fix(cpp): treat a member-function declaration as a method, not a field

[rajatnagda45]

Problem

Inside a class body a virtual double area() = 0; (pure virtual), a virtual void draw(); and a plain prototype int run(); all parse as a field_declaration whose declarator is a function_declarator. The field branch recognised that shape only to skip emitting a type reference — it still emitted a node for every declarator as a data member (a bare area label with a defines/field edge).

Consequences:

• a pure-virtual interface class had no method nodes, only phantom fields;
• a header-declared method and its out-of-line/overriding definition did not share one node (area field vs .area() method);
• a call to such a method resolved to a field-shaped node.

Implementation

In the field-declaration branch, emit a function_declarator with the same shape a defined method uses — a .name() label, a method edge, and callable — so the declaration and its definition collapse to one method node (parity with the Java/TypeScript/Scala abstract-method contract). A genuine data member still gets its defines field edge.

Verification

• tests/test_cpp_method_declarations.py (new): pure-virtual / declared-virtual / prototype members become method nodes; a data member stays a field; the decl/def merge keeps one node and a call through it resolves. All fail on v8, pass with the fix.
• This corrects the label of every header-declared method, so a few existing cross-file assertions that expected the old bare bar/run field label now expect .bar()/.run() (test_cpp_objc_cross_file_calls, test_languages). The resolved call edges (source, target, confidence) are unchanged, and the decl/def merge still produces exactly one node carrying the definition site.
• uv run pytest tests/ — full suite green (6175 passed, 15 skipped).
• uv run ruff check . clean; uv run pyright adds zero new errors.

[github-actions]

Thanks for the pull request, @rajatnagda45. A maintainer will review it soon.

Want to talk it through while it is in review? Come join us on our Discord server. For longer-form discussion there is also GitHub Discussions.

A couple of things that speed up review: make sure the test suite passes on Python 3.10 and 3.13, and that the change keeps extraction deterministic.

[graphify-labs]

Graphify reviewed this change.

Worth a look — the grounded gate found no coupling regressions or blocking issues, but 1 advisory finding(s) below merit a look before merge.

---

Graphify review — findings

Fixes C++ member-function declarations so that prototypes, virtual f(); and pure-virtual virtual f() = 0; inside a class body (including pointer/reference-returning ones) are emitted as callable .name() nodes with a method edge instead of bare-named defines fields. Pure-virtual interfaces now get real method nodes, and a declaration shares one node with its out-of-line or in-class definition, so calls resolve to .bar() rather than a phantom bar field. Genuine data members still get their defines field edge, and existing cross-file call tests now expect the .name() labels.

Worth a look

• C++ function-pointer data members are classified as methods — graphify/extractors/engine.py:5149 · Escalate · medium · 2 independent checks
  • agreed by 2 of 2 members but NOT verified (no proof, no reproducing execution) — consensus is not a verdict; needs human review

Analysis details — impact, health, verification

Impact & health

Graphify review

Impact — 1128 functions depend on the 871 functions this change touches.

Health — this change adds coupling hotspots:

• new: _extract_generic() — 18 callers, 29 callees
• new: extract_js() — 87 callers, 4 callees
• new: extract_xaml() — 19 callers, 17 callees
• new: extract_objc() — 27 callers, 9 callees
• new: extract_julia() — 19 callers, 7 callees
• new: extract_cpp() — 31 callers, 3 callees
• new: extract_vue() — 10 callers, 7 callees
• new: walk() — 1 callers, 66 callees
• …and 9 more — each is listed as a finding

Verification — 1128 functions in the blast radius were not formally verified this run (proofs are advisory here).

Gate & verification

graphify gate

PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.

Advisory (not blocking):

• verification_scope: 1082 function(s) in the blast radius were not formally verified this run

Test selection

Test selection

27 of 313 test file(s) selected (9%) via static blast radius.

• tests/test_astro_extraction.py — impact
• tests/test_build.py — impact
• tests/test_cjs_module_extension.py — impact
• tests/test_cpp_method_declarations.py — impact, changed-test
• tests/test_cpp_nested_and_cli.py — impact
• tests/test_cpp_objc_cross_file_calls.py — impact, changed-test
• tests/test_dotnet.py — impact
• tests/test_extract.py — impact
• tests/test_extract_php_closures.py — impact
• tests/test_import_extension_resolution.py — impact
• tests/test_indirect_call_block_scoped_shadow.py — impact
• tests/test_indirect_dispatch.py — impact
• tests/test_indirect_dispatch_assign_return.py — impact
• tests/test_indirect_dispatch_getattr.py — impact
• tests/test_js_exported_scalar_bindings.py — impact
• tests/test_languages.py — impact, changed-test
• tests/test_multilang.py — impact
• tests/test_python_underscore_resolution.py — impact
• tests/test_rationale.py — impact
• tests/test_ruby_resolution.py — impact
• tests/test_scala_self_type.py — impact
• tests/test_swift_computed_properties.py — impact
• tests/test_swift_protocol_requirements.py — impact
• tests/test_trailing_newline_not_a_syntax_error.py — impact
• tests/test_ts_new_expression_calls.py — impact
• tests/test_typescript_module_extensions.py — impact
• tests/test_vue_extraction.py — impact

Selection is safe under the controlled-regression assumption; always-run tests + a periodic full run are the backstops. Advisory — it never changes the check verdict.

· 17 more finding(s) on lines outside this diff (see the check run).