feat: syft 3435 - add file components to cyclonedx bom output when file metadata is available

syft/format/common/cyclonedxhelpers/to_format_model.go

@@ -6,19 +6,33 @@ import (
 	"strings"
 	"time"
 
-	"github.com/CycloneDX/cyclonedx-go"
+	cyclonedx "github.com/CycloneDX/cyclonedx-go"
 	"github.com/google/uuid"
 
+	stfile "github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/internal/log"
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/cpe"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/format/internal/cyclonedxutil/helpers"
 	"github.com/anchore/syft/syft/linux"
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/anchore/syft/syft/sbom"
 	"github.com/anchore/syft/syft/source"
 )
 
+var cycloneDXValidHash = map[string]cyclonedx.HashAlgorithm{
+	"sha1":       cyclonedx.HashAlgoSHA1,
+	"md5":        cyclonedx.HashAlgoMD5,
+	"sha256":     cyclonedx.HashAlgoSHA256,
+	"sha384":     cyclonedx.HashAlgoSHA384,
+	"sha512":     cyclonedx.HashAlgoSHA512,
+	"blake2b256": cyclonedx.HashAlgoBlake2b_256,
+	"blake2b384": cyclonedx.HashAlgoBlake2b_384,
+	"blake2b512": cyclonedx.HashAlgoBlake2b_512,
+	"blake3":     cyclonedx.HashAlgoBlake3,
+}
+
 func ToFormatModel(s sbom.SBOM) *cyclonedx.BOM {
 	cdxBOM := cyclonedx.NewBOM()
 
@@ -28,12 +42,51 @@ func ToFormatModel(s sbom.SBOM) *cyclonedx.BOM {
 	cdxBOM.SerialNumber = uuid.New().URN()
 	cdxBOM.Metadata = toBomDescriptor(s.Descriptor.Name, s.Descriptor.Version, s.Source)
 
+	// Packages
 	packages := s.Artifacts.Packages.Sorted()
 	components := make([]cyclonedx.Component, len(packages))
 	for i, p := range packages {
 		components[i] = helpers.EncodeComponent(p)
 	}
 	components = append(components, toOSComponent(s.Artifacts.LinuxDistribution)...)
+
+	// Files
+	artifacts := s.Artifacts
+	coordinates := s.AllCoordinates()
+	for _, coordinate := range coordinates {
+		var metadata *file.Metadata
+		// File Info
+		fileMetadata, exists := artifacts.FileMetadata[coordinate]
+		// no file metadata then don't include in SBOM
+		// the syft config allows for sometimes only capturing files owned by packages
+		// so there can be a map miss here where we have less metadata than all coordinates
+		if !exists {
+			continue
+		}
+		if fileMetadata.Type == stfile.TypeDirectory ||
+			fileMetadata.Type == stfile.TypeSocket ||
+			fileMetadata.Type == stfile.TypeSymLink {
+			// skip dir, symlinks and sockets for the final bom
+			continue
+		}
+		metadata = &fileMetadata
+
+		// Digests
+		var digests []file.Digest
+		if digestsForLocation, exists := artifacts.FileDigests[coordinate]; exists {
+			digests = digestsForLocation
+		}
+
+		// if cdx doesn't have an algorithm for the SBOM we need to drop the components
+		// since an empty hash field is not allowed: https://cyclonedx.org/docs/1.6/json/#components_items_hashes_items_alg
+		cdxHashes := digestsToHashes(digests)
+		components = append(components, cyclonedx.Component{
+			BOMRef: string(coordinate.ID()),
+			Type:   cyclonedx.ComponentTypeFile,
+			Name:   metadata.Path,
+			Hashes: &cdxHashes,
+		})
+	}
 	cdxBOM.Components = &components
 
 	dependencies := toDependencies(s.Relationships)
@@ -44,6 +97,29 @@ func ToFormatModel(s sbom.SBOM) *cyclonedx.BOM {
 	return cdxBOM
 }
 
+func digestsToHashes(digests []file.Digest) []cyclonedx.Hash {
+	var hashes []cyclonedx.Hash
+	for _, digest := range digests {
+		lookup := strings.ToLower(digest.Algorithm)
+		cdxAlgo, exists := cycloneDXValidHash[lookup]
+		if !exists {
+			continue
+		}
+		hashes = append(hashes, cyclonedx.Hash{
+			Algorithm: cdxAlgo,
+			Value:     digest.Value,
+		})
+	}
+	return hashes
+}
+
+// Not pulling it's weight, let's just do strings to lower on the map
+// supported algorithm in cycloneDX as of 1.4
+// "MD5", "SHA-1", "SHA-256", "SHA-384", "SHA-512",
+// "SHA3-256", "SHA3-384", "SHA3-512", "BLAKE2b-256", "BLAKE2b-384", "BLAKE2b-512", "BLAKE3"
+// syft supported digests: cmd/syft/cli/eventloop/tasks.go
+// MD5, SHA1, SHA256
+
 func toOSComponent(distro *linux.Release) []cyclonedx.Component {
 	if distro == nil {
 		return []cyclonedx.Component{}

syft/format/common/cyclonedxhelpers/to_format_model_test.go

@@ -9,7 +9,9 @@ import (
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
+	stfile "github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/format/internal/cyclonedxutil/helpers"
 	"github.com/anchore/syft/syft/linux"
 	"github.com/anchore/syft/syft/pkg"
@@ -143,6 +145,184 @@ func Test_relationships(t *testing.T) {
 	}
 }
 
+func Test_FileComponents(t *testing.T) {
+	p1 := pkg.Package{
+		Name: "p1",
+	}
+	tests := []struct {
+		name string
+		sbom sbom.SBOM
+		want []cyclonedx.Component
+	}{
+		{
+			name: "sbom coordinates with file metadata are serialized to cdx along with packages",
+			sbom: sbom.SBOM{
+				Artifacts: sbom.Artifacts{
+					Packages: pkg.NewCollection(p1),
+					FileMetadata: map[file.Coordinates]file.Metadata{
+						{RealPath: "/test"}: {Path: "/test", Type: stfile.TypeRegular},
+					},
+					FileDigests: map[file.Coordinates][]file.Digest{
+						{RealPath: "/test"}: {
+							{
+								Algorithm: "sha256",
+								Value:     "xyz12345",
+							},
+						},
+					},
+				},
+			},
+			want: []cyclonedx.Component{
+				{
+					BOMRef: "2a1fc74ade23e357",
+					Type:   cyclonedx.ComponentTypeLibrary,
+					Name:   "p1",
+				},
+				{
+					BOMRef: "3f31cb2d98be6c1e",
+					Name:   "/test",
+					Type:   cyclonedx.ComponentTypeFile,
+					Hashes: &[]cyclonedx.Hash{
+						{Algorithm: "SHA-256", Value: "xyz12345"},
+					},
+				},
+			},
+		},
+		{
+			name: "sbom coordinates that don't contain metadata are not added to the final output",
+			sbom: sbom.SBOM{
+				Artifacts: sbom.Artifacts{
+					FileMetadata: map[file.Coordinates]file.Metadata{
+						{RealPath: "/test"}: {Path: "/test", Type: stfile.TypeRegular},
+					},
+					FileDigests: map[file.Coordinates][]file.Digest{
+						{RealPath: "/test"}: {
+							{
+								Algorithm: "sha256",
+								Value:     "xyz12345",
+							},
+						},
+						{RealPath: "/test-2"}: {
+							{
+								Algorithm: "sha256",
+								Value:     "xyz678910",
+							},
+						},
+					},
+				},
+			},
+			want: []cyclonedx.Component{
+				{
+					BOMRef: "3f31cb2d98be6c1e",
+					Name:   "/test",
+					Type:   cyclonedx.ComponentTypeFile,
+					Hashes: &[]cyclonedx.Hash{
+						{Algorithm: "SHA-256", Value: "xyz12345"},
+					},
+				},
+			},
+		},
+		{
+			name: "sbom coordinates that return hashes not covered by cdx only include valid digests",
+			sbom: sbom.SBOM{
+				Artifacts: sbom.Artifacts{
+					FileMetadata: map[file.Coordinates]file.Metadata{
+						{RealPath: "/test"}: {Path: "/test", Type: stfile.TypeRegular},
+					},
+					FileDigests: map[file.Coordinates][]file.Digest{
+						{RealPath: "/test"}: {
+							{
+								Algorithm: "xxh64",
+								Value:     "xyz12345",
+							},
+							{
+								Algorithm: "sha256",
+								Value:     "xyz678910",
+							},
+						},
+					},
+				},
+			},
+			want: []cyclonedx.Component{
+				{
+					BOMRef: "3f31cb2d98be6c1e",
+					Name:   "/test",
+					Type:   cyclonedx.ComponentTypeFile,
+					Hashes: &[]cyclonedx.Hash{
+						{Algorithm: "SHA-256", Value: "xyz678910"},
+					},
+				},
+			},
+		},
+		{
+			name: "sbom coordinates who's metadata is directory or symlink are skipped",
+			sbom: sbom.SBOM{
+				Artifacts: sbom.Artifacts{
+					FileMetadata: map[file.Coordinates]file.Metadata{
+						{RealPath: "/testdir"}: {
+							Path: "/testdir",
+							Type: stfile.TypeDirectory,
+						},
+						{RealPath: "/testsym"}: {
+							Path: "/testsym",
+							Type: stfile.TypeSymLink,
+						},
+						{RealPath: "/test"}: {Path: "/test", Type: stfile.TypeRegular},
+					},
+					FileDigests: map[file.Coordinates][]file.Digest{
+						{RealPath: "/test"}: {
+							{
+								Algorithm: "sha256",
+								Value:     "xyz12345",
+							},
+						},
+					},
+				},
+			},
+			want: []cyclonedx.Component{
+				{
+					BOMRef: "3f31cb2d98be6c1e",
+					Name:   "/test",
+					Type:   cyclonedx.ComponentTypeFile,
+					Hashes: &[]cyclonedx.Hash{
+						{Algorithm: "SHA-256", Value: "xyz12345"},
+					},
+				},
+			},
+		},
+		{
+			name: "sbom with no files serialized correctly",
+			sbom: sbom.SBOM{
+				Artifacts: sbom.Artifacts{
+					Packages: pkg.NewCollection(p1),
+				},
+			},
+			want: []cyclonedx.Component{
+				{
+					BOMRef: "2a1fc74ade23e357",
+					Type:   cyclonedx.ComponentTypeLibrary,
+					Name:   "p1",
+				},
+			},
+		},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			cdx := ToFormatModel(test.sbom)
+			got := *cdx.Components
+			if diff := cmp.Diff(test.want, got); diff != "" {
+				t.Errorf("cdx file components mismatch (-want +got):\n%s", diff)
+			}
+		})
+	}
+}
+
+// mockFileInfo is a test struct that simulates fs.FileInfo
+type mockFileInfo struct {
+	isDir     bool
+	isSymlink bool
+}
+
 func Test_toBomDescriptor(t *testing.T) {
 	type args struct {
 		name        string