license parsing corner case

[spotrh]

I think the license regexes needs some adjustment for a corner case. In Fedora, we have a package pending review (sip6) that has this as a license tag:

License: (GPLv2 or GPLv3) and (GPLv2+ with exceptions)

rpmlint returns:

sip6.src: W: invalid-license-exception exceptions)
sip6.src: W: invalid-license (GPLv2+

[marxin]

Note that rpmlint uses SPDX License List as defined here: https://spdx.org/licenses/. So my guess is that you need something like GPL-2.0-with-autoconf-exception, in your case sip6..

[Conan-Kudo]

Not for Fedora, it doesn't. It still uses the traditional RPM shorthand identifiers.

[marxin]

Not for Fedora, it doesn't. It still uses the traditional RPM shorthand identifiers.

What do you suggest for the mentioned license (License: (GPLv2 or GPLv3) and (GPLv2+ with exceptions)) ?
What can we do?

[Conan-Kudo]

I'm not sure. I think we're probably not handling whitespace in the field properly.

[marxin]

Oh, you are right, I haven't noticed the split error messages.

[marxin]

I can confirm the parsing can be improved to the following tokens: GPLv2, GPLv3 and GPLv2+ (+ exception=exceptions). Anyway, I'm not planning to work on that right now..

[smooge]

I have closed #769 as this is the ticket I need to track

[xsuchy]

Note that for Fedora you can use BNF grammar that comes in fedora-license-data package in a file /usr/share/fedora-license-data/grammar.lark.