Allow string formatting of scalar DataArrays

[fmaussion]

• Closes Should str.format() work on xarray scalars? #5976
• Tests added
• Passes pre-commit run --all-files
• User visible changes (including notable bug fixes) are documented in whats-new.rst

This is a first try at formatting dataarray scalars. Here is the current behavior:

In [1]: import xarray as xr
   ...: import numpy as np

In [2]: a = np.array(1)
   ...: da = xr.DataArray(a)

In [3]: print(a)
1

In [4]: print(da)
<xarray.DataArray ()>
array(1)

In [5]: print('{}'.format(a))
1

In [6]: print('{}'.format(da))
<xarray.DataArray ()>
array(1)

In [7]: print('{:.3f}'.format(a))
1.000

In [8]: print('{:.3f}'.format(da))
1.000

In [9]: a = np.array([1, 2])
   ...: da = xr.DataArray(a)

In [10]: print('{}'.format(a))
[1 2]

In [11]: print('{}'.format(da))
<xarray.DataArray (dim_0: 2)>
array([1, 2])
Dimensions without coordinates: dim_0

In [12]: print('{:.3f}'.format(a))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-12-c5afc7863e89> in <module>
----> 1 print('{:.3f}'.format(a))

TypeError: unsupported format string passed to numpy.ndarray.__format__

In [13]: print('{:.3f}'.format(da))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-13-bddebd8462bd> in <module>
----> 1 print('{:.3f}'.format(da))

~/disk/Dropbox/HomeDocs/git/xarray/xarray/core/common.py in __format__(self, format_spec)
    162             return formatting.array_repr(self)
    163         # Else why fall back to numpy
--> 164         return self.values.__format__(format_spec)
    165 
    166     def _iter(self: Any) -> Iterator[Any]:

TypeError: unsupported format string passed to numpy.ndarray.__format__

I don't think there is any backwards compatibility issue but lets see if the tests pass

[fmaussion]

Note that there would be a way to change the behavior to:

print(da) -> lengthy repr
print('{}'.format(da) -> fall back to numpy.__format__

This would break backwards compatibility, but I think it would be my preference - there are some chances that it brakes some code (in documentation pages maybe?), but I don't think that many people rely on '{}'.format(da) to return __repr__ ...

[Illviljan]

Suggested change

	assert var.__format__("") == "<xarray.DataArray ()>\narray(0)"
	assert var.__format__("d") == "0"
	assert var.__format__(".2f") == "0.00"
	var = xr.DataArray([0])
	assert var.__format__("") == (
	"<xarray.DataArray (dim_0: 1)>\narray([0])"
	"\nDimensions without coordinates: dim_0"
	)
	assert var.__format__("") == var.__repr__()
	assert var.__format__("d") == "0"
	assert var.__format__(".2f") == "0.00"
	var = xr.DataArray([0])
	assert var.__format__("") == var.__repr__()

repr style can change over time. I think this should work.

[fmaussion]

Thanks that makes sense!

However I think I would like to simplify this even more and always call ndarray.__format__ when DataArray.__format__ is called. This won't be backwards compatible but I think this would be better and more predictable.

[fmaussion]

Side note: with this design (no call to repr() in format()) we would mimick numpy's behavior:

a = np.array([0.1, 0.2])
a.__format__("")
Out[16]: '[0.1 0.2]'
a.__repr__()
Out[17]: 'array([0.1, 0.2])'

However I'm quite surprised as to why .__format__("") works on non-scalar arrays but not with other specifiers - this is discussed somehow in numpy/numpy#5543 . I think it's okay for xarray to defer to whatever numpy is doing here, but of course we can also keep the status-quo, it's quite a small change after all.

[fmaussion]

@Illviljan and other devs - I still think this small change would be a nice addition for accessibility and teaching. Let me know if we can purse or if I should close this.

[Illviljan]

Thanks @fmaussion!