spdx-utils: Fix license choice order-dependence in compound expressions #11101
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cow-lang
force-pushed
the
fix/10888-license-choice-order-independence
branch
from
a1d9a52 to
3296ba8
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #11101 +/- ##
============================================
+ Coverage 57.38% 57.45% +0.06%
- Complexity 1703 1708 +5
============================================
Files 346 346
Lines 12825 12844 +19
Branches 1214 1218 +4
============================================
+ Hits 7360 7379 +19
Misses 4997 4997
Partials 468 468
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
cow-lang
force-pushed
the
fix/10888-license-choice-order-independence
branch
7 times, most recently
from
52b1218 to
1666e6e
Compare
sschuberth
reviewed
Nov 19, 2025
Member
sschuberth
left a comment
sschuberth
left a comment
There was a problem hiding this comment.
Basically LGTM, let me make some slight adjustments straight to your branch to get it merged soon.
Comment on lines
349
to
350
| this != subExpression && | ||
| children.any { it.isSubExpression(subExpression) } |
Member
There was a problem hiding this comment.
Ok, so these two lines seem to be the only different to the solution in my branch (for I created a draft PR now for easier comparison).
Comment on lines
202
to
239
|
|
||
| "apply choice regardless of license order in OR expression" { | ||
| val expression = "(GPL-2.0-only OR MIT) AND Apache-2.0".toSpdx() | ||
|
|
||
| val choices1 = listOf( | ||
| SpdxLicenseChoice("GPL-2.0-only OR MIT".toSpdx(), "MIT".toSpdx()) | ||
| ) | ||
| val choices2 = listOf( | ||
| SpdxLicenseChoice("MIT OR GPL-2.0-only".toSpdx(), "MIT".toSpdx()) | ||
| ) | ||
|
|
||
| val result1 = expression.applyChoices(choices1) | ||
| result1 shouldBe "MIT AND Apache-2.0".toSpdx() | ||
|
|
||
| val result2 = expression.applyChoices(choices2) | ||
| result2 shouldBe "MIT AND Apache-2.0".toSpdx() | ||
| } | ||
|
|
||
| "resolve the choice for a sub-expression in a compound expression" { | ||
| // This test ensures that when applying a choice to a subexpression that appears multiple times, | ||
| // only the matching occurrence is replaced, not similar-looking expressions with WITH exceptions. | ||
| // This prevents the over-simplification bug where string replacement would affect too much. | ||
| val expression = "(CDDL-1.1 OR GPL-2.0-only) AND (CDDL-1.1 OR GPL-2.0-only WITH Classpath-exception-2.0)" | ||
| .toSpdx() | ||
| val choices = listOf( | ||
| SpdxLicenseChoice("CDDL-1.1 OR GPL-2.0-only".toSpdx(), "CDDL-1.1".toSpdx()) | ||
| ) | ||
|
|
||
| val result = expression.applyChoices(choices) | ||
|
|
||
| // The second OR expression should remain unchanged because it has a WITH exception modifier | ||
| // and doesn't exactly match the subexpression being chosen. | ||
| result shouldBe "CDDL-1.1 AND (CDDL-1.1 OR GPL-2.0-only WITH Classpath-exception-2.0)".toSpdx() | ||
| } |
Member
There was a problem hiding this comment.
These tests are new as well, but in exchange mine is omitted.
ORT had an architectural mismatch where `SpdxCompoundExpression` used
order-invariant equality ("MIT OR Apache-2.0" == "Apache-2.0 OR
MIT"), but the license choice matching used order-dependent matching
based on strings.
That is, if the expression contained "MIT OR Apache-2.0" but the user
gave "Apache-2.0 OR MIT", the string match failed, triggering a
fallback that rebuilt the expression by OR-ing all valid choices.
With multiple license choices, this caused exponential expression
growth (> 1000 characters in length).
This solution replaces the string-based algorithm with set-based
operations in `replaceSubexpressionWithChoice()`:
1. Decompose expressions into a set using `validChoices()`.
2. Remove dismissed choices from the set (no string operations).
3. Rebuild from remaining choices with controlled reduction.
4. Only recurse to AND-children if the sub-expression is actually
contained in a child to prevent inappropriate recursion for the
entire expression or derived sub-expression choices.
Since sets are order-invariant ({A,B} == {B,A}), both orderings
execute identical code paths, guaranteeing deterministic behavior.
Fixes oss-review-toolkit#10888.
Signed-off-by: Antoni <168914426+cow-lang@users.noreply.github.com>
Co-authored-by: Sebastian Schuberth <sebastian@doubleopen.org>
sschuberth
force-pushed
the
fix/10888-license-choice-order-independence
branch
from
1666e6e to
63242d9
Compare
sschuberth
approved these changes
Nov 19, 2025
Closed
Member
|
Merging despite the unrelated test failure(s). |
This was referenced Nov 20, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
ORT has a bug where license choice application is non-deterministic depending on the order of licenses in SPDX expressions. When "MIT OR Apache-2.0" appears in the scan result but the user configured "Apache-2.0 OR MIT", string matching fails, triggering a fallback that causes exponential expression growth.
Fixes: #10888