refactor(cyclonedx): implement json.Unmarshaler (#2662)

* refactor(cyclonedx): implement json.Unmarshaler

* fix: use pointer

pkg/fanal/artifact/sbom/sbom.go

@@ -59,18 +59,11 @@ func (a Artifact) Inspect(_ context.Context) (types.ArtifactReference, error) {
 		return types.ArtifactReference{}, xerrors.Errorf("seek error: %w", err)
 	}
 
-	var unmarshaler sbom.Unmarshaler
-	switch format {
-	case sbom.FormatCycloneDXJSON:
-		unmarshaler = cyclonedx.NewJSONUnmarshaler()
-	default:
-		return types.ArtifactReference{}, xerrors.Errorf("%s scanning is not yet supported", format)
-
-	}
-	bom, err := unmarshaler.Unmarshal(f)
+	bom, err := a.Decode(f, format)
 	if err != nil {
-		return types.ArtifactReference{}, xerrors.Errorf("failed to unmarshal: %w", err)
+		return types.ArtifactReference{}, xerrors.Errorf("SBOM decode error: %w", err)
 	}
+
 	blobInfo := types.BlobInfo{
 		SchemaVersion: types.BlobJSONSchemaVersion,
 		OS:            bom.OS,
@@ -104,6 +97,30 @@ func (a Artifact) Inspect(_ context.Context) (types.ArtifactReference, error) {
 	}, nil
 }
 
+func (a Artifact) Decode(f io.Reader, format sbom.Format) (sbom.SBOM, error) {
+	var (
+		v       interface{}
+		bom     sbom.SBOM
+		decoder interface{ Decode(any) error }
+	)
+
+	switch format {
+	case sbom.FormatCycloneDXJSON:
+		v = &cyclonedx.CycloneDX{SBOM: &bom}
+		decoder = json.NewDecoder(f)
+	default:
+		return sbom.SBOM{}, xerrors.Errorf("%s scanning is not yet supported", format)
+
+	}
+
+	// Decode a file content into sbom.SBOM
+	if err := decoder.Decode(v); err != nil {
+		return sbom.SBOM{}, xerrors.Errorf("failed to decode: %w", err)
+	}
+
+	return bom, nil
+}
+
 func (a Artifact) Clean(reference types.ArtifactReference) error {
 	return a.cache.DeleteBlobs(reference.BlobIDs)
 }

pkg/sbom/cyclonedx/marshal.go

@@ -367,7 +367,7 @@ func (e *Marshaler) reportToCdxComponent(r types.Report) (*cdx.Component, error)
 	return component, nil
 }
 
-func (e Marshaler) resultToCdxComponent(r types.Result, osFound *ftypes.OS) cdx.Component {
+func (e *Marshaler) resultToCdxComponent(r types.Result, osFound *ftypes.OS) cdx.Component {
 	component := cdx.Component{
 		Name: r.Target,
 		Properties: &[]cdx.Property{

pkg/sbom/cyclonedx/unmarshal.go

@@ -1,7 +1,7 @@
 package cyclonedx
 
 import (
-	"io"
+	"bytes"
 	"sort"
 	"strconv"
 	"strings"
@@ -15,54 +15,46 @@ import (
 	"github.com/aquasecurity/trivy/pkg/sbom"
 )
 
-type Unmarshaler struct {
-	format cdx.BOMFileFormat
+type CycloneDX struct {
+	*sbom.SBOM
 
 	dependencies map[string][]string
 	components   map[string]cdx.Component
 }
 
-func NewJSONUnmarshaler() sbom.Unmarshaler {
-	return &Unmarshaler{
-		format: cdx.BOMFileFormatJSON,
+func (c *CycloneDX) UnmarshalJSON(b []byte) error {
+	if c.SBOM == nil {
+		c.SBOM = &sbom.SBOM{}
 	}
-}
-
-func (u *Unmarshaler) Unmarshal(r io.Reader) (sbom.SBOM, error) {
 	bom := cdx.NewBOM()
-	decoder := cdx.NewBOMDecoder(r, u.format)
+	decoder := cdx.NewBOMDecoder(bytes.NewReader(b), cdx.BOMFileFormatJSON)
 	if err := decoder.Decode(bom); err != nil {
-		return sbom.SBOM{}, xerrors.Errorf("CycloneDX decode error: %w", err)
+		return xerrors.Errorf("CycloneDX decode error: %w", err)
 	}
 
-	u.dependencies = dependencyMap(bom.Dependencies)
-	u.components = componentMap(bom.Metadata, bom.Components)
-
-	var (
-		osInfo   *ftypes.OS
-		apps     []ftypes.Application
-		pkgInfos []ftypes.PackageInfo
-		seen     = make(map[string]struct{})
-	)
-	for bomRef := range u.dependencies {
-		component := u.components[bomRef]
+	c.dependencies = dependencyMap(bom.Dependencies)
+	c.components = componentMap(bom.Metadata, bom.Components)
+
+	var seen = make(map[string]struct{})
+	for bomRef := range c.dependencies {
+		component := c.components[bomRef]
 		switch component.Type {
 		case cdx.ComponentTypeOS: // OS info and OS packages
-			osInfo = toOS(component)
-			pkgInfo, err := u.parseOSPkgs(component, seen)
+			c.OS = toOS(component)
+			pkgInfo, err := c.parseOSPkgs(component, seen)
 			if err != nil {
-				return sbom.SBOM{}, xerrors.Errorf("failed to parse os packages: %w", err)
+				return xerrors.Errorf("failed to parse os packages: %w", err)
 			}
-			pkgInfos = append(pkgInfos, pkgInfo)
+			c.Packages = append(c.Packages, pkgInfo)
 		case cdx.ComponentTypeApplication: // It would be a lock file in a CycloneDX report generated by Trivy
 			if lookupProperty(component.Properties, PropertyType) == "" {
 				continue
 			}
-			app, err := u.parseLangPkgs(component, seen)
+			app, err := c.parseLangPkgs(component, seen)
 			if err != nil {
-				return sbom.SBOM{}, xerrors.Errorf("failed to parse language packages: %w", err)
+				return xerrors.Errorf("failed to parse language packages: %w", err)
 			}
-			apps = append(apps, *app)
+			c.Applications = append(c.Applications, *app)
 		case cdx.ComponentTypeLibrary:
 			// It is an individual package not associated with any lock files and should be processed later.
 			// e.g. .gemspec, .egg and .wheel
@@ -71,7 +63,7 @@ func (u *Unmarshaler) Unmarshal(r io.Reader) (sbom.SBOM, error) {
 	}
 
 	var libComponents []cdx.Component
-	for ref, component := range u.components {
+	for ref, component := range c.components {
 		if _, ok := seen[ref]; ok {
 			continue
 		}
@@ -82,15 +74,15 @@ func (u *Unmarshaler) Unmarshal(r io.Reader) (sbom.SBOM, error) {
 
 	aggregatedApps, err := aggregateLangPkgs(libComponents)
 	if err != nil {
-		return sbom.SBOM{}, xerrors.Errorf("failed to aggregate packages: %w", err)
+		return xerrors.Errorf("failed to aggregate packages: %w", err)
 	}
-	apps = append(apps, aggregatedApps...)
+	c.Applications = append(c.Applications, aggregatedApps...)
 
-	sort.Slice(apps, func(i, j int) bool {
-		if apps[i].Type != apps[j].Type {
-			return apps[i].Type < apps[j].Type
+	sort.Slice(c.Applications, func(i, j int) bool {
+		if c.Applications[i].Type != c.Applications[j].Type {
+			return c.Applications[i].Type < c.Applications[j].Type
 		}
-		return apps[i].FilePath < apps[j].FilePath
+		return c.Applications[i].FilePath < c.Applications[j].FilePath
 	})
 
 	var metadata ftypes.Metadata
@@ -102,29 +94,24 @@ func (u *Unmarshaler) Unmarshal(r io.Reader) (sbom.SBOM, error) {
 	}
 
 	var components []ftypes.Component
-	for _, c := range lo.FromPtr(bom.Components) {
-		components = append(components, toTrivyCdxComponent(c))
+	for _, component := range lo.FromPtr(bom.Components) {
+		components = append(components, toTrivyCdxComponent(component))
 	}
 
-	return sbom.SBOM{
-		OS:           osInfo,
-		Packages:     pkgInfos,
-		Applications: apps,
-
-		// Keep the original SBOM
-		CycloneDX: &ftypes.CycloneDX{
-			BOMFormat:    bom.BOMFormat,
-			SpecVersion:  bom.SpecVersion,
-			SerialNumber: bom.SerialNumber,
-			Version:      bom.Version,
-			Metadata:     metadata,
-			Components:   components,
-		},
-	}, nil
+	// Keep the original SBOM
+	c.CycloneDX = &ftypes.CycloneDX{
+		BOMFormat:    bom.BOMFormat,
+		SpecVersion:  bom.SpecVersion,
+		SerialNumber: bom.SerialNumber,
+		Version:      bom.Version,
+		Metadata:     metadata,
+		Components:   components,
+	}
+	return nil
 }
 
-func (u *Unmarshaler) parseOSPkgs(component cdx.Component, seen map[string]struct{}) (ftypes.PackageInfo, error) {
-	components := u.walkDependencies(component.BOMRef)
+func (c *CycloneDX) parseOSPkgs(component cdx.Component, seen map[string]struct{}) (ftypes.PackageInfo, error) {
+	components := c.walkDependencies(component.BOMRef)
 	pkgs, err := parsePkgs(components, seen)
 	if err != nil {
 		return ftypes.PackageInfo{}, xerrors.Errorf("failed to parse os package: %w", err)
@@ -135,8 +122,8 @@ func (u *Unmarshaler) parseOSPkgs(component cdx.Component, seen map[string]struc
 	}, nil
 }
 
-func (u *Unmarshaler) parseLangPkgs(component cdx.Component, seen map[string]struct{}) (*ftypes.Application, error) {
-	components := u.walkDependencies(component.BOMRef)
+func (c *CycloneDX) parseLangPkgs(component cdx.Component, seen map[string]struct{}) (*ftypes.Application, error) {
+	components := c.walkDependencies(component.BOMRef)
 	components = lo.UniqBy(components, func(c cdx.Component) string {
 		return c.BOMRef
 	})
@@ -175,10 +162,10 @@ func parsePkgs(components []cdx.Component, seen map[string]struct{}) ([]ftypes.P
 //     - type: Application 3
 //       - type: Library D
 //       - type: Library E
-func (u *Unmarshaler) walkDependencies(rootRef string) []cdx.Component {
+func (c *CycloneDX) walkDependencies(rootRef string) []cdx.Component {
 	var components []cdx.Component
-	for _, dep := range u.dependencies[rootRef] {
-		component, ok := u.components[dep]
+	for _, dep := range c.dependencies[rootRef] {
+		component, ok := c.components[dep]
 		if !ok {
 			continue
 		}
@@ -188,7 +175,7 @@ func (u *Unmarshaler) walkDependencies(rootRef string) []cdx.Component {
 			components = append(components, component)
 		}
 
-		components = append(components, u.walkDependencies(dep)...)
+		components = append(components, c.walkDependencies(dep)...)
 	}
 	return components
 }

pkg/sbom/cyclonedx/unmarshal_test.go

@@ -1,6 +1,7 @@
 package cyclonedx_test
 
 import (
+	"encoding/json"
 	"os"
 	"testing"
 
@@ -196,15 +197,16 @@ func TestUnmarshaler_Unmarshal(t *testing.T) {
 			require.NoError(t, err)
 			defer f.Close()
 
-			unmarshaler := cyclonedx.NewJSONUnmarshaler()
-			got, err := unmarshaler.Unmarshal(f)
+			var cdx cyclonedx.CycloneDX
+			err = json.NewDecoder(f).Decode(&cdx)
 			if tt.wantErr != "" {
 				require.Error(t, err)
 				assert.Contains(t, err.Error(), tt.wantErr)
 				return
 			}
 
 			// Not compare the CycloneDX field
+			got := *cdx.SBOM
 			got.CycloneDX = nil
 
 			require.NoError(t, err)

pkg/sbom/sbom.go

@@ -19,18 +19,14 @@ type SBOM struct {
 	CycloneDX *types.CycloneDX
 }
 
-type Unmarshaler interface {
-	Unmarshal(io.Reader) (SBOM, error)
-}
-
 type Format string
 
 const (
-	FormatCycloneDXJSON = "cyclonedx-json"
-	FormatCycloneDXXML  = "cyclonedx-xml"
-	FormatSPDXJSON      = "spdx-json"
-	FormatSPDXXML       = "spdx-xml"
-	FormatUnknown       = "unknown"
+	FormatCycloneDXJSON Format = "cyclonedx-json"
+	FormatCycloneDXXML  Format = "cyclonedx-xml"
+	FormatSPDXJSON      Format = "spdx-json"
+	FormatSPDXXML       Format = "spdx-xml"
+	FormatUnknown       Format = "unknown"
 )
 
 func DetectFormat(r io.ReadSeeker) (Format, error) {