Merge origin/main into fix/swift-init-extraction

Signed-off-by: Martin Vogel <martin.vogel.tech@gmail.com>

internal/cbm/lang_specs.c

@@ -1040,6 +1040,7 @@ static const char *solidity_field_types[] = {"state_variable_declaration", "stru
 static const char *solidity_call_types[] = {"call_expression", "call", "new_expression", NULL};
 static const char *solidity_import_types[] = {"import_directive", "import", "using_directive",
                                               NULL};
+static const char *solidity_decorator_types[] = {"modifier_invocation", NULL};
 static const char *solidity_branch_types[] = {"if_statement",    "for_statement",
                                               "while_statement", "do_while_statement",
                                               "try_statement",   NULL};
@@ -2117,7 +2118,7 @@ static const CBMLangSpec lang_specs[CBM_LANG_COUNT] = {
                            solidity_field_types, solidity_module_types, solidity_call_types,
                            solidity_import_types, empty_types, solidity_branch_types,
                            solidity_var_types, solidity_assign_types, solidity_throw_types, NULL,
-                           empty_types, NULL, NULL, tree_sitter_solidity, NULL},
+                           solidity_decorator_types, NULL, NULL, tree_sitter_solidity, NULL},
 
     // CBM_LANG_TYPST
     [CBM_LANG_TYPST] = {CBM_LANG_TYPST, typst_func_types, empty_types, empty_types,

src/pipeline/pass_parallel.c

@@ -3791,10 +3791,9 @@ static void resolve_file_usages(resolve_ctx_t *rc, resolve_worker_state_t *ws,
             if (tgt && cbm_suppress_cross_language_ref(lang, tgt->file_path)) {
                 continue;
             }
-            /* #1942/#1962: a bare Go reference can never denote a struct
-             * field; the member half of a selector may. */
-            if (tgt && cbm_go_suppress_bare_field_ref(lang == CBM_LANG_GO, usage->is_member_access,
-                                                      tgt->label)) {
+            /* #1942/#1962/#2666: a bare Go or C reference can never denote
+             * a struct field; the member half of a selector may. */
+            if (tgt && cbm_suppress_bare_field_ref(lang, usage->is_member_access, tgt->label)) {
                 continue;
             }
             if (usage->semantic_reference_blocked && (usage->semantic_reference_local_shadow ||
@@ -3920,9 +3919,9 @@ static void resolve_file_rw(resolve_ctx_t *rc, resolve_worker_state_t *ws, CBMFi
         if (cbm_suppress_cross_language_ref(lang, tgt->file_path)) {
             continue;
         }
-        /* #1942/#1962: a bare Go reference can never denote a struct field;
-         * a selector-LHS write (`t.err = x`) may bind it. */
-        if (cbm_go_suppress_bare_field_ref(lang == CBM_LANG_GO, rw->is_member_access, tgt->label)) {
+        /* #1942/#1962/#2666: a bare Go or C reference can never denote a
+         * struct field; a selector-LHS write (`t.err = x`) may bind it. */
+        if (cbm_suppress_bare_field_ref(lang, rw->is_member_access, tgt->label)) {
             continue;
         }
         const char *etype = rw->is_write ? "WRITES" : "READS";

src/pipeline/pass_usages.c

@@ -283,10 +283,9 @@ static int resolve_usage_edges(cbm_pipeline_ctx_t *ctx, const CBMFileResult *res
             if (tgt && cbm_suppress_cross_language_ref(lang, tgt->file_path)) {
                 continue;
             }
-            /* #1942/#1962: a bare Go reference can never denote a struct
-             * field; the member half of a selector may. */
-            if (tgt && cbm_go_suppress_bare_field_ref(lang == CBM_LANG_GO, usage->is_member_access,
-                                                      tgt->label)) {
+            /* #1942/#1962/#2666: a bare Go or C reference can never denote
+             * a struct field; the member half of a selector may. */
+            if (tgt && cbm_suppress_bare_field_ref(lang, usage->is_member_access, tgt->label)) {
                 continue;
             }
             if (usage->semantic_reference_blocked && (usage->semantic_reference_local_shadow ||
@@ -416,9 +415,9 @@ static int resolve_rw_edges(cbm_pipeline_ctx_t *ctx, const CBMFileResult *result
         if (cbm_suppress_cross_language_ref(lang, tgt->file_path)) {
             continue;
         }
-        /* #1942/#1962: a bare Go reference can never denote a struct field;
-         * a selector-LHS write (`t.err = x`) may bind it. */
-        if (cbm_go_suppress_bare_field_ref(lang == CBM_LANG_GO, rw->is_member_access, tgt->label)) {
+        /* #1942/#1962/#2666: a bare Go or C reference can never denote a
+         * struct field; a selector-LHS write (`t.err = x`) may bind it. */
+        if (cbm_suppress_bare_field_ref(lang, rw->is_member_access, tgt->label)) {
             continue;
         }
 

src/pipeline/pipeline.h

@@ -467,13 +467,14 @@ bool cbm_suppress_cross_language_suffix_match(CBMLanguage caller_lang, const cha
  * unit-tested in test_registry.c. */
 bool cbm_suppress_cross_language_ref(CBMLanguage caller_lang, const char *target_file_path);
 
-/* #1942: a bare (dot-less) Go reference can never denote a struct field —
- * field access is always a selector expression, and selector references
- * resolve on the LSP path. Drops a READS/WRITES/USAGE bind whose target is a
- * Field when the reference text carries no '.'. Go only: other OO languages
- * legitimately reference their own members bare inside method bodies. Pure;
- * unit-tested in test_registry.c. */
-bool cbm_go_suppress_bare_field_ref(bool is_go, bool is_member_access, const char *target_label);
+/* #1942/#2666: a bare Go or C reference can never denote a struct field —
+ * field access is always a selector expression (or, in C, a designator), and
+ * the member half of one is recorded as is_member_access. Drops a
+ * READS/WRITES/USAGE bind whose target is a Field when the reference was not
+ * the member half of a selector. Go and C only: other OO languages (and C++,
+ * which every `.h` is parsed as) legitimately reference their own members
+ * bare inside method bodies. Pure; unit-tested in test_registry.c. */
+bool cbm_suppress_bare_field_ref(CBMLanguage lang, bool is_member_access, const char *target_label);
 
 /* What to do with a call whose registry resolution landed on a Field node.
  * Pure; unit-tested in test_registry.c. */

src/pipeline/registry.c

@@ -1276,7 +1276,8 @@ bool cbm_suppress_cross_language_ref(CBMLanguage caller_lang, const char *target
     return true;
 }
 
-bool cbm_go_suppress_bare_field_ref(bool is_go, bool is_member_access, const char *target_label) {
+bool cbm_suppress_bare_field_ref(CBMLanguage lang, bool is_member_access,
+                                 const char *target_label) {
     /* #1942/#1962: a bare Go identifier can never denote a struct field —
      * field access is always a selector expression (x.f). The extractor
      * strips the receiver before the resolver runs (resolve_lhs_write_name
@@ -1284,10 +1285,20 @@ bool cbm_go_suppress_bare_field_ref(bool is_go, bool is_member_access, const cha
      * field_identifier), so the reference TEXT is always dot-less and cannot
      * carry the distinction — the recorded is_member_access shape can. Only a
      * reference that was never the member half of a selector is refused a
-     * Field bind. Go-gated: a C#/Java/C++/Python method body legitimately
+     * Field bind.
+     *
+     * #2666: C has the same rule. A struct member is named only as the member
+     * half of `a.b`, `a->b` or a designator `.b`, all recorded as
+     * field_identifier; a bare C identifier (a macro argument such as
+     * RUN_SELECTED_SUITE(mcp), a stray name in an unexpanded macro body) is
+     * never one. The call path already drops a bare C call onto a Field
+     * (cbm_call_onto_field_policy).
+     *
+     * Go- and C-gated: a C#/Java/C++/Python method body legitimately
      * references its own members bare (cp_reads_writes_cs_static_field pins
-     * that shape as required), so a global veto would break those languages. */
-    if (!is_go || is_member_access || !target_label) {
+     * that shape as required), so a global veto would break those languages.
+     * A `.h` is parsed as C++ and keeps that behavior. */
+    if ((lang != CBM_LANG_GO && lang != CBM_LANG_C) || is_member_access || !target_label) {
         return false;
     }
     return strcmp(target_label, "Field") == 0;

tests/test_extraction.c

@@ -5807,6 +5807,24 @@ TEST(extract_python_bare_decorator_route_rules) {
     PASS();
 }
 
+/* Issue #2090: Solidity modifier_invocation tracked as decorator. */
+TEST(solidity_modifier_decorator) {
+    CBMFileResult *r = extract("pragma solidity ^0.8.0;\n"
+                               "contract Vault {\n"
+                               "    modifier onlyOwner() { _; }\n"
+                               "    function withdraw() public onlyOwner {\n"
+                               "    }\n"
+                               "}\n",
+                               CBM_LANG_SOLIDITY, "t", "Vault.sol");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *m = find_def_by_name(r, "withdraw");
+    ASSERT_NOT_NULL(m);
+    ASSERT(decorators_contain(m, "onlyOwner"));
+    cbm_free_result(r);
+    PASS();
+}
+
 /* ── ArkTS (HarmonyOS .ets) ─────────────────────────────────────── */
 
 TEST(arkts_component_struct) {
@@ -14093,6 +14111,7 @@ SUITE(extraction) {
     RUN_TEST(extract_java_method_annotations_issue382);
     RUN_TEST(extract_python_mock_patch_is_not_route);
     RUN_TEST(extract_python_bare_decorator_route_rules);
+    RUN_TEST(solidity_modifier_decorator);
     RUN_TEST(arkts_component_struct);
     RUN_TEST(arkts_exported_struct_decorators);
     RUN_TEST(arkts_member_decorators);

tests/test_pipeline.c

@@ -11516,6 +11516,93 @@ TEST(pipeline_go_bare_ref_never_binds_field_parallel) {
     PASS();
 }
 
+/* #2666: tests/test_main.c runs every suite through a macro,
+ * RUN_SELECTED_SUITE(mcp); the argument is a bare C identifier that names no
+ * declaration, and the registry fallback bound it to whichever struct member
+ * was called `mcp`. toggle() writes a bare name the extractor cannot see
+ * declared (a macro-made global, say) that a struct member also carries.
+ * state.c holds the genuine member uses that must keep their edges.
+ * pad_files pushes the run past MIN_FILES_FOR_PARALLEL for the parallel twin. */
+static void write_c_bare_field_fixture(const char *tmp, int pad_files) {
+    write_temp_file(tmp, "include/suite.h",
+                    "typedef struct suite_state {\n"
+                    "    char mcp[64];\n"
+                    "    int count;\n"
+                    "    int enabled;\n"
+                    "} suite_state_t;\n");
+    write_temp_file(tmp, "tests/main.c",
+                    "#include \"suite.h\"\n"
+                    "#define RUN_SELECTED_SUITE(name) run_suite(#name)\n"
+                    "int run_suite(const char *name);\n"
+                    "int main(void) {\n"
+                    "    RUN_SELECTED_SUITE(mcp);\n"
+                    "    RUN_SELECTED_SUITE(count);\n"
+                    "    return 0;\n"
+                    "}\n"
+                    "void toggle(void) {\n"
+                    "    enabled = 1;\n"
+                    "}\n");
+    write_temp_file(tmp, "src/state.c",
+                    "#include \"suite.h\"\n"
+                    "int state_count(const suite_state_t *s) { return s->count; }\n"
+                    "void state_reset(suite_state_t *s) { s->enabled = 0; }\n");
+    for (int i = 0; i < pad_files; i++) {
+        char name[64];
+        char body[128];
+        snprintf(name, sizeof(name), "pad/filler%d.c", i);
+        snprintf(body, sizeof(body), "int filler%d(void) { return %d; }\n", i, i);
+        write_temp_file(tmp, name, body);
+    }
+}
+
+static int run_c_bare_field_case(const char *tmpl, int pad_files) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "%s", tmpl);
+    if (!cbm_mkdtemp(tmp)) {
+        FAIL("tmpdir");
+    }
+    write_c_bare_field_fixture(tmp, pad_files);
+
+    char db_path[512];
+    snprintf(db_path, sizeof(db_path), "%s/c_bare.db", tmp);
+    cbm_pipeline_t *p = cbm_pipeline_new(tmp, db_path, CBM_MODE_FULL);
+    ASSERT_NOT_NULL(p);
+    ASSERT_EQ(cbm_pipeline_run(p), 0);
+    const char *project = cbm_pipeline_project_name(p);
+
+    cbm_store_t *s = cbm_store_open_path(db_path);
+    ASSERT_NOT_NULL(s);
+
+    /* The fields must exist for the probes to mean anything. */
+    ASSERT_GTE(fixture_node_count(s, project, "include/suite.h", "mcp", "Field"), 1);
+    ASSERT_GTE(fixture_node_count(s, project, "include/suite.h", "count", "Field"), 1);
+    ASSERT_GTE(fixture_node_count(s, project, "include/suite.h", "enabled", "Field"), 1);
+    /* Reproduce-first: RED before the fix — the macro arguments bind the
+     * same-named struct members, and the bare write binds `enabled`. */
+    ASSERT_FALSE(cross_file_edge_exists(s, project, "main", "mcp", "USAGE"));
+    ASSERT_FALSE(cross_file_edge_exists(s, project, "main", "count", "USAGE"));
+    ASSERT_FALSE(cross_file_edge_exists(s, project, "toggle", "enabled", "WRITES"));
+    /* The member half of `s->count` / `s->enabled = 0` still binds the field. */
+    ASSERT_TRUE(cross_file_edge_exists(s, project, "state_count", "count", "USAGE"));
+    ASSERT_TRUE(cross_file_edge_exists(s, project, "state_reset", "enabled", "WRITES"));
+
+    cbm_store_close(s);
+    cbm_pipeline_free(p);
+    th_rmtree(tmp);
+    PASS();
+}
+
+TEST(pipeline_c_bare_ref_never_binds_field) {
+    /* Sequential path: resolve_usage_edges / resolve_rw_edges. */
+    return run_c_bare_field_case("/tmp/cbm_c_bare_XXXXXX", 0);
+}
+
+TEST(pipeline_c_bare_ref_never_binds_field_parallel) {
+    /* Parallel twin: resolve_file_usages / resolve_file_rw must consult the
+     * same predicate (#1928's lesson). */
+    return run_c_bare_field_case("/tmp/cbm_c_barep_XXXXXX", 52);
+}
+
 /* A C caller of a header function that shares its name with a struct member
  * elsewhere (kernel: include/linux/mm.h put_page beside the arm64
  * kvm_pgtable_mm_ops.put_page member). pad_files pushes the run past
@@ -28933,6 +29020,8 @@ SUITE(pipeline) {
     RUN_TEST(pipeline_rust_std_receiver_never_binds_project_method_parallel);
     RUN_TEST(pipeline_go_bare_ref_never_binds_field);
     RUN_TEST(pipeline_go_bare_ref_never_binds_field_parallel);
+    RUN_TEST(pipeline_c_bare_ref_never_binds_field);
+    RUN_TEST(pipeline_c_bare_ref_never_binds_field_parallel);
     RUN_TEST(pipeline_c_call_reaches_header_function_beside_field);
     RUN_TEST(pipeline_c_call_reaches_header_function_beside_field_parallel);
     RUN_TEST(pipeline_c_member_access_binds_field_by_object_type);

tests/test_registry.c

@@ -1245,17 +1245,36 @@ TEST(go_bare_ref_never_binds_field) {
      * trailing name; is_reference_node records the inner field_identifier),
      * so the selector-vs-bare distinction arrives as the recorded
      * is_member_access signal, never as a dot in the reference text. */
-    ASSERT_TRUE(cbm_go_suppress_bare_field_ref(true, false, "Field"));
+    ASSERT_TRUE(cbm_suppress_bare_field_ref(CBM_LANG_GO, false, "Field"));
     /* The member half of a selector may bind a field. */
-    ASSERT_FALSE(cbm_go_suppress_bare_field_ref(true, true, "Field"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_GO, true, "Field"));
     /* Bare references to non-fields are untouched. */
-    ASSERT_FALSE(cbm_go_suppress_bare_field_ref(true, false, "Variable"));
-    ASSERT_FALSE(cbm_go_suppress_bare_field_ref(true, false, "Function"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_GO, false, "Variable"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_GO, false, "Function"));
     /* Other languages reference their own members bare inside methods —
      * never suppressed (cp_reads_writes_cs_static_field pins the C# shape). */
-    ASSERT_FALSE(cbm_go_suppress_bare_field_ref(false, false, "Field"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_CSHARP, false, "Field"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_JAVA, false, "Field"));
     /* Degenerate input → nothing to judge. */
-    ASSERT_FALSE(cbm_go_suppress_bare_field_ref(true, false, NULL));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_GO, false, NULL));
+    PASS();
+}
+
+TEST(c_bare_ref_never_binds_field) {
+    /* #2666: a C struct member is named only as the member half of `a.b`,
+     * `a->b` or a designator `.b` (field_identifier → is_member_access). A
+     * bare C identifier — a macro argument like RUN_SELECTED_SUITE(mcp) — is
+     * never one, so it must not bind a same-named Field. */
+    ASSERT_TRUE(cbm_suppress_bare_field_ref(CBM_LANG_C, false, "Field"));
+    /* The member half still binds (through the C owner rows upstream). */
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_C, true, "Field"));
+    /* A bare C name keeps binding globals, functions and types. */
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_C, false, "Variable"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_C, false, "Function"));
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_C, false, "Type"));
+    /* C++ (and every `.h`, parsed as C++): a method body names its own
+     * members bare, so the bind is kept. */
+    ASSERT_FALSE(cbm_suppress_bare_field_ref(CBM_LANG_CPP, false, "Field"));
     PASS();
 }
 
@@ -2571,6 +2590,7 @@ SUITE(registry) {
     RUN_TEST(registry_tie_prefers_csharp_implementation_over_ref_stub);
     RUN_TEST(cross_language_ref_drops_go_vs_c);
     RUN_TEST(go_bare_ref_never_binds_field);
+    RUN_TEST(c_bare_ref_never_binds_field);
     RUN_TEST(call_onto_field_follows_language_and_shape);
     RUN_TEST(c_bare_call_does_not_weigh_field_namesakes);
     RUN_TEST(c_arrow_member_call_is_c_family_and_arrow_only);