def test_errors_multiple_script_subtags(self):
tag = Tag('mt-Arab-Arab')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_SCRIPT)
self.assertEqual(err.message, 'Extra script subtag \'Arab\' found.')
tag = Tag('en-Cyrl-Latn')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_SCRIPT)
self.assertEqual(err.message, 'Extra script subtag \'Latn\' found.')
# First error should be regarding suppress-script, second should be regarding extra script.
tag = Tag('en-Latn-Cyrl')
errs = tag.errors
self.assertEqual(len(errs), 2)
err = errs[0]
self.assertEqual(err.code, tag.ERR_SUPPRESS_SCRIPT)
self.assertEqual(
err.message,
'The script subtag \'Latn\' is the same as the language suppress-script.'
)
err = errs[1]
self.assertEqual(err.code, tag.ERR_EXTRA_SCRIPT)
self.assertEqual(err.message, 'Extra script subtag \'Cyrl\' found.')
def test_perferred(self):
tag = Tag('zh-cmn-Hant')
self.assertEqual(tag.type, 'redundant')
self.assertIsNotNone(tag.deprecated)
self.assertIsNotNone(tag.preferred)
self.assertEqual(tag.preferred.format, 'cmn-Hant')
self.assertIsNone(Tag('cmn-Hant').preferred)
def test_subtags_correct_type(self):
tag = Tag('aa')
subtags = tag.subtags
self.assertEqual(len(subtags), 1)
self.assertEqual(subtags[0].type, 'language')
self.assertEqual(subtags[0].format, 'aa')
# Lowercase - lookup should be case insensitive.
tag = Tag('en-mt')
subtags = tag.subtags
self.assertEqual(len(subtags), 2)
self.assertEqual(subtags[0].type, 'language')
self.assertEqual(subtags[0].format, 'en')
self.assertEqual(subtags[1].type, 'region')
self.assertEqual(subtags[1].format, 'MT')
tag = Tag('en-mt-arab')
subtags = tag.subtags
self.assertEqual(len(subtags), 3)
self.assertEqual(subtags[0].type, 'language')
self.assertEqual(subtags[0].format, 'en')
self.assertEqual(subtags[1].type, 'region')
self.assertEqual(subtags[1].format, 'MT')
self.assertEqual(subtags[2].type, 'script')
self.assertEqual(subtags[2].format, 'Arab')
def test_valid_tag(self):
self.assertTrue(Tag('en').valid)
self.assertTrue(Tag('en-GB').valid)
self.assertTrue(Tag('gsw').valid)
self.assertTrue(Tag('de-CH').valid)
# returns true for subtag followed by private tag
self.assertTrue(Tag('en-x-whatever').valid)
def test_description(self):
tag = Tag('en-GB-oed')
self.assertEqual(tag.type, 'grandfathered')
self.assertEqual('2015-04-17', tag.deprecated)
self.assertEqual(tag.descriptions,
['English, Oxford English Dictionary spelling'])
self.assertEqual(Tag('en').descriptions, [])
def test_only_existent_subtags(self):
tag = Tag('hello')
self.assertEqual(tag.subtags, [])
tag = Tag('en-hello')
subtags = tag.subtags
self.assertEqual(len(subtags), 1)
self.assertEqual(subtags[0].type, 'language')
self.assertEqual(subtags[0].format, 'en')
def test_valid_tag_redundant(self):
# Redundant but not deprecated, therefore valid.
tag = Tag('zh-Hans')
self.assertEqual(tag.type, 'redundant')
self.assertIsNone(tag.deprecated)
self.assertTrue(tag.valid)
tag = Tag('es-419')
self.assertEqual(tag.type, 'redundant')
self.assertIsNone(tag.deprecated)
self.assertTrue(tag.valid)
def test_non_valid_deprecated_redundant(self):
# Redundant and deprecated, therefore invalid.
tag = Tag('zh-cmn')
self.assertEqual(tag.type, 'redundant')
self.assertIsNotNone(tag.deprecated)
self.assertFalse(tag.valid)
tag = Tag('zh-cmn-Hans')
self.assertEqual(tag.type, 'redundant')
self.assertIsNotNone(tag.deprecated)
self.assertFalse(tag.valid)
def test_errors_length_private_use_subtags(self):
tag = Tag('en-x-more-than-eight-chars')
errs = tag.errors
self.assertEqual(len(errs), 0)
tag = Tag('en-x-morethaneightchars')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_TOO_LONG)
self.assertEqual(
err.message,
'The private-use subtag \'morethaneightchars\' is too long.')
def test_errors_multiple_region_subtags(self):
tag = Tag('en-GB-GB')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_REGION)
self.assertEqual(err.message, 'Extra region subtag \'GB\' found.')
tag = Tag('ko-mt-mt')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_REGION)
self.assertEqual(err.message, 'Extra region subtag \'MT\' found.')
def test_format(self):
self.assertEqual(Tag('en').format, 'en')
self.assertEqual(
Tag('en-x-more-than-eight-chars').format,
'en-x-more-than-eight-chars')
self.assertEqual(Tag('En').format, 'en')
self.assertEqual(Tag('EN').format, 'en')
self.assertEqual(Tag('eN').format, 'en')
self.assertEqual(Tag('en-gb').format, 'en-GB')
self.assertEqual(Tag('en-gb-oed').format, 'en-GB-oed')
self.assertEqual(Tag('az-latn').format, 'az-Latn')
self.assertEqual(Tag('ZH-hant-hK').format, 'zh-Hant-HK')
def test_errors_front_no_language_tag(self):
tag = Tag('GB-en')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_NO_LANGUAGE)
self.assertEqual(err.message, 'Missing language tag in \'gb-en\'.')
def test_errors_deprecated_subtags(self):
# Moldovan (mo) is deprecated as a language.
tag = Tag('mo')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_SUBTAG_DEPRECATED)
self.assertEqual(err.message, 'The subtag \'mo\' is deprecated.')
# Neutral Zone (NT) is deprecated as a region.
tag = Tag('en-NT')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_SUBTAG_DEPRECATED)
self.assertEqual(err.message, 'The subtag \'NT\' is deprecated.')
def test_errors_multiple_extlang_subtags(self):
tag = Tag('en-asp-bog')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_EXTLANG)
self.assertEqual(err.message, 'Extra extlang subtag \'bog\' found.')
def test_subtags_of_private_tags(self):
tag = Tag('en-GB-x-Beano')
subtags = tag.subtags
self.assertEqual(len(subtags), 2)
self.assertEqual(subtags[0].type, 'language')
self.assertEqual(subtags[0].format, 'en')
self.assertEqual(subtags[1].type, 'region')
self.assertEqual(subtags[1].format, 'GB')
def check(tag):
"""
Check if a string (hyphen-separated) tag is valid.
:param str tag: (hyphen-separated) tag.
:return: bool -- True if valid.
"""
return Tag(tag).valid
def tag(tag):
"""
Get a :class:`language_tags.Tag.Tag` of a string (hyphen-separated) tag.
:param str tag: (hyphen-separated) tag.
:return: :class:`language_tags.Tag.Tag`.
"""
return Tag(tag)
def test_errors_duplicate_variant_subtags(self):
tag = Tag('ca-valencia-valencia')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_DUPLICATE_VARIANT)
self.assertEqual(err.message,
'Duplicate variant subtag \'valencia\' found.')
def test_errors_wrong_order_subtags(self):
tag = Tag('mt-MT-Arab')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_WRONG_ORDER)
self.assertEqual(
err.message,
'The subtag \'MT\' should not appear before \'Arab\'.')
def test_errors_supress_script(self):
tag = Tag('gsw-Latn')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_SUPPRESS_SCRIPT)
self.assertEqual(
err.message,
'The script subtag \'Latn\' is the same as the language suppress-script.'
)
tag = Tag('en-Latn-GB')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_SUPPRESS_SCRIPT)
self.assertEqual(
err.message,
'The script subtag \'Latn\' is the same as the language suppress-script.'
)
def test_errors_deprecated_redundant(self):
# Redundant and deprecated, therefore invalid.
tag = Tag('zh-cmn')
self.assertEqual(tag.type, 'redundant')
self.assertIsNotNone(tag.deprecated)
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_DEPRECATED)
self.assertEqual(err.tag, 'zh-cmn')
self.assertEqual(err.message,
'The tag zh-cmn is deprecated. Use \'cmn\' instead.')
def test_errors_multiple_language_subtags(self):
tag = Tag('en-en')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_LANGUAGE)
self.assertEqual(err.message, 'Extra language subtag \'en\' found.')
tag = Tag('en-en-GB')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_LANGUAGE)
self.assertEqual(err.message, 'Extra language subtag \'en\' found.')
tag = Tag('ko-en')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_EXTRA_LANGUAGE)
self.assertEqual(err.message, 'Extra language subtag \'en\' found.')
def test_deprecated(self):
# Redundant and deprecated.
tag = Tag('zh-cmn-Hant')
self.assertEqual(tag.type, 'redundant')
self.assertEqual(tag.deprecated, '2009-07-29')
# Redundant but not deprecated.
tag = Tag('zh-Hans')
self.assertEqual(tag.type, 'redundant')
self.assertIsNone(tag.deprecated)
# Grandfathered and deprecated.
tag = Tag('zh-xiang')
self.assertEqual(tag.type, 'grandfathered')
self.assertEqual(tag.deprecated, '2009-07-29')
# Grandfathered but not deprecated.
tag = Tag('i-default')
self.assertEqual(tag.type, 'grandfathered')
self.assertIsNone(tag.deprecated)
self.assertIsNone(Tag('en').deprecated)
def test_errors_no_language_tag(self):
# test with empty tag.
tag = Tag('')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_NO_LANGUAGE)
self.assertEqual(err.message, 'Empty tag.')
tag = Tag('IQ-Arab')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_NO_LANGUAGE)
self.assertEqual(err.message, 'Missing language tag in \'iq-arab\'.')
tag = Tag('419')
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_NO_LANGUAGE)
self.assertEqual(err.message, 'Missing language tag in \'419\'.')
def description(tag):
"""
Gets a list of descriptions given the tag.
:param str tag: (hyphen-separated) tag.
:return: list of string descriptions. The return list can be empty.
"""
tag_object = Tag(tag)
results = tag_object.descriptions
subtags = tag_object.subtags
for subtag in subtags:
results += subtag.description
return results
def test_added(self):
# Redundant and deprecated.
tag = Tag('zh-cmn-Hant')
self.assertEqual(tag.type, 'redundant')
self.assertEqual(tag.added, '2005-07-15')
# Redundant but not deprecated.
tag = Tag('zh-Hans')
self.assertEqual(tag.type, 'redundant')
self.assertIsNone(tag.deprecated)
self.assertEqual(tag.added, '2003-05-30')
# Grandfathered and deprecated.
tag = Tag('zh-xiang')
self.assertEqual(tag.type, 'grandfathered')
self.assertEqual(tag.added, '1999-12-18')
# Grandfathered but not deprecated.
tag = Tag('i-default')
self.assertEqual(tag.type, 'grandfathered')
self.assertIsNone(tag.deprecated)
self.assertEqual(tag.added, '1998-03-10')
self.assertIsNone(Tag('en').added)
def test_errors_deprecated_grandfathered(self):
# Grandfathered and deprecated, therefore invalid.
tag = Tag('art-lojban')
self.assertEqual(tag.type, 'grandfathered')
self.assertIsNotNone(tag.deprecated)
errs = tag.errors
self.assertEqual(len(errs), 1)
err = errs[0]
self.assertEqual(err.code, tag.ERR_DEPRECATED)
self.assertEqual(err.tag, 'art-lojban')
self.assertEqual(
err.message,
'The tag art-lojban is deprecated. Use \'jbo\' instead.')
self.assertIn(str(err.code), err.__str__())
self.assertIn(err.message, err.__str__())
self.assertIn(err.tag, err.__str__())
self.assertIn(str(err.subtag), err.__str__())
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 test_script_subtags(self):
self.assertEqual(Tag('en-mt-arab').script.format, 'Arab')
self.assertIsNone(Tag('en-mt').script)
def test_region_subtags(self):
self.assertEqual(Tag('nl-BE').region.format, 'BE')
self.assertIsNone(Tag('nl').region)
