def test__make_anonymization_mappings(self, mock_table):
"""Test that ``_make_anonymization_mappings`` creates the expected mappings.
The ``_make_anonymization_mappings`` method should map values in the original
data to fake values for non-id fields that are labeled pii.
Setup:
- Create a Table that has metadata about three fields (one pii field, one id field,
and one non-pii field).
Input:
- Data that contains a pii field, an id field, and a non-pii field.
Side Effects:
- Expect ``_get_fake_values`` to be called with the number of unique values of the
pii field.
- Expect the resulting `_ANONYMIZATION_MAPPINGS` field to contain the pii field, with
the correct number of mappings and keys.
"""
# Setup
metadata = Mock()
metadata._ANONYMIZATION_MAPPINGS = {}
foo_metadata = {
'type': 'categorical',
'pii': True,
'pii_category': 'email',
}
metadata._fields_metadata = {
'foo': foo_metadata,
'bar': {
'type': 'categorical',
},
'baz': {
'type': 'id',
}
}
foo_values = [
'*****@*****.**', '*****@*****.**', '*****@*****.**'
]
data = pd.DataFrame({
'foo': foo_values,
'bar': ['a', 'b', 'c'],
'baz': [1, 2, 3],
})
# Run
Table._make_anonymization_mappings(metadata, data)
# Assert
assert mock_table._get_fake_values.called_once_with(foo_metadata, 3)
mappings = metadata._ANONYMIZATION_MAPPINGS[id(metadata)]
assert len(mappings) == 1
foo_mappings = mappings['foo']
assert len(foo_mappings) == 3
assert list(foo_mappings.keys()) == foo_values
def test__make_anonymization_mappings_unique_faked_value_in_field(
self, mock_table):
"""Test that ``_make_anonymization_mappings`` method creates mappings for anonymized values.
The ``_make_anonymization_mappings`` method should map equal values in the original data
to the same faked value.
Input:
- DataFrame with a field that should be anonymized based on the metadata description.
Side Effect:
- Mappings are created from the original values to faked values.
"""
# Setup
metadata = Mock()
metadata._ANONYMIZATION_MAPPINGS = {}
foo_metadata = {
'type': 'categorical',
'pii': True,
'pii_category': 'email'
}
metadata._fields_metadata = {'foo': foo_metadata}
data = pd.DataFrame({
'foo':
['*****@*****.**', '*****@*****.**', '*****@*****.**']
})
# Run
Table._make_anonymization_mappings(metadata, data)
# Assert
assert mock_table._get_fake_values.called_once_with(foo_metadata, 2)
mappings = metadata._ANONYMIZATION_MAPPINGS[id(metadata)]
assert len(mappings) == 1
foo_mappings = mappings['foo']
assert len(foo_mappings) == 2
assert list(
foo_mappings.keys()) == ['*****@*****.**', '*****@*****.**']
