fix: continue file tree searches after link cycles

pkg/filetree/depth_first_path_walker.go

@@ -6,6 +6,7 @@ import (
 	"sort"
 	"strings"
 
+	"github.com/anchore/stereoscope/internal/log"
 	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/stereoscope/pkg/filetree/filenode"
 )
@@ -78,6 +79,10 @@ func (w *DepthFirstPathWalker) Walk(from file.Path) (file.Path, *filenode.FileNo
 		currentPath = w.pathStack.Pop()
 
 		currentNode, err = w.tree.node(currentPath, linkStrat)
+		if errors.Is(err, ErrLinkCycleDetected) {
+			log.WithFields("path", currentPath, "error", err).Warn("skipping path with link cycle during file tree walk")
+			continue
+		}
 		if err != nil {
 			return "", nil, err
 		}

pkg/filetree/depth_first_path_walker_test.go

@@ -120,6 +120,48 @@ func TestDFS_WalkAll(t *testing.T) {
 	assertExpectedTraversal(t, possiblePaths, actualPaths)
 }
 
+func TestDFS_WalkAll_SkipsLinkCycles(t *testing.T) {
+	tr := New()
+
+	for _, p := range []file.Path{
+		"/usr/bin/a-before",
+		"/usr/bin/zz-after",
+	} {
+		_, err := tr.AddFile(p)
+		if err != nil {
+			t.Fatalf("failed to add path %q: %+v", p, err)
+		}
+	}
+
+	_, err := tr.AddSymLink("/usr/bin/xz", "/usr/bin/xzcat")
+	if err != nil {
+		t.Fatalf("could not setup link: %+v", err)
+	}
+	_, err = tr.AddSymLink("/usr/bin/xzcat", "/usr/bin/xz")
+	if err != nil {
+		t.Fatalf("could not setup link: %+v", err)
+	}
+
+	visited := file.NewPathSet()
+	walker := NewDepthFirstPathWalker(tr, func(path file.Path, _ filenode.FileNode) error {
+		visited.Add(path)
+		return nil
+	}, nil)
+
+	if err := walker.WalkAll(); err != nil {
+		t.Fatalf("could not walk: %+v", err)
+	}
+
+	for _, p := range []file.Path{
+		"/usr/bin/a-before",
+		"/usr/bin/zz-after",
+	} {
+		if !visited.Contains(p) {
+			t.Errorf("did not visit path %q", p)
+		}
+	}
+}
+
 func TestDFS_WalkAll_EarlyTermination(t *testing.T) {
 	tr, possiblePaths := dfsTestTree(t)
 

pkg/filetree/search.go

@@ -1,6 +1,7 @@
 package filetree
 
 import (
+	"errors"
 	"fmt"
 	"path"
 	"sort"
@@ -223,6 +224,10 @@ func (sc searchContext) firstMatchingReferences(glob string, entries []IndexEntr
 	var references []file.Resolution
 	for _, entry := range entries {
 		ref, err := sc.firstMatchingReference(glob, string(entry.RealPath))
+		if errors.Is(err, ErrLinkCycleDetected) {
+			log.WithFields("path", entry.RealPath, "error", err).Warn("skipping path with link cycle during file tree search")
+			continue
+		}
 		if err != nil {
 			return nil, err
 		}

pkg/filetree/search_test.go

@@ -406,6 +406,45 @@ func Test_searchContext_SearchByMIMEType(t *testing.T) {
 	}
 }
 
+func Test_searchContext_SearchByMIMEType_SkipsLinkCycles(t *testing.T) {
+	tree := New()
+	idx := NewIndex()
+	const mimeType = "application/x-executable"
+
+	for _, p := range []file.Path{
+		"/usr/bin/a-before",
+		"/usr/bin/zz-after",
+	} {
+		ref, err := tree.AddFile(p)
+		require.NoError(t, err)
+		require.NotNil(t, ref)
+		idx.Add(*ref, file.Metadata{MIMEType: mimeType})
+	}
+
+	for path, destination := range map[file.Path]file.Path{
+		"/usr/bin/xz":    "/usr/bin/xzcat",
+		"/usr/bin/xzcat": "/usr/bin/xz",
+	} {
+		ref, err := tree.AddSymLink(path, destination)
+		require.NoError(t, err)
+		require.NotNil(t, ref)
+		idx.Add(*ref, file.Metadata{MIMEType: mimeType, Type: file.TypeSymLink})
+	}
+
+	results, err := NewSearchContext(tree, idx).SearchByMIMEType(mimeType)
+	require.NoError(t, err)
+	require.Len(t, results, 2)
+
+	var paths []string
+	for _, result := range results {
+		paths = append(paths, string(result.RealPath))
+	}
+	require.ElementsMatch(t, []string{
+		"/usr/bin/a-before",
+		"/usr/bin/zz-after",
+	}, paths)
+}
+
 func Test_complexSymlinkPerformance(t *testing.T) {
 	tr := New()
 	idx := NewIndex()