fix(parsers/c): guard macro-alias resolution against cyclic #defines

[gadievron]

fix(parsers/c): guard macro-alias resolution against cyclic #defines

Base: pr/zig-php-generic-anon · Depends-on: #127 · Type: bug fix · Finding: F4 (HIGH)

Stacked on #127 because that branch's C-parser rework already restructures
_resolve_call (adds receiver_type/is_member, _resolve_same_file); the
macro-alias recursion it carries is still unguarded. Targeting master instead
would hard-conflict with #127 in this same function. This layers only the cycle
guard on top of #127's signature.

What

parsers/c/call_graph_builder.py — _resolve_call threads an _alias_chain
visited-set through its macro-alias recursion, so a cyclic alias resolves to
None instead of recursing forever.

Why

macro_aliases is built from function-like #defines. A cyclic pair —
#define A(x) B(x) / #define B(x) A(x) → {"A":"B","B":"A"} — made
_resolve_call("A") recurse A→B→A→… with no guard → RecursionError.
build_call_graph iterates every function with no try/except, so one poisoned
define pair aborts the entire repository's C call-graph build (seen on vendored
LLVM in ziglang/zig). A self-alias was already safe via the resolved_name != call_name short-circuit; only 2+ node cycles triggered it.

Tests

tests/test_c_macro_alias_cycle.py (new): 2-node cycle, 3-node cycle, self-alias.

• RED (on the fix(zig,php parsers): attribute generic-container & anonymous-class methods correctly #127 base): RecursionError.
• GREEN: 3 passed; C call-graph suite 15 passed / 2 skipped.

Upstream coordination

The cyclic-macro recursion is live on master, on the C staging stack (#114/#121),
and on #127 — all unguarded. This guard is non-overlapping with those refactors;
sequence after #127 (or the C stack), re-anchoring onto the post-rework
_resolve_call.

Author notes

• Fix line: the if resolved_name not in _alias_chain: guard around the
  recursive _resolve_call(..., _alias_chain=_alias_chain).
• Input it now handles: cyclic #define macro pairs that crashed the whole-repo
  parse.
• Likely pushback: "set vs depth cap?" — a visited-set terminates on the actual
  cycle with no arbitrary limit, mirroring the file's other visited guards.

[gadievron]

Finding F4 (HIGH). ⚠️ Stacked on #127 (base pr/zig-php-generic-anon). The cyclic-macro RecursionError it fixes is live on master, the C staging stack (#114/#121), AND #127's C rework — all carry the unguarded _resolve_call macro-alias recursion. This adds only the _alias_chain cycle guard. Merge after #127, or rebase onto master.

[gadievron]

Rebased onto origin/master (post-release #133 / Wave 1). Conflicts resolved by UNION — both this PR's fix and the Wave-1 parser fixes are preserved; no Wave-1 change reverted (independently judge-verified). Force-pushed with --force-with-lease.

Rebased onto the newly-rebased #127 (kept stacked, base pr/zig-php-generic-anon). Merge AFTER #127. C tests green (16).

[gadievron]

Closing as already resolved on current master. This fix's behavior shipped via the parser-fix-stack release (#133/#134); confirmed by independent reproduction — the original bug (reconstructed from its bug-pipeline investigation) no longer manifests on master, not merely by line-presence. Part of a full validated sweep of the open-PR corpus.