Fix group kwarg #338
Fix group kwarg #338
Conversation
TomNicholas
added
Kerchunk
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
for more information, see https://pre-commit.ci
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
| for key in refs.keys(): | ||
| # has to capture "foo/.zarray", but ignore ".zgroup", ".zattrs", and "subgroup/bar/.zarray" | ||
| # TODO this might be a sign that we should introduce a KerchunkGroupRefs type and cut down the references before getting to this point... | ||
| if key not in (".zgroup", ".zattrs", ".zmetadata"): |
There was a problem hiding this comment.
@norlandrhagen the only reason I needed the ".zmetadata" key in the check here is to make one kerchunk parquet test pass. Is it really the case that kerchunk parquet references use different keys than kerchunk json references? Or is that just a mistake in the fake kerchunk parquet data that we create and use in
There was a problem hiding this comment.
Hmm, I think we were relying on the parquet directory to have a .zmetadata file so that we could identify the directory containing parquet's as parquet. #278 (comment)
There was a problem hiding this comment.
Okay let's merge then.
There was a problem hiding this comment.
Oh I have another bug to fix first...
| found_var_names = [] | ||
| for key in refs.keys(): | ||
| # has to capture "foo/.zarray", but ignore ".zgroup", ".zattrs", and "subgroup/bar/.zarray" | ||
| # TODO this might be a sign that we should introduce a KerchunkGroupRefs type and cut down the references before getting to this point... |
There was a problem hiding this comment.
This would be cleaner, but is a refactor that can be done afterwards.
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
TomNicholas
temporarily deployed
to
test-release
GitHub Actions
Inactive
This also changes the expected behaviour of
open_virtual_dataset(file, group=None)(i.e. the default value of thegroupkwarg) when multiple groups are encountered. It now just opens the root group instead of raising, which is more consistent withxr.open_dataset.open_virtual_datasetfails when there is a subgroup #336docs/releases.rstNew functions/methods are listed inapi.rstNew functionality has documentation