def test_compute_divisions():
from dask.dataframe.shuffle import compute_divisions
df = pd.DataFrame(
{
"x": [1, 2, 3, 4],
"y": [10, 20, 20, 40],
"z": [4, 3, 2, 1]
},
index=[1, 3, 10, 20],
)
a = dd.from_pandas(df, 2, sort=False)
assert not a.known_divisions
divisions = compute_divisions(a)
b = copy(a)
b.divisions = divisions
assert_eq(a, b, check_divisions=False)
assert b.known_divisions
c = dd.from_pandas(df.set_index("y"), 2, sort=False)
# Partitions overlap warning
with pytest.warns(UserWarning):
compute_divisions(c)
def test_compute_divisions():
from dask.dataframe.shuffle import compute_divisions
df = pd.DataFrame({'x': [1, 2, 3, 4],
'y': [10, 20, 30, 40],
'z': [4, 3, 2, 1]},
index=[1, 3, 10, 20])
a = dd.from_pandas(df, 2, sort=False)
assert not a.known_divisions
divisions = compute_divisions(a)
b = copy(a)
b.divisions = divisions
assert_eq(a, b, check_divisions=False)
assert b.known_divisions
def test_compute_divisions():
from dask.dataframe.shuffle import compute_divisions
df = pd.DataFrame({'x': [1, 2, 3, 4],
'y': [10, 20, 30, 40],
'z': [4, 3, 2, 1]},
index=[1, 3, 10, 20])
a = dd.from_pandas(df, 2, sort=False)
assert not a.known_divisions
divisions = compute_divisions(a)
b = copy(a)
b.divisions = divisions
assert_eq(a, b, check_divisions=False)
assert b.known_divisions
def test_compute_divisions():
from dask.dataframe.shuffle import compute_divisions
df = pd.DataFrame(
{
"x": [1, 2, 3, 4],
"y": [10, 20, 20, 40],
"z": [4, 3, 2, 1]
},
index=[1, 3, 10, 20],
)
a = dd.from_pandas(df, 2, sort=False)
assert not a.known_divisions
b = compute_divisions(copy(a))
assert_eq(a, b, check_divisions=False)
assert b.known_divisions
