Does it matter if this internal function is used for contains_properly? It seems like it can be applied generically to all other predicates.

It is only used in contains_properly. I will move it to binpred_utils.py if another predicate depends on it.

pli? Please avoid unnecessary abbreviation.

It also seems to me that the other intersection should be called _basic_point_intersects since that computes point intersections while this is _basic_intersects.

pli: pairwise_linestring_intersection. This returns the unmodified results of the pairwise_linestring_intersection call, hence pli. I'll think about some renamings.

I think it would be nice to separate this file into test_[name_of_binpreds] for each tested binpreds.

This file only covers an extremely small range of within, overlaps, and intersects tests. I'm happy to write more tests in specific files once test dispatching with passing tests is merged in #1064.

Can we remove this now that #1093 is merged?

Probably, I'd like to leave that for a future issue/PR.

Created an issue to track this #1113

Can't this be a lot simpler? Any geometry, regardless of complexity, stores a flat array of coordinates. The number of points is just the length of the xy points array, isn't it?

The reason for this logic is because sizes returns the size of each row of the GeoSeries. For example

x = cuspatial.GeoSeries([
    LineString([(0, 0), (1, 1)]),
    LineString([(0, 0), (1, 1), (2, 2)]),
    LineString([(0, 0), (1, 1), (2, 2), (3, 3)]),
])
print(x.sizes)

0    2
1    3
2    4
dtype: int32

The .xy array in this case contains all of the points for all of the features in the GeoSeries:

x.lines.xy

0     0.0
1     0.0
2     1.0
3     1.0
4     0.0
5     0.0
6     1.0
7     1.0
8     2.0
9     2.0
10    0.0
11    0.0
12    1.0
13    1.0
14    2.0
15    2.0
16    3.0
17    3.0
dtype: float64

But has no information about the lengths of the individual geometries.

sum(x.sizes) == 9 and len(x.lines.xy) / 2 == 9. So why doesn't the latter serve the purpose?

For example, the sizes are used for each row of the GeoSeries to determine if the .contains (boundary inclusive) predicate has been met:

return contains + intersects >= rhs.sizes

Where contains is a cudf.Series of the number of points that satisfied .contains_properly for each row in the GeoSeries, and intersects is a cudf.Series of the number of points that satisfied .intersects for those same rows in the GeoSeries. So if I had a GeoSeries of a length-three polygon and a length 4 polygon, sizes would be

0    3
1    4
dtype: int32

and in the cast of a self-contains test contains would be

0    0
1    0
dtype: int32

and intersects would be

0    3
0    4
dtype: int32

so the predicate result of .contains is
`contains_properly + intersects >= rhs.sizes ==

0    True
1    True
dtype: bool

Does that properly answer your question? I'm not using the sum of the sizes in the GeoSeries, but each individually.