def test_candidates_white_space_and_underscores(self):
# Any white space and '_' can be used to separate keywords in the
# query string:
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('gatto_ sorride')[0][:2],
('😺', 'gatto che sorride'))
self.assertEqual(
mq.candidates('nerd glasses')[0][:2],
('🤓', 'nerd face'))
self.assertEqual(
mq.candidates('smiling face sun glasses')[0][:2],
('😎', 'smiling face with sunglasses'))
def test_candidates_some_letters(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('a')[0][:2],
('🅰\ufe0f', 'negative squared latin capital letter a'))
self.assertEqual(
mq.candidates('squared a')[0][:2],
('🅰\ufe0f', 'negative squared latin capital letter a'))
self.assertEqual(
mq.candidates('squared capital a')[0][:2],
('🅰\ufe0f', 'negative squared latin capital letter a'))
self.assertEqual(
mq.candidates('c')[0][:2],
('©️', 'copyright sign'))
def test_candidates_skin_tones(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('man tone5')[0][:2],
('👨🏿', 'man: dark skin tone “man tone5”'))
self.assertEqual(
mq.candidates('skin tone')[0][:2],
('🏻', 'emoji modifier fitzpatrick type-1-2 “tone1” [skin tone]'))
self.assertEqual(
mq.candidates('tone1')[0][:2],
('🏻', 'emoji modifier fitzpatrick type-1-2 “tone1”'))
self.assertEqual(
mq.candidates('tone5')[0][:2],
('🏿', 'emoji modifier fitzpatrick type-6 “tone5”'))
def test_candidates_animals(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('animal')[0][:2],
('🐵', 'cara de mono [animal]'))
self.assertEqual(
mq.candidates('dromedary animal')[0][:2],
('🐪', 'dromedary camel'))
self.assertEqual(
mq.candidates('camel')[0][:2],
('🐫', 'bactrian camel'))
self.assertEqual(
mq.candidates('nature')[0][:2],
('🙈', 'see-no-evil monkey {nature}'))
def test_candidates_code_point_input(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages = ['fr_FR'])
self.assertEqual(
mq.candidates('2019'),
[('’', 'U+2019 apostrophe droite', 2000)])
self.assertEqual(
mq.candidates('41'),
[('A', 'U+41 latin capital letter a', 2000)])
self.assertEqual(
mq.candidates('2a'),
[('*', 'U+2A astérisque', 2000)])
self.assertEqual(
mq.candidates('1b'),
[('\x1b', 'U+1B', 2000), ('🧔🏻\u200d♂️', 'man: light skin tone, beard', 44), ('🧔🏻\u200d♀️', 'woman: light skin tone, beard', 44), ('🧑🏻\u200d🦲', 'person: light skin tone, bald', 44)])
def test_candidates_code_point_input(self):
mq = itb_emoji.EmojiMatcher(
languages = ['fr_FR'])
self.assertEqual(
mq.candidates('2019'),
[('’', 'U+2019 right single quotation mark', 200)])
self.assertEqual(
mq.candidates('41'),
[('A', 'U+41 latin capital letter a', 200)])
self.assertEqual(
mq.candidates('2a'),
[('*', 'U+2A asterisk', 200)])
self.assertEqual(
mq.candidates('1b'),
[('\x1b', 'U+1B', 200)])
def test_candidates_symbols(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('symbol')[0][:2],
('🔣', 'input symbol for symbols {Symbol}'))
self.assertEqual(
mq.candidates('atomsymbol')[0][:2],
('⚛\ufe0f', 'atom symbol'))
self.assertEqual(
mq.candidates('peacesymbol')[0][:2],
('☮\ufe0f', 'peace symbol'))
self.assertEqual(
mq.candidates('peace symbol')[0][:2],
('☮\ufe0f', 'peace symbol {Symbol}'))
def test_candidates_skin_tones(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages = ['ja_JP'])#, 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('man tone5')[0][:2],
('👨🏿', 'man: dark skin tone “man tone5”'))
self.assertEqual(
mq.candidates('skin tone')[0][:2],
('🧑🏾\u200d🤝\u200d🧑🏼', 'people holding hands: medium-dark skin tone, medium-light skin tone “people holding hands medium dark skin tone medium light skin tone”'))
self.assertEqual(
mq.candidates('tone1')[0][:2],
('🏻', 'emoji modifier fitzpatrick type-1-2 “tone1”'))
self.assertEqual(
mq.candidates('tone5')[0][:2],
('🏿', 'emoji modifier fitzpatrick type-6 “tone5”'))
def test_similar_white_smiling_face_en_US(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages=['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(mq.similar(
'☺', match_limit=5
), [(
'☺️',
'white smiling face [☺️, So, people, face, outlined, relaxed, smile, uc1, smiling face]',
9),
('😙', 'kissing face with smiling eyes [So, people, face, smile]',
4),
('😍',
'smiling face with heart-shaped eyes [So, people, face, smile]',
4),
('😋', 'face savouring delicious food [So, people, face, smile]',
4), ('😇', 'smiling face with halo [So, people, face, smile]', 4)])
def test_candidates_pykakasi_missing_ja_JP(self) -> None:
mq = itb_emoji.EmojiMatcher(languages=['ja_JP'])
self.assertEqual(mq.similar('🐤', match_limit=5),
[('🐤', 'ひよこ [🐤, ひな, ひよこ, 動物, 横を向いているひよこ, 顔, 鳥]', 7),
('🐣', '卵からかえったひよこ [ひな, ひよこ, 動物, 顔, 鳥]', 5),
('🐥', '前を向いているひよこ [ひな, ひよこ, 動物, 鳥]', 4),
('🐦', '鳥 [動物, 顔, 鳥]', 3),
('🐔', 'にわとり [動物, 顔, 鳥]', 3)])
self.assertEqual(mq.similar('🐌', match_limit=5),
[('🐌', 'かたつむり [🐌, かたつむり, でんでん虫, 虫]', 4),
('🦋', 'チョウ [虫]', 1), ('🐛', '毛虫 [虫]', 1),
('🐜', 'アリ [虫]', 1), ('🐝', 'ミツバチ [虫]', 1)])
self.assertEqual(mq.similar('😱', match_limit=5),
[('😱', '恐怖 [😱, がーん, ショック, 叫び, 恐怖, 顔]', 6),
('🙀', '絶望する猫 [がーん, ショック, 顔]', 3),
('😨', '青ざめ [がーん, 顔]', 2), ('🤯', '頭爆発 [ショック, 顔]', 2),
('😭', '大泣き [顔]', 1)])
def test_candidates_various_unicode_chars(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('euro sign')[0][:2],
('€', 'euro sign'))
self.assertEqual(
mq.candidates('superscript one')[0][:2],
('¹', 'superscript one'))
self.assertEqual(
mq.candidates('currency')[0][:2],
('₳', 'austral sign {Currency} [currency]'))
self.assertEqual(
mq.candidates('connector')[0][:2],
('﹎', 'centreline low line {Connector}'))
self.assertEqual(
mq.candidates('dash')[0][:2],
('💨', 'dash symbol'))
self.assertEqual(
mq.candidates('close')[0][:2],
('〉', 'right angle bracket “close angle bracket” {Close}'))
self.assertEqual(
mq.candidates('punctuation')[0][:2],
('‼\ufe0f', 'double exclamation mark {Punctuation} [punctuation]'))
self.assertEqual(
mq.candidates('final quote')[0][:2],
('’', 'right single quotation mark {Final quote}'))
self.assertEqual(
mq.candidates('initial quote')[0][:2],
('‘', 'left single quotation mark {Initial quote}'))
self.assertEqual(
mq.candidates('modifier')[0][:2],
('🏻', 'emoji modifier fitzpatrick type-1-2 {Modifier}'))
self.assertEqual(
mq.candidates('math')[0][:2],
('𝜵', 'mathematical bold italic nabla {Math}'))
self.assertEqual(
mq.candidates('separator line')[0][:2],
('
', 'U+2028 line separator {Line}'))
self.assertEqual(
mq.candidates('separator paragraph')[0][:2],
('
', 'U+2029 paragraph separator {Paragraph}'))
self.assertEqual(
mq.candidates('separator space')[0][:2],
(' ', 'U+20 space {Space}'))
def test_candidates_flags(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages = ['en_US'])
self.assertEqual(
mq.candidates('us')[0][:2],
('🇺🇸', 'flag: united states “us”'))
self.assertEqual(
mq.candidates('flag us')[0][:2],
('🇺🇸', 'flag: united states “us”'))
self.assertEqual(
mq.candidates('united nations')[0][:2],
('🇺🇳', 'flag: united nations'))
self.assertEqual(
mq.candidates('united')[0][:2],
('🇺🇳', 'flag: united nations'))
self.assertEqual(
mq.candidates('outlying islands')[0][:2],
('🇺🇲', 'flag: u.s. outlying islands'))
self.assertEqual(
mq.candidates('flag united arab')[0][:2],
('🇦🇪', 'flag: united arab emirates “flag ae”'))
self.assertEqual(
mq.candidates('mm')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “mm”'))
self.assertEqual(
mq.candidates('flag mm')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “mm”'))
self.assertEqual(
mq.candidates('myanmar')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “flag: myanmar burma”'))
self.assertEqual(
mq.candidates('sj')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “sj”'))
self.assertEqual(
mq.candidates('flag sj')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “sj”'))
self.assertEqual(
mq.candidates('svalbard')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “flag: svalbard & jan mayen”'))
self.assertEqual(
mq.candidates('jan mayen')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “flag: svalbard & jan mayen”'))
self.assertEqual(
mq.candidates('mayen')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “flag: svalbard & jan mayen”'))
def test_candidates_flags(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('us')[0][:2],
('🇺🇸', 'flag: united states “us”'))
self.assertEqual(
mq.candidates('flag us')[0][:2],
('🇺🇸', 'flag: united states “us”'))
self.assertEqual(
mq.candidates('united states')[0][:2],
('🇺🇸', 'flag: united states'))
self.assertEqual(
mq.candidates('united')[0][:2],
('🇦🇪', 'flag: united arab emirates'))
self.assertEqual(
mq.candidates('united minor outlying islands')[0][:2],
('🇺🇲', 'flag: u.s. outlying islands'))
self.assertEqual(
mq.candidates('united arab')[0][:2],
('🇦🇪', 'flag: united arab emirates'))
self.assertEqual(
mq.candidates('mm')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “mm”'))
self.assertEqual(
mq.candidates('flag mm')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “mm”'))
self.assertEqual(
mq.candidates('myanmar')[0][:2],
('🇲🇲', 'flag: myanmar (burma) “myanmar burma”'))
self.assertEqual(
mq.candidates('sj')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “sj”'))
self.assertEqual(
mq.candidates('flag sj')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen “sj”'))
self.assertEqual(
mq.candidates('svalbard')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen'))
self.assertEqual(
mq.candidates('jan mayen')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen'))
self.assertEqual(
mq.candidates('mayen')[0][:2],
('🇸🇯', 'flag: svalbard & jan mayen'))
def test_candidates_spellchecking(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages = ['en_US'])
self.assertEqual(
('\U0001f98b', 'butterfly'),
mq.candidates('buterfly')[0][:2])
self.assertEqual(
('🏸', 'badminton racquet and shuttlecock'),
mq.candidates('badminton')[0][:2])
self.assertEqual(
('🏸', 'badminton racquet and shuttlecock'),
mq.candidates('badmynton')[0][:2])
self.assertEqual(
('🏸', 'badminton racquet and shuttlecock'),
mq.candidates('padminton')[0][:2])
self.assertEqual(
('🦔', 'hedgehog'),
mq.candidates('hedgehgo')[0][:2])
def test_candidates_spellchecking(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('buterfly')[0][:2],
('\U0001f98b', 'butterfly'))
self.assertEqual(
mq.candidates('badminton')[0][:2],
('🏸', 'badminton racquet and shuttlecock'))
self.assertEqual(
mq.candidates('badmynton')[0][:2],
('🏸', 'badminton racquet and shuttlecock'))
self.assertEqual(
mq.candidates('padminton')[0][:2],
('🏸', 'badminton racquet and shuttlecock'))
self.assertEqual(
mq.candidates('fery')[0][:2],
('⛴\ufe0f', 'ferry'))
def test_candidates_pykakasi_available_ja_JP(self):
mq = itb_emoji.EmojiMatcher(
languages = ['ja_JP'])
self.assertEqual(
mq.candidates('katatsumuri')[0][:2],
('🐌', 'かたつむり “katatsumuri”'))
self.assertEqual(
mq.candidates('かたつむり')[0][:2],
('🐌', 'かたつむり'))
self.assertEqual(
mq.candidates('かたつむり_')[0][:2],
('🐌', 'かたつむり'))
self.assertEqual(
mq.candidates('かたつむり＿')[0][:2],
('🐌', 'かたつむり'))
self.assertEqual(
0, len(mq.candidates('カタツムリ')))
self.assertEqual(
0, len(mq.candidates('カタツムリ_')))
self.assertEqual(
0, len(mq.candidates('カタツムリ＿')))
self.assertEqual(
mq.candidates('ネコ')[0][:2],
('🐈', 'ネコ'))
self.assertEqual(
mq.candidates('ネコ_')[0][:2],
('🐈', 'ネコ'))
self.assertEqual(
mq.candidates('ネコ＿')[0][:2],
('🐈', 'ネコ'))
self.assertEqual(
mq.candidates('ねこ')[0][:2],
('🐈', 'ネコ “ねこ”'))
self.assertEqual(
mq.candidates('ねこ_')[0][:2],
('🐈', 'ネコ “ねこ”'))
self.assertEqual(
mq.candidates('ねこ＿')[0][:2],
('🐈', 'ネコ “ねこ”'))
def test_candidates_multilingual(self):
mq = itb_emoji.EmojiMatcher(
languages = ['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(
mq.candidates('ant')[0][:2],
('🐜', 'ant'))
self.assertEqual(
mq.candidates('ameise')[0][:2],
('🐜', 'Ameise'))
self.assertEqual(
mq.candidates('Ameise')[0][:2],
('🐜', 'Ameise'))
self.assertEqual(
mq.candidates('formica')[0][:2],
('🐜', 'formica'))
self.assertEqual(
mq.candidates('hormiga')[0][:2],
('🐜', 'hormiga'))
self.assertEqual(
mq.candidates('cacca')[0][:2],
('💩', 'cacca'))
self.assertEqual(
mq.candidates('orso')[0][:2],
('🐻', 'orso'))
self.assertEqual(
mq.candidates('lupo')[0][:2],
('🐺', 'lupo'))
self.assertEqual(
mq.candidates('gatto')[0][:2],
('🐈', 'gatto'))
self.assertEqual(
mq.candidates('gatto sorride')[0][:2],
('😺', 'gatto che sorride'))
self.assertEqual(
mq.candidates('halo')[0][:2],
('😇', 'smiling face with halo'))
self.assertEqual(
mq.candidates('factory')[0][:2],
('🏭', 'factory'))
def test_similar_query_is_not_an_emoji(self) -> None:
mq = itb_emoji.EmojiMatcher(
languages=['en_US', 'it_IT', 'es_MX', 'es_ES', 'de_DE', 'ja_JP'])
self.assertEqual(mq.similar('this is not an emoji', match_limit=5), [])
def test_similar_horse_racing_pinyin_available_zh_TW(self) -> None:
mq = itb_emoji.EmojiMatcher(languages=['zh_TW'])
self.assertEqual(mq.similar('🏇', match_limit=1),
[('🏇', '賽馬 [🏇, 賽馬, sàimǎ, 騎馬, qímǎ]', 5)])
def test_similar_horse_racing_pinyin_missing_zh_TW(self) -> None:
mq = itb_emoji.EmojiMatcher(languages=['zh_TW'])
self.assertEqual(mq.similar('🏇', match_limit=1),
[('🏇', '賽馬 [🏇, 賽馬, 騎馬]', 3)])
def test_similar_horse_racing_pinyin_available_zh_CN(self) -> None:
mq = itb_emoji.EmojiMatcher(languages=['zh_CN'])
self.assertEqual(mq.similar('🏇', match_limit=3),
[('🏇', '赛马 [🏇, 赛马, sàimǎ, 马, mǎ]', 5),
('🏇🏻', '赛马: 较浅肤色 [赛马, sàimǎ, 马, mǎ]', 4),
('🏇🏼', '赛马: 中等-浅肤色 [赛马, sàimǎ, 马, mǎ]', 4)])
def test_similar_horse_racing_pinyin_missing_zh_CN(self) -> None:
mq = itb_emoji.EmojiMatcher(languages=['zh_CN'])
self.assertEqual(mq.similar('🏇', match_limit=3),
[('🏇', '赛马 [🏇, 赛马, 马]', 3),
('🏇🏻', '赛马: 较浅肤色 [赛马, 马]', 2),
('🏇🏼', '赛马: 中等-浅肤色 [赛马, 马]', 2)])
def test_unicode_version_emoji_data_file(self):
mq = itb_emoji.EmojiMatcher(languages=['en_US'])
self.assertEqual(mq.unicode_version('☺'), '2.0')
self.assertEqual(mq.unicode_version('🤿'), '12.1')
self.assertEqual(mq.unicode_version('⚧'), '13.0')
def test_candidates_empty_query(self):
mq = itb_emoji.EmojiMatcher(
languages = ['fr_FR'])
self.assertEqual(mq.candidates(''), [])
def test_candidates_french_similar(self):
mq = itb_emoji.EmojiMatcher(
languages = ['fr_FR'])
self.assertEqual(
mq.candidates('🤔', match_limit = 3),
[('🤔', 'visage en pleine réflexion [🤔, réflexion, visage, visage en pleine réflexion]', 4), ('🤐', 'visage avec bouche fermeture éclair [visage]', 1), ('🤗', 'visage qui fait un câlin [visage]', 1)])
