def test_ngram_blocks(self):
blocking = NGramBlocking(ref_attr_index=1, target_attr_index=1)
blocking.fit(SOUNDEX_REFSET, SOUNDEX_TARGETSET)
blocks = list(blocking.iter_id_blocks())
self.assertEqual(len(blocks), 3)
self.assertIn((['a3'], ['b1', 'b2']), blocks)
self.assertIn((['a5'], ['b4']), blocks)
self.assertIn((['a1', 'a4'], ['b3']), blocks)
def test_ngram_blocks_2(self):
refset = [['3', 'ccdd', 'aabb'],
['4', 'ccdd', 'bbaa']]
targetset = [['c', 'ccdd', 'aabb'],
['d', 'ccdd', 'bbaa']]
true_pairs = [('3', 'c'), ('4', 'd')]
blocking = NGramBlocking(ref_attr_index=2, target_attr_index=2,
ngram_size=2, depth=1)
blocking.fit(refset, targetset)
pairs = list(blocking.iter_id_pairs())
self.assertEqual(len(pairs), len(true_pairs))