Add Union-style indexing to .points, .multipoints, .lines, and …

…`.polygons` `GeoSeries` accessors (#685)

This PR uses the `column._meta` object to slice the sub-geometry accessors named in the title so that `.x`, `.y`, and `.xy` uses reflect any previous slicing that has occurred to the `GeoSeries`.

Fixes #683

Authors:
  - H. Thomson Comer (https://github.com/thomcom)

Approvers:
  - Michael Wang (https://github.com/isVoid)

URL: #685

python/cuspatial/cuspatial/core/geoseries.py

@@ -102,7 +102,9 @@ def __init__(
                 index = data.index
         if index is None:
             index = cudf.RangeIndex(0, len(column))
-        super().__init__(column, index, dtype, name, nan_as_null)
+        super().__init__(
+            column, index, dtype=dtype, name=name, nan_as_null=nan_as_null
+        )
 
     @property
     def type(self):
@@ -120,28 +122,42 @@ def type(self):
         return result
 
     class GeoColumnAccessor:
-        def __init__(self, list_series):
+        def __init__(self, list_series, meta):
             self._series = list_series
             self._col = self._series._column
+            self._meta = meta
+            self._type = Feature_Enum.POINT
 
         @property
         def x(self):
-            return cudf.Series(self._col.leaves().values[0::2])
+            return self.xy[::2].reset_index(drop=True)
 
         @property
         def y(self):
-            return cudf.Series(self._col.leaves().values[1::2])
+            return self.xy[1::2].reset_index(drop=True)
 
         @property
         def xy(self):
-            return cudf.Series(self._col.leaves().values)
+            types = self._meta.input_types
+            offsets = self._meta.union_offsets
+            indices = offsets[types == self._type.value]
+            result = self._col.take(indices._column).leaves().values
+            return cudf.Series(result)
 
     class MultiPointGeoColumnAccessor(GeoColumnAccessor):
+        def __init__(self, list_series, meta):
+            super().__init__(list_series, meta)
+            self._type = Feature_Enum.MULTIPOINT
+
         @property
         def geometry_offset(self):
             return cudf.Series(self._col.offsets.values)
 
     class LineStringGeoColumnAccessor(GeoColumnAccessor):
+        def __init__(self, list_series, meta):
+            super().__init__(list_series, meta)
+            self._type = Feature_Enum.LINESTRING
+
         @property
         def geometry_offset(self):
             return cudf.Series(self._col.offsets.values)
@@ -151,6 +167,10 @@ def part_offset(self):
             return cudf.Series(self._col.elements.offsets.values)
 
     class PolygonGeoColumnAccessor(GeoColumnAccessor):
+        def __init__(self, list_series, meta):
+            super().__init__(list_series, meta)
+            self._type = Feature_Enum.POLYGON
+
         @property
         def geometry_offset(self):
             return cudf.Series(self._col.offsets.values)
@@ -168,28 +188,34 @@ def points(self):
         """
         Access the `PointsArray` of the underlying `GeoArrowBuffers`.
         """
-        return self.GeoColumnAccessor(self._column.points)
+        return self.GeoColumnAccessor(self._column.points, self._column._meta)
 
     @property
     def multipoints(self):
         """
         Access the `MultiPointArray` of the underlying `GeoArrowBuffers`.
         """
-        return self.MultiPointGeoColumnAccessor(self._column.mpoints)
+        return self.MultiPointGeoColumnAccessor(
+            self._column.mpoints, self._column._meta
+        )
 
     @property
     def lines(self):
         """
         Access the `LineArray` of the underlying `GeoArrowBuffers`.
         """
-        return self.LineStringGeoColumnAccessor(self._column.lines)
+        return self.LineStringGeoColumnAccessor(
+            self._column.lines, self._column._meta
+        )
 
     @property
     def polygons(self):
         """
         Access the `PolygonArray` of the underlying `GeoArrowBuffers`.
         """
-        return self.PolygonGeoColumnAccessor(self._column.polygons)
+        return self.PolygonGeoColumnAccessor(
+            self._column.polygons, self._column._meta
+        )
 
     def __repr__(self):
         # TODO: Implement Iloc with slices so that we can use `Series.__repr__`
@@ -274,9 +300,11 @@ def __getitem__(self, item):
             )
 
             if isinstance(item, Integral):
-                return GeoSeries(column).to_shapely()
+                return GeoSeries(column, name=self._sr.name).to_shapely()
             else:
-                return GeoSeries(column, index=self._sr.index[indexes])
+                return GeoSeries(
+                    column, index=self._sr.index[indexes], name=self._sr.name
+                )
 
     def from_arrow(union):
         column = GeoColumn(
@@ -318,6 +346,7 @@ def to_geopandas(self, nullable=False):
         return gpGeoSeries(
             final_union_slice.to_shapely(),
             index=self.index.to_pandas(),
+            name=self.name,
         )
 
     def to_pandas(self):

python/cuspatial/cuspatial/tests/test_geodataframe.py

@@ -135,13 +135,19 @@ def test_interleaved_point(gpdf, polys):
     cugs = cugpdf["geometry"]
     gs = gpdf["geometry"]
     pd.testing.assert_series_equal(
-        cugs.points.x.to_pandas(),
+        cugs.points.x.to_pandas().reset_index(drop=True),
         gs[gs.type == "Point"].x.reset_index(drop=True),
     )
     pd.testing.assert_series_equal(
-        cugs.points.y.to_pandas(),
+        cugs.points.y.to_pandas().reset_index(drop=True),
         gs[gs.type == "Point"].y.reset_index(drop=True),
     )
+
+
+def test_interleaved_multipoint(gpdf, polys):
+    cugpdf = cuspatial.from_geopandas(gpdf)
+    cugs = cugpdf["geometry"]
+    gs = gpdf["geometry"]
     cudf.testing.assert_series_equal(
         cudf.Series.from_arrow(cugs.multipoints.x.to_arrow()),
         cudf.Series(
@@ -164,6 +170,11 @@ def test_interleaved_point(gpdf, polys):
             ).flatten()
         ),
     )
+
+
+def test_interleaved_lines(gpdf, polys):
+    cugpdf = cuspatial.from_geopandas(gpdf)
+    cugs = cugpdf["geometry"]
     cudf.testing.assert_series_equal(
         cudf.Series.from_arrow(cugs.lines.x.to_arrow()),
         cudf.Series(
@@ -178,6 +189,11 @@ def test_interleaved_point(gpdf, polys):
             dtype="float64",
         ),
     )
+
+
+def test_interleaved_polygons(gpdf, polys):
+    cugpdf = cuspatial.from_geopandas(gpdf)
+    cugs = cugpdf["geometry"]
     cudf.testing.assert_series_equal(
         cudf.Series.from_arrow(cugs.polygons.x.to_arrow()),
         cudf.Series(polys[:, 0], dtype="float64"),

python/cuspatial/cuspatial/tests/test_geoseries.py

@@ -347,6 +347,157 @@ def test_size(gs, series_slice):
     assert len(gi) == len(cugs)
 
 
+def test_geometry_point_slicing(gs):
+    cugs = cuspatial.from_geopandas(gs)
+    assert (cugs[:1].points.x == cudf.Series([-1])).all()
+    assert (cugs[:1].points.y == cudf.Series([0])).all()
+    assert (cugs[:1].points.xy == cudf.Series([-1, 0])).all()
+    assert (cugs[3:].points.x == cudf.Series([9])).all()
+    assert (cugs[3:].points.y == cudf.Series([10])).all()
+    assert (cugs[3:].points.xy == cudf.Series([9, 10])).all()
+    assert (cugs[0:4].points.x == cudf.Series([-1, 9])).all()
+    assert (cugs[0:4].points.y == cudf.Series([0, 10])).all()
+    assert (cugs[0:4].points.xy == cudf.Series([-1, 0, 9, 10])).all()
+
+
+def test_geometry_multipoint_slicing(gs):
+    cugs = cuspatial.from_geopandas(gs)
+    assert (cugs[:2].multipoints.x == cudf.Series([1, 3])).all()
+    assert (cugs[:2].multipoints.y == cudf.Series([2, 4])).all()
+    assert (cugs[:2].multipoints.xy == cudf.Series([1, 2, 3, 4])).all()
+    assert (cugs[2:].multipoints.x == cudf.Series([5, 7])).all()
+    assert (cugs[2:].multipoints.y == cudf.Series([6, 8])).all()
+    assert (cugs[2:].multipoints.xy == cudf.Series([5, 6, 7, 8])).all()
+    assert (cugs[0:4].multipoints.x == cudf.Series([1, 3, 5, 7])).all()
+    assert (cugs[0:4].multipoints.y == cudf.Series([2, 4, 6, 8])).all()
+    assert (
+        cugs[0:4].multipoints.xy == cudf.Series([1, 2, 3, 4, 5, 6, 7, 8])
+    ).all()
+
+
+def test_geometry_linestring_slicing(gs):
+    cugs = cuspatial.from_geopandas(gs)
+    assert (cugs[:5].lines.x == cudf.Series([11, 13])).all()
+    assert (cugs[:5].lines.y == cudf.Series([12, 14])).all()
+    assert (cugs[:5].lines.xy == cudf.Series([11, 12, 13, 14])).all()
+    assert (cugs[:6].lines.x == cudf.Series([11, 13, 15, 17, 19, 21])).all()
+    assert (cugs[:6].lines.y == cudf.Series([12, 14, 16, 18, 20, 22])).all()
+    assert (
+        cugs[:6].lines.xy
+        == cudf.Series([11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22])
+    ).all()
+    assert (cugs[7:].lines.x == cudf.Series([31, 33])).all()
+    assert (cugs[7:].lines.y == cudf.Series([32, 34])).all()
+    assert (cugs[7:].lines.xy == cudf.Series([31, 32, 33, 34])).all()
+    assert (cugs[6:].lines.x == cudf.Series([23, 25, 27, 29, 31, 33])).all()
+    assert (cugs[6:].lines.y == cudf.Series([24, 26, 28, 30, 32, 34])).all()
+    assert (
+        cugs[6:].lines.xy
+        == cudf.Series([23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34])
+    ).all()
+
+
+def test_geometry_polygon_slicing(gs):
+    cugs = cuspatial.from_geopandas(gs)
+    assert (cugs[:9].polygons.x == cudf.Series([35, 37, 39, 41, 35])).all()
+    assert (cugs[:9].polygons.y == cudf.Series([36, 38, 40, 42, 36])).all()
+    assert (
+        cugs[:9].polygons.xy
+        == cudf.Series([35, 36, 37, 38, 39, 40, 41, 42, 35, 36])
+    ).all()
+    assert (
+        cugs[:10].polygons.x
+        == cudf.Series(
+            [
+                35,
+                37,
+                39,
+                41,
+                35,
+                43,
+                45,
+                47,
+                43,
+                49,
+                51,
+                53,
+                49,
+                55,
+                57,
+                59,
+                55,
+                61,
+                63,
+                65,
+                61,
+            ]
+        )
+    ).all()
+    assert (
+        cugs[:10].polygons.y
+        == cudf.Series(
+            [
+                36,
+                38,
+                40,
+                42,
+                36,
+                44,
+                46,
+                48,
+                44,
+                50,
+                52,
+                54,
+                50,
+                56,
+                58,
+                60,
+                56,
+                62,
+                64,
+                66,
+                62,
+            ]
+        )
+    ).all()
+    assert (
+        cugs[11:].polygons.x
+        == cudf.Series([97, 99, 102, 101, 97, 106, 108, 110, 113, 106])
+    ).all()
+    assert (
+        cugs[11:].polygons.y
+        == cudf.Series([98, 101, 103, 108, 98, 107, 109, 111, 108, 107])
+    ).all()
+    assert (
+        cugs[11:].polygons.xy
+        == cudf.Series(
+            [
+                97,
+                98,
+                99,
+                101,
+                102,
+                103,
+                101,
+                108,
+                97,
+                98,
+                106,
+                107,
+                108,
+                109,
+                110,
+                111,
+                113,
+                108,
+                106,
+                107,
+            ]
+        )
+    ).all()
+
+
 def test_loc(gs):
     index = ["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l"]
     gs.index = index