fix: separate golang license caches from mod dir (#2852)

Signed-off-by: Keith Zantow <kzantow@gmail.com>

syft/source/directorysource/cache_excludes.go

@@ -0,0 +1,33 @@
+package directorysource
+
+import (
+	"os"
+	"strings"
+
+	"github.com/anchore/syft/internal/cache"
+	"github.com/anchore/syft/internal/log"
+	"github.com/anchore/syft/syft/internal/fileresolver"
+)
+
+// we do not want to cache things and then subsequently scan them, if, for example a user runs `syft /` twice
+func excludeCachePathVisitors() []fileresolver.PathIndexVisitor {
+	var out []fileresolver.PathIndexVisitor
+	for _, dir := range cache.GetManager().RootDirs() {
+		out = append(out, excludeCacheDirPathVisitor{
+			dir: dir,
+		}.excludeCacheDir)
+	}
+	return out
+}
+
+type excludeCacheDirPathVisitor struct {
+	dir string
+}
+
+func (d excludeCacheDirPathVisitor) excludeCacheDir(_, path string, _ os.FileInfo, _ error) error {
+	if strings.HasPrefix(path, d.dir) {
+		log.Tracef("skipping cache path: %s", path)
+		return fileresolver.ErrSkipPath
+	}
+	return nil
+}

syft/source/directorysource/directory_source.go

@@ -142,6 +142,9 @@ func (s *directorySource) FileResolver(_ source.Scope) (file.Resolver, error) {
 			return nil, err
 		}
 
+		// this should be the only file resolver that might have overlap with where files are cached
+		exclusionFunctions = append(exclusionFunctions, excludeCachePathVisitors()...)
+
 		res, err := fileresolver.NewFromDirectory(s.config.Path, s.config.Base, exclusionFunctions...)
 		if err != nil {
 			return nil, fmt.Errorf("unable to create directory resolver: %w", err)