['script_code'] = script_code
result['variant_code'] = variant_code
result['language_id'] = language_id
result['script_id'] = script_id
result['country_id'] = country_id
supplementalPath = dir_name + "/../supplemental/supplementalData.xml"
currencies = findTagsInFile(supplementalPath, "currencyData/region[iso3166=%s]"%country_code);
result['currencyIsoCode'] = ''
result['currencyDigits'] = 2
result['currencyRounding'] = 1
if currencies:
for e in currencies:
if e[0] == 'currency':
tender = True
t = filter(lambda x: x[0] == 'tender', e[1])
if t and t[0][1] == 'false':
tender = False;
if tender and not filter(lambda x: x[0] == 'to', e[1]):
result['currencyIsoCode'] = filter(lambda x: x[0] == 'iso4217', e[1])[0][1]
break
if result['currencyIsoCode']:
t = findTagsInFile(supplementalPath, "currencyData/fractions/info[iso4217=%s]"%result['currencyIsoCode']);
if t and t[0][0] == 'info':
result['currencyDigits'] = int(filter(lambda x: x[0] == 'digits', t[0][1])[0][1])
result['currencyRounding'] = int(filter(lambda x: x[0] == 'rounding', t[0][1])[0][1])
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def findEntryDef(path, xpath, value=''):
try:
return findEntry(path, xpath)
except xpathlite.Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
result['list'] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result['percent'] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
result['zero'] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result['quotationStart'] = findEntry(path, "delimiters/quotationStart")
result['quotationEnd'] = findEntry(path, "delimiters/quotationEnd")
result['alternateQuotationStart'] = findEntry(path, "delimiters/alternateQuotationStart")
result['alternateQuotationEnd'] = findEntry(path, "delimiters/alternateQuotationEnd")
result['listPatternPartStart'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[start]"))
result['listPatternPartMiddle'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[middle]"))
result['listPatternPartEnd'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[end]"))
result['listPatternPartTwo'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[2]"))
result['am'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]", draft)
result['pm'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]", draft)
result['longDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"))
result['shortDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"))
result['longTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"))
result['shortTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"))
endonym = None
if country_code and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s_%s]" % (language_code, script_code, country_code))
if not endonym and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, country_code))
if not endonym:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s]" % (language_code))
result['language_endonym'] = endonym
result['country_endonym'] = findEntryDef(path, "localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(path, "numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern", numbering_system)
currency_format = parse_number_format(currency_format, result)
result['currencyFormat'] = currency_format[0]
result['currencyNegativeFormat'] = ''
if len(currency_format) > 1:
result['currencyNegativeFormat'] = currency_format[1]
result['currencySymbol'] = ''
result['currencyDisplayName'] = ''
if result['currencyIsoCode']:
result['currencySymbol'] = findEntryDef(path, "numbers/currencies/currency[%s]/symbol" % result['currencyIsoCode'])
display_name_path = "numbers/currencies/currency[%s]/displayName" % result['currencyIsoCode']
result['currencyDisplayName'] \
= findEntryDef(path, display_name_path) + ";" \
+ findEntryDef(path, display_name_path + "[count=zero]") + ";" \
+ findEntryDef(path, display_name_path + "[count=one]") + ";" \
+ findEntryDef(path, display_name_path + "[count=two]") + ";" \
+ findEntryDef(path, display_name_path + "[count=few]") + ";" \
+ findEntryDef(path, display_name_path + "[count=many]") + ";" \
+ findEntryDef(path, display_name_path + "[count=other]") + ";"
standalone_long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[wide]/month"
result['standaloneLongMonths'] \
= findEntry(path, standalone_long_month_path + "[1]") + ";" \
+ findEntry(path, standalone_long_month_path + "[2]") + ";" \
+ findEntry(path, standalone_long_month_path + "[3]") + ";" \
+ findEntry(path, standalone_long_month_path + "[4]") + ";" \
+ findEntry(path, standalone_long_month_path + "[5]") + ";" \
+ findEntry(path, standalone_long_month_path + "[6]") + ";" \
+ findEntry(path, standalone_long_month_path + "[7]") + ";" \
+ findEntry(path, standalone_long_month_path + "[8]") + ";" \
+ findEntry(path, standalone_long_month_path + "[9]") + ";" \
+ findEntry(path, standalone_long_month_path + "[10]") + ";" \
+ findEntry(path, standalone_long_month_path + "[11]") + ";" \
+ findEntry(path, standalone_long_month_path + "[12]") + ";"
standalone_short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[abbreviated]/month"
result['standaloneShortMonths'] \
= findEntry(path, standalone_short_month_path + "[1]") + ";" \
+ findEntry(path, standalone_short_month_path + "[2]") + ";" \
+ findEntry(path, standalone_short_month_path + "[3]") + ";" \
+ findEntry(path, standalone_short_month_path + "[4]") + ";" \
+ findEntry(path, standalone_short_month_path + "[5]") + ";" \
+ findEntry(path, standalone_short_month_path + "[6]") + ";" \
+ findEntry(path, standalone_short_month_path + "[7]") + ";" \
+ findEntry(path, standalone_short_month_path + "[8]") + ";" \
+ findEntry(path, standalone_short_month_path + "[9]") + ";" \
+ findEntry(path, standalone_short_month_path + "[10]") + ";" \
+ findEntry(path, standalone_short_month_path + "[11]") + ";" \
+ findEntry(path, standalone_short_month_path + "[12]") + ";"
standalone_narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[narrow]/month"
result['standaloneNarrowMonths'] \
= findEntry(path, standalone_narrow_month_path + "[1]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[2]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[3]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[4]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[5]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[6]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[7]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[8]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[9]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[10]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[11]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[12]") + ";"
long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[wide]/month"
result['longMonths'] \
= findEntry(path, long_month_path + "[1]") + ";" \
+ findEntry(path, long_month_path + "[2]") + ";" \
+ findEntry(path, long_month_path + "[3]") + ";" \
+ findEntry(path, long_month_path + "[4]") + ";" \
+ findEntry(path, long_month_path + "[5]") + ";" \
+ findEntry(path, long_month_path + "[6]") + ";" \
+ findEntry(path, long_month_path + "[7]") + ";" \
+ findEntry(path, long_month_path + "[8]") + ";" \
+ findEntry(path, long_month_path + "[9]") + ";" \
+ findEntry(path, long_month_path + "[10]") + ";" \
+ findEntry(path, long_month_path + "[11]") + ";" \
+ findEntry(path, long_month_path + "[12]") + ";"
short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[abbreviated]/month"
result['shortMonths'] \
= findEntry(path, short_month_path + "[1]") + ";" \
+ findEntry(path, short_month_path + "[2]") + ";" \
+ findEntry(path, short_month_path + "[3]") + ";" \
+ findEntry(path, short_month_path + "[4]") + ";" \
+ findEntry(path, short_month_path + "[5]") + ";" \
+ findEntry(path, short_month_path + "[6]") + ";" \
+ findEntry(path, short_month_path + "[7]") + ";" \
+ findEntry(path, short_month_path + "[8]") + ";" \
+ findEntry(path, short_month_path + "[9]") + ";" \
+ findEntry(path, short_month_path + "[10]") + ";" \
+ findEntry(path, short_month_path + "[11]") + ";" \
+ findEntry(path, short_month_path + "[12]") + ";"
narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[narrow]/month"
result['narrowMonths'] \
= findEntry(path, narrow_month_path + "[1]") + ";" \
+ findEntry(path, narrow_month_path + "[2]") + ";" \
+ findEntry(path, narrow_month_path + "[3]") + ";" \
+ findEntry(path, narrow_month_path + "[4]") + ";" \
+ findEntry(path, narrow_month_path + "[5]") + ";" \
+ findEntry(path, narrow_month_path + "[6]") + ";" \
+ findEntry(path, narrow_month_path + "[7]") + ";" \
+ findEntry(path, narrow_month_path + "[8]") + ";" \
+ findEntry(path, narrow_month_path + "[9]") + ";" \
+ findEntry(path, narrow_month_path + "[10]") + ";" \
+ findEntry(path, narrow_month_path + "[11]") + ";" \
+ findEntry(path, narrow_month_path + "[12]") + ";"
long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[wide]/day"
result['longDays'] \
= findEntry(path, long_day_path + "[sun]") + ";" \
+ findEntry(path, long_day_path + "[mon]") + ";" \
+ findEntry(path, long_day_path + "[tue]") + ";" \
+ findEntry(path, long_day_path + "[wed]") + ";" \
+ findEntry(path, long_day_path + "[thu]") + ";" \
+ findEntry(path, long_day_path + "[fri]") + ";" \
+ findEntry(path, long_day_path + "[sat]") + ";"
short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[abbreviated]/day"
result['shortDays'] \
= findEntry(path, short_day_path + "[sun]") + ";" \
+ findEntry(path, short_day_path + "[mon]") + ";" \
+ findEntry(path, short_day_path + "[tue]") + ";" \
+ findEntry(path, short_day_path + "[wed]") + ";" \
+ findEntry(path, short_day_path + "[thu]") + ";" \
+ findEntry(path, short_day_path + "[fri]") + ";" \
+ findEntry(path, short_day_path + "[sat]") + ";"
narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[narrow]/day"
result['narrowDays'] \
= findEntry(path, narrow_day_path + "[sun]") + ";" \
+ findEntry(path, narrow_day_path + "[mon]") + ";" \
+ findEntry(path, narrow_day_path + "[tue]") + ";" \
+ findEntry(path, narrow_day_path + "[wed]") + ";" \
+ findEntry(path, narrow_day_path + "[thu]") + ";" \
+ findEntry(path, narrow_day_path + "[fri]") + ";" \
+ findEntry(path, narrow_day_path + "[sat]") + ";"
standalone_long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[wide]/day"
result['standaloneLongDays'] \
= findEntry(path, standalone_long_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_long_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_long_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_long_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_long_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_long_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_long_day_path + "[sat]") + ";"
standalone_short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[abbreviated]/day"
result['standaloneShortDays'] \
= findEntry(path, standalone_short_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_short_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_short_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_short_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_short_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_short_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_short_day_path + "[sat]") + ";"
standalone_narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[narrow]/day"
result['standaloneNarrowDays'] \
= findEntry(path, standalone_narrow_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[sat]") + ";"
return result
def generateLocaleInfo(path):
(dir_name, file_name) = os.path.split(path)
if not path.endswith(".xml"):
return {}
language_code = findEntryInFile(path, "identity/language", attribute="type")[0]
if language_code == 'root':
# just skip it
return {}
country_code = findEntryInFile(path, "identity/territory", attribute="type")[0]
script_code = findEntryInFile(path, "identity/script", attribute="type")[0]
variant_code = findEntryInFile(path, "identity/variant", attribute="type")[0]
# we should handle fully qualified names with the territory
if not country_code:
return {}
# we do not support scripts and variants
if variant_code or script_code:
return {}
language_id = enumdata.languageCodeToId(language_code)
if language_id == -1:
sys.stderr.write("unnknown language code \"" + language_code + "\"\n")
return {}
language = enumdata.language_list[language_id][0]
country_id = enumdata.countryCodeToId(country_code)
country = ""
if country_id != -1:
country = enumdata.country_list[country_id][0]
if country == "":
sys.stderr.write("unnknown country code \"" + country_code + "\"\n")
return {}
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = {}
result['language'] = language
result['country'] = country
result['language_code'] = language_code
result['country_code'] = country_code
result['script_code'] = script_code
result['variant_code'] = variant_code
result['language_id'] = language_id
result['country_id'] = country_id
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
result['list'] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result['percent'] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
result['zero'] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result['am'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]", draft)
result['pm'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]", draft)
result['longDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"))
result['shortDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"))
result['longTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"))
result['shortTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"))
standalone_long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[wide]/month"
result['standaloneLongMonths'] \
= findEntry(path, standalone_long_month_path + "[1]") + ";" \
+ findEntry(path, standalone_long_month_path + "[2]") + ";" \
+ findEntry(path, standalone_long_month_path + "[3]") + ";" \
+ findEntry(path, standalone_long_month_path + "[4]") + ";" \
+ findEntry(path, standalone_long_month_path + "[5]") + ";" \
+ findEntry(path, standalone_long_month_path + "[6]") + ";" \
+ findEntry(path, standalone_long_month_path + "[7]") + ";" \
+ findEntry(path, standalone_long_month_path + "[8]") + ";" \
+ findEntry(path, standalone_long_month_path + "[9]") + ";" \
+ findEntry(path, standalone_long_month_path + "[10]") + ";" \
+ findEntry(path, standalone_long_month_path + "[11]") + ";" \
+ findEntry(path, standalone_long_month_path + "[12]") + ";"
standalone_short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[abbreviated]/month"
result['standaloneShortMonths'] \
= findEntry(path, standalone_short_month_path + "[1]") + ";" \
+ findEntry(path, standalone_short_month_path + "[2]") + ";" \
+ findEntry(path, standalone_short_month_path + "[3]") + ";" \
+ findEntry(path, standalone_short_month_path + "[4]") + ";" \
+ findEntry(path, standalone_short_month_path + "[5]") + ";" \
+ findEntry(path, standalone_short_month_path + "[6]") + ";" \
+ findEntry(path, standalone_short_month_path + "[7]") + ";" \
+ findEntry(path, standalone_short_month_path + "[8]") + ";" \
+ findEntry(path, standalone_short_month_path + "[9]") + ";" \
+ findEntry(path, standalone_short_month_path + "[10]") + ";" \
+ findEntry(path, standalone_short_month_path + "[11]") + ";" \
+ findEntry(path, standalone_short_month_path + "[12]") + ";"
standalone_narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[narrow]/month"
result['standaloneNarrowMonths'] \
= findEntry(path, standalone_narrow_month_path + "[1]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[2]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[3]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[4]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[5]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[6]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[7]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[8]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[9]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[10]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[11]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[12]") + ";"
long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[wide]/month"
result['longMonths'] \
= findEntry(path, long_month_path + "[1]") + ";" \
+ findEntry(path, long_month_path + "[2]") + ";" \
+ findEntry(path, long_month_path + "[3]") + ";" \
+ findEntry(path, long_month_path + "[4]") + ";" \
+ findEntry(path, long_month_path + "[5]") + ";" \
+ findEntry(path, long_month_path + "[6]") + ";" \
+ findEntry(path, long_month_path + "[7]") + ";" \
+ findEntry(path, long_month_path + "[8]") + ";" \
+ findEntry(path, long_month_path + "[9]") + ";" \
+ findEntry(path, long_month_path + "[10]") + ";" \
+ findEntry(path, long_month_path + "[11]") + ";" \
+ findEntry(path, long_month_path + "[12]") + ";"
short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[abbreviated]/month"
result['shortMonths'] \
= findEntry(path, short_month_path + "[1]") + ";" \
+ findEntry(path, short_month_path + "[2]") + ";" \
+ findEntry(path, short_month_path + "[3]") + ";" \
+ findEntry(path, short_month_path + "[4]") + ";" \
+ findEntry(path, short_month_path + "[5]") + ";" \
+ findEntry(path, short_month_path + "[6]") + ";" \
+ findEntry(path, short_month_path + "[7]") + ";" \
+ findEntry(path, short_month_path + "[8]") + ";" \
+ findEntry(path, short_month_path + "[9]") + ";" \
+ findEntry(path, short_month_path + "[10]") + ";" \
+ findEntry(path, short_month_path + "[11]") + ";" \
+ findEntry(path, short_month_path + "[12]") + ";"
narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[narrow]/month"
result['narrowMonths'] \
= findEntry(path, narrow_month_path + "[1]") + ";" \
+ findEntry(path, narrow_month_path + "[2]") + ";" \
+ findEntry(path, narrow_month_path + "[3]") + ";" \
+ findEntry(path, narrow_month_path + "[4]") + ";" \
+ findEntry(path, narrow_month_path + "[5]") + ";" \
+ findEntry(path, narrow_month_path + "[6]") + ";" \
+ findEntry(path, narrow_month_path + "[7]") + ";" \
+ findEntry(path, narrow_month_path + "[8]") + ";" \
+ findEntry(path, narrow_month_path + "[9]") + ";" \
+ findEntry(path, narrow_month_path + "[10]") + ";" \
+ findEntry(path, narrow_month_path + "[11]") + ";" \
+ findEntry(path, narrow_month_path + "[12]") + ";"
long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[wide]/day"
result['longDays'] \
= findEntry(path, long_day_path + "[sun]") + ";" \
+ findEntry(path, long_day_path + "[mon]") + ";" \
+ findEntry(path, long_day_path + "[tue]") + ";" \
+ findEntry(path, long_day_path + "[wed]") + ";" \
+ findEntry(path, long_day_path + "[thu]") + ";" \
+ findEntry(path, long_day_path + "[fri]") + ";" \
+ findEntry(path, long_day_path + "[sat]") + ";"
short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[abbreviated]/day"
result['shortDays'] \
= findEntry(path, short_day_path + "[sun]") + ";" \
+ findEntry(path, short_day_path + "[mon]") + ";" \
+ findEntry(path, short_day_path + "[tue]") + ";" \
+ findEntry(path, short_day_path + "[wed]") + ";" \
+ findEntry(path, short_day_path + "[thu]") + ";" \
+ findEntry(path, short_day_path + "[fri]") + ";" \
+ findEntry(path, short_day_path + "[sat]") + ";"
narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[narrow]/day"
result['narrowDays'] \
= findEntry(path, narrow_day_path + "[sun]") + ";" \
+ findEntry(path, narrow_day_path + "[mon]") + ";" \
+ findEntry(path, narrow_day_path + "[tue]") + ";" \
+ findEntry(path, narrow_day_path + "[wed]") + ";" \
+ findEntry(path, narrow_day_path + "[thu]") + ";" \
+ findEntry(path, narrow_day_path + "[fri]") + ";" \
+ findEntry(path, narrow_day_path + "[sat]") + ";"
standalone_long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[wide]/day"
result['standaloneLongDays'] \
= findEntry(path, standalone_long_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_long_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_long_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_long_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_long_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_long_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_long_day_path + "[sat]") + ";"
standalone_short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[abbreviated]/day"
result['standaloneShortDays'] \
= findEntry(path, standalone_short_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_short_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_short_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_short_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_short_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_short_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_short_day_path + "[sat]") + ";"
standalone_narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[narrow]/day"
result['standaloneNarrowDays'] \
= findEntry(path, standalone_narrow_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[sat]") + ";"
return result
def generateLocaleInfo(path):
(dir_name, file_name) = os.path.split(path)
if not path.endswith(".xml"):
return {}
language_code = findEntryInFile(path,
"identity/language",
attribute="type")[0]
if language_code == 'root':
# just skip it
return {}
country_code = findEntryInFile(path,
"identity/territory",
attribute="type")[0]
script_code = findEntryInFile(path, "identity/script", attribute="type")[0]
variant_code = findEntryInFile(path, "identity/variant",
attribute="type")[0]
# we should handle fully qualified names with the territory
if not country_code:
return {}
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
return {}
language_id = enumdata.languageCodeToId(language_code)
if language_id == -1:
sys.stderr.write("unnknown language code \"" + language_code + "\"\n")
return {}
language = enumdata.language_list[language_id][0]
script_id = enumdata.scriptCodeToId(script_code)
if script_code == -1:
sys.stderr.write("unnknown script code \"" + script_code + "\"\n")
return {}
script = "AnyScript"
if script_id != -1:
script = enumdata.script_list[script_id][0]
country_id = enumdata.countryCodeToId(country_code)
country = ""
if country_id != -1:
country = enumdata.country_list[country_id][0]
if country == "":
sys.stderr.write("unnknown country code \"" + country_code + "\"\n")
return {}
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = {}
result['language'] = language
result['script'] = script
result['country'] = country
result['language_code'] = language_code
result['country_code'] = country_code
result['script_code'] = script_code
result['variant_code'] = variant_code
result['language_id'] = language_id
result['script_id'] = script_id
result['country_id'] = country_id
supplementalPath = dir_name + "/../supplemental/supplementalData.xml"
currencies = findTagsInFile(
supplementalPath, "currencyData/region[iso3166=%s]" % country_code)
result['currencyIsoCode'] = ''
result['currencyDigits'] = 2
result['currencyRounding'] = 1
if currencies:
for e in currencies:
if e[0] == 'currency':
tender = True
t = filter(lambda x: x[0] == 'tender', e[1])
if t and t[0][1] == 'false':
tender = False
if tender and not filter(lambda x: x[0] == 'to', e[1]):
result['currencyIsoCode'] = filter(
lambda x: x[0] == 'iso4217', e[1])[0][1]
break
if result['currencyIsoCode']:
t = findTagsInFile(
supplementalPath, "currencyData/fractions/info[iso4217=%s]" %
result['currencyIsoCode'])
if t and t[0][0] == 'info':
result['currencyDigits'] = int(
filter(lambda x: x[0] == 'digits', t[0][1])[0][1])
result['currencyRounding'] = int(
filter(lambda x: x[0] == 'rounding', t[0][1])[0][1])
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def findEntryDef(path, xpath, value=''):
try:
return findEntry(path, xpath)
except xpathlite.Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(
path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal",
numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group",
numbering_system)
result['list'] = get_number_in_system(path, "numbers/symbols/list",
numbering_system)
result['percent'] = get_number_in_system(path,
"numbers/symbols/percentSign",
numbering_system)
result['zero'] = get_number_in_system(path,
"numbers/symbols/nativeZeroDigit",
numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign",
numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign",
numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential",
numbering_system).lower()
result['quotationStart'] = findEntry(path, "delimiters/quotationStart")
result['quotationEnd'] = findEntry(path, "delimiters/quotationEnd")
result['alternateQuotationStart'] = findEntry(
path, "delimiters/alternateQuotationStart")
result['alternateQuotationEnd'] = findEntry(
path, "delimiters/alternateQuotationEnd")
result['listPatternPartStart'] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[start]"))
result['listPatternPartMiddle'] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[middle]"))
result['listPatternPartEnd'] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[end]"))
result['listPatternPartTwo'] = parse_list_pattern_part_format(
findEntry(path, "listPatterns/listPattern/listPatternPart[2]"))
result['am'] = findEntry(
path,
"dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]",
draft)
result['pm'] = findEntry(
path,
"dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]",
draft)
result['longDateFormat'] = convert_date(
findEntry(
path,
"dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"
))
result['shortDateFormat'] = convert_date(
findEntry(
path,
"dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"
))
result['longTimeFormat'] = convert_date(
findEntry(
path,
"dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"
))
result['shortTimeFormat'] = convert_date(
findEntry(
path,
"dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"
))
endonym = None
if country_code and script_code:
endonym = findEntryDef(
path, "localeDisplayNames/languages/language[type=%s_%s_%s]" %
(language_code, script_code, country_code))
if not endonym and script_code:
endonym = findEntryDef(
path, "localeDisplayNames/languages/language[type=%s_%s]" %
(language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(
path, "localeDisplayNames/languages/language[type=%s_%s]" %
(language_code, country_code))
if not endonym:
endonym = findEntryDef(
path,
"localeDisplayNames/languages/language[type=%s]" % (language_code))
result['language_endonym'] = endonym
result['country_endonym'] = findEntryDef(
path,
"localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(
path,
"numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern",
numbering_system)
currency_format = parse_number_format(currency_format, result)
result['currencyFormat'] = currency_format[0]
result['currencyNegativeFormat'] = ''
if len(currency_format) > 1:
result['currencyNegativeFormat'] = currency_format[1]
result['currencySymbol'] = ''
result['currencyDisplayName'] = ''
if result['currencyIsoCode']:
result['currencySymbol'] = findEntryDef(
path, "numbers/currencies/currency[%s]/symbol" %
result['currencyIsoCode'])
display_name_path = "numbers/currencies/currency[%s]/displayName" % result[
'currencyIsoCode']
result['currencyDisplayName'] \
= findEntryDef(path, display_name_path) + ";" \
+ findEntryDef(path, display_name_path + "[count=zero]") + ";" \
+ findEntryDef(path, display_name_path + "[count=one]") + ";" \
+ findEntryDef(path, display_name_path + "[count=two]") + ";" \
+ findEntryDef(path, display_name_path + "[count=few]") + ";" \
+ findEntryDef(path, display_name_path + "[count=many]") + ";" \
+ findEntryDef(path, display_name_path + "[count=other]") + ";"
standalone_long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[wide]/month"
result['standaloneLongMonths'] \
= findEntry(path, standalone_long_month_path + "[1]") + ";" \
+ findEntry(path, standalone_long_month_path + "[2]") + ";" \
+ findEntry(path, standalone_long_month_path + "[3]") + ";" \
+ findEntry(path, standalone_long_month_path + "[4]") + ";" \
+ findEntry(path, standalone_long_month_path + "[5]") + ";" \
+ findEntry(path, standalone_long_month_path + "[6]") + ";" \
+ findEntry(path, standalone_long_month_path + "[7]") + ";" \
+ findEntry(path, standalone_long_month_path + "[8]") + ";" \
+ findEntry(path, standalone_long_month_path + "[9]") + ";" \
+ findEntry(path, standalone_long_month_path + "[10]") + ";" \
+ findEntry(path, standalone_long_month_path + "[11]") + ";" \
+ findEntry(path, standalone_long_month_path + "[12]") + ";"
standalone_short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[abbreviated]/month"
result['standaloneShortMonths'] \
= findEntry(path, standalone_short_month_path + "[1]") + ";" \
+ findEntry(path, standalone_short_month_path + "[2]") + ";" \
+ findEntry(path, standalone_short_month_path + "[3]") + ";" \
+ findEntry(path, standalone_short_month_path + "[4]") + ";" \
+ findEntry(path, standalone_short_month_path + "[5]") + ";" \
+ findEntry(path, standalone_short_month_path + "[6]") + ";" \
+ findEntry(path, standalone_short_month_path + "[7]") + ";" \
+ findEntry(path, standalone_short_month_path + "[8]") + ";" \
+ findEntry(path, standalone_short_month_path + "[9]") + ";" \
+ findEntry(path, standalone_short_month_path + "[10]") + ";" \
+ findEntry(path, standalone_short_month_path + "[11]") + ";" \
+ findEntry(path, standalone_short_month_path + "[12]") + ";"
standalone_narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[stand-alone]/monthWidth[narrow]/month"
result['standaloneNarrowMonths'] \
= findEntry(path, standalone_narrow_month_path + "[1]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[2]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[3]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[4]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[5]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[6]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[7]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[8]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[9]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[10]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[11]") + ";" \
+ findEntry(path, standalone_narrow_month_path + "[12]") + ";"
long_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[wide]/month"
result['longMonths'] \
= findEntry(path, long_month_path + "[1]") + ";" \
+ findEntry(path, long_month_path + "[2]") + ";" \
+ findEntry(path, long_month_path + "[3]") + ";" \
+ findEntry(path, long_month_path + "[4]") + ";" \
+ findEntry(path, long_month_path + "[5]") + ";" \
+ findEntry(path, long_month_path + "[6]") + ";" \
+ findEntry(path, long_month_path + "[7]") + ";" \
+ findEntry(path, long_month_path + "[8]") + ";" \
+ findEntry(path, long_month_path + "[9]") + ";" \
+ findEntry(path, long_month_path + "[10]") + ";" \
+ findEntry(path, long_month_path + "[11]") + ";" \
+ findEntry(path, long_month_path + "[12]") + ";"
short_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[abbreviated]/month"
result['shortMonths'] \
= findEntry(path, short_month_path + "[1]") + ";" \
+ findEntry(path, short_month_path + "[2]") + ";" \
+ findEntry(path, short_month_path + "[3]") + ";" \
+ findEntry(path, short_month_path + "[4]") + ";" \
+ findEntry(path, short_month_path + "[5]") + ";" \
+ findEntry(path, short_month_path + "[6]") + ";" \
+ findEntry(path, short_month_path + "[7]") + ";" \
+ findEntry(path, short_month_path + "[8]") + ";" \
+ findEntry(path, short_month_path + "[9]") + ";" \
+ findEntry(path, short_month_path + "[10]") + ";" \
+ findEntry(path, short_month_path + "[11]") + ";" \
+ findEntry(path, short_month_path + "[12]") + ";"
narrow_month_path = "dates/calendars/calendar[gregorian]/months/monthContext[format]/monthWidth[narrow]/month"
result['narrowMonths'] \
= findEntry(path, narrow_month_path + "[1]") + ";" \
+ findEntry(path, narrow_month_path + "[2]") + ";" \
+ findEntry(path, narrow_month_path + "[3]") + ";" \
+ findEntry(path, narrow_month_path + "[4]") + ";" \
+ findEntry(path, narrow_month_path + "[5]") + ";" \
+ findEntry(path, narrow_month_path + "[6]") + ";" \
+ findEntry(path, narrow_month_path + "[7]") + ";" \
+ findEntry(path, narrow_month_path + "[8]") + ";" \
+ findEntry(path, narrow_month_path + "[9]") + ";" \
+ findEntry(path, narrow_month_path + "[10]") + ";" \
+ findEntry(path, narrow_month_path + "[11]") + ";" \
+ findEntry(path, narrow_month_path + "[12]") + ";"
long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[wide]/day"
result['longDays'] \
= findEntry(path, long_day_path + "[sun]") + ";" \
+ findEntry(path, long_day_path + "[mon]") + ";" \
+ findEntry(path, long_day_path + "[tue]") + ";" \
+ findEntry(path, long_day_path + "[wed]") + ";" \
+ findEntry(path, long_day_path + "[thu]") + ";" \
+ findEntry(path, long_day_path + "[fri]") + ";" \
+ findEntry(path, long_day_path + "[sat]") + ";"
short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[abbreviated]/day"
result['shortDays'] \
= findEntry(path, short_day_path + "[sun]") + ";" \
+ findEntry(path, short_day_path + "[mon]") + ";" \
+ findEntry(path, short_day_path + "[tue]") + ";" \
+ findEntry(path, short_day_path + "[wed]") + ";" \
+ findEntry(path, short_day_path + "[thu]") + ";" \
+ findEntry(path, short_day_path + "[fri]") + ";" \
+ findEntry(path, short_day_path + "[sat]") + ";"
narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[format]/dayWidth[narrow]/day"
result['narrowDays'] \
= findEntry(path, narrow_day_path + "[sun]") + ";" \
+ findEntry(path, narrow_day_path + "[mon]") + ";" \
+ findEntry(path, narrow_day_path + "[tue]") + ";" \
+ findEntry(path, narrow_day_path + "[wed]") + ";" \
+ findEntry(path, narrow_day_path + "[thu]") + ";" \
+ findEntry(path, narrow_day_path + "[fri]") + ";" \
+ findEntry(path, narrow_day_path + "[sat]") + ";"
standalone_long_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[wide]/day"
result['standaloneLongDays'] \
= findEntry(path, standalone_long_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_long_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_long_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_long_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_long_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_long_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_long_day_path + "[sat]") + ";"
standalone_short_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[abbreviated]/day"
result['standaloneShortDays'] \
= findEntry(path, standalone_short_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_short_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_short_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_short_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_short_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_short_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_short_day_path + "[sat]") + ";"
standalone_narrow_day_path = "dates/calendars/calendar[gregorian]/days/dayContext[stand-alone]/dayWidth[narrow]/day"
result['standaloneNarrowDays'] \
= findEntry(path, standalone_narrow_day_path + "[sun]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[mon]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[tue]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[wed]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[thu]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[fri]") + ";" \
+ findEntry(path, standalone_narrow_day_path + "[sat]") + ";"
return result
def _generateLocaleInfo(path, language_code, script_code, country_code, variant_code=""):
if not path.endswith(".xml"):
return {}
if language_code == 'root':
# just skip it
return {}
# we do not support variants
# ### actually there is only one locale with variant: en_US_POSIX
# does anybody care about it at all?
if variant_code:
raise xpathlite.Error('we do not support variants ("%s")' % variant_code)
language_id = enumdata.languageCodeToId(language_code)
if language_id <= 0:
raise xpathlite.Error('unknown language code "%s"' % language_code)
script_id = enumdata.scriptCodeToId(script_code)
if script_id == -1:
raise xpathlite.Error('unknown script code "%s"' % script_code)
# we should handle fully qualified names with the territory
if not country_code:
return {}
country_id = enumdata.countryCodeToId(country_code)
if country_id <= 0:
raise xpathlite.Error('unknown country code "%s"' % country_code)
# So we say we accept only those values that have "contributed" or
# "approved" resolution. see http://www.unicode.org/cldr/process.html
# But we only respect the resolution for new datas for backward
# compatibility.
draft = DraftResolution.contributed
result = dict(
language=enumdata.language_list[language_id][0],
language_code=language_code, language_id=language_id,
script=enumdata.script_list[script_id][0],
script_code=script_code, script_id=script_id,
country=enumdata.country_list[country_id][0],
country_code=country_code, country_id=country_id,
variant_code=variant_code)
(dir_name, file_name) = os.path.split(path)
def from_supplement(tag,
path=os.path.join(dir_name, '..', 'supplemental',
'supplementalData.xml')):
return findTagsInFile(path, tag)
currencies = from_supplement('currencyData/region[iso3166=%s]' % country_code)
result['currencyIsoCode'] = ''
result['currencyDigits'] = 2
result['currencyRounding'] = 1
if currencies:
for e in currencies:
if e[0] == 'currency':
t = [x[1] == 'false' for x in e[1] if x[0] == 'tender']
if t and t[0]:
pass
elif not any(x[0] == 'to' for x in e[1]):
result['currencyIsoCode'] = (x[1] for x in e[1] if x[0] == 'iso4217').next()
break
if result['currencyIsoCode']:
t = from_supplement("currencyData/fractions/info[iso4217=%s]"
% result['currencyIsoCode'])
if t and t[0][0] == 'info':
result['currencyDigits'] = (int(x[1]) for x in t[0][1] if x[0] == 'digits').next()
result['currencyRounding'] = (int(x[1]) for x in t[0][1] if x[0] == 'rounding').next()
numbering_system = None
try:
numbering_system = findEntry(path, "numbers/defaultNumberingSystem")
except:
pass
def findEntryDef(path, xpath, value=''):
try:
return findEntry(path, xpath)
except xpathlite.Error:
return value
def get_number_in_system(path, xpath, numbering_system):
if numbering_system:
try:
return findEntry(path, xpath + "[numberSystem=" + numbering_system + "]")
except xpathlite.Error:
# in CLDR 1.9 number system was refactored for numbers (but not for currency)
# so if previous findEntry doesn't work we should try this:
try:
return findEntry(path, xpath.replace("/symbols/", "/symbols[numberSystem=" + numbering_system + "]/"))
except xpathlite.Error:
# fallback to default
pass
return findEntry(path, xpath)
result['decimal'] = get_number_in_system(path, "numbers/symbols/decimal", numbering_system)
result['group'] = get_number_in_system(path, "numbers/symbols/group", numbering_system)
result['list'] = get_number_in_system(path, "numbers/symbols/list", numbering_system)
result['percent'] = get_number_in_system(path, "numbers/symbols/percentSign", numbering_system)
try:
numbering_systems = {}
for ns in findTagsInFile(os.path.join(cldr_dir, '..', 'supplemental',
'numberingSystems.xml'),
'numberingSystems'):
tmp = {}
id = ""
for data in ns[1:][0]: # ns looks like this: [u'numberingSystem', [(u'digits', u'0123456789'), (u'type', u'numeric'), (u'id', u'latn')]]
tmp[data[0]] = data[1]
if data[0] == u"id":
id = data[1]
numbering_systems[id] = tmp
result['zero'] = numbering_systems[numbering_system][u"digits"][0]
except e:
sys.stderr.write("Native zero detection problem:\n" + str(e) + "\n")
result['zero'] = get_number_in_system(path, "numbers/symbols/nativeZeroDigit", numbering_system)
result['minus'] = get_number_in_system(path, "numbers/symbols/minusSign", numbering_system)
result['plus'] = get_number_in_system(path, "numbers/symbols/plusSign", numbering_system)
result['exp'] = get_number_in_system(path, "numbers/symbols/exponential", numbering_system).lower()
result['quotationStart'] = findEntry(path, "delimiters/quotationStart")
result['quotationEnd'] = findEntry(path, "delimiters/quotationEnd")
result['alternateQuotationStart'] = findEntry(path, "delimiters/alternateQuotationStart")
result['alternateQuotationEnd'] = findEntry(path, "delimiters/alternateQuotationEnd")
result['listPatternPartStart'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[start]"))
result['listPatternPartMiddle'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[middle]"))
result['listPatternPartEnd'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[end]"))
result['listPatternPartTwo'] = parse_list_pattern_part_format(findEntry(path, "listPatterns/listPattern/listPatternPart[2]"))
result['am'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[am]", draft)
result['pm'] = findEntry(path, "dates/calendars/calendar[gregorian]/dayPeriods/dayPeriodContext[format]/dayPeriodWidth[wide]/dayPeriod[pm]", draft)
result['longDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[full]/dateFormat/pattern"))
result['shortDateFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/dateFormats/dateFormatLength[short]/dateFormat/pattern"))
result['longTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[full]/timeFormat/pattern"))
result['shortTimeFormat'] = convert_date(findEntry(path, "dates/calendars/calendar[gregorian]/timeFormats/timeFormatLength[short]/timeFormat/pattern"))
endonym = None
if country_code and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s_%s]" % (language_code, script_code, country_code))
if not endonym and script_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, script_code))
if not endonym and country_code:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s_%s]" % (language_code, country_code))
if not endonym:
endonym = findEntryDef(path, "localeDisplayNames/languages/language[type=%s]" % (language_code))
result['language_endonym'] = endonym
result['country_endonym'] = findEntryDef(path, "localeDisplayNames/territories/territory[type=%s]" % (country_code))
currency_format = get_number_in_system(path, "numbers/currencyFormats/currencyFormatLength/currencyFormat/pattern", numbering_system)
currency_format = parse_number_format(currency_format, result)
result['currencyFormat'] = currency_format[0]
result['currencyNegativeFormat'] = ''
if len(currency_format) > 1:
result['currencyNegativeFormat'] = currency_format[1]
result['currencySymbol'] = ''
result['currencyDisplayName'] = ''
if result['currencyIsoCode']:
result['currencySymbol'] = findEntryDef(path, "numbers/currencies/currency[%s]/symbol" % result['currencyIsoCode'])
result['currencyDisplayName'] = ';'.join(
findEntryDef(path, 'numbers/currencies/currency[' + result['currencyIsoCode']
+ ']/displayName' + tail)
for tail in ['',] + [
'[count=%s]' % x for x in ('zero', 'one', 'two', 'few', 'many', 'other')
]) + ';'
def findUnitDef(path, stem, fallback=''):
# The displayName for a quantified unit in en.xml is kByte
# instead of kB (etc.), so prefer any unitPattern provided:
for count in ('many', 'few', 'two', 'other', 'zero', 'one'):
try:
ans = findEntry(path, stem + 'unitPattern[count=%s]' % count)
except xpathlite.Error:
continue
# TODO: epxloit count-handling, instead of discarding placeholders
if ans.startswith('{0}'):
ans = ans[3:].lstrip()
if ans:
return ans
return findEntryDef(path, stem + 'displayName', fallback)
# First without quantifier, then quantified each way:
result['byte_unit'] = findEntryDef(
path, 'units/unitLength[type=long]/unit[type=digital-byte]/displayName',
'bytes')
stem = 'units/unitLength[type=short]/unit[type=digital-%sbyte]/'
known = [] # cases where we *do* have a given version:
result['byte_si_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem, 'B', known))
# IEC 60027-2
# http://physics.nist.gov/cuu/Units/binary.html
result['byte_iec_quantified'] = ';'.join(unit_quantifiers(findUnitDef, path, stem % '%sbi', 'iB', known))
# Used for month and day data:
namings = (
('standaloneLong', 'stand-alone', 'wide'),
('standaloneShort', 'stand-alone', 'abbreviated'),
('standaloneNarrow', 'stand-alone', 'narrow'),
('long', 'format', 'wide'),
('short', 'format', 'abbreviated'),
('narrow', 'format', 'narrow'),
)
# Month data:
for cal in ('gregorian',): # We shall want to add to this
stem = 'dates/calendars/calendar[' + cal + ']/months/'
for (key, mode, size) in namings:
prop = 'monthContext[' + mode + ']/monthWidth[' + size + ']/'
result[key + 'Months'] = ';'.join(
findEntry(path, stem + prop + "month[%d]" % i)
for i in range(1, 13)) + ';'
# Day data (for Gregorian, at least):
stem = 'dates/calendars/calendar[gregorian]/days/'
days = ('sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat')
for (key, mode, size) in namings:
prop = 'dayContext[' + mode + ']/dayWidth[' + size + ']/day'
result[key + 'Days'] = ';'.join(
findEntry(path, stem + prop + '[' + day + ']')
for day in days) + ';'
return Locale(result)