C++: Add model validation for constructor summary models - #21426
Conversation
Sorry, something went wrong.
There was a problem hiding this comment.
Pull request overview
Adds model validation for constructor summary models in C++, ensuring that the output column uses Argument[this] (or Argument[-1]) instead of ReturnValue, consistent with a similar validation already added for Java.
Changes:
- Adds getIncorrectConstructorSummaryOutput() predicate to detect constructor summary models that incorrectly use ReturnValue in the output column.
- Integrates the new predicate into the existing invalidModelRow query.
Sorry, something went wrong.
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
| ) | ||
| } | ||
|
|
||
| string getIncorrectConstructorSummaryOutput() { |
There was a problem hiding this comment.
Is there a reason this predicate is not private?
Sorry, something went wrong.
There was a problem hiding this comment.
Nope, I should have done that. I've done it now, and taken the liberty of fixing it for Java and C# in this PR as well. (I had to merge main in first.)
Sorry, something went wrong.
The output column of a summary model for a constructor should use Argument[this] instead of ReturnValue.
This didn't find anything, but it did for java (#21415). I also manually tested it by changing two constructor summary models (with a generic type and one with a non-generic type) to use ReturnValue and it was correctly identified when I ran cpp/ql/test/library-tests/dataflow/external-models/validatemodels.ql.