feat: add identifiable field to source object

syft/source/source.go

@@ -15,6 +15,7 @@ import (
 
 	"github.com/bmatcuk/doublestar/v4"
 	"github.com/mholt/archiver/v3"
+	digest "github.com/opencontainers/go-digest"
 	"github.com/spf13/afero"
 
 	"github.com/anchore/stereoscope"
@@ -25,6 +26,7 @@ import (
 // Source is an object that captures the data source to be cataloged, configuration, and a specific resolver used
 // in cataloging (based on the data source and configuration)
 type Source struct {
+	id                string
 	Image             *image.Image // the image object to be cataloged (image only)
 	Metadata          Metadata
 	directoryResolver *directoryResolver
@@ -304,6 +306,50 @@ func NewFromImage(img *image.Image, userImageStr string) (Source, error) {
 	}, nil
 }
 
+func (s *Source) ID() string {
+	return s.id
+}
+
+func (s *Source) SetID() {
+	if s.Metadata.Scheme != ImageScheme {
+		// How do we generate ID for non-image sources?
+		s.id = digest.FromString(s.Metadata.Path).String()
+		return
+	}
+
+	// calcuate chain ID for image sources
+	// https://github.com/opencontainers/image-spec/blob/main/config.md#layer-chainid
+	s.id = calculateChainID(s.Image)
+
+	if s.id == "" {
+		// TODO what happens here if image has no layers?
+		s.id = digest.FromString(s.Metadata.ImageMetadata.UserInput).String()
+	}
+	return
+}
+
+func calculateChainID(img *image.Image) string {
+	if len(img.Layers) < 1 {
+		return ""
+	}
+
+	// DiffID(L0) = digest of layer 0
+	// https://github.com/anchore/stereoscope/blob/1b1b744a919964f38d14e1416fb3f25221b761ce/pkg/image/layer_metadata.go#L19-L32
+	chainID := img.Layers[0].Metadata.Digest
+	id := chain(chainID, img.Layers[1:])
+
+	return id
+}
+
+func chain(chainID string, layers []*image.Layer) string {
+	if len(layers) < 1 {
+		return chainID
+	}
+
+	chainID = digest.FromString(layers[0].Metadata.Digest + " " + chainID).String()
+	return chain(chainID, layers[1:])
+}
+
 func (s *Source) FileResolver(scope Scope) (FileResolver, error) {
 	switch s.Metadata.Scheme {
 	case DirectoryScheme, FileScheme:

syft/source/source_test.go

@@ -65,6 +65,44 @@ func TestNewFromImageFails(t *testing.T) {
 	})
 }
 
+func TestSetID(t *testing.T) {
+	img := imagetest.GetFixtureImage(t, "oci-archive", "image-simple")
+	tests := []struct {
+		name     string
+		input    *Source
+		expected string
+	}{
+		{
+			name: "source.SetID sets the ID for non image sources",
+			input: &Source{
+				Metadata: Metadata{
+					Scheme: FileScheme,
+					Path:   "test-fixtures/image-simple/file-1.txt",
+				},
+			},
+			expected: "sha256:fbfb0730f4306b27c118715998ba58f1ad350f0451513c36c267dc4b9d3b688d",
+		},
+		{
+			name: "source.SetID sets the ID for image sources",
+			input: &Source{
+				Image: img,
+				Metadata: Metadata{
+					Scheme:        ImageScheme,
+					ImageMetadata: NewImageMetadata(img, "image-simple"),
+				},
+			},
+			expected: "sha256:e6d9f87981af1a1007a42be43b21ba6abe7c1608b1541e877c69052af5356669",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			test.input.SetID()
+			assert.Equal(t, test.expected, test.input.ID())
+		})
+	}
+}
+
 func TestNewFromImage(t *testing.T) {
 	layer := image.NewLayer(nil)
 	img := image.Image{