def test_invalid_date_column(self):
currency_converter = ConvertCurrencyScrubber('value', 'invalid', 'USD',
'invalid_date')
with self.assertRaises(RuntimeError):
self.assertEqual(False,
currency_converter.validate(self._data_model))
def test_valid_metadata(self):
self._data_model.metadata.define_numerical_columns(['value'])
self._data_model.metadata.define_categorical_columns(['currency'])
currency_converter = ConvertCurrencyScrubber('value', 'value',
'currency', 'USD', 'date')
currency_converter.validate_metadata(self._data_model.metadata)
def test_invalid_metadata(self):
self._data_model.metadata.define_numerical_columns(
['value', 'currency'])
currency_converter = ConvertCurrencyScrubber('value', 'currency',
'USD', 'date')
with self.assertRaises(RuntimeError):
currency_converter.validate_metadata(self._data_model.metadata)
def test_converting_without_date_column(self):
currency_converter = ConvertCurrencyScrubber('value', 'value',
'currency', 'USD')
currency_converter.scrub(self._data_model)
df = self._data_model.get_dataframe()
result = df['value']
self.assertEqual(0.0, result[0])
self.assertEqual(2.0, result[1])
self.assertEqual(116, round(result[2]))
def test_multiple_validation(self):
average_column_scrubber = AverageColumnScrubber(('value_1', 'value_2'),
'value')
convert_currency_scrubber = ConvertCurrencyScrubber(
'value', 'new_value', 'currency', 'USD', 'date')
self.and_scrubber = AndScrubber(average_column_scrubber,
convert_currency_scrubber)
self.and_scrubber.validate_metadata(self.data_model.metadata)
def test_second_invalid(self):
average_column_scrubber = AverageColumnScrubber(('value_1', 'value_2'),
'value')
convert_currency_scrubber = ConvertCurrencyScrubber(
'value', 'invalid', 'USD', 'date')
self.and_scrubber = AndScrubber(average_column_scrubber,
convert_currency_scrubber)
with self.assertRaises(RuntimeError):
self.and_scrubber.validate_metadata(self.data_model.metadata)
def test_multiple_scrubbing(self):
average_column_scrubber = AverageColumnScrubber(('value_1', 'value_2'),
'value')
convert_currency_scrubber = ConvertCurrencyScrubber(
'value', 'new_value', 'currency', 'USD', 'date')
self.and_scrubber = AndScrubber(average_column_scrubber,
convert_currency_scrubber)
self.and_scrubber.scrub(self.data_model)
result = self.data_model.get_dataframe()['new_value']
self.assertEqual(0.0, result[0])
self.assertEqual(2.0, result[1])
self.assertEqual(105, round(result[2]))
def test_valid_columns_without_date(self):
currency_converter = ConvertCurrencyScrubber('value', 'value',
'currency', 'USD')
currency_converter.validate(self._data_model)
def test_valid_columns(self):
currency_converter = ConvertCurrencyScrubber('value', 'currency',
'USD', 'date')
with self.assertRaises(RuntimeError):
currency_converter.validate(self._data_model)