def setUpClass(cls):
tokenizer = Tokenizer()
cls.bj = bag_jag()
cls.bd0 = bag_dict().read(tokenizer, {
'_id': '0',
'title': 'テストデータ',
'body': 'テスト',
'anchor': 'モニタ',
})
cls.bj.append(cls.bd0)
cls.bd1 = bag_dict().read(tokenizer, {
'_id': '1',
'title': 'テストデータ',
'body': 'テスト',
})
cls.bj.append(cls.bd1)
cls.bd2 = bag_dict().read(tokenizer, {
'_id': '2',
'body': 'テスト',
})
cls.bj.append(cls.bd2)
cls.bd3 = bag_dict().read(tokenizer, {
'_id': '3',
})
cls.bj.append(cls.bd3)
cls.query = bag_of_words()
cls.query['テスト'] = 1
cls.query['モニタ'] = 1
def test_bag_jag_rw_continuous(self):
expect = bag_jag()
bd = bag_dict().read(self.tokenizer, {
'_id': 'テスト用のデータ001',
'~pv': 123.0,
})
expect.append(bd)
d = tempfile.TemporaryDirectory()
p = '%s/tmp.txt' % d.name
expect.write(p)
actual = bag_jag()
actual.read(p)
d.cleanup()
self.assertEqual(expect.body, actual.body)
self.assertEqual(expect.df, actual.df)
self.assertTrue(123.0 not in expect.df)
self.assertTrue('123.0' not in actual.df)
self.assertEqual(expect.total_len, actual.total_len)
def test_bag_jag_rw(self):
expect = self.bj
d = tempfile.TemporaryDirectory()
p = '%s/tmp.txt' % d.name
expect.write(p)
actual = bag_jag()
actual.read(p)
d.cleanup()
self.assertEqual(expect.body, actual.body)
self.assertEqual(expect.df, actual.df)
self.assertEqual(expect.total_len, actual.total_len)
def test_weight_continuous(self):
tokenizer = Tokenizer()
bj = bag_jag()
bd0 = bag_dict().read(tokenizer, {'~pv': 1})
bj.append(bd0)
bd1 = bag_dict().read(tokenizer, {'~pv': 10})
bj.append(bd1)
bd2 = bag_dict().read(tokenizer, {'~pv': 100})
bj.append(bd2)
self.assertEqual((1 * 1.0), weight('ダミー', bd0, bj))
self.assertEqual((10 * 1.0), weight('ダミー', bd1, bj))
self.assertEqual((100 * 1.0), weight('ダミー', bd2, bj))
def test_bag_jag_iadd(self):
bj = bag_jag()
bj += self.bj
bj += self.bj
self.assertEqual(8, len(bj))
self.assertEqual(6, bj.df['テスト'])
self.assertEqual(4, bj.df['データ'])
self.assertEqual(2, bj.df['モニタ'])
self.assertEqual(0, bj.df['ダミー'])
self.assertEqual(8, bj.total_len['title'])
self.assertEqual(6, bj.total_len['body'])
self.assertEqual(2, bj.total_len['anchor'])
def setUpClass(cls):
cls.tokenizer = Tokenizer()
cls.bj = bag_jag()
bd0 = bag_dict().read(cls.tokenizer, {
'title': 'テストデータ',
'body': 'テスト',
'anchor': 'モニタ',
})
bd1 = bag_dict().read(cls.tokenizer, {
'title': 'テストデータ',
'body': 'テスト',
})
bd2 = bag_dict().read(cls.tokenizer, {
'body': 'テスト',
})
bd3 = bag_dict().read(cls.tokenizer, {})
cls.bj.append(bd0).append(bd1).append(bd2).append(bd3)