Support `<LITERAL ELLIPSIS>` in `doctest` module

[user202729]

Feature or enhancement

Proposal:

Currently, with # doctest: +ELLIPSIS, a ... can match anything. Sometimes, we want to test that a test matches a literal ellipsis.

I propose something similar to <LITERAL ELLIPSIS>. Possibly guarded behind another doctest flag.

'''
>>> '123...456' # doctest: +ELLIPSIS
'123<LITERAL ELLIPSIS>456'
'''

Motivation: SageMath doctest. See sagemath/sage#41276, sagemath/sage#41182.

Has this already been discussed elsewhere?

This is a minor feature, which does not need previous discussion elsewhere

Links to previous discussion of this feature:

No response

[picnixz]

Would this literal ellipsis only be effective with the ellipsis flag? becausr otherwise we could assume that \.\.\. is the literal ellipsis. That would avoid to parse and find a good name for the new token (and it is a long name)

[user202729]

Does \.\.\. work at the moment? I don't think so.

Plus, if \.\.\. matches literal ..., then what should match literal \.\.\.?

<LITERAL ELLIPSIS> is just a suggestion, similar to the existing <BLANKLINE>. I think it's descriptive, although might be a bit too long. (if it's not frequently needed then it doesn't really matter)

[picnixz]

Does ... work at the moment? I don't think so.

Note that I never said it worked, it was just a suggestion.

Plus, if ... matches literal ..., then what should match literal ...?

Well, you escape \ once again. \\.\\.\\. In general, we don't want to expand doctest so that it becomes too complicate. But now that I think about, \ may not be preferrable as a SyntaxWarning will be emitted as the docstring would contain backslashes and this would be very annoying, so a special token is likely better actualy (and wouldn't conflict with people using r-strings for their docstrings). Now, I think <ELLIPSIS> would be sufficient.