def test_basic(npartitions):
df = gpd.GeoDataFrame({'x': [1, 2], 'geometry': triangles})
ddf = dg.from_pandas(df, npartitions=npartitions)
assert ddf.npartitions == npartitions
assert_eq(ddf, df)
assert_eq(ddf.x, df.x)
assert_eq(ddf.x + 1, df.x + 1)
def test_unary_method(npartitions, op, args):
df = gpd.GeoDataFrame({'x': [1, 2], 'geometry': triangles})
ddf = dg.from_pandas(df, npartitions=npartitions)
def func(a):
return getattr(a, op)(*args)
point = Point(0.2, 0.2)
assert_eq(func(ddf), func(df))
assert_eq(func(ddf.geometry), func(df.geometry))
def test_binary_singleton(npartitions, op):
df = gpd.GeoDataFrame({'x': [1, 2], 'geometry': triangles})
ddf = dg.from_pandas(df, npartitions=npartitions)
def func(a, b):
return getattr(a, op)(b)
point = Point(0.2, 0.2)
result = func(ddf, point)
assert_eq(func(ddf, point), func(df, point))
assert_eq(func(ddf.geometry, point), func(df.geometry, point))
def test_repartition_polys():
with dask.set_options(get=dask.get):
df = dg.from_pandas(grid_df, npartitions=3)
assert len(df) == len(grid_df)
df = dg.repartition(df, triangles)
# assert len(df) == len(grid_df) # fails because touches fails
df = df.persist()
assert df.npartitions == 2
for i in range(df.npartitions):
part = df.get_partition(i)
geoms = part.geometry.compute()
assert geoms.within(part._regions.iloc[0]).all()
def test_drop():
df = dg.from_pandas(points_df, npartitions=4)
assert_eq(df.drop('value', axis=1), points_df.drop('value', axis=1))
assert isinstance(df.drop('geometry', axis=1), dd.DataFrame)
assert_eq(df.drop('geometry', axis=1), points_df.drop('geometry', axis=1))
def test_sjoin_dask_normal():
df = dg.from_pandas(points_df, npartitions=4)
assert_eq(df.index, points_df.index)
def test_len():
df = points_df
ddf = dg.from_pandas(df, npartitions=3)
assert len(ddf) == len(df)
def test_repr(func):
ddf = dg.from_pandas(points_df, npartitions=5)
for x in [ddf, ddf.value, ddf.geometry]:
text = func(x)
assert '5' in text
assert type(x).__name__ in text
def test_persist():
df = dg.from_pandas(points_df, npartitions=5)
df = dg.repartition(df, triangles)
df2 = df.persist()
assert not any(map(dask.core.istask, df2.dask.values()))
def test_head():
ddf = dg.from_pandas(points_df, npartitions=5)
head = ddf.head(5)
assert isinstance(head, gpd.GeoDataFrame)
assert_eq(head, points_df.head(5))