def test_soundex_couples(self):
blocking = SoundexBlocking(ref_attr_index=1, target_attr_index=1,
language='english')
blocking.fit(SOUNDEX_REFSET, SOUNDEX_TARGETSET)
pairs = list(blocking.iter_id_pairs())
self.assertEqual(len(pairs), 8)
for pair in SOUNDEX_PAIRS:
self.assertIn(pair, pairs)
def test_soundex_blocks(self):
blocking = SoundexBlocking(ref_attr_index=1, target_attr_index=1,
language='english')
blocking.fit(SOUNDEX_REFSET, SOUNDEX_TARGETSET)
blocks = list(blocking.iter_id_blocks())
self.assertEqual(len(blocks), 3)
self.assertIn((['a1', 'a7'], ['b3', 'b6']), blocks)
self.assertIn((['a2', 'a5'], ['b4']), blocks)
self.assertIn((['a3'], ['b1', 'b2']), blocks)