Subset should always return None if no value matches are found

[rcomer]

🚀 Pull Request

Description

Getting a subcube from a cube based on a coordinates values will usually return None if there is no value match. However, if the source coordinate is a vector, subset currently throws an exception:

import iris.cube
import iris.coords


vector_cube = iris.cube.Cube([0, 2], long_name="spam")
coord = iris.coords.DimCoord([0, 1], long_name="eggs")
vector_cube.add_dim_coord(coord, 0)

scalar_cube = vector_cube[0]

constraint = iris.Constraint(eggs=5)
target_coord = iris.coords.DimCoord(5, long_name="eggs")


for cube in [scalar_cube, vector_cube]:
    print(cube.extract(constraint))
    print(cube.subset(target_coord))

With main branch:

None
None
None
Traceback (most recent call last):
  File "extract_subset_nomatch.py", line 17, in <module>
    print(cube.subset(target_coord))
  File "[git-path]/iris/lib/iris/cube.py", line 2588, in subset
    result = self[full_slice]
  File "[git-path]/iris/lib/iris/cube.py", line 2457, in __getitem__
    dimension_mapping, data = iris.util._slice_data_with_keys(
  File "[git-path]/iris/lib/iris/util.py", line 778, in _slice_data_with_keys
    data = data[this_slice]
IndexError: arrays used as indices must be of integer (or boolean) type

This PR modifies subset so that the 4th case will also return None.

---

Consult Iris pull request check list

[trexfeathers]

LGTM, thanks @rcomer!

[rcomer]

Thanks @trexfeathers 🎉