`or`-ed license check breaks on licenses with "or" in the name about python-license-check HOT 6 CLOSED

@MartijnVdS I've opened a new PR (#104) which handles the situation you raised. Do you have any concerns with the tests/implementation there?

from python-license-check.

Based on the actions run prompting this issue, it appears that @MartijnVdS would also appreciate having GNU Lesser General Public License v2 or later (LGPLv2+) in the proposed whitelist.

from python-license-check.

Oops. Reverting.

from python-license-check.

@MartijnVdS Thanks for identifying this issue.

Concerns

However, I'm concerned that a patch version was used to revert #100, which is itself a breaking change. I sympathize with the criticism that the current or split is too broad, but an immediate revert disregards the or-split's value in resolving the issue of combinatorial explosion of multiple licenses for most projects.

Furthermore, OR is the keyword for separating license identifiers in the SPDX format which has become an international standard for tracking license requirements of software dependencies.

Proposed Solution

Surely a more productive fix would have been to add a test for GNU Library or Lesser General Public License (LGPL), then add it to a whitelist of licenses which would not be split by the character sequence or. Such an approach would satisfy both needs without too much difficulty.

from python-license-check.

Reading the standard, it only specifies OR in capital letters. Maybe not matching lowercase or would do the trick?

from python-license-check.

Maybe not matching lowercase or would do the trick?

That could work. The existing code normalizes the casing to lowercase before checking licenses, but the OR split* could happen before that normalization occurs.

• Note the spaces surrounding OR.

from python-license-check.