Rust: Account for attribute expansions in path resolution

[paldepind]

Note, since getImmediateParent finds the enclosing ItemNode wherever it may be, simply removing items that have an expanded macro attribute is enough to enable path resolution to find things in inside the expansion.

Going per-commit shows a bug that my first approach introduced, and then a test and fix for that.

[Copilot]

Pull Request Overview

This PR improves Rust path resolution by accounting for attribute macro expansions. When items have attribute macros applied, they may be expanded/transformed, and the original items should be excluded from path resolution to prevent duplicate or incorrect resolution targets.

• Introduces a new predicate to identify items superceded by attribute macro expansions
• Filters out superceded items from the ItemNode class hierarchy
• Updates test expectations to reflect improved path resolution accuracy

Reviewed Changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
rust/ql/lib/codeql/rust/internal/PathResolution.qll	Adds logic to exclude items superceded by attribute macro expansions from path resolution
rust/ql/lib/utils/test/PathResolutionInlineExpectationsTest.qll	Removes macro filtering from test expectations
rust/ql/test/library-tests/path-resolution/main.rs	Adds test cases for attribute macro scenarios
rust/ql/test/library-tests/path-resolution/proc_macro.rs	Adds identity macro for testing
.expected files	Updated test expectations reflecting improved path resolution

[Copilot]

The word 'superceded' should be spelled 'superseded' in the predicate name, comments, and documentation.

Suggested change

	*Holdsif `n` is supercededby an attribute macro expansion. Thatis, `n` is
	*anitemoratransitivechildofanitemwithanattributemacro expansion.
	*/
	predicatesupercededByAttributeMacroExpansion(AstNoden){
	n.(Item).hasAttributeMacroExpansion()
	or
	exists(AstNodeparent|
	n.getParentNode()=parent and
	supercededByAttributeMacroExpansion(parent)and
	// Don't exclude expansions themselves as they supercede other nodes.
	notn=parent.(Item).getAttributeMacroExpansion() and
	// Don't consider attributes themselves to be superceded. E.g., in `#[a] fn
	// f() {}` the macro expansion supercedes `fn f() {}` but not `#[a]`.
	*Holdsif `n` is supersededby an attribute macro expansion. Thatis, `n` is
	*anitemoratransitivechildofanitemwithanattributemacro expansion.
	*/
	predicatesupersededByAttributeMacroExpansion(AstNoden){
	n.(Item).hasAttributeMacroExpansion()
	or
	exists(AstNodeparent|
	n.getParentNode()=parent and
	supersededByAttributeMacroExpansion(parent)and
	// Don't exclude expansions themselves as they supersede other nodes.
	notn=parent.(Item).getAttributeMacroExpansion() and
	// Don't consider attributes themselves to be superseded. E.g., in `#[a] fn
	// f() {}` the macro expansion supersedes `fn f() {}` but not `#[a]`.

[Copilot]

The function name 'commmentAt' has an extra 'm' and should be 'commentAt'.

[hvitved]

nit: superseded

[hvitved]

Actually, I think it is easier just to make getRelevantLocatable public (I already do that in on of my PRs).

[paldepind]

In this case I wanted to see the ItemNodes in a given range. getRelevantLocatable would give all Locateables wouldn't it?

[hvitved]

Yes; perhaps your use case was different from mine; I would add calls to Debug::getRelevantLocatable throughout the source code when debugging.

[hvitved]

We should do the same for the CFG (on a separate PR).