Add bundled or derived role predicate

README.md

@@ -99,6 +99,11 @@ and notice files. Roles follow a fixed order; evidence follows ancestor order
 and then filename matches. A `source` role describes layout and does not
 establish compilability or authorship.
 
+A role set cannot establish whether a file is first-party code. `main.go`
+has no roles, and `scripts/deploy.py` has `tooling` without `source`.
+Requiring `source` would drop both from analysis, even when they were written
+for the project.
+
 ## Usage
 
 ```go
@@ -136,6 +141,14 @@ no file reads or evidence allocations and does not require a second traversal.
 `Match` supports concurrent calls; content analysis and directory exclusion
 policies remain with the caller.
 
+`Set.BundledOrDerived()` reports whether any of `vendor`, `fixture`, `cache`
+or `build-output` is present. Callers can use this grouping when filtering
+project evidence. `generated` alone does not qualify, so generated lockfiles
+remain eligible as dependency evidence. Test and tooling paths also remain
+eligible unless they carry one of those four roles. A false result does not
+establish first-party ownership, and callers scanning licences or security
+issues may still need files for which it returns true.
+
 Paths use `/` separators. A trailing slash denotes a directory, so `vendor/`
 is a vendor directory while `vendor` alone is a filename. Empty paths,
 absolute paths, NUL bytes, repeated separators and `.` or `..` components

roles.go

@@ -72,6 +72,13 @@ func (s *Set) UnmarshalJSON(data []byte) error {
 // Has reports whether the set contains role. Unknown role names return false.
 func (s Set) Has(role Role) bool { return s&roleBit(role) != 0 }
 
+// BundledOrDerived reports whether the set contains Vendor, Fixture, Cache or
+// BuildOutput. Generated alone does not qualify. The result does not establish
+// ownership or whether a file is safe to skip.
+func (s Set) BundledOrDerived() bool {
+	return s.Has(Vendor) || s.Has(Fixture) || s.Has(Cache) || s.Has(BuildOutput)
+}
+
 // List returns the roles in deterministic order.
 func (s Set) List() []Role {
 	if s == 0 {

roles_test.go

@@ -140,6 +140,56 @@ func TestEmptyResultJSON(t *testing.T) {
 	}
 }
 
+func TestBundledOrDerived(t *testing.T) {
+	for _, tc := range []struct {
+		path string
+		want bool
+	}{
+		{"vendor/sqlite/LICENSE", true},
+		{"vendor/lib/main.go", true},
+		{"fixtures/input.json", true},
+		{"testdata/package-lock.json", true},
+		{".pytest_cache/v/cache/nodeids", true},
+		{"CMakeFiles/app.dir/main.o", true},
+		{"main.go", false},
+		{"src/parser.go", false},
+		{"alembic.ini", false},
+		{"scripts/deploy.py", false},
+		{"tests/test_app.py", false},
+		{"internal/parser_test.go", false},
+		{"examples/client/main.go", false},
+		{"service.pb.go", false},
+		{"package-lock.json", false},
+		{"src/app.min.js", false},
+	} {
+		t.Run(tc.path, func(t *testing.T) {
+			set, err := roles.Match(tc.path)
+			if err != nil {
+				t.Fatal(err)
+			}
+			if got := set.BundledOrDerived(); got != tc.want {
+				t.Errorf("BundledOrDerived() = %v for roles %v, want %v", got, set.List(), tc.want)
+			}
+		})
+	}
+}
+
+func TestBundledOrDerivedVendorContext(t *testing.T) {
+	classifier, err := roles.New([]roles.VendorRoot{{Path: "deps/local"}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, path := range []string{"deps/local/", "deps/local/src/lib.rs"} {
+		set, err := classifier.Match(path)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if !set.BundledOrDerived() {
+			t.Errorf("configured vendor path %q did not qualify: %v", path, set.List())
+		}
+	}
+}
+
 func TestSetJSON(t *testing.T) {
 	set, err := roles.Match("vendor/LICENSES/NOTICE")
 	if err != nil {