def test_to_numpy_dtype_integer(self):
mock_name = Mock(name='name')
mock_statistics = Mock(name='statistics')
mock_representations = Mock(name='representations')
mock_field_type = 'integer'
expectations = {
(-100, 100): np.int8,
(0, 240): np.uint8,
(-10000, 10000): np.int16,
(500, 40000): np.uint16,
(-200000000, 200000000): np.int32,
(25, 4000000000): np.uint32,
(-9000000000000000000, 2000000000): np.int64,
(25, 10000000000000000000): np.uint64,
(25, 1000000000000000000000000000): np.float128,
(None, None): np.int64
}
for (mock_min, mock_max), expected_pandas_type in expectations.items():
mock_constraints = Mock(name='constraints')
mock_constraints.min_ = mock_min
mock_constraints.max_ = mock_max
field = RecordsSchemaField(name=mock_name,
field_type=mock_field_type,
constraints=mock_constraints,
statistics=mock_statistics,
representations=mock_representations)
out = field.to_numpy_dtype()
self.assertEqual(out, expected_pandas_type, f"min={mock_min}, max={mock_max}")
def test_to_numpy_dtype_decimal_float(self):
mock_name = Mock(name='name')
mock_statistics = Mock(name='statistics')
mock_representations = Mock(name='representations')
mock_field_type = 'decimal'
expectations = {
(8, 4): np.float16,
(20, 10): np.float32,
(40, 20): np.float64,
(80, 64): np.float128,
(500, 250): np.float128,
(None, None): np.float64,
}
for (fp_total_bits, fp_significand_bits), expected_pandas_type in expectations.items():
mock_constraints = Mock(name='constraints')
mock_constraints.fixed_precision = None
mock_constraints.fixed_scale = None
mock_constraints.fp_total_bits = fp_total_bits
mock_constraints.fp_significand_bits = fp_significand_bits
field = RecordsSchemaField(name=mock_name,
field_type=mock_field_type,
constraints=mock_constraints,
statistics=mock_statistics,
representations=mock_representations)
out = field.to_numpy_dtype()
self.assertEqual(out, expected_pandas_type)
def test_to_numpy_dtype_decimal_no_constraints(self):
mock_name = Mock(name='name')
mock_statistics = Mock(name='statistics')
mock_representations = Mock(name='representations')
mock_field_type = 'decimal'
field = RecordsSchemaField(name=mock_name,
field_type=mock_field_type,
constraints=None,
statistics=mock_statistics,
representations=mock_representations)
out = field.to_numpy_dtype()
self.assertEqual(out, np.float64)
def test_to_numpy_dtype_fixed_precision_(self):
mock_name = Mock(name='name')
mock_statistics = Mock(name='statistics')
mock_representations = Mock(name='representations')
mock_constraints = Mock(name='constraints')
mock_constraints.fixed_precision = 1
mock_constraints.fixed_scale = 1
mock_field_type = 'decimal'
field = RecordsSchemaField(name=mock_name,
field_type=mock_field_type,
constraints=mock_constraints,
statistics=mock_statistics,
representations=mock_representations)
out = field.to_numpy_dtype()
self.assertEqual(out, np.float64)
def test_to_numpy_dtype_misc(self):
mock_name = Mock(name='name')
mock_constraints = Mock(name='constraints')
mock_statistics = Mock(name='statistics')
mock_representations = Mock(name='representations')
expectations = {
'boolean': np.bool_,
'string': np.object_,
'date': np.object_,
'datetime': 'datetime64[ns]',
'datetimetz': 'datetime64[ns, UTC]',
'time': np.object_,
}
for field_type, expected_pandas_type in expectations.items():
field = RecordsSchemaField(name=mock_name,
field_type=field_type,
constraints=mock_constraints,
statistics=mock_statistics,
representations=mock_representations)
out = field.to_numpy_dtype()
self.assertEqual(out, expected_pandas_type)