def test_formatter() -> None:
_ = str(object_formatter)
t = Tafra({'x': Decimal(1.2345)})
assert t._dtypes['x'] == 'float64'
assert t['x'].dtype == np.dtype(float)
object_formatter['Decimal'] = lambda x: x.astype(int)
t = Tafra({'x': Decimal(1.2345)})
if platform.system() == 'Windows':
assert t._dtypes['x'] == 'int32'
elif platform.system() == 'Linux':
assert t._dtypes['x'] == 'int64'
assert t['x'].dtype == np.dtype(int)
_ = str(object_formatter)
for fmt in object_formatter:
pass
_ = object_formatter.copy()
del object_formatter['Decimal']
with pytest.raises(ValueError) as e:
object_formatter['Decimal'] = lambda x: 'int' # type: ignore
_ = str(object_formatter)
def test_cross_join() -> None:
l = Tafra({
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1])
})
r = Tafra({
'a':
np.array([1, 2, 3, 4, 5, 6]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.cross_join(r)
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 2, 2, 3, 3]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.cross_join(r)
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 1, 2, 2, 2]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.cross_join(r)
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 1, 2, 2, 2]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.cross_join(r, select=['x', 'z', 'a', 'c'])
check_tafra(t)
with pytest.raises(IndexError) as e:
t = l.cross_join(r, select=['x', 'z'])
with pytest.raises(IndexError) as e:
t = l.cross_join(r, select=['a', 'c'])
def test_coalesce() -> None:
t = Tafra({'x': np.array([1, 2, None, 4, None])})
t['x'] = t.coalesce(
'x',
[[1, 2, 3, None, 5], [None, None, None, None, 'five']]) # type: ignore
t['y'] = t.coalesce(
'y',
[[1, 2, 3, None, 5], [None, None, None, None, 'five']]) # type: ignore
assert np.all(t['x'] != np.array(None))
assert t['y'][3] == np.array(None)
check_tafra(t)
t = Tafra({'x': np.array([1, 2, None, 4, None])})
t.coalesce_inplace(
'x',
[[1, 2, 3, None, 5], [None, None, None, None, 'five']]) # type: ignore
t.coalesce_inplace(
'y',
[[1, 2, 3, None, 5], [None, None, None, None, 'five']]) # type: ignore
assert np.all(t['x'] != np.array(None))
assert t['y'][3] == np.array(None)
check_tafra(t)
t = Tafra({'x': np.array([None])})
t.coalesce('x', [[1], [None]]) # type: ignore
check_tafra(t)
def test_left_join_equi() -> None:
l = Tafra({
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1])
})
r = Tafra({
'a':
np.array([1, 2, 3, 4, 5, 6]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.left_join(r, [('x', 'a', '==')], ['x', 'y', 'a', 'b'])
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 1, 2, 2, 2]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([2, 2, 2, 3, 3, 3])
})
t = l.left_join(r, [('x', 'a', '=='), ('z', 'c', '==')],
['x', 'y', 'a', 'b'])
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 1, 2, 2, 2]),
'_a':
np.array([1, 1, 2, 2, 3, 3]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.left_join(r, [('x', 'a', '=='), ('x', '_a', '==')],
['x', 'y', 'a', 'b'])
check_tafra(t)
r = Tafra({
'a':
np.array([1, 1, 2, 2, 3, 3]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
t = l.left_join(r, [('x', 'a', '<')], ['x', 'y', 'a', 'b'])
check_tafra(t)
def test_left_join_invalid() -> None:
l = Tafra({
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1])
})
r = Tafra({
'a':
np.array([1, 2, 3, 4, 5, 6]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
with pytest.raises(TypeError) as e:
t = l.left_join(r, [('x', 'a', '===')], ['x', 'y', 'a', 'b'])
r = Tafra({
'a':
np.array([1, 2, 3, 4, 5, 6], dtype='float'),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
with pytest.raises(TypeError) as e:
t = l.left_join(r, [('x', 'a', '==')], ['x', 'y', 'a', 'b'])
r = Tafra({
'a':
np.array([1, 2, 3, 4, 5, 6]),
'b':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'c':
np.array([0, 0, 0, 1, 1, 1])
})
l._dtypes['x'] = 'float'
with pytest.raises(TypeError) as e:
t = l.left_join(r, [('x', 'a', '==')], ['x', 'y', 'a', 'b'])
def build_tafra() -> Tafra:
return Tafra({
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'], dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1])
})
def test_destructors() -> None:
def gen_values() -> Iterator[Dict[str, np.ndarray]]:
yield {'x': np.arange(6)}
yield {'y': np.arange(6)}
t = Tafra(gen_values())
check_tafra(t)
t = build_tafra()
t = t.update_dtypes({'x': 'float'})
t.data['x'][2] = np.nan
check_tafra(t)
_ = tuple(t.to_records())
_ = tuple(t.to_records(columns='x'))
_ = tuple(t.to_records(columns=['x']))
_ = tuple(t.to_records(columns=['x', 'y']))
_ = tuple(t.to_records(cast_null=False))
_ = tuple(t.to_records(columns='x', cast_null=False))
_ = tuple(t.to_records(columns=['x'], cast_null=False))
_ = tuple(t.to_records(columns=['x', 'y'], cast_null=False))
_ = t.to_list()
_ = t.to_list(columns='x')
_ = t.to_list(columns=['x'])
_ = t.to_list(columns=['x', 'y'])
_ = t.to_list(inner=True)
_ = t.to_list(columns='x', inner=True)
_ = t.to_list(columns=['x'], inner=True)
_ = t.to_list(columns=['x', 'y'], inner=True)
_ = t.to_tuple()
_ = t.to_tuple(columns='x')
_ = t.to_tuple(columns=['x'])
_ = t.to_tuple(columns=['x', 'y'])
_ = t.to_tuple(inner=True)
_ = t.to_tuple(columns='x', inner=True)
_ = t.to_tuple(columns=['x'], inner=True)
_ = t.to_tuple(columns=['x', 'y'], inner=True)
_ = t.to_array()
_ = t.to_array(columns='x')
_ = t.to_array(columns=['x'])
_ = t.to_array(columns=['x', 'y'])
_ = t.to_pandas()
_ = t.to_pandas(columns='x')
_ = t.to_pandas(columns=['x'])
_ = t.to_pandas(columns=['x', 'y'])
filepath = Path('test/test_to_csv.csv')
t.to_csv(filepath)
t.to_csv(filepath, columns='x')
t.to_csv(filepath, columns=['x'])
t.to_csv(filepath, columns=['x', 'y'])
def test_map() -> None:
t = build_tafra()
_ = list(t.row_map(np.repeat, 6))
_ = list(t.tuple_map(np.repeat, 6))
_ = list(t.col_map(np.repeat, repeats=6))
_ = Tafra(t.key_map(np.repeat, repeats=6))
def test_constructions() -> None:
t = build_tafra()
check_tafra(t)
t = Tafra(
{
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1])
},
validate=False)
check_tafra(t)
t = Tafra(
{
'x':
np.array([1, 2, 3, 4, 5, 6]),
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object'),
'z':
np.array([0, 0, 0, 1, 1, 1, 2, 2, 2])
},
validate=False,
check_rows=False)
check_tafra(t, check_rows=False)
with pytest.raises(TypeError) as e:
t = Tafra() # type: ignore # noqa
with pytest.raises(ValueError) as e:
t = Tafra({})
t = Tafra({'x': None})
with warnings.catch_warnings(record=True) as w:
check_tafra(t)
t = Tafra({'x': Decimal('1.23456')})
check_tafra(t)
t = Tafra({'x': np.array(1)})
check_tafra(t)
t = Tafra({'x': np.array([1])})
check_tafra(t)
t = Tafra({'x': [True, False]})
check_tafra(t)
t = Tafra({'x': 'test'})
check_tafra(t)
t = Tafra((('x', np.arange(6)), ))
check_tafra(t)
t = Tafra([('x', np.arange(6))])
check_tafra(t)
t = Tafra([['x', np.arange(6)]])
check_tafra(t)
t = Tafra([(np.array('x'), np.arange(6))])
check_tafra(t)
t = Tafra([(np.array(['x']), np.arange(6))])
check_tafra(t)
t = Tafra([('x', np.arange(6)), ('y', np.linspace(0, 1, 6))])
check_tafra(t)
t = Tafra([['x', np.arange(6)], ('y', np.linspace(0, 1, 6))])
check_tafra(t)
t = Tafra([('x', np.arange(6)), ['y', np.linspace(0, 1, 6)]])
check_tafra(t)
t = Tafra([['x', np.arange(6)], ['y', np.linspace(0, 1, 6)]])
check_tafra(t)
t = Tafra([{'x': np.arange(6)}, {'y': np.linspace(0, 1, 6)}])
check_tafra(t)
t = Tafra(iter([{'x': np.arange(6)}, {'y': np.linspace(0, 1, 6)}]))
check_tafra(t)
def iterator() -> Iterator[Dict[str, np.ndarray]]:
yield {'x': np.array([1, 2, 3, 4, 5, 6])}
yield {
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object')
}
yield {'z': np.array([0, 0, 0, 1, 1, 1])}
t = Tafra(iterator())
check_tafra(t)
class DictIterable:
def __iter__(self) -> Iterator[Dict[str, np.ndarray]]:
yield {'x': np.array([1, 2, 3, 4, 5, 6])}
yield {
'y':
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object')
}
yield {'z': np.array([0, 0, 0, 1, 1, 1])}
t = Tafra(DictIterable())
check_tafra(t)
t = Tafra(iter(DictIterable()))
check_tafra(t)
class SequenceIterable:
def __iter__(self) -> Iterator[Any]:
yield ('x', np.array([1, 2, 3, 4, 5, 6]))
yield [
'y',
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object')
]
yield ('z', np.array([0, 0, 0, 1, 1, 1]))
t = Tafra(SequenceIterable())
check_tafra(t)
class SequenceIterable2:
def __iter__(self) -> Iterator[Any]:
yield (np.array(['x']), np.array([1, 2, 3, 4, 5, 6]))
yield [
np.array(['y']),
np.array(['one', 'two', 'one', 'two', 'one', 'two'],
dtype='object')
]
yield (np.array(['z']), np.array([0, 0, 0, 1, 1, 1]))
t = Tafra(SequenceIterable2())
check_tafra(t)
t = Tafra(iter(SequenceIterable2()))
check_tafra(t)
t = Tafra(enumerate(np.arange(6)))
check_tafra(t)
t = build_tafra()
df = pd.DataFrame(t.data)
_ = Tafra.from_series(df['x'])
check_tafra(_)
_ = Tafra.from_dataframe(df)
check_tafra(_)
_ = Tafra.as_tafra(df)
check_tafra(_)
_ = Tafra.as_tafra(df['x'])
check_tafra(_)
_ = Tafra.as_tafra(t)
check_tafra(_)
_ = Tafra.as_tafra({'x': np.array(1)})
check_tafra(_)
_ = Tafra.from_series(Series())
check_tafra(_)
_ = Tafra.as_tafra(Series())
check_tafra(_)
_ = Tafra.from_dataframe(DataFrame()) # type: ignore
check_tafra(_)
_ = Tafra.as_tafra(DataFrame())
check_tafra(_)
with pytest.raises(TypeError) as e:
t = Tafra([{1, 2}]) # type: ignore
class BadIterable:
def __iter__(self) -> Iterator[Any]:
yield {1, 2}
yield {3.1412159, .5772156}
with pytest.raises(TypeError) as e:
t = Tafra(BadIterable())
with pytest.raises(TypeError) as e:
t = Tafra(iter(BadIterable()))
with pytest.raises(TypeError) as e:
_ = Tafra(np.arange(6))
with pytest.raises(TypeError) as e:
_ = Tafra.as_tafra(np.arange(6))
with pytest.raises(ValueError) as e:
t = Tafra({'x': np.array([1, 2]), 'y': np.array([3., 4., 5.])})
def test_constructions() -> None:
with pytest.raises(TypeError) as e:
t = Tafra() # type: ignore # noqa
with pytest.raises(ValueError) as e:
t = Tafra({}) # type: ignore
t = Tafra({'x': None})
check_tafra(t)
t = Tafra({'x': Decimal('1.23456')})
check_tafra(t)
t = Tafra({'x': np.array(1)})
check_tafra(t)
t = Tafra({'x': np.array([1])})
check_tafra(t)
t = Tafra({'x': [True, False]})
check_tafra(t)
t = Tafra({'x': 'test'})
check_tafra(t)
t.update_dtypes_inplace({'x': 'O'})
check_tafra(t)
t = Tafra(enumerate(np.arange(6)))
check_tafra(t)
with pytest.raises(ValueError) as e:
t = Tafra({'x': np.array([1, 2]), 'y': np.array([3., 4., 5.])})
def gen_values() -> Iterator[Dict[str, np.ndarray]]:
yield {'x': np.arange(6)}
yield {'y': np.arange(6)}
t = Tafra(gen_values())
check_tafra(t)
t = build_tafra()
t = t.update_dtypes({'x': 'float'})
t.data['x'][2] = np.nan
check_tafra(t)
_ = tuple(t.to_records())
_ = tuple(t.to_records(columns='x'))
_ = tuple(t.to_records(columns=['x']))
_ = tuple(t.to_records(columns=['x', 'y']))
_ = tuple(t.to_records(cast_null=False))
_ = tuple(t.to_records(columns='x', cast_null=False))
_ = tuple(t.to_records(columns=['x'], cast_null=False))
_ = tuple(t.to_records(columns=['x', 'y'], cast_null=False))
_ = t.to_list()
_ = t.to_list(columns='x')
_ = t.to_list(columns=['x'])
_ = t.to_list(columns=['x', 'y'])
_ = t.to_list(inner=True)
_ = t.to_list(columns='x', inner=True)
_ = t.to_list(columns=['x'], inner=True)
_ = t.to_list(columns=['x', 'y'], inner=True)
_ = t.to_array()
_ = t.to_array(columns='x')
_ = t.to_array(columns=['x'])
_ = t.to_array(columns=['x', 'y'])
t = build_tafra()
df = pd.DataFrame(t.data)
_ = Tafra.from_series(df['x'])
check_tafra(_)
_ = Tafra.from_dataframe(df)
check_tafra(_)
_ = Tafra.as_tafra(df)
check_tafra(_)
_ = Tafra.as_tafra(df['x'])
check_tafra(_)
_ = Tafra.as_tafra(t)
check_tafra(_)
_ = Tafra.as_tafra({'x': np.array(1)})
check_tafra(_)
_ = Tafra.from_series(Series())
check_tafra(_)
_ = Tafra.as_tafra(Series())
check_tafra(_)
_ = Tafra.from_dataframe(DataFrame()) # type: ignore
check_tafra(_)
_ = Tafra.as_tafra(DataFrame())
check_tafra(_)
with pytest.raises(TypeError) as e:
_ = Tafra(np.arange(6))
with pytest.raises(TypeError) as e:
_ = Tafra.as_tafra(np.arange(6))
def test_map() -> None:
t = build_tafra()
_ = list(t.row_map(np.repeat, 6))
_ = Tafra(t.col_map(np.repeat, name=True, repeats=6))
_ = list(t.col_map(np.repeat, name=False, repeats=6))