def test_preferred(self):
# Extlang
subtag = Subtag('vsv', 'extlang')
preferred = subtag.preferred
self.assertIsNotNone(preferred)
self.assertEqual(preferred.type, 'language')
self.assertEqual(preferred.format, 'vsv')
# Language
# Moldovan -> Romanian
subtag = Subtag('mo', 'language')
preferred = subtag.preferred
self.assertIsNotNone(preferred)
self.assertEqual(preferred.type, 'language')
self.assertEqual(preferred.format, 'ro')
# Region
# Burma -> Myanmar
subtag = Subtag('BU', 'region')
preferred = subtag.preferred
self.assertIsNotNone(preferred)
self.assertEqual(preferred.type, 'region')
self.assertEqual(preferred.format, 'MM')
# Variant
subtag = Subtag('heploc', 'variant')
preferred = subtag.preferred
self.assertIsNotNone(preferred)
self.assertEqual(preferred.type, 'variant')
self.assertEqual(preferred.format, 'alalc97')
# Should return None if no preferred value.
# Latin America and the Caribbean
subtag = Subtag('419', 'region')
self.assertIsNone(subtag.preferred)
def test_script(self):
subtag = Subtag('en', 'language')
script = subtag.script
self.assertIsNotNone(script)
self.assertEqual(script.type, 'script')
self.assertEqual(script.format, 'Latn')
# Should return null if no script.
# A macrolanguage like 'zh' should have no suppress-script.
subtag = Subtag('zh', 'language')
script = subtag.script
self.assertIsNone(script)
def test_errors(self):
self.assertRaises(Exception, Subtag, '123', 'region')
with self.assertRaises(Exception) as context:
Subtag('123', 'region')
self.assertIn('Non-existent subtag 123.', context.exception.message)
self.assertIn('Non-existent subtag 123.', context.exception.__str__())
self.assertRaises(Exception, Subtag, 'art-lojban', 'grandfathered')
with self.assertRaises(Exception) as context:
Subtag('art-lojban', 'grandfathered')
self.assertIn('art-lojban is a grandfathered tag',
context.exception.message)
self.assertRaises(Exception, Subtag, 'nl', 'variant')
with self.assertRaises(Exception) as context:
Subtag('nl', 'variant')
self.assertIn('Non-existent subtag nl of type variant.',
context.exception.message)
def type(subtag, type):
"""
Get a :class:`language_tags.Subtag.Subtag` by subtag and type. Can be None if not exists.
:param str subtag: subtag.
:param str type: type of the subtag.
:return: :class:`language_tags.Subtag.Subtag` if exists, otherwise None.
"""
subtag = subtag.lower()
if subtag in index:
types = index[subtag]
if type in types:
return Subtag(subtag, type)
return None
def subtags(subtags):
"""
Get a list of existing :class:`language_tags.Subtag.Subtag` objects given the input subtag(s).
:param subtags: string subtag or list of string subtags.
:return: a list of existing :class:`language_tags.Subtag.Subtag` objects. The return list can be empty.
"""
result = []
if not isinstance(subtags, list):
subtags = [subtags]
for subtag in subtags:
for type in tags.types(subtag):
result.append(Subtag(subtag, type))
return result
def test_format(self):
# Language
self.assertEqual(Subtag('en', 'language').format, 'en')
self.assertEqual(Subtag('EN', 'language').format, 'en')
# Region
self.assertEqual(Subtag('GB', 'region').format, 'GB')
self.assertEqual(Subtag('gb', 'region').format, 'GB')
# Script
self.assertEqual(Subtag('Latn', 'script').format, 'Latn')
self.assertEqual(Subtag('latn', 'script').format, 'Latn')
def search(description, all=False):
"""
Gets a list of :class:`language_tags.Subtag.Subtag` objects where the description matches.
:param description: a string or compiled regular expression. For example: ``search(re.compile('\d{4}'))`` if the
description of the returned subtag must contain four contiguous numerical digits.
:type description: str or RegExp
:param all: If set on True grandfathered and redundant tags will be included in the return
list.
:type all: bool, optional
:return: list of :class:`language_tags.Subtag.Subtag` objects each including the description.
The return list can be empty.
"""
# If the input query is all lowercase, make a case-insensitive match.
if isinstance(description, str):
list_to_string = lambda l: ', '.join(l).lower(
) if description.lower() == description else ', '.join(l)
def test(record):
return description in list_to_string(record['Description'])
elif hasattr(description.search, '__call__'):
def test(record):
return description.search(', '.join(
record['Description'])) is not None
records = filter(
lambda r: False
if ('Subtag' not in r and not all) else test(r), registry)
if six.PY3:
records = list(records)
# Sort by matched description string length. This is a quick way to push precise matches towards the top.
results = sorted(records, key=lambda r: min([abs(len(r_description) - len(description))
for r_description in r['Description']])) \
if isinstance(description, str) else records
return [
Subtag(r['Subtag'], r['Type']) if 'Subtag' in r else Tag(['Tag'])
for r in results
]
def languages(macrolanguage):
"""
Get a list of :class:`language_tags.Subtag.Subtag` objects given the string macrolanguage.
:param string macrolanguage: subtag macrolanguage.
:return: a list of the macrolanguage :class:`language_tags.Subtag.Subtag` objects.
:raise Exception: if the macrolanguage does not exists.
"""
results = []
macrolanguage = macrolanguage.lower()
macrolanguage_data = data.get('macrolanguage')
if macrolanguage not in macrolanguage_data:
raise Exception('\'' + macrolanguage +
'\' is not a macrolanguage.')
for registry_item in registry:
record = registry_item
if 'Macrolanguage' in record:
if record['Macrolanguage'] == macrolanguage:
results.append(Subtag(record['Subtag'], record['Type']))
return results
def subtags(self):
"""
Get the :class:`language_tags.Subtag.Subtag` objects of the tag.
:return: list of :class:`language_tags.Subtag.Subtag` objects that are part of the tag.
The return list can be empty.
"""
data = self.data
subtags = []
# if tag is grandfathered return no subtags
if 'record' in data and self.data['record']['Type'] == 'grandfathered':
return subtags
codes = data['tag'].split('-')
# Try and find the language tag.
for i, code in enumerate(codes):
# Singletons and anything after are unhandled.
if len(code) == 1:
#Stop the loop (stop processing after a singleton).
break
# Check for non-existent tag.
if code not in index:
continue
types = index[code]
# Language subtags may only appear at the beginning of the tag, otherwise the subtag type is indeterminate.
if 'language' in types and i == 0:
subtags.append(Subtag(code, 'language'))
continue
if len(code) == 2:
# Should be a region
if 'region' in types:
subtags.append(Subtag(code, 'region'))
# Error case: language subtag in the wrong place.
elif 'language' in types:
subtags.append(Subtag(code, 'language'))
elif len(code) == 3:
# Could be a numeric region code e.g. '001' for 'World'.
if 'region' in types:
subtags.append(Subtag(code, 'region'))
elif 'extlang' in types:
subtags.append(Subtag(code, 'extlang'))
# Error case: language subtag in the wrong place.
elif 'language' in types:
subtags.append(Subtag(code, 'language'))
elif len(code) == 4:
# Could be a numeric variant
if 'variant' in types:
subtags.append(Subtag(code, 'variant'))
elif 'script' in types:
subtags.append(Subtag(code, 'script'))
else:
# Should be a variant
if 'variant' in types:
subtags.append(Subtag(code, 'variant'))
return subtags
def test_comments(self):
self.assertEqual(
Subtag('YU', 'region').comments, ['see BA, HR, ME, MK, RS, or SI'])
def test_added(self):
self.assertEqual(Subtag('DD', 'region').added, '2005-10-16')
self.assertEqual(Subtag('DG', 'region').added, '2009-07-29')
def test_deprecated(self):
self.assertEqual(Subtag('DD', 'region').deprecated, '1990-10-30')
self.assertIsNone(Subtag('DE', 'region').deprecated)
def test_scope(self):
self.assertEqual(Subtag('zh', 'language').scope, 'macrolanguage')
self.assertEqual(Subtag('nah', 'language').scope, 'collection')
self.assertIsNone(Subtag('en', 'language').scope)
self.assertIsNone(Subtag('IQ', 'region').scope)
def test_type(self):
self.assertEqual(Subtag('zh', 'language').type, 'language')
self.assertEqual(Subtag('IQ', 'region').type, 'region')
def test_description(self):
self.assertEqual(Subtag('IQ', 'region').description, ['Iraq'])
self.assertEqual(
Subtag('vsv', 'extlang').description,
['Valencian Sign Language', 'Llengua de signes valenciana'])
def test_repr(self):
print(repr(Subtag('aae', 'language')))
self.assertIn('aae', repr(Subtag('aae', 'language')))
self.assertIn('Arbëreshë Albanian', repr(Subtag('aae', 'language')))
