TagsCheck: handle license exception in first item of a grouping

[tmzullinger]

The change in 7d707f7 (TagsCheck: handle license exception in grouping, 2023-03-03) is insufficient when the exception is on the first item in a grouping, e.g.:

License: (Apache-2.0 WITH LLVM-exception OR NCSA) AND BSD-3-Clause

Adjust license_exception_regex to exclude a leading '(' as well as a trailing ')'.

This fixes a variation of the same issue which #1014 intended to fix. I don't know why I thought a leading parenthesis would be handle by the license_regex and not needed in license_exception_regex. (Just wishful thinking, perhaps.)

Here is the spec file used to create the additional test srpm:
valid-exception-begin-grouping.spec.txt

If this doesn't warrant a separate srpm from the previous test, I could replace the srpm in the existing test case. This new srpm has an exception at the start and end of the grouping (by virtue of being a grouping with only one entry).

As always, I'm happy to adjust things as needed. Thanks!

[danigm]

Looks good, but maybe it's worth to keep the \s exception in the regular expression? After the usage of this regex there's a str.strip so it's working as expected, but maybe it's a better regex this one:

r'([^(\s]+)\s(?:WITH|with)\s([^)\s]+)'

[tmzullinger]

Looks good, but maybe it's worth to keep the \s exception in the regular expression? After the usage of this regex there's a str.strip so it's working as expected, but maybe it's a better regex this one:

r'([^(\s]+)\s(?:WITH|with)\s([^)\s]+)'

Yes, I think you're right about that. Relying on a later strip() to clean up the whitespace has the potential to cause problems if the code is re-arranged. That's worth avoiding. I added a preparatory commit which restores the \s exclusion to the existing regex at the trailing end and then amended the commit which aims to fix the beginning to include it.

Thanks!