def test_lovins_snowball(self):
"""Test abydos.stemmer.lovins (Snowball testset).
These test cases are from
https://github.com/snowballstem/snowball-data/tree/master/lovins
"""
# Snowball Lovins test set
with codecs.open(TESTDIR + '/corpora/snowball_lovins.csv',
encoding='utf-8') as snowball_testset:
next(snowball_testset)
for line in snowball_testset:
if line[0] != '#':
line = line.strip().split(',')
word, stem = line[0], line[1]
self.assertEqual(lovins(word), stem.lower())
def test_lovins_snowball(self):
"""test abydos.stemmer.lovins (Snowball testset)
These test cases are from
https://github.com/snowballstem/snowball-data/tree/master/lovins
"""
# Snowball Lovins test set
with codecs.open(TESTDIR+'/corpora/snowball_lovins.csv',
encoding='utf-8') as snowball_testset:
next(snowball_testset)
for line in snowball_testset:
if line[0] != '#':
line = line.strip().split(',')
word, stem = line[0], line[1]
self.assertEqual(lovins(word), stem.lower())
def test_lovins(self):
"""Test abydos.stemmer.Lovins."""
# base case
self.assertEqual(self.stmr.stem(''), '')
# test cases from Lovins' "Development of a Stemming Algorithm":
# http://www.mt-archive.info/MT-1968-Lovins.pdf
self.assertEqual(self.stmr.stem('magnesia'), 'magnes')
self.assertEqual(self.stmr.stem('magnesite'), 'magnes')
self.assertEqual(self.stmr.stem('magnesian'), 'magnes')
self.assertEqual(self.stmr.stem('magnesium'), 'magnes')
self.assertEqual(self.stmr.stem('magnet'), 'magnet')
self.assertEqual(self.stmr.stem('magnetic'), 'magnet')
self.assertEqual(self.stmr.stem('magneto'), 'magnet')
self.assertEqual(self.stmr.stem('magnetically'), 'magnet')
self.assertEqual(self.stmr.stem('magnetism'), 'magnet')
self.assertEqual(self.stmr.stem('magnetite'), 'magnet')
self.assertEqual(self.stmr.stem('magnetitic'), 'magnet')
self.assertEqual(self.stmr.stem('magnetizable'), 'magnet')
self.assertEqual(self.stmr.stem('magnetization'), 'magnet')
self.assertEqual(self.stmr.stem('magnetize'), 'magnet')
self.assertEqual(self.stmr.stem('magnetometer'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetometric'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetometry'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetomotive'), 'magnetomot')
self.assertEqual(self.stmr.stem('magnetron'), 'magnetron')
self.assertEqual(self.stmr.stem('metal'), 'metal')
self.assertEqual(self.stmr.stem('metall'), 'metal')
self.assertEqual(self.stmr.stem('metallically'), 'metal')
self.assertEqual(self.stmr.stem('metalliferous'), 'metallifer')
self.assertEqual(self.stmr.stem('metallize'), 'metal')
self.assertEqual(self.stmr.stem('metallurgical'), 'metallurg')
self.assertEqual(self.stmr.stem('metallurgy'), 'metallurg')
self.assertEqual(self.stmr.stem('induction'), 'induc')
self.assertEqual(self.stmr.stem('inductance'), 'induc')
self.assertEqual(self.stmr.stem('induced'), 'induc')
self.assertEqual(self.stmr.stem('angular'), 'angl')
self.assertEqual(self.stmr.stem('angle'), 'angl')
# missed branch test cases
self.assertEqual(self.stmr.stem('feminism'), 'fem')
# Test wrapper
self.assertEqual(lovins('magnesia'), 'magnes')
def test_lovins(self):
"""Test abydos.stemmer.Lovins."""
# base case
self.assertEqual(self.stmr.stem(''), '')
# test cases from Lovins' "Development of a Stemming Algorithm":
# http://www.mt-archive.info/MT-1968-Lovins.pdf
self.assertEqual(self.stmr.stem('magnesia'), 'magnes')
self.assertEqual(self.stmr.stem('magnesite'), 'magnes')
self.assertEqual(self.stmr.stem('magnesian'), 'magnes')
self.assertEqual(self.stmr.stem('magnesium'), 'magnes')
self.assertEqual(self.stmr.stem('magnet'), 'magnet')
self.assertEqual(self.stmr.stem('magnetic'), 'magnet')
self.assertEqual(self.stmr.stem('magneto'), 'magnet')
self.assertEqual(self.stmr.stem('magnetically'), 'magnet')
self.assertEqual(self.stmr.stem('magnetism'), 'magnet')
self.assertEqual(self.stmr.stem('magnetite'), 'magnet')
self.assertEqual(self.stmr.stem('magnetitic'), 'magnet')
self.assertEqual(self.stmr.stem('magnetizable'), 'magnet')
self.assertEqual(self.stmr.stem('magnetization'), 'magnet')
self.assertEqual(self.stmr.stem('magnetize'), 'magnet')
self.assertEqual(self.stmr.stem('magnetometer'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetometric'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetometry'), 'magnetometer')
self.assertEqual(self.stmr.stem('magnetomotive'), 'magnetomot')
self.assertEqual(self.stmr.stem('magnetron'), 'magnetron')
self.assertEqual(self.stmr.stem('metal'), 'metal')
self.assertEqual(self.stmr.stem('metall'), 'metal')
self.assertEqual(self.stmr.stem('metallically'), 'metal')
self.assertEqual(self.stmr.stem('metalliferous'), 'metallifer')
self.assertEqual(self.stmr.stem('metallize'), 'metal')
self.assertEqual(self.stmr.stem('metallurgical'), 'metallurg')
self.assertEqual(self.stmr.stem('metallurgy'), 'metallurg')
self.assertEqual(self.stmr.stem('induction'), 'induc')
self.assertEqual(self.stmr.stem('inductance'), 'induc')
self.assertEqual(self.stmr.stem('induced'), 'induc')
self.assertEqual(self.stmr.stem('angular'), 'angl')
self.assertEqual(self.stmr.stem('angle'), 'angl')
# missed branch test cases
self.assertEqual(self.stmr.stem('feminism'), 'fem')
# Test wrapper
self.assertEqual(lovins('magnesia'), 'magnes')