Support glob patterns in open_datatree(group=...) for selective group loading

[aladinor]

Summary

When the group parameter contains glob metacharacters (*, ?, [), filter which groups are opened instead of re-rooting the tree. This avoids loading the entire hierarchy when only a subset is needed.

Use cases

• Radar data: xr.open_datatree("radar.nc", group="*/sweep_0") — load only the lowest elevation sweep from each volume scan
• CMIP archives: xr.open_datatree("cmip.zarr", group="*/historical/tas") — load only temperature across all models

Changes

• Added shared utilities _is_glob_pattern, _filter_group_paths, and _resolve_group_and_filter in common.py
• Updated NetCDF4, H5NetCDF, and Zarr backends to use a discover → filter → open pipeline
• Uses the same matching engine as DataTree.match() (PurePosixPath.match)
• Root (/) and all ancestors of matched nodes are always included to form a valid tree

Behavior summary

group value	Behavior
None	Load all groups (unchanged)
"VCP-34" (no glob chars)	Root selection (unchanged)
"*/sweep_0" (glob chars)	Filter mode — only matched groups + ancestors
Pattern matches nothing	Root-only tree

• Closes Support glob patterns in open_datatree(group=...) for selective group loading #11196
• Tests added
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

Test plan

• 27 new tests covering all backends (netCDF4, h5netcdf, zarr v2/v3)
• Unit tests for _is_glob_pattern, _filter_group_paths, _resolve_group_and_filter with *, ?, []
• Integration tests: glob match, no-match, data preservation, open_groups API
• Full test_backends_datatree.py suite passes (228 passed, 0 failures)
• Pre-commit checks pass

[kmuehlbauer]

@aladinor Thanks, that's great a feature. I'd instantly use it.

There might be some pitfalls if group names are containing one or more of the glob meta characters. Will this be handled, too?

my_nifty_group_with_a_star_*_01
my_nifty_group_with_a_star_*_11
my_nifty_group_with_a_star_*_12

[kmuehlbauer]

XRef: h5py/h5py#2059 for discussion of adding globbing in h5py

[aladinor]

@kmuehlbauer, thanks for taking the time to check this out.

my_nifty_group_with_a_star_01
my_nifty_group_with_a_star11
my_nifty_group_with_a_star*_12

This seems to be a strange way to name a group, but yes. It will work via the same character-class escape that fnmatch / PurePath.match supports.

For example, if we have something like this

  paths = ['/my_nifty_group_with_a_star_*_01',
           '/my_nifty_group_with_a_star_*_11',                                                                                                                                                                         
           '/my_nifty_group_with_a_star_*_12']      

We can use this pattern to get those groups "*star_[*]_*". This will match all 3. literal * via [*]

[kmuehlbauer]

@aladinor Thanks for adding the glob escapes. Is this ready from your side?

[aladinor]

Yep, it is ready to merge @kmuehlbauer

[kmuehlbauer]

This is looking good to me. Can't say much wrt typing, though.

[kmuehlbauer]

@pydata/xarray Another set of eyes much appreciated here. If there are no concerns, I'd move on and merge early next week. Thanks!

[shoyer]

I like the idea of this feature, but worry about ambiguity with the existing group argument -- are we sure that names with these characters are invalidate in netCDF/Zarr?

A safer strategy would be to make a new argument, something like group_filter.