Wildcard in dimension name?

[AndrewILWilliams]

Hi there!

I have a very weird dataset, where sometimes the time coordinate is labelled time2 and sometimes is labelled time3 (I don't know why).

Is there a way to use a wildcard in selecting this dimension? E.g. da.mean("time*")?

[max-sixty]

Hi @AndrewWilliams3142 !

There's no query functionality on var names at the moment.

It would be possible to rename them with a fairly simple function on the user side.

In concept we could allow supplying a re.match object, though balanced against the larger API surface.

[AndrewILWilliams]

Thanks @max-sixty ! Yes I realised after I wrote this I could just query the da.dims and check like that.

I suppose it might be tricky to allow this in xarray in case there are multiple dimensions that match the wildcard?

[ahuang11]

There's this: https://xarray.pydata.org/en/stable/generated/xarray.Dataset.filter_by_attrs.html

But pandas, you could filter columns by regex like
https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.filter.html

So maybe xarray could also have something similar

[jthielen]

While not a fully generic wildcard handling, this specific issue with time coordinates (i.e., handling multiple coordinates of the same "physical type") is one of the motivating use cases for cf-xarray. With cf-xarray, this would look something like

da.cf.mean("T")