Add options to control expand/collapse of sections in display of Dataset and DataArray

[tomwhite]

• Closes FR: Provide option for collapsing the HTML display in notebooks #4229
• Tests added
• Passes pre-commit run --all-files
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

[dcherian]

Thanks @tomwhite !

I don't know this part of the code at all but maybe @jsignell or @benbovy can take a look.

[jsignell]

This seems like a nice complete solution to me! I just found the one typo :)

[tomwhite]

Thanks for the review @jsignell!

[max-sixty]

@tomwhite would you like to add a note to whatsnew, and then we can merge?

[tomwhite]

@max-sixty yes - I've rebased and updated whats-new.

[max-sixty]

Great — looks like one small linting issue, and then ping when green and we can hit the big button

[keewis]

that linting issue is was on master, too (there's a new version of black), so we might not have to actually fix it here.

[max-sixty]

Thanks @keewis . I'll merge this and then #5220 on green too

[max-sixty]

Thanks for the PR @tomwhite , we're happy to have you as a contributor!

[tomwhite]

Thank you!

[keewis]

this raises for

ds = xr.tutorial.open_dataset("rasm")
with xr.set_options(display_expand_data=False):
    display(ds.Tair)

i.e. with the data being a MemoryCachedArray. Should this be something like

Suggested change

	data_repr = inline_variable_array_repr(arr, OPTIONS["display_width"])
	data_repr = inline_variable_array_repr(arr.variable, OPTIONS["display_width"])

or should we let display_expand_data have no effect for MemoryCachedArray (the expanded repr is already really small)? The difference would be: [2029500 values with dtype=float64] for display_expand_data=True vs ... for display_expand_data=False

[max-sixty]

Thanks a lot for spotting @keewis .

No strong view on which of those we go ahead with — @tomwhite are you familiar with this? Otherwise @keewis do you want to make a call?

@tomwhite how would you feel about following up your first PR with a second? :)

[keewis]

I think it's more helpful to always have the expanded repr, but I'm not sure how complicated that special-case would be.

[tomwhite]

Thanks for finding this @keewis. I can put together a PR - I'm just trying to find a simple test case that exposes the problem (the tests don't currently use the tutorial as far as I can tell).

[keewis]

this should work with any opened file, so the test case should use either open_dataset or open_dataarray (creating the MemoryCachedArray by hand would also work but might be a bit more difficult). Something like

@requires_netCDF4
def test_repr_memory_cached_collapsed(temp_path):
    arr = xr.DataArray([0, 1], dims="x")
    arr.to_netcdf(temp_path, engine="netcdf4")

    with xr.open_dataarray(temp_path) as arr, xr.set_options(display_expand_data=True):
        # check the result of repr
        ...

might work? (temp_path should be a builtin pytest fixture to create a temporary file and return the path, but I'm not sure I got the name right)

[tomwhite]

I've opened #5235 with a possible fix.