[BUG]: cuDF errors on GeoDataFrame Print

[jarmak-nv]

Version

23.04

On which installation method(s) does this occur?

Conda

Describe the issue

After generating a GDF - printing throws an error:

TypeError: data type 'geometry' not understood

Minimum reproducible example

import geopandas as gpd
import numpy as np
import cuspatial
from shapely.geometry import LineString

# Make some lines
xmin, ymin, xmax, ymax = -180, -90, 180, 90
n_points = np.random.randint(2, 100, size=1000)
x_points = np.random.uniform(xmin, xmax, size=(1000, np.max(n_points)))
y_points = np.random.uniform(ymin, ymax, size=(1000, np.max(n_points)))
lines = [LineString(np.column_stack((x[:n], y[:n]))) for x, y, n in zip(x_points, y_points, n_points)]

# Make a gdf, then cuspatial gdf
gdf = gpd.GeoDataFrame(geometry=lines)
cuspatial_gdf = cuspatial.from_geopandas(gdf)
print(cuspatial_gdf)

Relevant log output

TypeError                                 Traceback (most recent call last)
File ~/miniconda3/envs/rapids-23.04-release2/lib/python3.10/site-packages/IPython/core/formatters.py:344, in BaseFormatter.__call__(self, obj)
    342     method = get_real_method(obj, self.print_method)
    343     if method is not None:
--> 344         return method()
    345     return None
    346 else:

File ~/miniconda3/envs/rapids-23.04-release2/lib/python3.10/site-packages/nvtx/nvtx.py:101, in annotate.__call__..inner(*args, **kwargs)
     98 @wraps(func)
     99 def inner(*args, **kwargs):
    100     libnvtx_push_range(self.attributes, self.domain.handle)
--> 101     result = func(*args, **kwargs)
    102     libnvtx_pop_range(self.domain.handle)
    103     return result

File ~/miniconda3/envs/rapids-23.04-release2/lib/python3.10/site-packages/cudf/core/dataframe.py:1906, in DataFrame._repr_latex_(self)
   1904 @_cudf_nvtx_annotate
   1905 def _repr_latex_(self):
-> 1906     return self._get_renderable_dataframe().to_pandas()._repr_latex_()

File ~/miniconda3/envs/rapids-23.04-release2/lib/python3.10/site-packages/cudf/core/dataframe.py:1875, in DataFrame._get_renderable_dataframe(self)
   1873     upper = cudf.concat([upper_left, upper_right], axis=1)
   1874     lower = cudf.concat([lower_left, lower_right], axis=1)
...
File column.pyx:380, in cudf._lib.column.Column._view()

File types.pyx:241, in cudf._lib.types.dtype_to_data_type()

TypeError: data type 'geometry' not understood

Environment details

No response

Other/Misc.

After erroring, in a Jupyter Notebook it will still then print out the GDF info.

                                              geometry
0    LINESTRING (-45.35532 -6.81397, 20.22481 -61.0...
1    LINESTRING (8.74951 -82.15693, -76.63895 -31.9...
2    LINESTRING (-69.24599 -77.87897, -42.02738 -11...
3    LINESTRING (-84.89946 -44.72954, -29.47745 -64...
4    LINESTRING (162.67521 -50.75755, -84.80769 82....
..                                                 ...
995  LINESTRING (-167.43410 19.66316, -158.42750 13...
996  LINESTRING (-9.51915 15.09780, 144.76755 68.10...
997  LINESTRING (61.13300 -80.36317, 28.90036 58.69...
998  LINESTRING (20.38292 -71.43508, -64.48469 11.0...
999  LINESTRING (-14.11138 30.30998, -88.91027 -88....

[1000 rows x 1 columns]
(GPU)

[thomcom]

This error has to do with how cudf slices up dataframes with more rows than pd.options.display.max_rows. We need to implement _get_renderable_dataframe to avoid the issue.

You can avoid the issue by simply slicing your own dataframe before printing it, or using .head() or .tail()