port javascript cataloger to new generic cataloger pattern (#1308)

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

syft/pkg/catalog.go

@@ -1,7 +1,6 @@
 package pkg
 
 import (
-	"sort"
 	"sync"
 
 	"github.com/jinzhu/copier"
@@ -199,24 +198,7 @@ func (c *Catalog) Sorted(types ...Type) (pkgs []Package) {
 		pkgs = append(pkgs, p)
 	}
 
-	sort.SliceStable(pkgs, func(i, j int) bool {
+	Sort(pkgs)
-		if pkgs[i].Name == pkgs[j].Name {
-			if pkgs[i].Version == pkgs[j].Version {
-				iLocations := pkgs[i].Locations.ToSlice()
-				jLocations := pkgs[j].Locations.ToSlice()
-				if pkgs[i].Type == pkgs[j].Type && len(iLocations) > 0 && len(jLocations) > 0 {
-					if iLocations[0].String() == jLocations[0].String() {
-						// compare IDs as a final fallback
-						return pkgs[i].ID() < pkgs[j].ID()
-					}
-					return iLocations[0].String() < jLocations[0].String()
-				}
-				return pkgs[i].Type < pkgs[j].Type
-			}
-			return pkgs[i].Version < pkgs[j].Version
-		}
-		return pkgs[i].Name < pkgs[j].Name
-	})
 
 	return pkgs
 }

syft/pkg/cataloger/internal/pkgtest/test_generic_parser.go

@@ -40,6 +40,19 @@ func NewCatalogTester() *CatalogTester {
 	}
 }
 
+func (p *CatalogTester) FromDirectory(t *testing.T, path string) *CatalogTester {
+	t.Helper()
+
+	s, err := source.NewFromDirectory(path)
+	require.NoError(t, err)
+
+	resolver, err := s.FileResolver(source.AllLayersScope)
+	require.NoError(t, err)
+
+	p.resolver = resolver
+	return p
+}
+
 func (p *CatalogTester) FromFile(t *testing.T, path string) *CatalogTester {
 	t.Helper()
 

syft/pkg/cataloger/javascript/cataloger.go

@@ -4,88 +4,24 @@ Package javascript provides a concrete Cataloger implementation for JavaScript e
 package javascript
 
 import (
-	"encoding/json"
-	"io"
-	"path"
-	"strings"
-
 	"github.com/anchore/syft/internal"
-	"github.com/anchore/syft/internal/log"
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/common"
 	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-	"github.com/anchore/syft/syft/source"
 )
 
 // NewJavascriptPackageCataloger returns a new JavaScript cataloger object based on detection of npm based packages.
-func NewJavascriptPackageCataloger() *common.GenericCataloger {
+func NewJavascriptPackageCataloger() *generic.Cataloger {
-	globParsers := map[string]common.ParserFn{
+	return generic.NewCataloger("javascript-package-cataloger").
-		"**/package.json": parsePackageJSON,
+		WithParserByGlobs(parsePackageJSON, "**/package.json")
-	}
-
-	return common.NewGenericCataloger(nil, globParsers, "javascript-package-cataloger")
 }
 
-// NewJavascriptLockCataloger returns a new Javascript cataloger object base on package lock files.
+func NewJavascriptLockCataloger() *generic.Cataloger {
-func NewJavascriptLockCataloger() *common.GenericCataloger {
+	return generic.NewCataloger("javascript-lock-cataloger").
-	globParsers := map[string]common.ParserFn{
+		WithParserByGlobs(parsePackageLock, "**/package-lock.json").
-		"**/package-lock.json": parsePackageLock,
+		WithParserByGlobs(parseYarnLock, "**/yarn.lock").
-		"**/yarn.lock":         parseYarnLock,
+		WithParserByGlobs(parsePnpmLock, "**/pnpm-lock.yaml")
-		"**/pnpm-lock.yaml":    parsePnpmLock,
-	}
-
-	return common.NewGenericCataloger(nil, globParsers, "javascript-lock-cataloger", addLicenses)
 }
 
 func NewNodeBinaryCataloger() *generic.Cataloger {
 	return generic.NewCataloger("node-binary-cataloger").
 		WithParserByMimeTypes(parseNodeBinary, internal.ExecutableMIMETypeSet.List()...)
 }
-
-func addLicenses(resolver source.FileResolver, location source.Location, p *pkg.Package) error {
-	dir := path.Dir(location.RealPath)
-	pkgPath := []string{dir, "node_modules"}
-	pkgPath = append(pkgPath, strings.Split(p.Name, "/")...)
-	pkgPath = append(pkgPath, "package.json")
-	pkgFile := path.Join(pkgPath...)
-	locations, err := resolver.FilesByPath(pkgFile)
-	if err != nil {
-		log.Debugf("an error occurred attempting to read: %s - %+v", pkgFile, err)
-		return nil
-	}
-
-	if len(locations) == 0 {
-		return nil
-	}
-
-	for _, location := range locations {
-		contentReader, err := resolver.FileContentsByLocation(location)
-		if err != nil {
-			log.Debugf("error getting file content reader for %s: %v", pkgFile, err)
-			return nil
-		}
-
-		contents, err := io.ReadAll(contentReader)
-		if err != nil {
-			log.Debugf("error reading file contents for %s: %v", pkgFile, err)
-			return nil
-		}
-
-		var pkgJSON packageJSON
-		err = json.Unmarshal(contents, &pkgJSON)
-		if err != nil {
-			log.Debugf("error parsing %s: %v", pkgFile, err)
-			return nil
-		}
-
-		licenses, err := pkgJSON.licensesFromJSON()
-		if err != nil {
-			log.Debugf("error getting licenses from %s: %v", pkgFile, err)
-			return nil
-		}
-
-		p.Licenses = append(p.Licenses, licenses...)
-	}
-
-	return nil
-}

syft/pkg/cataloger/javascript/cataloger_test.go

@@ -3,100 +3,120 @@ package javascript
 import (
 	"testing"
 
-	"github.com/stretchr/testify/require"
-
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
 	"github.com/anchore/syft/syft/source"
 )
 
 func Test_JavascriptCataloger(t *testing.T) {
-	expected := map[string]pkg.Package{
+	locationSet := source.NewLocationSet(source.NewLocation("package-lock.json"))
-		"@actions/core": {
+	expectedPkgs := []pkg.Package{
-			Name:     "@actions/core",
+		{
-			Version:  "1.6.0",
+			Name:      "@actions/core",
-			Language: pkg.JavaScript,
+			Version:   "1.6.0",
-			Type:     pkg.NpmPkg,
+			FoundBy:   "javascript-lock-cataloger",
-			Licenses: []string{"MIT"},
+			PURL:      "pkg:npm/%40actions/core@1.6.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+			Licenses:  []string{"MIT"},
 		},
-		"wordwrap": {
+		{
-			Name:     "wordwrap",
+			Name:      "ansi-regex",
-			Version:  "0.0.3",
+			Version:   "3.0.0",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/ansi-regex@3.0.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"get-stdin": {
+		{
-			Name:     "get-stdin",
+			Name:      "cowsay",
-			Version:  "5.0.1",
+			Version:   "1.4.0",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/cowsay@1.4.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+			Licenses:  []string{"MIT"},
 		},
-		"minimist": {
+		{
-			Name:     "minimist",
+			Name:      "get-stdin",
-			Version:  "0.0.10",
+			Version:   "5.0.1",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/get-stdin@5.0.1",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"optimist": {
+		{
-			Name:     "optimist",
+			Name:      "is-fullwidth-code-point",
-			Version:  "0.6.1",
+			Version:   "2.0.0",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/is-fullwidth-code-point@2.0.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"string-width": {
+		{
-			Name:     "string-width",
+			Name:      "minimist",
-			Version:  "2.1.1",
+			Version:   "0.0.10",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/minimist@0.0.10",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"strip-ansi": {
+		{
-			Name:     "strip-ansi",
+			Name:      "optimist",
-			Version:  "4.0.0",
+			Version:   "0.6.1",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/optimist@0.6.1",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"strip-eof": {
+		{
-			Name:     "wordwrap",
+			Name:      "string-width",
-			Version:  "1.0.0",
+			Version:   "2.1.1",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/string-width@2.1.1",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"ansi-regex": {
+		{
-			Name:     "ansi-regex",
+			Name:      "strip-ansi",
-			Version:  "3.0.0",
+			Version:   "4.0.0",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/strip-ansi@4.0.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"is-fullwidth-code-point": {
+		{
-			Name:     "is-fullwidth-code-point",
+			Name:      "strip-eof",
-			Version:  "2.0.0",
+			Version:   "1.0.0",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/strip-eof@1.0.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"cowsay": {
+		{
-			Name:     "cowsay",
+			Name:      "wordwrap",
-			Version:  "1.4.0",
+			Version:   "0.0.3",
-			Language: pkg.JavaScript,
+			FoundBy:   "javascript-lock-cataloger",
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/wordwrap@0.0.3",
-			Licenses: []string{"MIT"},
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
 	}
 
-	s, err := source.NewFromDirectory("test-fixtures/pkg-lock")
+	pkgtest.NewCatalogTester().
-	require.NoError(t, err)
+		FromDirectory(t, "test-fixtures/pkg-lock").
+		Expects(expectedPkgs, nil).
+		TestCataloger(t, NewJavascriptLockCataloger())
 
-	resolver, err := s.FileResolver(source.AllLayersScope)
-	require.NoError(t, err)
-
-	actual, _, err := NewJavascriptLockCataloger().Catalog(resolver)
-	if err != nil {
-		t.Fatalf("failed to parse package-lock.json: %+v", err)
-	}
-
-	var pkgs []*pkg.Package
-	for _, p := range actual {
-		p2 := p
-		pkgs = append(pkgs, &p2)
-	}
-
-	assertPkgsEqual(t, pkgs, expected)
 }

syft/pkg/cataloger/javascript/package.go

@@ -0,0 +1,176 @@
+package javascript
+
+import (
+	"encoding/json"
+	"io"
+	"path"
+	"strings"
+
+	"github.com/anchore/packageurl-go"
+	"github.com/anchore/syft/internal/log"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/source"
+)
+
+func newPackageJSONPackage(u packageJSON, locations ...source.Location) pkg.Package {
+	licenses, err := u.licensesFromJSON()
+	if err != nil {
+		log.Warnf("unable to extract licenses from javascript package.json: %+v", err)
+	}
+
+	p := pkg.Package{
+		Name:         u.Name,
+		Version:      u.Version,
+		Licenses:     licenses,
+		PURL:         packageURL(u.Name, u.Version),
+		Locations:    source.NewLocationSet(locations...),
+		Language:     pkg.JavaScript,
+		Type:         pkg.NpmPkg,
+		MetadataType: pkg.NpmPackageJSONMetadataType,
+		Metadata: pkg.NpmPackageJSONMetadata{
+			Name:     u.Name,
+			Version:  u.Version,
+			Author:   u.Author.AuthorString(),
+			Homepage: u.Homepage,
+			URL:      u.Repository.URL,
+			Licenses: licenses,
+			Private:  u.Private,
+		},
+	}
+
+	p.SetID()
+
+	return p
+}
+
+func newPackageLockPackage(resolver source.FileResolver, location source.Location, name string, u lockDependency, licenseMap map[string]string) pkg.Package {
+	var sb strings.Builder
+	sb.WriteString(u.Resolved)
+	sb.WriteString(u.Integrity)
+	var licenses []string
+	if l, exists := licenseMap[sb.String()]; exists {
+		licenses = append(licenses, l)
+	}
+
+	return finalizeLockPkg(
+		resolver,
+		location,
+		pkg.Package{
+			Name:      name,
+			Version:   u.Version,
+			Locations: source.NewLocationSet(location),
+			PURL:      packageURL(name, u.Version),
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+			Licenses:  licenses,
+		},
+	)
+}
+
+func newPnpmPackage(resolver source.FileResolver, location source.Location, name, version string) pkg.Package {
+	return finalizeLockPkg(
+		resolver,
+		location,
+		pkg.Package{
+			Name:      name,
+			Version:   version,
+			Locations: source.NewLocationSet(location),
+			PURL:      packageURL(name, version),
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+	)
+}
+
+func newYarnLockPackage(resolver source.FileResolver, location source.Location, name, version string) pkg.Package {
+	return finalizeLockPkg(
+		resolver,
+		location,
+		pkg.Package{
+			Name:      name,
+			Version:   version,
+			Locations: source.NewLocationSet(location),
+			PURL:      packageURL(name, version),
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+	)
+}
+
+func finalizeLockPkg(resolver source.FileResolver, location source.Location, p pkg.Package) pkg.Package {
+	p.Licenses = append(p.Licenses, addLicenses(p.Name, resolver, location)...)
+	p.SetID()
+	return p
+}
+
+func addLicenses(name string, resolver source.FileResolver, location source.Location) (allLicenses []string) {
+	if resolver == nil {
+		return allLicenses
+	}
+	dir := path.Dir(location.RealPath)
+	pkgPath := []string{dir, "node_modules"}
+	pkgPath = append(pkgPath, strings.Split(name, "/")...)
+	pkgPath = append(pkgPath, "package.json")
+	pkgFile := path.Join(pkgPath...)
+	locations, err := resolver.FilesByPath(pkgFile)
+
+	if err != nil {
+		log.Debugf("an error occurred attempting to read: %s - %+v", pkgFile, err)
+		return allLicenses
+	}
+
+	if len(locations) == 0 {
+		return allLicenses
+	}
+
+	for _, l := range locations {
+		contentReader, err := resolver.FileContentsByLocation(l)
+		if err != nil {
+			log.Debugf("error getting file content reader for %s: %v", pkgFile, err)
+			return allLicenses
+		}
+
+		contents, err := io.ReadAll(contentReader)
+		if err != nil {
+			log.Debugf("error reading file contents for %s: %v", pkgFile, err)
+			return allLicenses
+		}
+
+		var pkgJSON packageJSON
+		err = json.Unmarshal(contents, &pkgJSON)
+		if err != nil {
+			log.Debugf("error parsing %s: %v", pkgFile, err)
+			return allLicenses
+		}
+
+		licenses, err := pkgJSON.licensesFromJSON()
+		if err != nil {
+			log.Debugf("error getting licenses from %s: %v", pkgFile, err)
+			return allLicenses
+		}
+
+		allLicenses = append(allLicenses, licenses...)
+	}
+
+	return allLicenses
+}
+
+// packageURL returns the PURL for the specific NPM package (see https://github.com/package-url/purl-spec)
+func packageURL(name, version string) string {
+	var namespace string
+
+	fields := strings.SplitN(name, "/", 2)
+	if len(fields) > 1 {
+		namespace = fields[0]
+		name = fields[1]
+	}
+
+	return packageurl.NewPackageURL(
+		packageurl.TypeNPM,
+		namespace,
+		name,
+		version,
+		nil,
+		"",
+	).ToString()
+}

syft/pkg/cataloger/javascript/package_test.go

@@ -0,0 +1,58 @@
+package javascript
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/anchore/packageurl-go"
+)
+
+func Test_packageURL(t *testing.T) {
+
+	tests := []struct {
+		testName  string
+		name      string
+		version   string
+		expected  string
+		namespace string
+	}{
+		{
+			testName: "no namespace",
+			name:     "arborist",
+			version:  "2.6.2",
+			expected: "pkg:npm/arborist@2.6.2",
+		},
+		{
+			testName:  "split by namespace",
+			name:      "npmcli/arborist",
+			version:   "2.6.2",
+			expected:  "pkg:npm/npmcli/arborist@2.6.2",
+			namespace: "npmcli",
+		},
+		{
+			testName:  "encoding @ symobl",
+			name:      "@npmcli/arborist",
+			version:   "2.6.2",
+			expected:  "pkg:npm/%40npmcli/arborist@2.6.2",
+			namespace: "@npmcli",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.testName, func(t *testing.T) {
+			actual := packageURL(tt.name, tt.version)
+			assert.Equal(t, tt.expected, actual)
+			decoded, err := packageurl.FromString(actual)
+			require.NoError(t, err)
+			assert.Equal(t, tt.namespace, decoded.Namespace)
+			if decoded.Namespace != "" {
+				assert.Equal(t, tt.name, fmt.Sprintf("%s/%s", decoded.Namespace, decoded.Name))
+			} else {
+				assert.Equal(t, tt.name, decoded.Name)
+			}
+			assert.Equal(t, tt.version, decoded.Version)
+		})
+	}
+}

syft/pkg/cataloger/javascript/parse_package_json.go

@@ -13,11 +13,12 @@ import (
 	"github.com/anchore/syft/internal/log"
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/common"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+	"github.com/anchore/syft/syft/source"
 )
 
 // integrity check
-var _ common.ParserFn = parsePackageJSON
+var _ generic.Parser = parsePackageJSON
 
 // packageJSON represents a JavaScript package.json file
 type packageJSON struct {
@@ -49,6 +50,32 @@ type repository struct {
 // ---> name: "Isaac Z. Schlueter" email: "i@izs.me" url: "http://blog.izs.me"
 var authorPattern = regexp.MustCompile(`^\s*(?P<name>[^<(]*)(\s+<(?P<email>.*)>)?(\s\((?P<url>.*)\))?\s*$`)
 
+// parsePackageJSON parses a package.json and returns the discovered JavaScript packages.
+func parsePackageJSON(_ source.FileResolver, _ *generic.Environment, reader source.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	var pkgs []pkg.Package
+	dec := json.NewDecoder(reader)
+
+	for {
+		var p packageJSON
+		if err := dec.Decode(&p); err == io.EOF {
+			break
+		} else if err != nil {
+			return nil, nil, fmt.Errorf("failed to parse package.json file: %w", err)
+		}
+
+		if !p.hasNameAndVersionValues() {
+			log.Debugf("encountered package.json file without a name and/or version field, ignoring (path=%q)", reader.AccessPath())
+			return nil, nil, nil
+		}
+
+		pkgs = append(pkgs, newPackageJSONPackage(p, reader.Location))
+	}
+
+	pkg.Sort(pkgs)
+
+	return pkgs, nil, nil
+}
+
 func (a *author) UnmarshalJSON(b []byte) error {
 	var authorStr string
 	var fields map[string]string
@@ -172,55 +199,6 @@ func licensesFromJSON(b []byte) ([]license, error) {
 	return nil, errors.New("unmarshal failed")
 }
 
-// parsePackageJSON parses a package.json and returns the discovered JavaScript packages.
-func parsePackageJSON(path string, reader io.Reader) ([]*pkg.Package, []artifact.Relationship, error) {
-	var packages []*pkg.Package
-	dec := json.NewDecoder(reader)
-
-	for {
-		var p packageJSON
-		if err := dec.Decode(&p); err == io.EOF {
-			break
-		} else if err != nil {
-			return nil, nil, fmt.Errorf("failed to parse package.json file: %w", err)
-		}
-
-		if !p.hasNameAndVersionValues() {
-			log.Debugf("encountered package.json file without a name and/or version field, ignoring (path=%q)", path)
-			return nil, nil, nil
-		}
-
-		packages = append(packages, newPackageJSONPackage(p))
-	}
-
-	return packages, nil, nil
-}
-
-func newPackageJSONPackage(p packageJSON) *pkg.Package {
-	licenses, err := p.licensesFromJSON()
-	if err != nil {
-		log.Warnf("unable to extract licenses from javascript package.json: %+v", err)
-	}
-
-	return &pkg.Package{
-		Name:         p.Name,
-		Version:      p.Version,
-		Licenses:     licenses,
-		Language:     pkg.JavaScript,
-		Type:         pkg.NpmPkg,
-		MetadataType: pkg.NpmPackageJSONMetadataType,
-		Metadata: pkg.NpmPackageJSONMetadata{
-			Name:     p.Name,
-			Version:  p.Version,
-			Author:   p.Author.AuthorString(),
-			Homepage: p.Homepage,
-			URL:      p.Repository.URL,
-			Licenses: licenses,
-			Private:  p.Private,
-		},
-	}
-}
-
 func (p packageJSON) hasNameAndVersionValues() bool {
 	return p.Name != "" && p.Version != ""
 }

syft/pkg/cataloger/javascript/parse_package_json_test.go

@@ -1,13 +1,13 @@
 package javascript
 
 import (
-	"os"
 	"testing"
 
-	"github.com/go-test/deep"
 	"github.com/stretchr/testify/assert"
 
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
+	"github.com/anchore/syft/syft/source"
 )
 
 func TestParsePackageJSON(t *testing.T) {
@@ -20,6 +20,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"Artistic-2.0"},
 				Language:     pkg.JavaScript,
@@ -39,6 +40,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"ISC"},
 				Language:     pkg.JavaScript,
@@ -58,6 +60,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"MIT", "Apache-2.0"},
 				Language:     pkg.JavaScript,
@@ -77,6 +80,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     nil,
 				Language:     pkg.JavaScript,
@@ -96,6 +100,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{},
 				Language:     pkg.JavaScript,
@@ -115,6 +120,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"Artistic-2.0"},
 				Language:     pkg.JavaScript,
@@ -134,6 +140,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "function-bind",
 				Version:      "1.1.1",
+				PURL:         "pkg:npm/function-bind@1.1.1",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"MIT"},
 				Language:     pkg.JavaScript,
@@ -153,6 +160,7 @@ func TestParsePackageJSON(t *testing.T) {
 			ExpectedPkg: pkg.Package{
 				Name:         "npm",
 				Version:      "6.14.6",
+				PURL:         "pkg:npm/npm@6.14.6",
 				Type:         pkg.NpmPkg,
 				Licenses:     []string{"Artistic-2.0"},
 				Language:     pkg.JavaScript,
@@ -172,46 +180,16 @@ func TestParsePackageJSON(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.Fixture, func(t *testing.T) {
-			fixture, err := os.Open(test.Fixture)
+			test.ExpectedPkg.Locations.Add(source.NewLocation(test.Fixture))
-			if err != nil {
+			pkgtest.TestFileParser(t, test.Fixture, parsePackageJSON, []pkg.Package{test.ExpectedPkg}, nil)
-				t.Fatalf("failed to open fixture: %+v", err)
-			}
-
-			actual, _, err := parsePackageJSON("", fixture)
-			if err != nil {
-				t.Fatalf("failed to parse package-lock.json: %+v", err)
-			}
-			if len(actual) != 1 {
-				for _, a := range actual {
-					t.Log("   ", a)
-				}
-				t.Fatalf("unexpected package count: %d!=1", len(actual))
-			}
-
-			for _, d := range deep.Equal(actual[0], &test.ExpectedPkg) {
-
-				t.Errorf("diff: %+v", d)
-			}
 		})
 	}
 }
 
 func TestParsePackageJSON_Partial(t *testing.T) { // see https://github.com/anchore/syft/issues/311
 	const fixtureFile = "test-fixtures/pkg-json/package-partial.json"
-	fixture, err := os.Open(fixtureFile)
-	if err != nil {
-		t.Fatalf("failed to open fixture: %+v", err)
-	}
 
-	// TODO: no relationships are under test yet
+	pkgtest.TestFileParser(t, fixtureFile, parsePackageJSON, nil, nil)
-	actual, _, err := parsePackageJSON("", fixture)
-	if err != nil {
-		t.Fatalf("failed to parse package-lock.json: %+v", err)
-	}
-
-	if actualCount := len(actual); actualCount != 0 {
-		t.Errorf("no packages should've been returned (but got %d packages)", actualCount)
-	}
 }
 
 func Test_pathContainsNodeModulesDirectory(t *testing.T) {

syft/pkg/cataloger/javascript/parse_package_lock.go

@@ -8,28 +8,29 @@ import (
 
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/common"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+	"github.com/anchore/syft/syft/source"
 )
 
 // integrity check
-var _ common.ParserFn = parsePackageLock
+var _ generic.Parser = parsePackageLock
 
-// PackageLock represents a JavaScript package.lock json file
+// packageLock represents a JavaScript package.lock json file
-type PackageLock struct {
+type packageLock struct {
 	Requires        bool `json:"requires"`
 	LockfileVersion int  `json:"lockfileVersion"`
-	Dependencies    map[string]Dependency
+	Dependencies    map[string]lockDependency
-	Packages        map[string]Package
+	Packages        map[string]lockPackage
 }
 
-// Dependency represents a single package dependency listed in the package.lock json file
+// lockDependency represents a single package dependency listed in the package.lock json file
-type Dependency struct {
+type lockDependency struct {
 	Version   string `json:"version"`
 	Resolved  string `json:"resolved"`
 	Integrity string `json:"integrity"`
 }
 
-type Package struct {
+type lockPackage struct {
 	Version   string `json:"version"`
 	Resolved  string `json:"resolved"`
 	Integrity string `json:"integrity"`
@@ -37,18 +38,18 @@ type Package struct {
 }
 
 // parsePackageLock parses a package-lock.json and returns the discovered JavaScript packages.
-func parsePackageLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Relationship, error) {
+func parsePackageLock(resolver source.FileResolver, _ *generic.Environment, reader source.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 	// in the case we find package-lock.json files in the node_modules directories, skip those
 	// as the whole purpose of the lock file is for the specific dependencies of the root project
-	if pathContainsNodeModulesDirectory(path) {
+	if pathContainsNodeModulesDirectory(reader.AccessPath()) {
 		return nil, nil, nil
 	}
 
-	var packages []*pkg.Package
+	var pkgs []pkg.Package
 	dec := json.NewDecoder(reader)
 
 	for {
-		var lock PackageLock
+		var lock packageLock
 		if err := dec.Decode(&lock); err == io.EOF {
 			break
 		} else if err != nil {
@@ -63,22 +64,11 @@ func parsePackageLock(path string, reader io.Reader) ([]*pkg.Package, []artifact
 		}
 
 		for name, pkgMeta := range lock.Dependencies {
-			var sb strings.Builder
+			pkgs = append(pkgs, newPackageLockPackage(resolver, reader.Location, name, pkgMeta, licenseMap))
-			sb.WriteString(pkgMeta.Resolved)
-			sb.WriteString(pkgMeta.Integrity)
-			var licenses []string
-			if license, exists := licenseMap[sb.String()]; exists {
-				licenses = append(licenses, license)
-			}
-			packages = append(packages, &pkg.Package{
-				Name:     name,
-				Version:  pkgMeta.Version,
-				Language: pkg.JavaScript,
-				Type:     pkg.NpmPkg,
-				Licenses: licenses,
-			})
 		}
 	}
 
-	return packages, nil, nil
+	pkg.Sort(pkgs)
+
+	return pkgs, nil, nil
 }

syft/pkg/cataloger/javascript/parse_package_lock_test.go

@@ -1,157 +1,142 @@
 package javascript
 
 import (
-	"os"
 	"testing"
 
-	"github.com/stretchr/testify/assert"
+	"github.com/anchore/syft/syft/artifact"
-
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
+	"github.com/anchore/syft/syft/source"
 )
 
-func assertPkgsEqual(t *testing.T, actual []*pkg.Package, expected map[string]pkg.Package) {
-	t.Helper()
-	if len(actual) != len(expected) {
-		for _, a := range actual {
-			t.Log("   ", a)
-		}
-		t.Fatalf("unexpected package count: %d!=%d", len(actual), len(expected))
-	}
-
-	for _, a := range actual {
-		expectedPkg, ok := expected[a.Name]
-		assert.True(t, ok)
-		assert.Equal(t, expectedPkg.Version, a.Version, "bad version")
-		assert.Equal(t, expectedPkg.Language, a.Language, "bad language")
-		assert.Equal(t, expectedPkg.Type, a.Type, "bad type")
-		assert.Equal(t, expectedPkg.Licenses, a.Licenses, "bad license count")
-	}
-}
-
 func TestParsePackageLock(t *testing.T) {
-	expected := map[string]pkg.Package{
+	var expectedRelationships []artifact.Relationship
-		"@actions/core": {
+	expectedPkgs := []pkg.Package{
+		{
 			Name:     "@actions/core",
 			Version:  "1.6.0",
+			PURL:     "pkg:npm/%40actions/core@1.6.0",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 		},
-		"wordwrap": {
+		{
-			Name:     "wordwrap",
-			Version:  "0.0.3",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"get-stdin": {
-			Name:     "get-stdin",
-			Version:  "5.0.1",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"minimist": {
-			Name:     "minimist",
-			Version:  "0.0.10",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"optimist": {
-			Name:     "optimist",
-			Version:  "0.6.1",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"string-width": {
-			Name:     "string-width",
-			Version:  "2.1.1",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"strip-ansi": {
-			Name:     "strip-ansi",
-			Version:  "4.0.0",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"strip-eof": {
-			Name:     "wordwrap",
-			Version:  "1.0.0",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"ansi-regex": {
 			Name:     "ansi-regex",
 			Version:  "3.0.0",
+			PURL:     "pkg:npm/ansi-regex@3.0.0",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 		},
-		"is-fullwidth-code-point": {
+		{
-			Name:     "is-fullwidth-code-point",
-			Version:  "2.0.0",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"cowsay": {
 			Name:     "cowsay",
 			Version:  "1.4.0",
+			PURL:     "pkg:npm/cowsay@1.4.0",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "get-stdin",
+			Version:  "5.0.1",
+			PURL:     "pkg:npm/get-stdin@5.0.1",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "is-fullwidth-code-point",
+			Version:  "2.0.0",
+			PURL:     "pkg:npm/is-fullwidth-code-point@2.0.0",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "minimist",
+			Version:  "0.0.10",
+			PURL:     "pkg:npm/minimist@0.0.10",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "optimist",
+			Version:  "0.6.1",
+			PURL:     "pkg:npm/optimist@0.6.1",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "string-width",
+			Version:  "2.1.1",
+			PURL:     "pkg:npm/string-width@2.1.1",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "strip-ansi",
+			Version:  "4.0.0",
+			PURL:     "pkg:npm/strip-ansi@4.0.0",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "strip-eof",
+			Version:  "1.0.0",
+			PURL:     "pkg:npm/strip-eof@1.0.0",
+			Language: pkg.JavaScript,
+			Type:     pkg.NpmPkg,
+		},
+		{
+			Name:     "wordwrap",
+			Version:  "0.0.3",
+			PURL:     "pkg:npm/wordwrap@0.0.3",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 		},
 	}
-	fixture, err := os.Open("test-fixtures/pkg-lock/package-lock.json")
+	fixture := "test-fixtures/pkg-lock/package-lock.json"
-	if err != nil {
+	for i := range expectedPkgs {
-		t.Fatalf("failed to open fixture: %+v", err)
+		expectedPkgs[i].Locations.Add(source.NewLocation(fixture))
 	}
 
-	// TODO: no relationships are under test yet
+	pkgtest.TestFileParser(t, fixture, parsePackageLock, expectedPkgs, expectedRelationships)
-	actual, _, err := parsePackageLock(fixture.Name(), fixture)
-	if err != nil {
-		t.Fatalf("failed to parse package-lock.json: %+v", err)
-	}
-
-	assertPkgsEqual(t, actual, expected)
-
 }
 
 func TestParsePackageLockV2(t *testing.T) {
-	expected := map[string]pkg.Package{
+	fixture := "test-fixtures/pkg-lock/package-lock-2.json"
-		"@types/prop-types": {
+	var expectedRelationships []artifact.Relationship
+	expectedPkgs := []pkg.Package{
+		{
 			Name:     "@types/prop-types",
 			Version:  "15.7.5",
+			PURL:     "pkg:npm/%40types/prop-types@15.7.5",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 			Licenses: []string{"MIT"},
 		},
-		"@types/react": {
+		{
-			Name:     "@types/prop-types",
+			Name:     "@types/react",
 			Version:  "18.0.17",
+			PURL:     "pkg:npm/%40types/react@18.0.17",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 			Licenses: []string{"MIT"},
 		},
-		"@types/scheduler": {
+		{
 			Name:     "@types/scheduler",
 			Version:  "0.16.2",
+			PURL:     "pkg:npm/%40types/scheduler@0.16.2",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 			Licenses: []string{"MIT"},
 		},
-		"csstype": {
+		{
 			Name:     "csstype",
 			Version:  "3.1.0",
+			PURL:     "pkg:npm/csstype@3.1.0",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 			Licenses: []string{"MIT"},
 		},
 	}
-	fixture, err := os.Open("test-fixtures/pkg-lock/package-lock-2.json")
+	for i := range expectedPkgs {
-	if err != nil {
+		expectedPkgs[i].Locations.Add(source.NewLocation(fixture))
-		t.Fatalf("failed to open fixture: %+v", err)
 	}
-
+	pkgtest.TestFileParser(t, fixture, parsePackageLock, expectedPkgs, expectedRelationships)
-	actual, _, err := parsePackageLock(fixture.Name(), fixture)
-	if err != nil {
-		t.Fatalf("failed to parse package-lock.json: %+v", err)
-	}
-
-	assertPkgsEqual(t, actual, expected)
 }

syft/pkg/cataloger/javascript/parse_pnpm_lock.go

@@ -8,23 +8,24 @@ import (
 
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/common"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+	"github.com/anchore/syft/syft/source"
 )
 
 // integrity check
-var _ common.ParserFn = parsePnpmLock
+var _ generic.Parser = parsePnpmLock
 
 type pnpmLockYaml struct {
 	Dependencies map[string]string `json:"dependencies"`
 }
 
-func parsePnpmLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Relationship, error) {
+func parsePnpmLock(resolver source.FileResolver, _ *generic.Environment, reader source.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 	bytes, err := io.ReadAll(reader)
 	if err != nil {
 		return nil, nil, fmt.Errorf("failed to load pnpm-lock.yaml file: %w", err)
 	}
 
-	var pkgs []*pkg.Package
+	var pkgs []pkg.Package
 	var lockFile pnpmLockYaml
 
 	if err := yaml.Unmarshal(bytes, &lockFile); err != nil {
@@ -32,13 +33,10 @@ func parsePnpmLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Re
 	}
 
 	for name, version := range lockFile.Dependencies {
-		pkgs = append(pkgs, &pkg.Package{
+		pkgs = append(pkgs, newPnpmPackage(resolver, reader.Location, name, version))
-			Name:     name,
-			Version:  version,
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		})
 	}
 
+	pkg.Sort(pkgs)
+
 	return pkgs, nil, nil
 }

syft/pkg/cataloger/javascript/parse_pnpm_lock_test.go

@@ -1,59 +1,46 @@
 package javascript
 
 import (
-	"os"
-	"sort"
 	"testing"
 
-	"github.com/go-test/deep"
+	"github.com/anchore/syft/syft/artifact"
-
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
+	"github.com/anchore/syft/syft/source"
 )
 
-func fixtureP(str string) *string {
-	return &str
-}
-
 func TestParsePnpmLock(t *testing.T) {
-	expected := []*pkg.Package{
+	var expectedRelationships []artifact.Relationship
+	fixture := "test-fixtures/pnpm/pnpm-lock.yaml"
+
+	locationSet := source.NewLocationSet(source.NewLocation(fixture))
+
+	expectedPkgs := []pkg.Package{
 		{
-			Name:     "nanoid",
+			Name:      "nanoid",
-			Version:  "3.3.4",
+			Version:   "3.3.4",
-			Language: pkg.JavaScript,
+			PURL:      "pkg:npm/nanoid@3.3.4",
-			Type:     pkg.NpmPkg,
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
 		{
-			Name:     "picocolors",
+			Name:      "picocolors",
-			Version:  "1.0.0",
+			Version:   "1.0.0",
-			Language: pkg.JavaScript,
+			PURL:      "pkg:npm/picocolors@1.0.0",
-			Type:     pkg.NpmPkg,
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
 		{
-			Name:     "source-map-js",
+			Name:      "source-map-js",
-			Version:  "1.0.2",
+			Version:   "1.0.2",
-			Language: pkg.JavaScript,
+			PURL:      "pkg:npm/source-map-js@1.0.2",
-			Type:     pkg.NpmPkg,
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
 	}
 
-	fixture, err := os.Open("test-fixtures/pnpm/pnpm-lock.yaml")
+	pkgtest.TestFileParser(t, fixture, parsePnpmLock, expectedPkgs, expectedRelationships)
-	if err != nil {
-		t.Fatalf("failed to open fixture: %+v", err)
-	}
-
-	// TODO: no relationships are under test yet
-	actual, _, err := parsePnpmLock(fixture.Name(), fixture)
-	if err != nil {
-		t.Error(err)
-	}
-
-	// we have to sort this for expected to match actual since yaml maps are unordered
-	sort.Slice(actual, func(p, q int) bool {
-		return actual[p].Name < actual[q].Name
-	})
-
-	differences := deep.Equal(expected, actual)
-	if differences != nil {
-		t.Errorf("returned package list differed from expectation: %+v", differences)
-	}
 }

syft/pkg/cataloger/javascript/parse_yarn_lock.go

@@ -3,17 +3,17 @@ package javascript
 import (
 	"bufio"
 	"fmt"
-	"io"
 	"regexp"
 
 	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/common"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+	"github.com/anchore/syft/syft/source"
 )
 
 // integrity check
-var _ common.ParserFn = parseYarnLock
+var _ generic.Parser = parseYarnLock
 
 var (
 	// packageNameExp matches the name of the dependency in yarn.lock
@@ -42,14 +42,14 @@ const (
 	noVersion = ""
 )
 
-func parseYarnLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Relationship, error) {
+func parseYarnLock(resolver source.FileResolver, _ *generic.Environment, reader source.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 	// in the case we find yarn.lock files in the node_modules directories, skip those
 	// as the whole purpose of the lock file is for the specific dependencies of the project
-	if pathContainsNodeModulesDirectory(path) {
+	if pathContainsNodeModulesDirectory(reader.AccessPath()) {
 		return nil, nil, nil
 	}
 
-	var packages []*pkg.Package
+	var pkgs []pkg.Package
 	scanner := bufio.NewScanner(reader)
 	parsedPackages := internal.NewStringSet()
 	currentPackage := noPackage
@@ -61,7 +61,7 @@ func parseYarnLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Re
 		if packageName := findPackageName(line); packageName != noPackage {
 			// When we find a new package, check if we have unsaved identifiers
 			if currentPackage != noPackage && currentVersion != noVersion && !parsedPackages.Contains(currentPackage+"@"+currentVersion) {
-				packages = append(packages, newYarnLockPackage(currentPackage, currentVersion))
+				pkgs = append(pkgs, newYarnLockPackage(resolver, reader.Location, currentPackage, currentVersion))
 				parsedPackages.Add(currentPackage + "@" + currentVersion)
 			}
 
@@ -69,7 +69,7 @@ func parseYarnLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Re
 		} else if version := findPackageVersion(line); version != noVersion {
 			currentVersion = version
 		} else if packageName, version := findPackageAndVersion(line); packageName != noPackage && version != noVersion && !parsedPackages.Contains(packageName+"@"+version) {
-			packages = append(packages, newYarnLockPackage(packageName, version))
+			pkgs = append(pkgs, newYarnLockPackage(resolver, reader.Location, packageName, version))
 			parsedPackages.Add(packageName + "@" + version)
 
 			// Cleanup to indicate no unsaved identifiers
@@ -80,7 +80,7 @@ func parseYarnLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Re
 
 	// check if we have valid unsaved data after end-of-file has reached
 	if currentPackage != noPackage && currentVersion != noVersion && !parsedPackages.Contains(currentPackage+"@"+currentVersion) {
-		packages = append(packages, newYarnLockPackage(currentPackage, currentVersion))
+		pkgs = append(pkgs, newYarnLockPackage(resolver, reader.Location, currentPackage, currentVersion))
 		parsedPackages.Add(currentPackage + "@" + currentVersion)
 	}
 
@@ -88,7 +88,9 @@ func parseYarnLock(path string, reader io.Reader) ([]*pkg.Package, []artifact.Re
 		return nil, nil, fmt.Errorf("failed to parse yarn.lock file: %w", err)
 	}
 
-	return packages, nil, nil
+	pkg.Sort(pkgs)
+
+	return pkgs, nil, nil
 }
 
 func findPackageName(line string) string {
@@ -114,12 +116,3 @@ func findPackageAndVersion(line string) (string, string) {
 
 	return noPackage, noVersion
 }
-
-func newYarnLockPackage(name, version string) *pkg.Package {
-	return &pkg.Package{
-		Name:     name,
-		Version:  version,
-		Language: pkg.JavaScript,
-		Type:     pkg.NpmPkg,
-	}
-}

syft/pkg/cataloger/javascript/parse_yarn_lock_test.go

@@ -1,170 +1,184 @@
 package javascript
 
 import (
-	"os"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
 
+	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
+	"github.com/anchore/syft/syft/source"
 )
 
 func TestParseYarnBerry(t *testing.T) {
-	expected := map[string]pkg.Package{
+	var expectedRelationships []artifact.Relationship
-		"@babel/code-frame": {
+	fixture := "test-fixtures/yarn-berry/yarn.lock"
-			Name:     "@babel/code-frame",
+	locations := source.NewLocationSet(source.NewLocation(fixture))
-			Version:  "7.10.4",
+
-			Language: pkg.JavaScript,
+	expectedPkgs := []pkg.Package{
-			Type:     pkg.NpmPkg,
+		{
+			Name:      "@babel/code-frame",
+			Version:   "7.10.4",
+			Locations: locations,
+			PURL:      "pkg:npm/%40babel/code-frame@7.10.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"@types/minimatch": {
+		{
-			Name:     "@types/minimatch",
+			Name:      "@types/minimatch",
-			Version:  "3.0.3",
+			Version:   "3.0.3",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/%40types/minimatch@3.0.3",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"@types/qs": {
+		{
-			Name:     "@types/qs",
+			Name:      "@types/qs",
-			Version:  "6.9.4",
+			Version:   "6.9.4",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/%40types/qs@6.9.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"ajv": {
+		{
-			Name:     "ajv",
+			Name:      "ajv",
-			Version:  "6.12.3",
+			Version:   "6.12.3",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/ajv@6.12.3",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"atob": {
+		{
-			Name:     "atob",
+			Name:      "asn1.js",
-			Version:  "2.1.2",
+			Version:   "4.10.1",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/asn1.js@4.10.1",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"aws-sdk": {
+		{
-			Name:     "aws-sdk",
+			Name:      "atob",
-			Version:  "2.706.0",
+			Version:   "2.1.2",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/atob@2.1.2",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"jhipster-core": {
+		{
-			Name:     "jhipster-core",
+			Name:      "aws-sdk",
-			Version:  "7.3.4",
+			Version:   "2.706.0",
-			Language: pkg.JavaScript,
+			PURL:      "pkg:npm/aws-sdk@2.706.0",
-			Type:     pkg.NpmPkg,
+			Locations: locations,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"asn1.js": {
+		{
-			Name:     "asn1.js",
+			Name:      "c0n-fab_u.laTION",
-			Version:  "4.10.1",
+			Version:   "7.7.7",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/c0n-fab_u.laTION@7.7.7",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"c0n-fab_u.laTION": {
+		{
-			Name:     "c0n-fab_u.laTION",
+			Name:      "jhipster-core",
-			Version:  "7.7.7",
+			Version:   "7.3.4",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/jhipster-core@7.3.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
 	}
-	testFixtures := []string{
-		"test-fixtures/yarn-berry/yarn.lock",
-	}
 
-	for _, file := range testFixtures {
+	pkgtest.TestFileParser(t, fixture, parseYarnLock, expectedPkgs, expectedRelationships)
-		file := file
-		t.Run(file, func(t *testing.T) {
-			t.Parallel()
 
-			fixture, err := os.Open(file)
-			require.NoError(t, err)
-
-			// TODO: no relationships are under test yet
-			actual, _, err := parseYarnLock(fixture.Name(), fixture)
-			require.NoError(t, err)
-
-			assertPkgsEqual(t, actual, expected)
-		})
-	}
 }
 
 func TestParseYarnLock(t *testing.T) {
-	expected := map[string]pkg.Package{
+	var expectedRelationships []artifact.Relationship
-		"@babel/code-frame": {
+	fixture := "test-fixtures/yarn/yarn.lock"
-			Name:     "@babel/code-frame",
+	locations := source.NewLocationSet(source.NewLocation(fixture))
-			Version:  "7.10.4",
+
+	expectedPkgs := []pkg.Package{
+		{
+			Name:      "@babel/code-frame",
+			Version:   "7.10.4",
+			Locations: locations,
+			PURL:      "pkg:npm/%40babel/code-frame@7.10.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "@types/minimatch",
+			Version:   "3.0.3",
+			Locations: locations,
+			PURL:      "pkg:npm/%40types/minimatch@3.0.3",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "@types/qs",
+			Version:   "6.9.4",
+			Locations: locations,
+			PURL:      "pkg:npm/%40types/qs@6.9.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "ajv",
+			Version:   "6.12.3",
+			Locations: locations,
+			PURL:      "pkg:npm/ajv@6.12.3",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "asn1.js",
+			Version:   "4.10.1",
+			Locations: locations,
+			PURL:      "pkg:npm/asn1.js@4.10.1",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "atob",
+			Version:   "2.1.2",
+			Locations: locations,
+
+			PURL:     "pkg:npm/atob@2.1.2",
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 		},
-		"@types/minimatch": {
+		{
-			Name:     "@types/minimatch",
+			Name:      "aws-sdk",
-			Version:  "3.0.3",
+			Version:   "2.706.0",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/aws-sdk@2.706.0",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"@types/qs": {
+		{
-			Name:     "@types/qs",
+			Name:      "jhipster-core",
-			Version:  "6.9.4",
+			Version:   "7.3.4",
-			Language: pkg.JavaScript,
+			Locations: locations,
-			Type:     pkg.NpmPkg,
+			PURL:      "pkg:npm/jhipster-core@7.3.4",
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
 		},
-		"ajv": {
+
-			Name:     "ajv",
+		{
-			Version:  "6.12.3",
+			Name:      "something-i-made-up",
-			Language: pkg.JavaScript,
+			Version:   "7.7.7",
-			Type:     pkg.NpmPkg,
+			Locations: locations,
-		},
+			PURL:      "pkg:npm/something-i-made-up@7.7.7",
-		"atob": {
+			Language:  pkg.JavaScript,
-			Name:     "atob",
+			Type:      pkg.NpmPkg,
-			Version:  "2.1.2",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"aws-sdk": {
-			Name:     "aws-sdk",
-			Version:  "2.706.0",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"jhipster-core": {
-			Name:     "jhipster-core",
-			Version:  "7.3.4",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"asn1.js": {
-			Name:     "asn1.js",
-			Version:  "4.10.1",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
-		},
-		"something-i-made-up": {
-			Name:     "something-i-made-up",
-			Version:  "7.7.7",
-			Language: pkg.JavaScript,
-			Type:     pkg.NpmPkg,
 		},
 	}
 
-	testFixtures := []string{
+	pkgtest.TestFileParser(t, fixture, parseYarnLock, expectedPkgs, expectedRelationships)
-		"test-fixtures/yarn/yarn.lock",
-	}
 
-	for _, file := range testFixtures {
-		file := file
-		t.Run(file, func(t *testing.T) {
-			t.Parallel()
-
-			fixture, err := os.Open(file)
-			require.NoError(t, err)
-
-			// TODO: no relationships are under test yet
-			actual, _, err := parseYarnLock(fixture.Name(), fixture)
-			require.NoError(t, err)
-
-			assertPkgsEqual(t, actual, expected)
-		})
-	}
 }
 
 func TestParseYarnFindPackageNames(t *testing.T) {
@@ -227,7 +241,6 @@ func TestParseYarnFindPackageNames(t *testing.T) {
 	}
 
 	for _, test := range tests {
-		test := test
 		t.Run(test.expected, func(t *testing.T) {
 			t.Parallel()
 			actual := findPackageName(test.line)
@@ -316,7 +329,6 @@ func TestParseYarnFindPackageVersions(t *testing.T) {
 	}
 
 	for _, test := range tests {
-		test := test
 		t.Run(test.expected, func(t *testing.T) {
 			t.Parallel()
 			actual := findPackageVersion(test.line)

syft/pkg/npm_package_json_metadata.go

@@ -1,14 +1,5 @@
 package pkg
 
-import (
-	"strings"
-
-	"github.com/anchore/packageurl-go"
-	"github.com/anchore/syft/syft/linux"
-)
-
-var _ urlIdentifier = (*NpmPackageJSONMetadata)(nil)
-
 // NpmPackageJSONMetadata holds extra information that is used in pkg.Package
 type NpmPackageJSONMetadata struct {
 	Name        string   `mapstructure:"name" json:"name"`
@@ -21,24 +12,3 @@ type NpmPackageJSONMetadata struct {
 	URL         string   `mapstructure:"url" json:"url"`
 	Private     bool     `mapstructure:"private" json:"private"`
 }
-
-// PackageURL returns the PURL for the specific NPM package (see https://github.com/package-url/purl-spec)
-func (p NpmPackageJSONMetadata) PackageURL(_ *linux.Release) string {
-	var namespace string
-	name := p.Name
-
-	fields := strings.SplitN(p.Name, "/", 2)
-	if len(fields) > 1 {
-		namespace = fields[0]
-		name = fields[1]
-	}
-
-	return packageurl.NewPackageURL(
-		packageurl.TypeNPM,
-		namespace,
-		name,
-		p.Version,
-		nil,
-		"",
-	).ToString()
-}

syft/pkg/npm_package_json_metadata_test.go

@@ -1,63 +0,0 @@
-package pkg
-
-import (
-	"fmt"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/anchore/packageurl-go"
-)
-
-func TestNpmPackageJSONMetadata_PackageURL(t *testing.T) {
-
-	tests := []struct {
-		name      string
-		metadata  NpmPackageJSONMetadata
-		expected  string
-		namespace string
-	}{
-		{
-			name: "no namespace",
-			metadata: NpmPackageJSONMetadata{
-				Name:    "arborist",
-				Version: "2.6.2",
-			},
-			expected: "pkg:npm/arborist@2.6.2",
-		},
-		{
-			name: "split by namespace",
-			metadata: NpmPackageJSONMetadata{
-				Name:    "npmcli/arborist",
-				Version: "2.6.2",
-			},
-			expected:  "pkg:npm/npmcli/arborist@2.6.2",
-			namespace: "npmcli",
-		},
-		{
-			name: "encoding @ symobl",
-			metadata: NpmPackageJSONMetadata{
-				Name:    "@npmcli/arborist",
-				Version: "2.6.2",
-			},
-			expected:  "pkg:npm/%40npmcli/arborist@2.6.2",
-			namespace: "@npmcli",
-		},
-	}
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			actual := tt.metadata.PackageURL(nil)
-			assert.Equal(t, tt.expected, actual)
-			decoded, err := packageurl.FromString(actual)
-			require.NoError(t, err)
-			assert.Equal(t, tt.namespace, decoded.Namespace)
-			if decoded.Namespace != "" {
-				assert.Equal(t, tt.metadata.Name, fmt.Sprintf("%s/%s", decoded.Namespace, decoded.Name))
-			} else {
-				assert.Equal(t, tt.metadata.Name, decoded.Name)
-			}
-			assert.Equal(t, tt.metadata.Version, decoded.Version)
-		})
-	}
-}

syft/pkg/package.go

@@ -5,6 +5,7 @@ package pkg
 
 import (
 	"fmt"
+	"sort"
 
 	"github.com/anchore/syft/internal/log"
 	"github.com/anchore/syft/syft/artifact"
@@ -80,3 +81,24 @@ func (p *Package) merge(other Package) error {
 func IsValid(p *Package) bool {
 	return p != nil && p.Name != ""
 }
+
+func Sort(pkgs []Package) {
+	sort.SliceStable(pkgs, func(i, j int) bool {
+		if pkgs[i].Name == pkgs[j].Name {
+			if pkgs[i].Version == pkgs[j].Version {
+				iLocations := pkgs[i].Locations.ToSlice()
+				jLocations := pkgs[j].Locations.ToSlice()
+				if pkgs[i].Type == pkgs[j].Type && len(iLocations) > 0 && len(jLocations) > 0 {
+					if iLocations[0].String() == jLocations[0].String() {
+						// compare IDs as a final fallback
+						return pkgs[i].ID() < pkgs[j].ID()
+					}
+					return iLocations[0].String() < jLocations[0].String()
+				}
+				return pkgs[i].Type < pkgs[j].Type
+			}
+			return pkgs[i].Version < pkgs[j].Version
+		}
+		return pkgs[i].Name < pkgs[j].Name
+	})
+}