def test_fuzzy_soundex(self):
"""Test abydos.phonetic.FuzzySoundex."""
self.assertEqual(self.pa.encode(''), '00000')
# http://wayback.archive.org/web/20100629121128/http://www.ir.iit.edu/publications/downloads/IEEESoundexV5.pdf
self.assertEqual(self.pa.encode('Kristen'), 'K6935')
self.assertEqual(self.pa.encode('Krissy'), 'K6900')
self.assertEqual(self.pa.encode('Christen'), 'K6935')
# http://books.google.com/books?id=LZrT6eWf9NMC&lpg=PA76&ots=Tex3FqNwGP&dq=%22phonix%20algorithm%22&pg=PA75#v=onepage&q=%22phonix%20algorithm%22&f=false
self.assertEqual(self.pa_4.encode('peter'), 'P360')
self.assertEqual(self.pa_4.encode('pete'), 'P300')
self.assertEqual(self.pa_4.encode('pedro'), 'P360')
self.assertEqual(self.pa_4.encode('stephen'), 'S315')
self.assertEqual(self.pa_4.encode('steve'), 'S310')
self.assertEqual(self.pa_4.encode('smith'), 'S530')
self.assertEqual(self.pa_4.encode('smythe'), 'S530')
self.assertEqual(self.pa_4.encode('gail'), 'G400')
self.assertEqual(self.pa_4.encode('gayle'), 'G400')
self.assertEqual(self.pa_4.encode('christine'), 'K693')
self.assertEqual(self.pa_4.encode('christina'), 'K693')
self.assertEqual(self.pa_4.encode('kristina'), 'K693')
# etc. (for code coverage)
self.assertEqual(self.pa.encode('Wight'), 'W3000')
self.assertEqual(self.pa.encode('Hardt'), 'H6000')
self.assertEqual(self.pa.encode('Knight'), 'N3000')
self.assertEqual(self.pa.encode('Czech'), 'S7000')
self.assertEqual(self.pa.encode('Tsech'), 'S7000')
self.assertEqual(self.pa.encode('gnomic'), 'N5900')
self.assertEqual(self.pa.encode('Wright'), 'R3000')
self.assertEqual(self.pa.encode('Hrothgar'), 'R3760')
self.assertEqual(self.pa.encode('Hwaet'), 'W3000')
self.assertEqual(self.pa.encode('Grant'), 'G6300')
self.assertEqual(self.pa.encode('Hart'), 'H6000')
self.assertEqual(self.pa.encode('Hardt'), 'H6000')
# max_length bounds tests
self.assertEqual(
FuzzySoundex(max_length=-1).encode('Niall'),
'N400000000000000000000000000000000000000000000000000000000000000',
)
self.assertEqual(FuzzySoundex(max_length=0).encode('Niall'), 'N400')
# zero_pad tests
self.assertEqual(
FuzzySoundex(max_length=-1, zero_pad=False).encode('Niall'), 'N4')
self.assertEqual(
FuzzySoundex(max_length=0, zero_pad=False).encode('Niall'), 'N4')
self.assertEqual(
FuzzySoundex(max_length=0, zero_pad=True).encode('Niall'), 'N400')
self.assertEqual(
FuzzySoundex(max_length=4, zero_pad=False).encode(''), '0')
self.assertEqual(
FuzzySoundex(max_length=4, zero_pad=True).encode(''), '0000')
# encode_alpha
self.assertEqual(self.pa.encode_alpha('pete'), 'PT')
self.assertEqual(self.pa.encode_alpha('pedro'), 'PTR')
self.assertEqual(self.pa.encode_alpha('stephen'), 'STPN')
self.assertEqual(self.pa.encode_alpha('steve'), 'STP')
# Test wrapper
self.assertEqual(fuzzy_soundex('Kristen'), 'K6935')
'davidson':
Davidson().encode,
'dolby':
Dolby().encode,
'dolby_ml4':
Dolby(max_length=4).encode,
'dolby_vowels':
Dolby(keep_vowels=True).encode,
'double_metaphone':
lambda _: ', '.join(double_metaphone.encode(_)),
'eudex':
Eudex().encode,
'fonem':
FONEM().encode,
'fuzzy_soundex':
FuzzySoundex().encode,
'fuzzy_soundex_0pad_ml8':
FuzzySoundex(max_length=8, zero_pad=True).encode,
'haase_phonetik':
lambda _: ', '.join(haase.encode(_)),
'haase_phonetik_primary':
lambda _: haase_primary.encode(_)[0],
'henry_early':
HenryEarly().encode,
'henry_early_ml8':
HenryEarly(max_length=8).encode,
'koelner_phonetik':
koelner.encode,
'koelner_phonetik_num_to_alpha': (
lambda _: koelner._to_alpha(koelner.encode(_)) # noqa: SF01
),
'bmpm_gen_exact': BeiderMorse(match_mode='exact').encode, 'bmpm_ash_approx': BeiderMorse(name_mode='ash').encode, 'bmpm_ash_exact': BeiderMorse(name_mode='ash', match_mode='exact').encode, 'bmpm_sep_approx': BeiderMorse(name_mode='sep').encode, 'bmpm_sep_exact': BeiderMorse(name_mode='sep', match_mode='exact').encode, 'caverphone_1': Caverphone(version=1).encode, 'caverphone_2': Caverphone().encode, 'daitch_mokotoff_soundex': DaitchMokotoff().encode, 'davidson': Davidson().encode, 'dolby': Dolby().encode, 'dolby_ml4': Dolby(max_length=4).encode, 'dolby_vowels': Dolby(keep_vowels=True).encode, 'double_metaphone': DoubleMetaphone().encode, 'eudex': Eudex().encode, 'fonem': FONEM().encode, 'fuzzy_soundex': FuzzySoundex().encode, 'fuzzy_soundex_0pad_ml8': FuzzySoundex(max_length=8, zero_pad=True).encode, 'haase_phonetik': Haase().encode, 'haase_phonetik_primary': Haase(primary_only=True).encode, 'henry_early': HenryEarly().encode, 'henry_early_ml8': HenryEarly(max_length=8).encode, 'koelner_phonetik': Koelner().encode, 'koelner_phonetik_alpha': Koelner().encode_alpha, 'lein': LEIN().encode, 'lein_nopad_ml8': LEIN(max_length=8, zero_pad=False).encode, 'metasoundex': MetaSoundex().encode, 'metasoundex_es': MetaSoundex(lang='es').encode, 'metaphone': Metaphone().encode, 'mra': MRA().encode, 'norphone': Norphone().encode, 'nrl': NRL().encode,
SpanishMetaphone,
StatisticsCanada,
)
from . import EXTREME_TEST, _corpus_file, _fuzz, _random_char
alpha_sis = AlphaSIS()
bm = BeiderMorse()
caverphone = Caverphone()
davidson = Davidson()
dm = DaitchMokotoff()
dolby = Dolby()
double_metaphone = DoubleMetaphone()
eudex = Eudex()
fonem = FONEM()
fuzzy_soundex = FuzzySoundex()
haase = Haase()
henry_early = HenryEarly()
koelner = Koelner()
lein = Lein()
metaphone = Metaphone()
metasoundex = MetaSoundex()
mra = MRA()
norphone = Norphone()
nrl = NRL()
nysiis = NYSIIS()
onca = ONCA()
parmar_kumbharana = ParmarKumbharana()
phonem = Phonem()
phonet = Phonet()
phonetic_spanish = PhoneticSpanish()