Correct ID handling during Syft JSON decoding (#900)

internal/formats/syftjson/model/package.go

@@ -129,6 +129,7 @@ func (p *Package) UnmarshalJSON(b []byte) error {
 		if err := json.Unmarshal(unpacker.Metadata, &payload); err != nil {
 			return err
 		}
+		p.Metadata = payload
 	default:
 		log.Warnf("unknown package metadata type=%q for packageID=%q", p.MetadataType, p.ID)
 	}

internal/formats/syftjson/model/package_test.go

@@ -1,18 +1,18 @@
 package model
 
-import "testing"
+import (
+	"testing"
 
-func TestUnmarshalPackage(t *testing.T) {
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestUnmarshalPackageGolang(t *testing.T) {
 	tests := []struct {
 		name        string
 		p           *Package
 		packageData []byte
 	}{
-		{
-			name:        "Package.UnmarshalJSON will unmarshal blank json",
-			p:           &Package{},
-			packageData: []byte(`{}`),
-		},
 		{
 			name: "Package.UnmarshalJSON unmarshals PackageBasicData",
 			p:    &Package{},
@@ -47,6 +47,11 @@ func TestUnmarshalPackage(t *testing.T) {
 			if err != nil {
 				t.Fatalf("could not unmarshal packageData: %v", err)
 			}
+
+			assert.NotNil(t, test.p.Metadata)
+			golangMetadata := test.p.Metadata.(pkg.GolangBinMetadata)
+			assert.NotEmpty(t, golangMetadata)
+			assert.Equal(t, "go1.18", golangMetadata.GoCompiledVersion)
 		})
 	}
 }

internal/formats/syftjson/to_syft_model.go

@@ -12,7 +12,9 @@ import (
 )
 
 func toSyftModel(doc model.Document) (*sbom.SBOM, error) {
-	catalog := toSyftCatalog(doc.Artifacts)
+	idAliases := make(map[string]string)
+
+	catalog := toSyftCatalog(doc.Artifacts, idAliases)
 
 	return &sbom.SBOM{
 		Artifacts: sbom.Artifacts{
@@ -21,7 +23,7 @@ func toSyftModel(doc model.Document) (*sbom.SBOM, error) {
 		},
 		Source:        *toSyftSourceData(doc.Source),
 		Descriptor:    toSyftDescriptor(doc.Descriptor),
-		Relationships: toSyftRelationships(&doc, catalog, doc.ArtifactRelationships),
+		Relationships: toSyftRelationships(&doc, catalog, doc.ArtifactRelationships, idAliases),
 	}, nil
 }
 
@@ -46,7 +48,7 @@ func toSyftLinuxRelease(d model.LinuxRelease) *linux.Release {
 	}
 }
 
-func toSyftRelationships(doc *model.Document, catalog *pkg.Catalog, relationships []model.Relationship) []artifact.Relationship {
+func toSyftRelationships(doc *model.Document, catalog *pkg.Catalog, relationships []model.Relationship, idAliases map[string]string) []artifact.Relationship {
 	idMap := make(map[string]interface{})
 
 	for _, p := range catalog.Sorted() {
@@ -62,7 +64,7 @@ func toSyftRelationships(doc *model.Document, catalog *pkg.Catalog, relationship
 
 	var out []artifact.Relationship
 	for _, r := range relationships {
-		syftRelationship := toSyftRelationship(idMap, r)
+		syftRelationship := toSyftRelationship(idMap, r, idAliases)
 		if syftRelationship != nil {
 			out = append(out, *syftRelationship)
 		}
@@ -70,13 +72,20 @@ func toSyftRelationships(doc *model.Document, catalog *pkg.Catalog, relationship
 	return out
 }
 
-func toSyftRelationship(idMap map[string]interface{}, relationship model.Relationship) *artifact.Relationship {
+func toSyftRelationship(idMap map[string]interface{}, relationship model.Relationship, idAliases map[string]string) *artifact.Relationship {
-	from, ok := idMap[relationship.Parent].(artifact.Identifiable)
+	id := func(id string) string {
+		aliased, ok := idAliases[id]
+		if ok {
+			return aliased
+		}
+		return id
+	}
+	from, ok := idMap[id(relationship.Parent)].(artifact.Identifiable)
 	if !ok {
 		log.Warnf("relationship mapping from key %s is not a valid artifact.Identifiable type: %+v", relationship.Parent, idMap[relationship.Parent])
 		return nil
 	}
-	to, ok := idMap[relationship.Child].(artifact.Identifiable)
+	to, ok := idMap[id(relationship.Child)].(artifact.Identifiable)
 	if !ok {
 		log.Warnf("relationship mapping to key %s is not a valid artifact.Identifiable type: %+v", relationship.Child, idMap[relationship.Child])
 		return nil
@@ -128,15 +137,15 @@ func toSyftSourceData(s model.Source) *source.Metadata {
 	return nil
 }
 
-func toSyftCatalog(pkgs []model.Package) *pkg.Catalog {
+func toSyftCatalog(pkgs []model.Package, idAliases map[string]string) *pkg.Catalog {
 	catalog := pkg.NewCatalog()
 	for _, p := range pkgs {
-		catalog.Add(toSyftPackage(p))
+		catalog.Add(toSyftPackage(p, idAliases))
 	}
 	return catalog
 }
 
-func toSyftPackage(p model.Package) pkg.Package {
+func toSyftPackage(p model.Package, idAliases map[string]string) pkg.Package {
 	var cpes []pkg.CPE
 	for _, c := range p.CPEs {
 		value, err := pkg.NewCPE(c)
@@ -153,7 +162,7 @@ func toSyftPackage(p model.Package) pkg.Package {
 		locations[i] = source.NewLocationFromCoordinates(c)
 	}
 
-	return pkg.Package{
+	out := pkg.Package{
 		Name:         p.Name,
 		Version:      p.Version,
 		FoundBy:      p.FoundBy,
@@ -166,4 +175,13 @@ func toSyftPackage(p model.Package) pkg.Package {
 		MetadataType: p.MetadataType,
 		Metadata:     p.Metadata,
 	}
+
+	out.SetID()
+
+	id := string(out.ID())
+	if id != p.ID {
+		idAliases[p.ID] = id
+	}
+
+	return out
 }

internal/formats/syftjson/to_syft_model_test.go

@@ -4,6 +4,7 @@ import (
 	"testing"
 
 	"github.com/anchore/syft/internal/formats/syftjson/model"
+	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/source"
 	"github.com/scylladb/go-set/strset"
 	"github.com/stretchr/testify/assert"
@@ -79,3 +80,46 @@ func Test_toSyftSourceData(t *testing.T) {
 	// assert all possible schemes were under test
 	assert.ElementsMatch(t, allSchemes.List(), testedSchemes.List(), "not all source.Schemes are under test")
 }
+
+func Test_idsHaveChanged(t *testing.T) {
+	s, err := toSyftModel(model.Document{
+		Source: model.Source{
+			Type:   "file",
+			Target: "some/path",
+		},
+		Artifacts: []model.Package{
+			{
+				PackageBasicData: model.PackageBasicData{
+					ID:   "1",
+					Name: "pkg-1",
+				},
+			},
+			{
+				PackageBasicData: model.PackageBasicData{
+					ID:   "2",
+					Name: "pkg-2",
+				},
+			},
+		},
+		ArtifactRelationships: []model.Relationship{
+			{
+				Parent: "1",
+				Child:  "2",
+				Type:   string(artifact.ContainsRelationship),
+			},
+		},
+	})
+
+	assert.NoError(t, err)
+	assert.Len(t, s.Relationships, 1)
+
+	r := s.Relationships[0]
+
+	from := s.Artifacts.PackageCatalog.Package(r.From.ID())
+	assert.NotNil(t, from)
+	assert.Equal(t, "pkg-1", from.Name)
+
+	to := s.Artifacts.PackageCatalog.Package(r.To.ID())
+	assert.NotNil(t, to)
+	assert.Equal(t, "pkg-2", to.Name)
+}