fix: spdx and cyclonedx

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

syft/format/common/spdxhelpers/to_format_model.go

@@ -6,6 +6,8 @@ import (
 	"crypto/sha1"
 	"fmt"
 	"path"
+	"path/filepath"
+	"slices"
 	"sort"
 	"strings"
 	"time"
@@ -52,7 +54,8 @@ func ToFormatModel(s sbom.SBOM) *spdx.Document {
 	rels := relationship.NewIndex(s.Relationships...)
 	packages, otherLicenses := toPackages(rels, s.Artifacts.Packages, s)
 
-	files, archiveRelationships := toFiles(s)
+	sourceFile := toSourceFile(s)
+	files, archiveRelationships := toFiles(s, sourceFile)
 	allRelationships := append(toRelationships(rels.All()), archiveRelationships...)
 
 	// for valid SPDX we need a document describes relationship
@@ -62,7 +65,7 @@ func ToFormatModel(s sbom.SBOM) *spdx.Document {
 	if rootPackage != nil {
 		describesID = rootPackage.PackageSPDXIdentifier
 
-		allRelationships = append(allRelationships, toRootRelationships(rootPackage, packages, files, allRelationships)...)
+		allRelationships = append(allRelationships, toRootRelationships(rootPackage, sourceFile, packages, files, allRelationships)...)
 
 		// append the root package
 		packages = append(packages, rootPackage)
@@ -160,38 +163,58 @@ func ToFormatModel(s sbom.SBOM) *spdx.Document {
 	}
 }
 
-// toRootRelationships relates the root package to each package and file nothing else CONTAINS: what is
-// within an archive is reached only through the archive's file, and an owned file through its package.
-func toRootRelationships(rootPackage *spdx.Package, packages []*spdx.Package, files []*spdx.File, relationships []*spdx.Relationship) (out []*spdx.Relationship) {
-	contained := map[spdx.ElementID]bool{}
+// toRootRelationships relates the root package to every package, and to each file nothing else CONTAINS: what is
+// within an archive is reached only through the archive's file, and an owned file through its package. When the
+// source is itself a file, the root package DESCRIBES it, and a scanned directory CONTAINS those files instead. A
+// package containing files must have analyzed them, so the root package then gets a verification code over them.
+func toRootRelationships(rootPackage *spdx.Package, sourceFile *file.Coordinates, packages []*spdx.Package, files []*spdx.File, relationships []*spdx.Relationship) (out []*spdx.Relationship) {
+	contained := map[spdx.ElementID]struct{}{}
 	for _, r := range relationships {
 		if r.Relationship == string(helpers.ContainsRelationship) {
-			contained[r.RefB.ElementRefID] = true
+			contained[r.RefB.ElementRefID] = struct{}{}
+		}
+	}
+	relate := func(a spdx.ElementID, typ helpers.RelationshipType, b spdx.ElementID) {
+		out = append(out, &spdx.Relationship{
+			RefA:         spdx.DocElementID{ElementRefID: a},
+			Relationship: string(typ),
+			RefB:         spdx.DocElementID{ElementRefID: b},
+		})
+	}
+
+	fileContainer := rootPackage.PackageSPDXIdentifier
+	if sourceFile != nil {
+		id := toSPDXID(*sourceFile)
+		relate(rootPackage.PackageSPDXIdentifier, helpers.DescribesRelationship, id)
+		contained[id] = struct{}{}
+		if isScanRoot(*sourceFile) {
+			fileContainer = id
 		}
 	}
 
-	var ids []spdx.ElementID
 	for _, p := range packages {
-		ids = append(ids, p.PackageSPDXIdentifier)
+		relate(rootPackage.PackageSPDXIdentifier, helpers.ContainsRelationship, p.PackageSPDXIdentifier)
 	}
+	var sha1s []string
 	for _, f := range files {
-		ids = append(ids, f.FileSPDXIdentifier)
-	}
-	for _, id := range ids {
-		if contained[id] {
+		if _, has := contained[f.FileSPDXIdentifier]; has {
 			continue
 		}
-		out = append(out, &spdx.Relationship{
-			RefA: spdx.DocElementID{
-				ElementRefID: rootPackage.PackageSPDXIdentifier,
-			},
-			Relationship: string(helpers.ContainsRelationship),
-			RefB: spdx.DocElementID{
-				ElementRefID: id,
-			},
-		})
+		relate(fileContainer, helpers.ContainsRelationship, f.FileSPDXIdentifier)
+		if fileContainer != rootPackage.PackageSPDXIdentifier {
+			continue
+		}
+		for _, c := range f.Checksums {
+			if c.Algorithm == spdx.SHA1 {
+				sha1s = append(sha1s, c.Value)
+				break
+			}
+		}
 	}
-	return
+
+	rootPackage.PackageVerificationCode = toVerificationCode(sha1s)
+	rootPackage.FilesAnalyzed = rootPackage.PackageVerificationCode != nil
+	return out
 }
 
 //nolint:funlen
@@ -728,10 +751,19 @@ const layerIDCommentPrefix = "layerID: "
 
 // toFiles describes every file, including each archive holding one. A file within an archive is named by
 // its path in that archive, and the archive CONTAINS it.
-func toFiles(s sbom.SBOM) (results []*spdx.File, archiveRelationships []*spdx.Relationship) {
+func toFiles(s sbom.SBOM, sourceFile *file.Coordinates) (results []*spdx.File, archiveRelationships []*spdx.Relationship) {
 	_, coordinateSorter := formatInternal.GetLocationSorters(s)
 	top, entries := formatInternal.ArchiveEntries(s.AllCoordinates(), coordinateSorter)
 
+	// a file holding entries is an archive, whatever its content starts with (such as a launch script)
+	describe := func(c file.Coordinates, name string) *spdx.File {
+		f := toFile(s.Artifacts, c, name)
+		if len(entries[c]) > 0 {
+			f.FileTypes = withArchiveType(f.FileTypes)
+		}
+		return f
+	}
+
 	var addEntries func(archive file.Coordinates)
 	addEntries = func(archive file.Coordinates) {
 		for _, c := range entries[archive] {
@@ -740,7 +772,7 @@ func toFiles(s sbom.SBOM) (results []*spdx.File, archiveRelationships []*spdx.Re
 				Relationship: string(helpers.ContainsRelationship),
 				RefB:         spdx.DocElementID{ElementRefID: toSPDXID(c)},
 			})
-			results = append(results, toFile(s.Artifacts, c, strings.TrimPrefix(c.LocalPath(), "/")))
+			results = append(results, describe(c, strings.TrimPrefix(c.LocalPath(), "/")))
 			addEntries(c)
 		}
 	}
@@ -750,7 +782,24 @@ func toFiles(s sbom.SBOM) (results []*spdx.File, archiveRelationships []*spdx.Re
 			log.Debugf("unable to convert relative path '%s' to absolute path: %s", c.RealPath, err)
 			name = c.RealPath
 		}
-		results = append(results, toFile(s.Artifacts, c, name))
+		var archive *source.FileMetadata
+		if isScanRoot(c) {
+			// the scanned root is the root package, unless it is the file the source is
+			if sourceFile == nil || c != *sourceFile {
+				continue
+			}
+			name = "."
+			if m, ok := s.Source.Metadata.(source.FileMetadata); ok {
+				archive = &m
+				name = filepath.Base(m.Path)
+			}
+		}
+		f := describe(c, name)
+		if archive != nil {
+			// an extracted archive is the archive file, not the directory it was extracted to
+			f.FileTypes = withArchiveType(toFileTypes(&file.Metadata{MIMEType: archive.MIMEType}))
+		}
+		results = append(results, f)
 		addEntries(c)
 	}
 
@@ -764,6 +813,47 @@ func toFiles(s sbom.SBOM) (results []*spdx.File, archiveRelationships []*spdx.Re
 	return results, archiveRelationships
 }
 
+// toSourceFile finds the file the scanned source is: a directory, or an extracted archive, is the scanned root,
+// while a single file is indexed beside its parent directory, which was not scanned.
+func toSourceFile(s sbom.SBOM) *file.Coordinates {
+	var top []file.Coordinates
+	for _, c := range s.AllCoordinates() {
+		if c.FullNestedPath() == c.RealPath {
+			top = append(top, c)
+		}
+	}
+	match := isScanRoot
+	switch m := s.Source.Metadata.(type) {
+	case source.DirectoryMetadata:
+	case source.FileMetadata:
+		name := filepath.Base(m.Path)
+		single := func(c file.Coordinates) bool { return strings.TrimPrefix(c.RealPath, "/") == name }
+		if slices.ContainsFunc(top, single) {
+			match = single
+		}
+	default:
+		return nil
+	}
+	i := slices.IndexFunc(top, match)
+	if i < 0 {
+		return nil
+	}
+	return &top[i]
+}
+
+// withArchiveType adds the archive file type, in place of other
+func withArchiveType(types []string) []string {
+	if slices.Contains(types, string(helpers.ArchiveFileType)) {
+		return types
+	}
+	types = slices.DeleteFunc(types, func(t string) bool { return t == string(helpers.OtherFileType) })
+	return append(types, string(helpers.ArchiveFileType))
+}
+
+func isScanRoot(c file.Coordinates) bool {
+	return c.RealPath == "" || c.RealPath == "/"
+}
+
 func toFile(artifacts sbom.Artifacts, c file.Coordinates, name string) *spdx.File {
 	var metadata *file.Metadata
 	if metadataForLocation, exists := artifacts.FileMetadata[c]; exists {
@@ -879,7 +969,7 @@ func newPackageVerificationCode(rels *relationship.Index, p pkg.Package, sbom sb
 	// if a sha1 for a file is provided, then the validator will fail if the package does not have
 	// a package verification code
 	coordinates := rels.Coordinates(p, artifact.ContainsRelationship)
-	var digests []file.Digest
+	var sha1s []string
 	for _, c := range coordinates {
 		digest := sbom.Artifacts.FileDigests[c]
 		if len(digest) == 0 {
@@ -893,27 +983,23 @@ func newPackageVerificationCode(rels *relationship.Index, p pkg.Package, sbom sb
 				break
 			}
 		}
-		digests = append(digests, d)
+		sha1s = append(sha1s, d.Value)
 	}
+	return toVerificationCode(sha1s)
+}
 
-	if len(digests) == 0 {
+func toVerificationCode(sha1s []string) *spdx.PackageVerificationCode {
+	if len(sha1s) == 0 {
 		return nil
 	}
 
 	// sort templist in ascending order by SHA1 value
-	sort.SliceStable(digests, func(i, j int) bool {
-		return digests[i].Value < digests[j].Value
-	})
+	sort.Strings(sha1s)
 
 	// filelist = templist with "/n"s removed. /* ordered sequence of SHA1 values with no separators
-	var b strings.Builder
-	for _, digest := range digests {
-		b.WriteString(digest.Value)
-	}
-
 	//nolint:gosec
 	hasher := sha1.New()
-	_, _ = hasher.Write([]byte(b.String()))
+	_, _ = hasher.Write([]byte(strings.Join(sha1s, "")))
 	return &spdx.PackageVerificationCode{
 		// 7.9.1: Package Verification Code Value
 		// Cardinality: mandatory, one
@@ -934,10 +1020,6 @@ func trimPatchVersion(semver string) string {
 // spdx requires that the file name field is a relative filename
 // with the root of the package archive or directory
 func convertAbsoluteToRelative(absPath string) (string, error) {
-	// the root itself; an empty file name is not valid SPDX (tag-value omits it and cannot read it back)
-	if absPath == "" || absPath == "/" {
-		return ".", nil
-	}
 	// Ensure the absolute path is absolute (although it should already be)
 	if !path.IsAbs(absPath) {
 		// already relative

syft/format/common/spdxhelpers/to_format_model_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"regexp"
+	"slices"
 	"strings"
 	"testing"
 
@@ -557,7 +558,7 @@ func Test_toFiles(t *testing.T) {
 	tests := []struct {
 		name string
 		in   sbom.SBOM
-		want spdx.File
+		want []string
 	}{
 		{
 			name: "File paths are converted to relative in final SPDX collection",
@@ -585,31 +586,56 @@ func Test_toFiles(t *testing.T) {
 					},
 				},
 			},
-			want: spdx.File{
-				FileName: "some/path",
+			want: []string{"some/path"},
+		},
+		{
+			name: "the scanned root is the root package, not a file",
+			in: sbom.SBOM{Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
+				{RealPath: ""}:     {Type: stfile.TypeDirectory},
+				{RealPath: "/etc"}: {Type: stfile.TypeDirectory},
+			}}},
+			want: []string{"etc"},
+		},
+		{
+			name: "a scanned directory is a file",
+			in: sbom.SBOM{
+				Source: source.Description{Metadata: source.DirectoryMetadata{Path: "/dir"}},
+				Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
+					{RealPath: ""}:     {Type: stfile.TypeDirectory},
+					{RealPath: "/etc"}: {Type: stfile.TypeDirectory},
+				}},
 			},
+			want: []string{".", "etc"},
 		},
 		{
-			name: "the scanned root has a name",
-			in: sbom.SBOM{Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
-				{RealPath: ""}: {Type: stfile.TypeDirectory},
-			}}},
-			want: spdx.File{FileName: "."},
+			name: "a scanned file is indexed beside its parent directory, which was not scanned",
+			in: sbom.SBOM{
+				Source: source.Description{Metadata: source.FileMetadata{Path: "/dir/go.mod"}},
+				Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
+					{RealPath: ""}:        {Type: stfile.TypeDirectory},
+					{RealPath: "/go.mod"}: {Type: stfile.TypeRegular},
+				}},
+			},
+			want: []string{"go.mod"},
 		},
 		{
-			name: "the absolute root has a name",
+			name: "the absolute root is the root package, not a file",
 			in: sbom.SBOM{Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
-				{RealPath: "/"}: {Type: stfile.TypeDirectory},
+				{RealPath: "/"}:    {Type: stfile.TypeDirectory},
+				{RealPath: "/etc"}: {Type: stfile.TypeDirectory},
 			}}},
-			want: spdx.File{FileName: "."},
+			want: []string{"etc"},
 		},
 	}
 
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {
-			files, _ := toFiles(test.in)
-			require.Len(t, files, 1)
-			assert.Equal(t, test.want.FileName, files[0].FileName)
+			files, _ := toFiles(test.in, toSourceFile(test.in))
+			var names []string
+			for _, f := range files {
+				names = append(names, f.FileName)
+			}
+			assert.Equal(t, test.want, names)
 		})
 	}
 }
@@ -1179,93 +1205,279 @@ func Test_nestedArchiveFiles(t *testing.T) {
 	inB := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/WEB-INF/lib/b.jar!/odd:na%me.txt", FileSystemID: "sha256:layer"}
 	owned := file.NewCoordinates("/usr/bin/tool", "sha256:layer")
 	a := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/WEB-INF/lib/a.jar", FileSystemID: "sha256:layer"}
+	war := file.NewCoordinates("/app.war", "sha256:layer")
+	b := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/WEB-INF/lib/b.jar", FileSystemID: "sha256:layer"}
 	topPkg := pkg.Package{Name: "top"}
 	topPkg.SetID()
 	nestedPkg := pkg.Package{Name: "nested"}
 	nestedPkg.SetID()
-	s := sbom.SBOM{
-		Artifacts: sbom.Artifacts{
-			Packages:     pkg.NewCollection(topPkg, nestedPkg),
-			FileMetadata: map[file.Coordinates]file.Metadata{top: {}, inA: {}, inB: {}, owned: {}},
+
+	sources := []source.Description{
+		{Name: "img", Metadata: source.ImageMetadata{UserInput: "img:latest", ManifestDigest: "sha256:abc"}},
+		{Name: "dir", Metadata: source.DirectoryMetadata{Path: "/dir"}},
+	}
+	for _, src := range sources {
+		t.Run(src.Name, func(t *testing.T) {
+			s := sbom.SBOM{
+				Source: src,
+				Artifacts: sbom.Artifacts{
+					Packages:     pkg.NewCollection(topPkg, nestedPkg),
+					FileMetadata: map[file.Coordinates]file.Metadata{top: {}, inA: {}, inB: {}, owned: {}},
+				},
+				Relationships: []artifact.Relationship{
+					{From: topPkg, To: owned, Type: artifact.ContainsRelationship},
+					{From: a, To: nestedPkg, Type: artifact.ContainsRelationship},
+				},
+			}
+
+			doc := ToFormatModel(s)
+
+			// each archive holding a described file is described too, named by its path in its own container
+			names := map[spdx.ElementID]string{}
+			for _, f := range doc.Files {
+				names[f.FileSPDXIdentifier] = f.FileName
+				assert.Equal(t, "layerID: sha256:layer", f.FileComment)
+			}
+			assert.Equal(t, map[spdx.ElementID]string{
+				toSPDXID(top):   "etc/os-release",
+				toSPDXID(owned): "usr/bin/tool",
+				toSPDXID(war):   "app.war",
+				toSPDXID(a):     "WEB-INF/lib/a.jar",
+				toSPDXID(b):     "WEB-INF/lib/b.jar",
+				toSPDXID(inA):   "META-INF/MANIFEST.MF",
+				toSPDXID(inB):   "odd:na%me.txt",
+			}, names)
+
+			var described []spdx.ElementID
+			var contains [][2]spdx.ElementID
+			for _, r := range doc.Relationships {
+				switch r.Relationship {
+				case string(helpers.DescribesRelationship):
+					described = append(described, r.RefB.ElementRefID)
+				case string(helpers.ContainsRelationship):
+					contains = append(contains, [2]spdx.ElementID{r.RefA.ElementRefID, r.RefB.ElementRefID})
+				}
+			}
+			rootPkg := doc.Packages[len(doc.Packages)-1]
+			root := rootPkg.PackageSPDXIdentifier
+			assert.Equal(t, []spdx.ElementID{root}, described)
+
+			// the root contains every package, but only the files nothing else does: not what is within an archive,
+			// nor an owned file
+			assert.ElementsMatch(t, [][2]spdx.ElementID{
+				{root, toSPDXID(topPkg)},
+				{root, toSPDXID(nestedPkg)},
+				{root, toSPDXID(top)},
+				{root, toSPDXID(war)},
+				{toSPDXID(topPkg), toSPDXID(owned)},
+				{toSPDXID(a), toSPDXID(nestedPkg)},
+				{toSPDXID(war), toSPDXID(a)},
+				{toSPDXID(war), toSPDXID(b)},
+				{toSPDXID(a), toSPDXID(inA)},
+				{toSPDXID(b), toSPDXID(inB)},
+			}, contains)
+
+			// a package containing files must have analyzed them
+			assert.True(t, rootPkg.FilesAnalyzed)
+			assert.NotNil(t, rootPkg.PackageVerificationCode)
+
+			// other tools may instead have the document describe the top-level files directly
+			describesFiles := ToFormatModel(s)
+			for _, r := range describesFiles.Relationships {
+				if r.RefA.ElementRefID == root && (r.RefB.ElementRefID == toSPDXID(top) || r.RefB.ElementRefID == toSPDXID(war)) {
+					r.RefA.ElementRefID = "DOCUMENT"
+					r.Relationship = string(helpers.DescribesRelationship)
+				}
+			}
+
+			for _, d := range []struct {
+				name string
+				doc  *spdx.Document
+			}{
+				{name: "root contains files", doc: doc},
+				{name: "document describes files", doc: describesFiles},
+			} {
+				t.Run(d.name, func(t *testing.T) {
+					decoded, err := ToSyftModel(d.doc)
+					require.NoError(t, err)
+
+					assert.Equal(t, src.Name, decoded.Source.Name)
+					assert.IsType(t, src.Metadata, decoded.Source.Metadata)
+
+					// decoded paths are relative, as SPDX names are
+					relative := func(c file.Coordinates) file.Coordinates {
+						c.RealPath = strings.TrimPrefix(c.RealPath, "/")
+						c.FullPath = strings.TrimPrefix(c.FullPath, "/")
+						return c
+					}
+					var got []file.Coordinates
+					for c := range decoded.Artifacts.FileMetadata {
+						got = append(got, c)
+					}
+					assert.ElementsMatch(t, []file.Coordinates{relative(top), relative(owned), relative(war), relative(a), relative(b), relative(inA), relative(inB)}, got)
+
+					var gotContains []artifact.Relationship
+					for _, r := range decoded.Relationships {
+						_, fromFile := r.From.(file.Coordinates)
+						_, toFile := r.To.(file.Coordinates)
+						if fromFile && toFile {
+							gotContains = append(gotContains, r)
+						}
+					}
+					assert.ElementsMatch(t, []artifact.Relationship{
+						{From: relative(war), To: relative(a), Type: artifact.ContainsRelationship},
+						{From: relative(war), To: relative(b), Type: artifact.ContainsRelationship},
+						{From: relative(a), To: relative(inA), Type: artifact.ContainsRelationship},
+						{From: relative(b), To: relative(inB), Type: artifact.ContainsRelationship},
+					}, gotContains)
+				})
+			}
+		})
+	}
+}
+
+func Test_sourceFile(t *testing.T) {
+	root := file.NewCoordinates("", "")
+	war := file.NewCoordinates("/app.war", "")
+	jar := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/a.jar"}
+	etc := file.NewCoordinates("/etc/os-release", "")
+	p := pkg.Package{Name: "p"}
+	p.SetID()
+
+	tests := []struct {
+		name   string
+		source source.Description
+		files  []file.Coordinates
+		// relationships as "from type to", with the root package as "root"
+		want         []string
+		wantAnalyzed bool
+		// the file types of the file the root package describes
+		wantTypes []string
+		// decoded files, and the file relationships among them, as "from type to"
+		wantFiles    []string
+		wantContains []string
+	}{
+		{
+			name:   "a scanned directory is described, and contains what is at its root",
+			source: source.Description{Name: "dir", Metadata: source.DirectoryMetadata{Path: "/dir"}},
+			files:  []file.Coordinates{root, etc, war, jar},
+			want: []string{
+				"root DESCRIBES .",
+				"root CONTAINS p",
+				". CONTAINS etc/os-release",
+				". CONTAINS app.war",
+				"app.war CONTAINS a.jar",
+			},
+			// the scanned root decodes back to the scanned root, and holds nothing as an archive would
+			wantFiles:    []string{"", "etc/os-release", "app.war", "app.war!/a.jar"},
+			wantContains: []string{"app.war contains app.war!/a.jar"},
 		},
-		Relationships: []artifact.Relationship{
-			{From: topPkg, To: owned, Type: artifact.ContainsRelationship},
-			{From: a, To: nestedPkg, Type: artifact.ContainsRelationship},
+		{
+			name:   "an extracted archive is described, and contains what is at its root",
+			source: source.Description{Name: "t.tar", Metadata: source.FileMetadata{Path: "/dir/t.tar"}},
+			files:  []file.Coordinates{root, etc},
+			want: []string{
+				"root DESCRIBES t.tar",
+				"root CONTAINS p",
+				"t.tar CONTAINS etc/os-release",
+			},
+			// the scanned root is the directory the archive was extracted to, but describes the archive
+			wantTypes: []string{"ARCHIVE"},
+			// the described archive is the scanned root, holding what is at the root
+			wantFiles: []string{"", "etc/os-release"},
+		},
+		{
+			name:   "an extracted archive takes its file types from its MIME type",
+			source: source.Description{Name: "t.zip", Metadata: source.FileMetadata{Path: "/dir/t.zip", MIMEType: "application/zip"}},
+			files:  []file.Coordinates{root, etc},
+			want: []string{
+				"root DESCRIBES t.zip",
+				"root CONTAINS p",
+				"t.zip CONTAINS etc/os-release",
+			},
+			wantTypes: []string{"APPLICATION", "ARCHIVE"},
+			wantFiles: []string{"", "etc/os-release"},
+		},
+		{
+			name:   "a scanned file is described, but not its parent directory",
+			source: source.Description{Name: "app.war", Metadata: source.FileMetadata{Path: "/dir/app.war"}},
+			files:  []file.Coordinates{root, war, jar},
+			want: []string{
+				"root DESCRIBES app.war",
+				"root CONTAINS p",
+				"app.war CONTAINS a.jar",
+			},
+			// an archive read in place is the scanned root too, so decodes as if it had been extracted
+			wantFiles: []string{"", "a.jar"},
+		},
+		{
+			name:   "without the scanned root, the root package contains what is at its root",
+			source: source.Description{Name: "dir", Metadata: source.DirectoryMetadata{Path: "/dir"}},
+			files:  []file.Coordinates{etc},
+			want: []string{
+				"root CONTAINS p",
+				"root CONTAINS etc/os-release",
+			},
+			wantAnalyzed: true,
+			wantFiles:    []string{"etc/os-release"},
 		},
 	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			metadata := map[file.Coordinates]file.Metadata{}
+			for _, c := range tt.files {
+				metadata[c] = file.Metadata{}
+			}
+			s := sbom.SBOM{
+				Source:    tt.source,
+				Artifacts: sbom.Artifacts{Packages: pkg.NewCollection(p), FileMetadata: metadata},
+			}
 
-	doc := ToFormatModel(s)
+			doc := ToFormatModel(s)
 
-	// each archive holding a described file is described too, named by its path in its own container
-	war := file.NewCoordinates("/app.war", "sha256:layer")
-	b := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/WEB-INF/lib/b.jar", FileSystemID: "sha256:layer"}
-	names := map[spdx.ElementID]string{}
-	for _, f := range doc.Files {
-		names[f.FileSPDXIdentifier] = f.FileName
-		assert.Equal(t, "layerID: sha256:layer", f.FileComment)
+			rootPkg := doc.Packages[len(doc.Packages)-1]
+			names := map[spdx.ElementID]string{rootPkg.PackageSPDXIdentifier: "root", toSPDXID(p): "p"}
+			for _, f := range doc.Files {
+				names[f.FileSPDXIdentifier] = f.FileName
+			}
+			var got []string
+			for _, r := range doc.Relationships {
+				if r.RefA.ElementRefID == "DOCUMENT" {
+					continue
+				}
+				got = append(got, names[r.RefA.ElementRefID]+" "+r.Relationship+" "+names[r.RefB.ElementRefID])
+			}
+			assert.ElementsMatch(t, tt.want, got)
+			assert.Equal(t, tt.wantAnalyzed, rootPkg.FilesAnalyzed)
+			if tt.wantTypes != nil {
+				for _, r := range doc.Relationships {
+					if r.RefA.ElementRefID == rootPkg.PackageSPDXIdentifier && r.Relationship == string(helpers.DescribesRelationship) {
+						i := slices.IndexFunc(doc.Files, func(f *spdx.File) bool { return f.FileSPDXIdentifier == r.RefB.ElementRefID })
+						require.GreaterOrEqual(t, i, 0)
+						assert.Equal(t, tt.wantTypes, doc.Files[i].FileTypes)
+					}
+				}
+			}
+
+			decoded, err := ToSyftModel(doc)
+			require.NoError(t, err)
+
+			var gotFiles []string
+			for c := range decoded.Artifacts.FileMetadata {
+				gotFiles = append(gotFiles, c.FullNestedPath())
+			}
+			assert.ElementsMatch(t, tt.wantFiles, gotFiles)
+
+			var gotContains []string
+			for _, r := range decoded.Relationships {
+				if from, ok := r.From.(file.Coordinates); ok {
+					to, _ := r.To.(file.Coordinates)
+					gotContains = append(gotContains, from.FullNestedPath()+" "+string(r.Type)+" "+to.FullNestedPath())
+				}
+			}
+			assert.ElementsMatch(t, tt.wantContains, gotContains)
+		})
 	}
-	assert.Equal(t, map[spdx.ElementID]string{
-		toSPDXID(top):   "etc/os-release",
-		toSPDXID(owned): "usr/bin/tool",
-		toSPDXID(war):   "app.war",
-		toSPDXID(a):     "WEB-INF/lib/a.jar",
-		toSPDXID(b):     "WEB-INF/lib/b.jar",
-		toSPDXID(inA):   "META-INF/MANIFEST.MF",
-		toSPDXID(inB):   "odd:na%me.txt",
-	}, names)
-
-	var root spdx.ElementID
-	var contains [][2]spdx.ElementID
-	for _, r := range doc.Relationships {
-		switch r.Relationship {
-		case string(helpers.DescribesRelationship):
-			root = r.RefB.ElementRefID
-		case string(helpers.ContainsRelationship):
-			contains = append(contains, [2]spdx.ElementID{r.RefA.ElementRefID, r.RefB.ElementRefID})
-		}
-	}
-	// the root contains only what nothing else does: not what is within an archive, nor an owned file
-	assert.ElementsMatch(t, [][2]spdx.ElementID{
-		{root, toSPDXID(topPkg)},
-		{root, toSPDXID(top)},
-		{root, toSPDXID(war)},
-		{toSPDXID(topPkg), toSPDXID(owned)},
-		{toSPDXID(a), toSPDXID(nestedPkg)},
-		{toSPDXID(war), toSPDXID(a)},
-		{toSPDXID(war), toSPDXID(b)},
-		{toSPDXID(a), toSPDXID(inA)},
-		{toSPDXID(b), toSPDXID(inB)},
-	}, contains)
-
-	decoded, err := ToSyftModel(doc)
-	require.NoError(t, err)
-
-	// decoded paths are relative, as SPDX names are
-	relative := func(c file.Coordinates) file.Coordinates {
-		c.RealPath = strings.TrimPrefix(c.RealPath, "/")
-		c.FullPath = strings.TrimPrefix(c.FullPath, "/")
-		return c
-	}
-	var got []file.Coordinates
-	for c := range decoded.Artifacts.FileMetadata {
-		got = append(got, c)
-	}
-	assert.ElementsMatch(t, []file.Coordinates{relative(top), relative(owned), relative(war), relative(a), relative(b), relative(inA), relative(inB)}, got)
-
-	var gotContains []artifact.Relationship
-	for _, r := range decoded.Relationships {
-		_, fromFile := r.From.(file.Coordinates)
-		_, toFile := r.To.(file.Coordinates)
-		if fromFile && toFile {
-			gotContains = append(gotContains, r)
-		}
-	}
-	assert.ElementsMatch(t, []artifact.Relationship{
-		{From: relative(war), To: relative(a), Type: artifact.ContainsRelationship},
-		{From: relative(war), To: relative(b), Type: artifact.ContainsRelationship},
-		{From: relative(a), To: relative(inA), Type: artifact.ContainsRelationship},
-		{From: relative(b), To: relative(inB), Type: artifact.ContainsRelationship},
-	}, gotContains)
 }
 
 func Test_toRootPackage_name(t *testing.T) {
@@ -1313,3 +1525,27 @@ func Test_toRootPackage_name(t *testing.T) {
 		})
 	}
 }
+
+func Test_toFiles_archiveType(t *testing.T) {
+	// a file holding entries is an archive, even when its content starts like something else, such as a Spring Boot
+	// executable jar's launch script
+	jar := file.NewCoordinates("/app.jar", "")
+	s := sbom.SBOM{Artifacts: sbom.Artifacts{FileMetadata: map[file.Coordinates]file.Metadata{
+		jar:                                   {MIMEType: "text/x-shellscript"},
+		file.NewCoordinates("/plain.zip", ""): {MIMEType: "application/zip"},
+		{RealPath: "/app.jar", FullPath: "/app.jar!/lib/b.jar!/c.txt"}: {MIMEType: "text/plain"},
+	}}}
+
+	files, _ := toFiles(s, nil)
+	got := map[string][]string{}
+	for _, f := range files {
+		got[f.FileName] = f.FileTypes
+	}
+	assert.Equal(t, map[string][]string{
+		"app.jar":   {"TEXT", "ARCHIVE"},
+		"plain.zip": {"APPLICATION", "ARCHIVE"},
+		// added only to hold its entry, with no metadata of its own
+		"lib/b.jar": {"ARCHIVE"},
+		"c.txt":     {"TEXT"},
+	}, got)
+}

syft/format/common/spdxhelpers/to_syft_model.go

@@ -38,6 +38,9 @@ func ToSyftModel(doc *spdx.Document) (*sbom.SBOM, error) {
 
 	spdxIDMap := make(map[string]any)
 
+	// found before the root package and its relationships are removed
+	sourceFile := findSourceFile(doc)
+
 	s := &sbom.SBOM{
 		Source: extractSource(spdxIDMap, doc),
 		Artifacts: sbom.Artifacts{
@@ -50,7 +53,7 @@ func ToSyftModel(doc *spdx.Document) (*sbom.SBOM, error) {
 
 	collectSyftPackages(s, spdxIDMap, doc)
 
-	collectSyftFiles(s, spdxIDMap, doc)
+	collectSyftFiles(s, spdxIDMap, doc, sourceFile)
 
 	s.Relationships = toSyftRelationships(spdxIDMap, doc)
 
@@ -379,7 +382,32 @@ func collectSyftPackages(s *sbom.SBOM, spdxIDMap map[string]any, doc *spdx.Docum
 	}
 }
 
-func collectSyftFiles(s *sbom.SBOM, spdxIDMap map[string]any, doc *spdx.Document) {
+// findSourceFile returns the file the root package DESCRIBES when it CONTAINS files: an archive scanned as the source
+// is the scanned root, so what it contains is at the root rather than within an archive.
+func findSourceFile(doc *spdx.Document) spdx.ElementID {
+	roots := findRootPackages(doc)
+	if len(roots) != 1 {
+		return ""
+	}
+	var described spdx.ElementID
+	containers := map[spdx.ElementID]bool{}
+	for _, r := range doc.Relationships {
+		switch helpers.RelationshipType(r.Relationship) {
+		case helpers.DescribesRelationship:
+			if r.RefA.ElementRefID == roots[0].PackageSPDXIdentifier {
+				described = r.RefB.ElementRefID
+			}
+		case helpers.ContainsRelationship:
+			containers[r.RefA.ElementRefID] = true
+		}
+	}
+	if !containers[described] {
+		return ""
+	}
+	return described
+}
+
+func collectSyftFiles(s *sbom.SBOM, spdxIDMap map[string]any, doc *spdx.Document, sourceFile spdx.ElementID) {
 	files := map[spdx.ElementID]*spdx.File{}
 	for _, p := range doc.Packages {
 		for _, f := range p.Files {
@@ -394,6 +422,11 @@ func collectSyftFiles(s *sbom.SBOM, spdxIDMap map[string]any, doc *spdx.Document
 	for id, f := range files {
 		nested.own[id] = toSyftCoordinates(f)
 	}
+	if c, ok := nested.own[sourceFile]; ok {
+		// the scanned root
+		c.RealPath = ""
+		nested.own[sourceFile] = c
+	}
 	for _, r := range doc.Relationships {
 		container, entry := r.RefA.ElementRefID, r.RefB.ElementRefID
 		switch helpers.RelationshipType(r.Relationship) {
@@ -403,7 +436,8 @@ func collectSyftFiles(s *sbom.SBOM, spdxIDMap map[string]any, doc *spdx.Document
 		default:
 			continue
 		}
-		if files[container] != nil && files[entry] != nil {
+		// the scanned root holds what is at the root, not within an archive
+		if files[container] != nil && files[entry] != nil && !isScanRoot(nested.own[container]) {
 			nested.container[entry] = container
 		}
 	}
@@ -568,7 +602,7 @@ func fileContainsRelationship(typ helpers.RelationshipType, a, b any) (artifact.
 		return artifact.Relationship{}, false
 	}
 	container, ok := a.(file.Location)
-	if !ok {
+	if !ok || isScanRoot(container.Coordinates) {
 		return artifact.Relationship{}, false
 	}
 	var entry artifact.Identifiable
@@ -612,6 +646,10 @@ func toSyftCoordinates(f *spdx.File) file.Coordinates {
 }
 
 func coordinatesFromFile(name, comment, id string) file.Coordinates {
+	// the scanned root
+	if name == "." {
+		name = ""
+	}
 	c := file.NewCoordinates(name, "")
 	if v, ok := strings.CutPrefix(comment, layerIDCommentPrefix); ok {
 		c.FileSystemID = v

syft/format/common/spdxhelpers/to_syft_model_v3.go

@@ -31,6 +31,9 @@ func ToSyftModelV3(doc *spdx.Document) (*sbom.SBOM, error) {
 
 	spdxMap := ptrMap[any]{}
 
+	// found before the root package and its relationships are removed
+	sourceFile := v3findSourceFile(doc)
+
 	s := &sbom.SBOM{
 		Source: v3extractSource(spdxMap, doc),
 		Artifacts: sbom.Artifacts{
@@ -45,7 +48,7 @@ func ToSyftModelV3(doc *spdx.Document) (*sbom.SBOM, error) {
 
 	v3collectSyftPackages(s, spdxMap, relationships, doc)
 
-	v3collectSyftFiles(s, spdxMap, doc)
+	v3collectSyftFiles(s, spdxMap, doc, sourceFile)
 
 	s.Relationships = v3toSyftRelationships(spdxMap, doc)
 
@@ -282,7 +285,40 @@ func v3collectSyftPackages(s *sbom.SBOM, spdxMap ptrMap[any], relationships ptrM
 	}
 }
 
-func v3collectSyftFiles(s *sbom.SBOM, spdxMap ptrMap[any], doc *spdx.Document) {
+// v3findSourceFile returns the file the root package DESCRIBES when it CONTAINS files: an archive scanned as the
+// source is the scanned root, so what it contains is at the root rather than within an archive.
+func v3findSourceFile(doc *spdx.Document) spdx.AnyFile {
+	roots := v3findRootPackages(doc)
+	if len(roots) != 1 {
+		return nil
+	}
+	var described spdx.AnyFile
+	containers := ptrMap[bool]{}
+	for _, r := range doc.Elements.Relationships() {
+		from := r.GetFrom()
+		if from == nil {
+			continue
+		}
+		switch r.GetType() {
+		case spdx.RelationshipType_Describes:
+			if from == spdx.AnyElement(roots[0]) {
+				for _, to := range r.GetTo() {
+					if f, ok := to.(spdx.AnyFile); ok {
+						described = f
+					}
+				}
+			}
+		case spdx.RelationshipType_Contains:
+			containers.Set(from, true)
+		}
+	}
+	if described == nil || !containers.Get(described) {
+		return nil
+	}
+	return described
+}
+
+func v3collectSyftFiles(s *sbom.SBOM, spdxMap ptrMap[any], doc *spdx.Document, sourceFile spdx.AnyFile) {
 	nested := newNestedFiles[reflect.Value]()
 	var files []spdx.AnyFile
 	for _, elementList := range []spdx.ElementList{doc.Elements, doc.RootElements} {
@@ -294,6 +330,13 @@ func v3collectSyftFiles(s *sbom.SBOM, spdxMap ptrMap[any], doc *spdx.Document) {
 			files = append(files, f)
 		}
 	}
+	if sourceFile != nil {
+		if c, ok := nested.own[ptrTo(sourceFile)]; ok {
+			// the scanned root
+			c.RealPath = ""
+			nested.own[ptrTo(sourceFile)] = c
+		}
+	}
 	for _, r := range doc.Elements.Relationships() {
 		container, ok := r.GetFrom().(spdx.AnyFile)
 		if !ok || r.GetType() != spdx.RelationshipType_Contains {
@@ -306,7 +349,7 @@ func v3collectSyftFiles(s *sbom.SBOM, spdxMap ptrMap[any], doc *spdx.Document) {
 			}
 			_, knownContainer := nested.own[ptrTo(container)]
 			_, knownEntry := nested.own[ptrTo(entry)]
-			if knownContainer && knownEntry {
+			if knownContainer && knownEntry && !isScanRoot(nested.own[ptrTo(container)]) {
 				nested.container[ptrTo(entry)] = ptrTo(container)
 			}
 		}

syft/format/spdxjson/encoder_test.go

@@ -15,6 +15,7 @@ import (
 	"github.com/anchore/syft/syft/format/internal/testutil"
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/anchore/syft/syft/sbom"
+	"github.com/anchore/syft/syft/source"
 )
 
 var updateSnapshot = flag.Bool("update-spdx-json", false, "update the *.golden files for spdx-json encoders")
@@ -281,3 +282,83 @@ func defaultFormatEncoders() []sbom.FormatEncoder {
 	}
 	return encs
 }
+
+func TestEncodeDecodeTopLevelFiles(t *testing.T) {
+	// the root package contains the top-level files, or the scanned root does when it is described
+	root := file.NewCoordinates("", "")
+	etc := file.NewCoordinates("/etc/os-release", "")
+	jar := file.Coordinates{RealPath: "/app.war", FullPath: "/app.war!/WEB-INF/lib/a.jar"}
+	war := file.NewCoordinates("/app.war", "")
+	tests := []struct {
+		name      string
+		src       source.Description
+		files     []file.Coordinates
+		wantPaths []string
+		wantRels  []string
+	}{
+		{
+			name:      "img",
+			src:       source.Description{Name: "img", Metadata: source.ImageMetadata{UserInput: "img:latest", ManifestDigest: "sha256:abc"}},
+			files:     []file.Coordinates{etc, jar},
+			wantPaths: []string{"etc/os-release", "app.war", "app.war!/WEB-INF/lib/a.jar"},
+			wantRels:  []string{"app.war contains app.war!/WEB-INF/lib/a.jar"},
+		},
+		{
+			name:      "dir",
+			src:       source.Description{Name: "dir", Metadata: source.DirectoryMetadata{Path: "dir"}},
+			files:     []file.Coordinates{etc, jar},
+			wantPaths: []string{"etc/os-release", "app.war", "app.war!/WEB-INF/lib/a.jar"},
+			wantRels:  []string{"app.war contains app.war!/WEB-INF/lib/a.jar"},
+		},
+		{
+			name:      "dir-root",
+			src:       source.Description{Name: "dir", Metadata: source.DirectoryMetadata{Path: "dir"}},
+			files:     []file.Coordinates{root, etc, jar},
+			wantPaths: []string{"", "etc/os-release", "app.war", "app.war!/WEB-INF/lib/a.jar"},
+			wantRels:  []string{"app.war contains app.war!/WEB-INF/lib/a.jar"},
+		},
+		{
+			// a scanned archive is the scanned root, holding what is at the root
+			name:      "file",
+			src:       source.Description{Name: "app.war", Metadata: source.FileMetadata{Path: "dir/app.war"}},
+			files:     []file.Coordinates{war, jar},
+			wantPaths: []string{"", "WEB-INF/lib/a.jar"},
+		},
+	}
+	for _, tt := range tests {
+		src := tt.src
+		for _, enc := range defaultFormatEncoders() {
+			t.Run(tt.name+"_"+enc.Version(), func(t *testing.T) {
+				metadata := map[file.Coordinates]file.Metadata{}
+				for _, c := range tt.files {
+					metadata[c] = file.Metadata{}
+				}
+				var buf bytes.Buffer
+				require.NoError(t, enc.Encode(&buf, sbom.SBOM{
+					Source:    src,
+					Artifacts: sbom.Artifacts{FileMetadata: metadata},
+				}))
+
+				s, _, _, err := NewFormatDecoder().Decode(bytes.NewReader(buf.Bytes()))
+				require.NoError(t, err, "encoded document:\n%s", buf.String())
+
+				assert.Equal(t, src.Name, s.Source.Name)
+				assert.IsType(t, src.Metadata, s.Source.Metadata)
+
+				var paths []string
+				for c := range s.Artifacts.FileMetadata {
+					paths = append(paths, c.FullNestedPath())
+				}
+				assert.ElementsMatch(t, tt.wantPaths, paths)
+
+				var rels []string
+				for _, r := range s.Relationships {
+					from, _ := r.From.(file.Coordinates)
+					to, _ := r.To.(file.Coordinates)
+					rels = append(rels, from.FullNestedPath()+" "+string(r.Type)+" "+to.FullNestedPath())
+				}
+				assert.ElementsMatch(t, tt.wantRels, rels)
+			})
+		}
+	}
+}

syft/format/spdxtagvalue/encoder_test.go

@@ -231,7 +231,7 @@ func TestEncodeDecodeSourceWithoutName(t *testing.T) {
 }
 
 func TestEncodeDecodeRootDirectoryFile(t *testing.T) {
-	// a directory scan with all file metadata describes the scanned root, whose relative path is empty
+	// a directory scan with all file metadata includes the scanned root, which has no relative path of its own
 	root := file.NewCoordinates("", "")
 	etc := file.NewCoordinates("/etc", "")
 	for _, enc := range defaultFormatEncoders() {
@@ -252,7 +252,7 @@ func TestEncodeDecodeRootDirectoryFile(t *testing.T) {
 			for c := range s.Artifacts.FileMetadata {
 				names = append(names, c.RealPath)
 			}
-			assert.ElementsMatch(t, []string{".", "etc"}, names)
+			assert.ElementsMatch(t, []string{"", "etc"}, names)
 		})
 	}
 }