def test_longer_table():
tube = tubes.Count(5).multi(lambda x: (tubes.Count(), x, x.to(object).to(
str), x.gt(1000))).first(500000)
table = tube.to_pyarrow(('index', 'src', 'strval', 'is_big'))
assert isinstance(table, pa.Table)
assert [str(c.type)
for c in table.columns] == ['int64', 'int64', 'string', 'bool']
assert table.shape == (500000, 4)
assert dict(table.to_pandas().iloc[-1]) == {
'index': 499999,
'src': 500004,
'strval': '500004',
'is_big': True
}
def test_doubles():
tube = tubes.Count(2).to(float).first(4)
table = tube.to_pyarrow(('a', ))
assert isinstance(table, pa.Table)
assert str(table.columns[0].type) == 'double'
assert table.to_pandas().to_dict() == {
'a': {0: 2, 1: 3, 2: 4, 3: 5},
}
def test_enumerate():
tube = tubes.Count(2).enumerate().first(4)
table = tube.to_pyarrow(('a', 'b'))
assert isinstance(table, pa.Table)
assert str(table.columns[0].type) == 'int64'
assert str(table.columns[1].type) == 'int64'
assert table.to_pandas().to_dict() == {
'a': {0: 0, 1: 1, 2: 2, 3: 3},
'b': {0: 2, 1: 3, 2: 4, 3: 5}
}
def test_skip_if_lambda():
tube = tubes.Count().first(6).skip_if(lambda x: x.lt(3))
assert list(tube) == [3, 4, 5]
def test_skip_if():
tube = tubes.Count().first(5).skip_if(
tubes.Each([False, True, False, True]).to(bool))
assert list(tube) == [0, 2]
def test_skip_unless_lambda():
tube = tubes.Count().skip_unless(lambda x: x.lt(3).first(5))
assert list(tube) == [0, 1, 2]
def test_skip_unless():
tube = tubes.Count().skip_unless(
tubes.Each([False, True, False, True]).to(bool))
assert list(tube) == [1, 3]
def test_fill_ndarray_same_type_fields():
nd = tubes.Count(5).first(4).enumerate().ndarray(fields=True)
assert nd.shape == (4, )
assert [tuple(x) for x in nd] == [(0, 5), (1, 6), (2, 7), (3, 8)]
def test_fill_ndarray_integers():
nd = tubes.Count().first(100).to(int).ndarray(estimated_rows=2)
assert nd.shape == (100, )
assert list(nd) == list(range(100))
def test_int():
tube = tubes.Count().to(int).first(20_000)
table = tube.to_pyarrow(('a', ))
assert isinstance(table, pa.Table)
assert len(table) == 20_000
assert list(table.to_pandas()['a']) == list(range(20_000))
def test_enumerate():
c = tubes.Count(2).enumerate().first(4)
assert list(c) == [(0, 2), (1, 3), (2, 4), (3, 5)]