def test_add_relationship_parent_no_exist(self):
"""Add relationship table no exist"""
# Setup
metadata = Mock(spec=Metadata)
metadata.get_tables.return_value = ['a_table']
# Run
with pytest.raises(ValueError):
Metadata.add_relationship(metadata, 'a_table', 'b_table')
def test_add_relationship_already_exist(self):
"""Add relationship already exist"""
# Setup
metadata = Mock(spec=Metadata)
metadata.get_tables.return_value = ['a_table', 'b_table']
metadata.get_parents.return_value = set(['b_table'])
# Run
with pytest.raises(ValueError):
Metadata.add_relationship(metadata, 'a_table', 'b_table')
def test_hma1_single_child_row_single_parent_row():
"""Test that ``HMA1`` supports a single child row per single parent row.
``HMA1`` doesn't learn the distribution of the values for a child row when those
are equal to 1. This is because those values will be equal to ``0`` and alter the
``std`` by a lot.
Setup:
- Create a dataset that has 1 child row per single parent row.
- Create the ``sdv.Metadata`` for that dataset.
- Create an instance of ``HMA1``.
Input:
- ``dataset``
- ``sdv.Metadata``
Output:
- ``dict`` with synthetic data.
"""
# Setup
parent_a = pd.DataFrame({
'parent_id': range(5),
'value': range(5)
})
child = pd.DataFrame({
'parent_a': range(5),
'value_a': range(5),
})
tables = {
'parent_a': parent_a,
'child': child
}
metadata = Metadata()
metadata.add_table('parent_a', parent_a, primary_key='parent_id')
metadata.add_table('child', child)
metadata.add_relationship('parent_a', 'child', 'parent_a')
model = HMA1(metadata)
# Run
model.fit(tables)
sampled = model.sample(num_rows=10)
# Assert
assert len(sampled) == 2
assert len(sampled['parent_a']) == 10
assert len(sampled['child']) == 10
assert len(sampled['parent_a']['parent_id'].unique()) == 10
assert len(sampled['child']['parent_a'].unique()) == 10
def test_add_relationship_parent_no_primary_key(self):
"""Add relationship parent no primary key"""
# Setup
metadata = Mock(spec=Metadata)
metadata.get_tables.return_value = ['a_table', 'b_table']
metadata.get_parents.return_value = set()
metadata.get_children.return_value = set()
metadata.get_primary_key.return_value = None
# Run
with pytest.raises(ValueError):
Metadata.add_relationship(metadata, 'a_table', 'b_table')