def test_bishingle(self):
expected = [
'A_B', 'A_C', 'A_a', 'A_b', 'A_c', 'B_C', 'B_a', 'B_b', 'B_c',
'C_a', 'C_b', 'C_c', 'a_b', 'a_c', 'b_c'
]
t = KShingleTransformer(Sentinel(), k=2)
self.assertListEqual(
expected, list(t.k_shingle(TestKShingleTransformer.segments)))
def test_bishingle(self):
expected = ['A_B', 'A_C', 'A_a', 'A_b', 'A_c',
'B_C', 'B_a', 'B_b', 'B_c',
'C_a', 'C_b', 'C_c',
'a_b', 'a_c',
'b_c']
t = KShingleTransformer(Sentinel(), k=2)
self.assertListEqual(expected, list(t.k_shingle(
TestKShingleTransformer.segments
)))
def test_iter(self):
expected = [
1,
TestKShingleTransformer.tokensA + TestKShingleTransformer.tokensB
]
n = -1
rows = Rows([[1, TestKShingleTransformer.segments]] * 3)
for n, row in enumerate(KShingleTransformer(rows)):
self.assertListEqual(expected, row)
self.assertEqual(2, n)
def test_unishingle(self):
expected = []
t = KShingleTransformer(Sentinel(), k=1)
self.assertListEqual(expected,
list(t.k_shingle(TestKShingleTransformer.segments)))
def test_unishingle(self):
expected = []
t = KShingleTransformer(Sentinel(), k=1)
self.assertListEqual(
expected, list(t.k_shingle(TestKShingleTransformer.segments)))
def test_setup(self):
s = Sentinel()
t = KShingleTransformer(s)
self.assertEqual(s, t.rows)
self.assertEqual(1, t.K)
self.assertEqual(None, t.text_columns)