def test__valid_separator_non_valid_separator_contained(self):
"""Test ``_valid_separator`` passing a column that contains the separator.
If any of the columns contains the separator string, result is ``False``.
Input:
- Table data (pandas.DataFrame) with a column that contains the separator string ('#')
Output:
- False (bool).
"""
# Setup
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance._separator = '#'
# Run
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', '#', 'f'],
'c': ['g', 'h', 'i']
})
is_valid = instance._valid_separator(table_data)
# Assert
assert not is_valid
def test_is_valid_false(self):
"""Test the ``UniqueCombinations.is_valid`` method.
If the input data doesn't satisfy the constraint, result is a series of ``False`` values.
Input:
- Table data (pandas.DataFrame), which does not satisfy the constraint.
Output:
- Series of ``False`` values (pandas.Series)
Side effects:
- Since the ``is_valid`` method needs ``self._combinations``, method ``fit``
must be called as well.
"""
# Setup
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance.fit(table_data)
# Run
incorrect_table = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['D', 'E', 'F'],
'c': ['g', 'h', 'i']
})
out = instance.is_valid(incorrect_table)
# Assert
expected_out = pd.Series([False, False, False], name='b#c')
pd.testing.assert_series_equal(expected_out, out)
def test__valid_separator_valid(self):
"""Test ``_valid_separator`` for a valid separator.
If the separator and data are valid, result is ``True``.
Input:
- Table data (pandas.DataFrame)
Output:
- True (bool).
"""
# Setup
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance._separator = '#'
# Run
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
is_valid = instance._valid_separator(table_data)
# Assert
assert is_valid
def reverse_transform(self):
"""Test the ``UniqueCombinations.reverse_transform`` method.
It is expected to return the original data separating the concatenated columns.
Input:
- Table data transformed (pandas.DataFrame)
Output:
- Original table data, with the concatenated columns separated (pandas.DataFrame)
Side effects:
- Since the ``transform`` method needs ``self._joint_column``, method ``fit``
must be called as well.
"""
# Setup
transformed_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b#c': ['d#g', 'e#h', 'f#i']
})
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance.fit(transformed_data)
# Run
out = instance.reverse_transform(transformed_data)
# Assert
expected_out = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
pd.testing.assert_frame_equal(expected_out, out)
def test_is_valid_true(self):
"""Test the ``UniqueCombinations.is_valid`` method.
If the input data satisfies the constraint, result is a series of ``True`` values.
Input:
- Table data (pandas.DataFrame), satisfying the constraint.
Output:
- Series of ``True`` values (pandas.Series)
Side effects:
- Since the ``is_valid`` method needs ``self._combinations``, method ``fit``
must be called as well.
"""
# Setup
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance.fit(table_data)
# Run
out = instance.is_valid(table_data)
expected_out = pd.Series([True, True, True], name='b#c')
pd.testing.assert_series_equal(expected_out, out)
def test_fit(self):
"""Test the ``UniqueCombinations.fit`` method.
The ``UniqueCombinations.fit`` method is expected to:
- Call ``UniqueCombinations._valid_separator``.
- Find a valid separator for the data and generate the joint column name.
Input:
- Table data (pandas.DataFrame)
"""
# Setup
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
# Run
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
instance.fit(table_data)
# Asserts
expected_combinations = pd.DataFrame({
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
assert instance._separator == '#'
assert instance._joint_column == 'b#c'
pd.testing.assert_frame_equal(instance._combinations,
expected_combinations)
def test__valid_separator_non_valid_name_joined_exists(self):
"""Test ``_valid_separator`` passing a column whose name is obtained after joining
the column names using the separator.
If the column name obtained after joining the column names using the separator
already exists, result is ``False``.
Input:
- Table data (pandas.DataFrame) with a column name that will be obtained by joining
the column names and the separator.
Output:
- False (bool).
"""
# Setup
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance._separator = '#'
# Run
table_data = pd.DataFrame({
'b#c': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
is_valid = instance._valid_separator(table_data)
# Assert
assert not is_valid
def test_to_dict(self):
"""Test the ``Constraint.to_dict`` method.
The ``Constraint.to_dict`` method is expected to return a dict
containting the FQN of the constraint instance and all the
required arguments rebuild it.
Output:
- Dict with the right values.
"""
# Run
instance = UniqueCombinations(columns=['a', 'b'],
handling_strategy='transform')
constraint_dict = instance.to_dict()
# Assert
expected_dict = {
'constraint': 'sdv.constraints.tabular.UniqueCombinations',
'handling_strategy': 'transform',
'columns': ['a', 'b'],
}
assert constraint_dict == expected_dict
def test___init__(self):
"""Test the ``UniqueCombinations.__init__`` method.
It is expected to create a new Constraint instance and receiving the names of
the columns that need to produce unique combinations.
Side effects:
- instance._colums == columns
"""
# Setup
columns = ['b', 'c']
# Run
instance = UniqueCombinations(columns=columns)
# Assert
assert instance._columns == columns
def test_transform_not_all_columns_provided(self):
"""Test the ``UniqueCombinations.transform`` method.
If some of the columns needed for the transform are missing, it will raise
a ``MissingConstraintColumnError``.
Input:
- Table data (pandas.DataFrame)
Output:
- Raises ``MissingConstraintColumnError``.
"""
# Setup
table_data = pd.DataFrame({
'a': ['a', 'b', 'c'],
'b': ['d', 'e', 'f'],
'c': ['g', 'h', 'i']
})
columns = ['b', 'c']
instance = UniqueCombinations(columns=columns)
instance.fit(table_data)
# Run/Assert
with pytest.raises(MissingConstraintColumnError):
instance.transform(pd.DataFrame({'a': ['a', 'b', 'c']}))