change dir resolver to include virtual path (#2259)

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

syft/file/cataloger/internal/all_regular_files_test.go

@@ -48,7 +48,7 @@ func Test_allRegularFiles(t *testing.T) {
 				return r
 			},
 			wantRealPaths:    strset.New("file1.txt", "nested/file2.txt"),
-			wantVirtualPaths: strset.New("nested/linked-file1.txt"),
+			wantVirtualPaths: strset.New("file1.txt", "nested/file2.txt", "nested/linked-file1.txt"),
 		},
 	}
 	for _, tt := range tests {

syft/file/location.go

@@ -68,6 +68,7 @@ func NewLocation(realPath string) Location {
 			Coordinates: Coordinates{
 				RealPath: realPath,
 			},
+			VirtualPath: realPath,
 		},
 		LocationMetadata: LocationMetadata{
 			Annotations: map[string]string{},
@@ -94,6 +95,7 @@ func NewLocationFromCoordinates(coordinates Coordinates) Location {
 	return Location{
 		LocationData: LocationData{
 			Coordinates: coordinates,
+			VirtualPath: coordinates.RealPath,
 		},
 		LocationMetadata: LocationMetadata{
 			Annotations: map[string]string{},
@@ -137,6 +139,7 @@ func NewLocationFromDirectory(responsePath string, ref file.Reference) Location
 			Coordinates: Coordinates{
 				RealPath: responsePath,
 			},
+			VirtualPath: responsePath,
 			ref:         ref,
 		},
 		LocationMetadata: LocationMetadata{
@@ -147,9 +150,6 @@ func NewLocationFromDirectory(responsePath string, ref file.Reference) Location
 
 // NewVirtualLocationFromDirectory creates a new Location representing the given path (extracted from the Reference) relative to the given directory with a separate virtual access path.
 func NewVirtualLocationFromDirectory(responsePath, virtualResponsePath string, ref file.Reference) Location {
-	if responsePath == virtualResponsePath {
-		return NewLocationFromDirectory(responsePath, ref)
-	}
 	return Location{
 		LocationData: LocationData{
 			Coordinates: Coordinates{

syft/internal/fileresolver/container_image_all_layers_test.go

@@ -93,9 +93,7 @@ func TestAllLayersResolver_FilesByPath(t *testing.T) {
 			img := imagetest.GetFixtureImage(t, "docker-archive", "image-symlinks")
 
 			resolver, err := NewFromContainerImageAllLayers(img)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not create resolver: %+v", err)
-			}
 
 			hasPath := resolver.HasPath(c.linkPath)
 			if !c.forcePositiveHasPath {
@@ -109,9 +107,7 @@ func TestAllLayersResolver_FilesByPath(t *testing.T) {
 			}
 
 			refs, err := resolver.FilesByPath(c.linkPath)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not use resolver: %+v", err)
-			}
 
 			if len(refs) != len(c.resolutions) {
 				t.Fatalf("unexpected number of resolutions: %d", len(refs))
@@ -207,14 +203,10 @@ func TestAllLayersResolver_FilesByGlob(t *testing.T) {
 			img := imagetest.GetFixtureImage(t, "docker-archive", "image-symlinks")
 
 			resolver, err := NewFromContainerImageAllLayers(img)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not create resolver: %+v", err)
-			}
 
 			refs, err := resolver.FilesByGlob(c.glob)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not use resolver: %+v", err)
-			}
 
 			if len(refs) != len(c.resolutions) {
 				t.Fatalf("unexpected number of resolutions: %d", len(refs))

syft/internal/fileresolver/container_image_squash_test.go

@@ -74,9 +74,7 @@ func TestImageSquashResolver_FilesByPath(t *testing.T) {
 			img := imagetest.GetFixtureImage(t, "docker-archive", "image-symlinks")
 
 			resolver, err := NewFromContainerImageSquash(img)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not create resolver: %+v", err)
-			}
 
 			hasPath := resolver.HasPath(c.linkPath)
 			if !c.forcePositiveHasPath {
@@ -90,9 +88,7 @@ func TestImageSquashResolver_FilesByPath(t *testing.T) {
 			}
 
 			refs, err := resolver.FilesByPath(c.linkPath)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not use resolver: %+v", err)
-			}
 
 			expectedRefs := 1
 			if c.resolvePath == "" {
@@ -187,14 +183,10 @@ func TestImageSquashResolver_FilesByGlob(t *testing.T) {
 			img := imagetest.GetFixtureImage(t, "docker-archive", "image-symlinks")
 
 			resolver, err := NewFromContainerImageSquash(img)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not create resolver: %+v", err)
-			}
 
 			refs, err := resolver.FilesByGlob(c.glob)
-			if err != nil {
+			require.NoError(t, err)
-				t.Fatalf("could not use resolver: %+v", err)
-			}
 
 			expectedRefs := 1
 			if c.resolvePath == "" {

syft/internal/fileresolver/directory.go

@@ -264,8 +264,9 @@ func (r *Directory) FilesByMIMEType(types ...string) ([]file.Location, error) {
 		if uniqueFileIDs.Contains(*refVia.Reference) {
 			continue
 		}
-		location := file.NewLocationFromDirectory(
+		location := file.NewVirtualLocationFromDirectory(
 			r.responsePath(string(refVia.Reference.RealPath)),
+			r.responsePath(string(refVia.RequestPath)),
 			*refVia.Reference,
 		)
 		uniqueFileIDs.Add(*refVia.Reference)

syft/internal/fileresolver/directory_test.go

@@ -71,7 +71,7 @@ func TestDirectoryResolver_FilesByPath_request_response(t *testing.T) {
 		base                string
 		input               string
 		expectedRealPath    string
-		expectedVirtualPath string
+		expectedVirtualPath string // note: if empty it will be assumed to match the expectedRealPath
 	}{
 		{
 			name:             "relative root, relative request, direct",
@@ -489,6 +489,9 @@ func TestDirectoryResolver_FilesByPath_request_response(t *testing.T) {
 	}
 	for _, c := range cases {
 		t.Run(c.name, func(t *testing.T) {
+			if c.expectedVirtualPath == "" {
+				c.expectedVirtualPath = c.expectedRealPath
+			}
 
 			// we need to mimic a shell, otherwise we won't get a path within a symlink
 			targetPath := filepath.Join(testDir, c.cwd)

syft/internal/fileresolver/unindexed_directory.go

@@ -261,13 +261,14 @@ func (u UnindexedDirectory) Write(location file.Location, reader io.Reader) erro
 func (u UnindexedDirectory) newLocation(filePath string, resolveLinks bool) *file.Location {
 	filePath = path.Clean(filePath)
 
-	virtualPath := ""
+	virtualPath := filePath
 	realPath := filePath
 
 	if resolveLinks {
 		paths := u.resolveLinks(filePath)
 		if len(paths) > 1 {
 			realPath = paths[len(paths)-1]
+			// TODO: this is not quite correct, as the equivalent of os.EvalSymlinks needs to be done (in the context of afero)
 			if realPath != path.Clean(filePath) {
 				virtualPath = paths[0]
 			}

syft/internal/fileresolver/unindexed_directory_test.go

@@ -96,7 +96,7 @@ func Test_UnindexDirectoryResolver_FilesByPath_request_response(t *testing.T) {
 		base                string
 		input               string
 		expectedRealPath    string
-		expectedVirtualPath string
+		expectedVirtualPath string // if empty, the virtual path should be the same as the real path
 	}{
 		{
 			name:             "relative root, relative request, direct",
@@ -498,6 +498,9 @@ func Test_UnindexDirectoryResolver_FilesByPath_request_response(t *testing.T) {
 	}
 	for _, c := range cases {
 		t.Run(c.name, func(t *testing.T) {
+			if c.expectedVirtualPath == "" {
+				c.expectedVirtualPath = c.expectedRealPath
+			}
 
 			// we need to mimic a shell, otherwise we won't get a path within a symlink
 			targetPath := filepath.Join(testDir, c.cwd)
@@ -1090,7 +1093,7 @@ func Test_UnindexedDirectoryResolver_resolvesLinks(t *testing.T) {
 				file.NewLocation("file-1.txt"),
 				file.NewLocation("file-2.txt"),
 				file.NewLocation("file-3.txt"),
-				file.NewLocation("parent/file-4.txt"),
+				file.NewVirtualLocation("parent/file-4.txt", "parent-link/file-4.txt"),
 			},
 		},
 		{
@@ -1120,7 +1123,7 @@ func Test_UnindexedDirectoryResolver_resolvesLinks(t *testing.T) {
 				file.NewLocation("file-1.txt"),
 				file.NewLocation("file-2.txt"),
 				file.NewLocation("file-3.txt"),
-				file.NewLocation("parent/file-4.txt"),
+				file.NewVirtualLocation("parent/file-4.txt", "parent-link/file-4.txt"),
 			},
 		},
 		{