def test_sjoin(translate):
if translate:
triangles2 = [
shapely.affinity.translate(t, 0.00001, 0) for t in triangles
]
else:
triangles2 = triangles
l = dg.repartition(points_df, triangles2)
r = dg.repartition(grid_df, coarse_grid)
result = dg.sjoin(l, r, how='inner', op='intersects')
expected = gpd.sjoin(points_df, grid_df, how='inner', op='intersects')
assert len(result) == len(expected)
assert (result._regions.area < (coarse_delta + 0.1)**2).all()
def test_repartition_single_element():
df = gpd.GeoDataFrame({
'x': [1, 2],
'geometry': [Point(0, 0), Point(1, 1)]
})
gdf = dg.repartition(df, triangles)
assert len(df) == 2
def test_sjoin_dask_normal():
df = dg.repartition(points_df, triangles)
result = dg.sjoin(df, grid_df, buffer=0)
expected = gpd.sjoin(points_df, grid_df)
assert_eq(result, expected)
df._regions.name = None
assert_eq(result._regions, df._regions)
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_repartition_pandas_expands_regions(translate):
if translate:
triangles2 = [
shapely.affinity.translate(t, 0.00001, 0) for t in triangles
]
else:
triangles2 = triangles
df = dg.repartition(grid_df, triangles2)
for i in range(df.npartitions):
part = df.get_partition(i)
geoms = part.geometry.compute()
assert geoms.within(part._regions.iloc[0]).all()
assert len(df) == len(grid_df)
def test_repartition():
df = points_df
df = dg.repartition(df, triangles).persist()
assert df.npartitions == 2
assert len(df) == len(points_df)
assert df._regions.iloc[0].equals(triangles[0])
assert df._regions.iloc[1].equals(triangles[1])
for i in range(df.npartitions):
part = df.get_partition(i)
geoms = part.geometry.compute()
assert geoms.within(part._regions.iloc[0]).all()
df2 = df.repartition(grid).persist()
assert len(df2) == len(points_df)
assert df2.npartitions == len(grid)
for i in range(df2.npartitions):
part = df2.get_partition(i)
geoms = part.geometry.compute()
assert geoms.within(part._regions.iloc[0]).all()
def test_sjoin_dask_normal():
df = dg.repartition(points_df, triangles)
assert isinstance(df.value, dd.Series)
assert isinstance(df[['value']], dd.DataFrame)
assert_eq((df.value > 0.5).value_counts(),
(points_df.value > 0.5).value_counts())
def test_repartition_trim():
polys = triangles + [Polygon([(10, 10), (10, 20), (20, 10)])]
df = dg.repartition(grid_df, polys)
assert df.npartitions == 2
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()))