| 1 | import unittest2 as unittest |
|---|
| 2 | import numpy as np |
|---|
| 3 | from scipy.sparse import coo_matrix |
|---|
| 4 | from vsm.model import tfidf |
|---|
| 5 | from vsm.model import tf |
|---|
| 6 | |
|---|
| 7 | |
|---|
| 8 | class TestTfIdf(unittest.TestCase): |
|---|
| 9 | |
|---|
| 10 | def setUp(self): |
|---|
| 11 | self.corpus = np.array([0, 1, 3, 1, 1, 0, 3, 0, 3, |
|---|
| 12 | 3, 0, 1, 0, |
|---|
| 13 | 3, |
|---|
| 14 | 1, 3]) |
|---|
| 15 | self.docs = [slice(0,9), slice(9,13), |
|---|
| 16 | slice(13,14), slice(14,16)] |
|---|
| 17 | self.V = 4 |
|---|
| 18 | |
|---|
| 19 | self.tf_mat = coo_matrix(np.array([[3, 2, 0, 0], |
|---|
| 20 | [3, 1, 0, 1], |
|---|
| 21 | [0, 0, 0, 0], |
|---|
| 22 | [3, 1, 1, 1]], dtype=np.int)) |
|---|
| 23 | self.tfidf_mat = np.array(\ |
|---|
| 24 | [[2.0794415, 1.3862944, 0, 0], |
|---|
| 25 | [0.86304623, 0.28768209, 0, 0.28768209], |
|---|
| 26 | [0, 0, 0, 0], |
|---|
| 27 | [0, 0, 0, 0]]) |
|---|
| 28 | self.undefined_rows = [2] |
|---|
| 29 | |
|---|
| 30 | |
|---|
| 31 | def test_TfIdf_train(self): |
|---|
| 32 | m = tfidf.TfIdf() |
|---|
| 33 | m.train() |
|---|
| 34 | self.assertTrue(m.matrix.size == 0) |
|---|
| 35 | self.assertTrue(len(m.undefined_rows) == 0) |
|---|
| 36 | |
|---|
| 37 | m = tfidf.TfIdf(tf_matrix=self.tf_mat) |
|---|
| 38 | m.train() |
|---|
| 39 | np.testing.assert_almost_equal(self.tfidf_mat, m.matrix.toarray()) |
|---|
| 40 | self.assertEqual(m.undefined_rows, self.undefined_rows) |
|---|
| 41 | |
|---|
| 42 | def test_TfIdf_from_tf(self): |
|---|
| 43 | tf_model = tf.TF() |
|---|
| 44 | tf_model.corpus = self.corpus |
|---|
| 45 | tf_model.docs = self.docs |
|---|
| 46 | tf_model.V = self.V |
|---|
| 47 | tf_model.train() |
|---|
| 48 | self.assertTrue((self.tf_mat == tf_model.matrix.toarray()).all()) |
|---|
| 49 | |
|---|
| 50 | m = tfidf.TfIdf.from_tf(tf_model) |
|---|
| 51 | self.assertTrue((m.matrix == tf_model.matrix.toarray()).all()) |
|---|
| 52 | m.train() |
|---|
| 53 | np.testing.assert_almost_equal(self.tfidf_mat, m.matrix.toarray()) |
|---|
| 54 | self.assertEqual(m.undefined_rows, self.undefined_rows) |
|---|
| 55 | |
|---|
| 56 | |
|---|
| 57 | #Define and run test suite |
|---|
| 58 | suite = unittest.TestLoader().loadTestsFromTestCase(TestTfIdf) |
|---|
| 59 | unittest.TextTestRunner(verbosity=2).run(suite) |
|---|
