From 651a4e15ac8fdb85dcc7e73ab9b4fbf27da46af3 Mon Sep 17 00:00:00 2001
From: Gergo Magyar <gergomagyar@icloud.com>
Date: Sat, 9 May 2026 06:49:18 +0100
Subject: [PATCH] fix(group): drop macro-style #include from consumer contracts

Tree-sitter's `(_) @import.source` wildcard matches the identifier node
of `#include PLATFORM_HEADER`, so the cleaned value `PLATFORM_HEADER`
slipped past the system-header / `..` filters and was emitted as a
permanently orphaned consumer contract (no file is named after a macro
identifier, so no provider can ever match). Add a shape guard that
skips cleaned values lacking both a path separator and an extension
dot, plus regression tests for single and multi-macro files.

Also document `IncludeExtractor.canExtract()` as unused by sync.ts
(gated via `config.detect.includes` instead) and kept solely for
ContractExtractor interface uniformity.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 .../group/extractors/include-extractor.ts     | 18 +++++++++
 .../test/unit/group/include-extractor.test.ts | 38 +++++++++++++++++++
 2 files changed, 56 insertions(+)

diff --git a/gitnexus/src/core/group/extractors/include-extractor.ts b/gitnexus/src/core/group/extractors/include-extractor.ts
index c27212a93..7bbfd61ed 100644
--- a/gitnexus/src/core/group/extractors/include-extractor.ts
+++ b/gitnexus/src/core/group/extractors/include-extractor.ts
@@ -296,6 +296,11 @@ function isLocalInclude(cleaned: string, suffixIndex: SuffixIndex): boolean {
 export class IncludeExtractor implements ContractExtractor {
   type = 'include' as const;
 
+  /**
+   * Always returns `true`. NOT called by `sync.ts`, which gates extraction via
+   * `config.detect.includes` instead (see `sync.ts:174`). Kept solely to satisfy
+   * the `ContractExtractor` interface so the type stays uniform across extractors.
+   */
   async canExtract(_repo: RepoHandle): Promise<boolean> {
     return true;
   }
@@ -544,6 +549,19 @@ export class IncludeExtractor implements ContractExtractor {
         // spurious consumer contracts.)
         if (cleaned.startsWith('../') || cleaned.startsWith('..\\')) continue;
 
+        // Skip macro-style includes: `#include PLATFORM_HEADER` parses as an
+        // identifier under tree-sitter's `(_) @import.source` wildcard. The
+        // identifier text passes the strip/clean step unchanged, so without
+        // this guard we would emit `include::platform_header` as a consumer
+        // contract — and no provider in any repo will ever expose a contract
+        // for a macro identifier (no file is named `PLATFORM_HEADER`). The
+        // contract would sit permanently orphaned in the registry. Real
+        // header references always contain a path separator (`/`, `\`) or an
+        // extension dot (`foo.h`), so an absent both is a reliable signal we
+        // are looking at a macro identifier. (PR #1156 follow-up review:
+        // macro includes emit orphaned consumer contracts.)
+        if (!/[./\\]/.test(cleaned)) continue;
+
         // Local resolution (PR #1156 review finding #4): only accept an
         // exact-suffix match on the *full* include path. The generic
         // suffixResolve() iterates all truncated suffixes, which would
diff --git a/gitnexus/test/unit/group/include-extractor.test.ts b/gitnexus/test/unit/group/include-extractor.test.ts
index 0a5a11bd3..3956cd5f2 100644
--- a/gitnexus/test/unit/group/include-extractor.test.ts
+++ b/gitnexus/test/unit/group/include-extractor.test.ts
@@ -398,6 +398,44 @@ int main() { return 0; }`,
     });
   });
 
+  // ---- PR #1156 follow-up: macro-style includes ----
+
+  describe('follow-up: macro-style #include emits no consumer contract', () => {
+    it('does not emit a consumer contract for `#include PLATFORM_HEADER` (no separator, no dot)', async () => {
+      // `#include PLATFORM_HEADER` parses under tree-sitter as an identifier
+      // node, slips past the existing system-header / `..` filters, and used
+      // to leak through as a permanently orphaned consumer contract because
+      // no file is ever named `PLATFORM_HEADER`. Verify the macro guard
+      // suppresses it while preserving the real cross-repo include.
+      writeFile(
+        'src/main.cpp',
+        `#include PLATFORM_HEADER
+#include "real/api.h"
+int main() { return 0; }`,
+      );
+
+      const contracts = await extractor.extract(null, tmpDir, makeRepo(tmpDir));
+      const consumers = contracts.filter((c) => c.role === 'consumer');
+
+      expect(consumers.map((c) => c.contractId)).toEqual(['include::real/api.h']);
+    });
+
+    it('skips multiple macro identifiers in the same translation unit', async () => {
+      writeFile(
+        'src/cfg.cpp',
+        `#include CONFIG_HEADER
+#include PLATFORM_HEADER
+#include ASSERT_H_
+int main(){return 0;}`,
+      );
+
+      const contracts = await extractor.extract(null, tmpDir, makeRepo(tmpDir));
+      const consumers = contracts.filter((c) => c.role === 'consumer');
+
+      expect(consumers).toHaveLength(0);
+    });
+  });
+
   // ---- PR #1156 follow-up: graph provider absolute paths ----
 
   describe('follow-up: extractProvidersGraph strips repo root from absolute paths', () => {