fix(native): skip callback-reference calls for this()/super() call expressions

[carlos-alm]

Summary

• In handle_call_expr (Rust JS extractor), when the callee node is this or super used as a function (this(b), super(a)), extract_callback_reference_calls was running and emitting identifier arguments as spurious dynamic calls
• The pts resolver then matched those names globally, producing false cross-file call edges (e.g. fun/fun.js:foo → spread/spread.js:b, classes2/classes2.js:G.constructor → prototypes3/prototypes3.js:a)
• Fix: add an early return in handle_call_expr for fn_node.kind() == "this" || fn_node.kind() == "super", mirroring the existing guard in the TS handleCallExpr (javascript.ts:1135)
• Add two regression tests covering both this(b) and super(a) argument non-emission

Root cause

The JS extractor (javascript.ts) has an early return at line 1135 when the function node type is this:

if (fn.type === 'this') {
  ctx.calls.push({ name: 'this', line: nodeStartLine(node) });
  return; // no further processing needed for this()-style calls
}

The Rust extractor (javascript.rs) only checked for fn_node.kind() == "import" before calling extract_callback_reference_calls, so this(b) / super(a) calls would emit b and a as dynamic calls. The pts resolver then resolved those names globally, finding same-named functions in sibling fixture files.

Test plan

• Two new Rust unit tests pass: this_call_args_do_not_emit_callback_reference_calls, super_call_args_do_not_emit_callback_reference_calls
• All 408 Rust unit tests pass
• All 180 JS test files pass (3048 tests)
• node scripts/parity-compare.mjs --langs jelly-micro eliminates all 6 targeted false-positive edges (13 → 12 divergences; the 6 fixed are the exact ones listed in fix(parity): native emits extra dynamic CHA edges for fun/classes2 fixtures #1543; the remaining 12 are pre-existing issues)

Closes #1543

[greptile-apps]

Greptile Summary

This PR addresses a previous review comment about a vacuously true assertion in the super_call_args_do_not_emit_callback_reference_calls test, and corrects an issue-reference comment from #1511 to #1543. The underlying production fix (early return in handle_call_expr for this/super callees) was already committed.

• The super(a, b) fixture replaces the old super(a); this.bb = b; fixture so that b is a genuine second argument to super() — both assertions now meaningfully exercise the guard.
• The parent class constructor is updated accordingly (constructor(c, d), this.dd = d) to remain a valid parse target.
• The issue reference in the this_call_args doc comment is corrected from #1511 to #1543.

Confidence Score: 5/5

Safe to merge — changes are test-only, directly address the prior review comment, and do not touch production logic.

The diff is entirely within the test module: one fixture update that turns a vacuously true assertion into a meaningful one, and one corrected issue-reference in a doc comment. The production guard was already in place; these changes only tighten the test coverage around it.

No files require special attention.

Important Files Changed

Filename	Overview
crates/codegraph-core/src/extractors/javascript.rs	Test-only changes: corrects a vacuously true assertion in the super test fixture by using super(a, b) so both arguments are real call args, and fixes an issue-reference comment (#1511 → #1543). No production logic changed in this diff.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[handle_call_expr] --> B{fn_node kind}
    B -->|import| C[handle_dynamic_import then return]
    B -->|this or super| D[early return - no callback-reference calls emitted]
    B -->|other| E[extract_call_info - push call record]
    E --> F[extract_callback_definition]
    F --> G[extract_callback_reference_calls]

    H[match_js_pts_bindings walk] -->|call_expression| I[collect_this_call_and_bindings]
    I -->|fn is this| J[push Call named this]
    I -->|fn is super| K[no call recorded - constructor delegation]

Loading

_{Reviews (3): Last reviewed commit: "test(native): strengthen super-call test..." | Re-trigger Greptile}

[greptile-apps]

Second assertion is vacuously true — doesn't exercise the fix

The second assert! checks that b is not emitted as a dynamic call, but b only appears in the assignment this.bb = b, not in any call expression. extract_callback_reference_calls only walks call_expression arguments, so b would never be emitted regardless of this fix. If the intent is to verify that both arguments to a multi-argument super call are suppressed, the fixture should use super(a, b) instead.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[carlos-alm]

Fixed — changed the fixture from super(a); this.bb = b; to super(a, b); so that b is a genuine call argument rather than an assignment target. Both assertions now actually exercise the fix by verifying that neither argument of a two-argument super() call is emitted as a dynamic callback-reference call.

[github-actions]

Codegraph Impact Analysis

1 functions changed → 0 callers affected across 0 files

• super_call_args_do_not_emit_callback_reference_calls in crates/codegraph-core/src/extractors/javascript.rs:4261 (0 transitive callers)

[carlos-alm]

@greptileai