Raise error for invalid reference date for encoding time units

[gcaria]

• Closes Better error message for setting encoding["units"]  #5199
• Tests added
• Passes pre-commit run --all-files
• User visible changes (including notable bug fixes) are documented in whats-new.rst

Although the error raised by this commit does not include the whole units string, I believe it is actually more useful and specific since it focuses on the part (reference date) that's actually causing the problem.
Also, the reference date is the only information available in coding.times._ensure_padded_year, so this is the simplest way of raising an error.

I had a look in tests/test_coding_times.py and could not find a test for this related error raising (since I'd put a test for this commit around there)

xarray/xarray/coding/times.py

Lines 127 to 129 in 234b40a

	matches = re.match(r"(.+) since (.+)", units)
	if not matches:
	raise ValueError(f"invalid time units: {units}")

Have I missed it?

EDIT: I've tried to substitute the error raise on line 129 with a pass and all the tests passed anyway.

[mathause]

Looks good, thanks. As usual a test would be superb! You can add a new test in

xarray/xarray/tests/test_coding_times.py

Line 816 in 44f4ae1

def test_encode_cf_datetime_overflow(shape):

It can be as easy as

with pytest.raises(ValueError, match="invalid reference date"):
    xr.coding.times.encode_cf_datetime([1, 2, 3], units='days since NO_YEAR')

[pep8speaks]

Hello @gcaria! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-05-13 17:43:10 UTC

[gcaria]

Done :) Since it was very little effort, I've also added a test for the other related error raising.

[mathause]

Sorry I wasn't entirely clear - I meant "the tests belong in this general vicinity" - could you put them in their own function? You can also give yourself credit by adding something to whats-new (up to you).

[dcherian]

Thanks @gcaria!

[gcaria]

Hi @dcherian, thanks for the merge, however I think there was one last edit left for this PR, i.e. put the tests in their own function. How should I go about this?

[max-sixty]

@gcaria no problem at all to submit another PR — the more the merrier