Fix Broken CI for 24.04 (#1338)

This PR fixes recent breakage due to pandas 2.2 upgrade.

Fixes #1341.

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

Approvers:
  - Mark Harris (https://github.com/harrism)
  - Paul Taylor (https://github.com/trxcllnt)

URL: #1338

python/cuspatial/cuspatial/core/binops/intersection.py

@@ -53,15 +53,16 @@ def pairwise_linestring_intersection(
 
     if len(linestrings1) == 0 and len(linestrings2) == 0:
         return (
-            cudf.Series([0]),
+            cudf.Series([0], dtype="i4"),
             GeoSeries([]),
             cudf.DataFrame(
                 {
                     "lhs_linestring_id": [],
                     "lhs_segment_id": [],
                     "rhs_linestring_id": [],
                     "rhs_segment_id": [],
-                }
+                },
+                dtype="i4",
             ),
         )
 

python/cuspatial/cuspatial/core/geoseries.py

@@ -698,7 +698,11 @@ def from_points_xy(cls, points_xy):
         GeoSeries:
             A GeoSeries made of the points.
         """
-        return cls(GeoColumn._from_points_xy(as_column(points_xy)))
+        coords_dtype = _check_coords_dtype(points_xy)
+
+        return cls(
+            GeoColumn._from_points_xy(as_column(points_xy, dtype=coords_dtype))
+        )
 
     @classmethod
     def from_multipoints_xy(cls, multipoints_xy, geometry_offset):
@@ -707,8 +711,9 @@ def from_multipoints_xy(cls, multipoints_xy, geometry_offset):
 
         Parameters
         ----------
-        points_xy: array-like
-            Coordinates of the points, interpreted as interleaved x-y coords.
+        multipoints_xy: array-like
+            Coordinates of the multipoints, interpreted as interleaved x-y
+            coords.
         geometry_offset: array-like
             Offsets indicating the starting index of the multipoint. Multiply
             the index by 2 results in the starting index of the coordinate.
@@ -729,9 +734,10 @@ def from_multipoints_xy(cls, multipoints_xy, geometry_offset):
         1    MULTIPOINT (2.00000 2.00000, 3.00000 3.00000)
         dtype: geometry
         """
+        coords_dtype = coords_dtype = _check_coords_dtype(multipoints_xy)
         return cls(
             GeoColumn._from_multipoints_xy(
-                as_column(multipoints_xy),
+                as_column(multipoints_xy, dtype=coords_dtype),
                 as_column(geometry_offset, dtype="int32"),
             )
         )
@@ -746,7 +752,8 @@ def from_linestrings_xy(
         Parameters
         ----------
         linestrings_xy : array-like
-            Coordinates of the points, interpreted as interleaved x-y coords.
+            Coordinates of the linestring, interpreted as interleaved x-y
+            coords.
         geometry_offset : array-like
             Offsets of the first coordinate of each geometry. The length of
             this array is the number of geometries.  Offsets with a difference
@@ -773,9 +780,10 @@ def from_linestrings_xy(
         0    LINESTRING (0 0, 1 1, 2 2, 3 3, 4 4, 5 5)
         dtype: geometry
         """
+        coords_dtype = _check_coords_dtype(linestrings_xy)
         return cls(
             GeoColumn._from_linestrings_xy(
-                as_column(linestrings_xy),
+                as_column(linestrings_xy, dtype=coords_dtype),
                 as_column(part_offset, dtype="int32"),
                 as_column(geometry_offset, dtype="int32"),
             )
@@ -822,9 +830,10 @@ def from_polygons_xy(
         0    POLYGON (0 0, 1 1, 2 2, 3 3, 4 4, 5 5)
         dtype: geometry
         """
+        coords_dtype = _check_coords_dtype(polygons_xy)
         return cls(
             GeoColumn._from_polygons_xy(
-                as_column(polygons_xy),
+                as_column(polygons_xy, dtype=coords_dtype),
                 as_column(ring_offset, dtype="int32"),
                 as_column(part_offset, dtype="int32"),
                 as_column(geometry_offset, dtype="int32"),
@@ -1001,15 +1010,14 @@ def reset_index(
                 # The columns of the `cudf.DataFrame` are the new
                 # columns of the `GeoDataFrame`.
                 columns = {
-                    col: cudf_result[col] for col in cudf_result.columns
+                    col: cudf_result[col]
+                    for col in cudf_result.columns
+                    if col is not None
                 }
                 geo_result = GeoDataFrame(columns)
                 geo_series.index = geo_result.index
                 # Add the original `GeoSeries` as a column.
-                if name:
-                    geo_result[name] = geo_series
-                else:
-                    geo_result[0] = geo_series
+                geo_result[name] = geo_series
                 return geo_result
         else:
             self.index = cudf_series.index
@@ -1483,3 +1491,10 @@ def distance(self, other, align=True):
         if other_is_scalar:
             res.index = self.index
         return res
+
+
+def _check_coords_dtype(coords):
+    if hasattr(coords, "dtype"):
+        return coords.dtype
+    else:
+        return "f8" if len(coords) == 0 else None

python/cuspatial/cuspatial/core/spatial/nearest_points.py

@@ -1,5 +1,6 @@
 import cupy as cp
 
+import cudf
 from cudf.core.column import as_column
 
 import cuspatial._lib.nearest_points as nearest_points
@@ -51,9 +52,9 @@ def pairwise_point_linestring_nearest_points(
 
     if len(points) == 0:
         data = {
-            "point_geometry_id": [],
-            "linestring_geometry_id": [],
-            "segment_id": [],
+            "point_geometry_id": cudf.Series([], dtype="i4"),
+            "linestring_geometry_id": cudf.Series([], dtype="i4"),
+            "segment_id": cudf.Series([], dtype="i4"),
             "geometry": GeoSeries([]),
         }
         return GeoDataFrame._from_data(data)

python/cuspatial/cuspatial/core/trajectory.py

@@ -122,6 +122,16 @@ def trajectory_bounding_boxes(num_trajectories, object_ids, points: GeoSeries):
     1     1.0     1.0     3.0     3.0
     """
 
+    if len(points) == 0:
+        return DataFrame(
+            {
+                "x_min": Series([], dtype=points.points.x.dtype),
+                "y_min": Series([], dtype=points.points.x.dtype),
+                "x_max": Series([], dtype=points.points.x.dtype),
+                "y_max": Series([], dtype=points.points.x.dtype),
+            }
+        )
+
     if len(points) > 0 and not contains_only_points(points):
         raise ValueError("`points` must only contain point geometries.")
 

python/cuspatial/cuspatial/tests/basicpreds/test_intersections.py

@@ -22,15 +22,16 @@ def test_empty():
     s1 = gpd.GeoSeries([])
     s2 = gpd.GeoSeries([])
 
-    expect_offset = pd.Series([0])
+    expect_offset = pd.Series([0], dtype="i4")
     expect_geom = gpd.GeoSeries([])
     expect_ids = pd.DataFrame(
         {
             "lhs_linestring_id": [],
             "lhs_segment_id": [],
             "rhs_linestring_id": [],
             "rhs_segment_id": [],
-        }
+        },
+        dtype="i4",
     )
 
     run_test(s1, s2, expect_offset, expect_geom, expect_ids)

python/cuspatial/cuspatial/tests/spatial/bounding/test_linestring_bounding_boxes.py

@@ -18,10 +18,10 @@ def test_linestring_bounding_boxes_empty():
         result,
         cudf.DataFrame(
             {
-                "minx": cudf.Series([]),
-                "miny": cudf.Series([]),
-                "maxx": cudf.Series([]),
-                "maxy": cudf.Series([]),
+                "minx": cudf.Series([], dtype=np.float64),
+                "miny": cudf.Series([], dtype=np.float64),
+                "maxx": cudf.Series([], dtype=np.float64),
+                "maxy": cudf.Series([], dtype=np.float64),
             }
         ),
     )

python/cuspatial/cuspatial/tests/spatial/join/test_point_in_polygon.py

@@ -247,7 +247,7 @@ def test_pip_bitmap_column_to_binary_array():
     )
     np.testing.assert_array_equal(got.copy_to_host(), expected)
 
-    col = cudf.Series([])._column
+    col = cudf.Series([], dtype="i8")._column
     got = pip_bitmap_column_to_binary_array(col, width=0)
     expected = np.array([], dtype="int8").reshape(0, 0)
     np.testing.assert_array_equal(got.copy_to_host(), expected)

python/cuspatial/cuspatial/tests/spatial/nearest_points/test_nearest_points.py

@@ -1,4 +1,5 @@
 import geopandas as gpd
+import pandas as pd
 import pytest
 import shapely
 from geopandas.testing import assert_geodataframe_equal
@@ -73,9 +74,9 @@ def test_empty_input():
     )
     expected = gpd.GeoDataFrame(
         {
-            "point_geometry_id": [],
-            "linestring_geometry_id": [],
-            "segment_id": [],
+            "point_geometry_id": pd.Series([], dtype="i4"),
+            "linestring_geometry_id": pd.Series([], dtype="i4"),
+            "segment_id": pd.Series([], dtype="i4"),
             "geometry": gpd.GeoSeries(),
         }
     )

python/cuspatial/cuspatial/tests/test_geodataframe.py

@@ -435,8 +435,20 @@ def test_reset_index(level, drop, inplace, col_level, col_fill):
         index=midx,
     )
     gdf = cuspatial.from_geopandas(gpdf)
-    expected = gpdf.reset_index(level, drop, inplace, col_level, col_fill)
-    got = gdf.reset_index(level, drop, inplace, col_level, col_fill)
+    expected = gpdf.reset_index(
+        level=level,
+        drop=drop,
+        inplace=inplace,
+        col_level=col_level,
+        col_fill=col_fill,
+    )
+    got = gdf.reset_index(
+        level=level,
+        drop=drop,
+        inplace=inplace,
+        col_level=col_level,
+        col_fill=col_fill,
+    )
     if inplace:
         expected = gpdf
         got = gdf

python/cuspatial/cuspatial/tests/test_geoseries.py

@@ -576,8 +576,10 @@ def test_reset_index(level, drop, name, inplace):
         [Point(0, 0), Point(0, 1), Point(2, 2), Point(3, 3)], index=midx
     )
     gs = cuspatial.from_geopandas(gps)
-    expected = gps.reset_index(level, drop, name, inplace)
-    got = gs.reset_index(level, drop, name, inplace)
+    expected = gps.reset_index(
+        level=level, drop=drop, name=name, inplace=inplace
+    )
+    got = gs.reset_index(level=level, drop=drop, name=name, inplace=inplace)
     if inplace:
         expected = gps
         got = gs