encode upstream qualifier on os package pURLs (anchore#769)

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

syft/pkg/apk_metadata.go

@@ -49,6 +49,14 @@ type ApkFileRecord struct {
 
 // PackageURL returns the PURL for the specific Alpine package (see https://github.com/package-url/purl-spec)
 func (m ApkMetadata) PackageURL(distro *linux.Release) string {
+	qualifiers := map[string]string{
+		purlArchQualifier: m.Architecture,
+	}
+
+	if m.OriginPackage != "" {
+		qualifiers[purlUpstreamQualifier] = m.OriginPackage
+	}
+
 	return packageurl.NewPackageURL(
 		// note: this is currently a candidate and not technically within spec
 		// see https://github.com/package-url/purl-spec#other-candidate-types-to-define
@@ -57,9 +65,7 @@ func (m ApkMetadata) PackageURL(distro *linux.Release) string {
 		m.Package,
 		m.Version,
 		purlQualifiers(
-			map[string]string{
-				purlArchQualifier: m.Architecture,
-			},
+			qualifiers,
 			distro,
 		),
 		"",

syft/pkg/apk_metadata_test.go

@@ -31,7 +31,7 @@ func TestApkMetadata_pURL(t *testing.T) {
 			expected: "pkg:alpine/p@v?arch=a&distro=alpine-3.4.6",
 		},
 		{
-			name: "missing architecure",
+			name: "missing architecture",
 			metadata: ApkMetadata{
 				Package: "p",
 				Version: "v",
@@ -67,6 +67,20 @@ func TestApkMetadata_pURL(t *testing.T) {
 			},
 			expected: "pkg:alpine/g%20plus%20plus@v84?arch=am86&distro=alpine-3.15.0",
 		},
+		{
+			name: "add source information as qualifier",
+			metadata: ApkMetadata{
+				Package:       "p",
+				Version:       "v",
+				Architecture:  "a",
+				OriginPackage: "origin",
+			},
+			distro: linux.Release{
+				ID:        "alpine",
+				VersionID: "3.4.6",
+			},
+			expected: "pkg:alpine/p@v?arch=a&upstream=origin&distro=alpine-3.4.6",
+		},
 	}
 
 	for _, test := range tests {

syft/pkg/dpkg_metadata.go

@@ -1,6 +1,7 @@
 package pkg
 
 import (
+	"fmt"
 	"sort"
 
 	"github.com/anchore/syft/syft/file"
@@ -43,6 +44,19 @@ func (m DpkgMetadata) PackageURL(distro *linux.Release) string {
 	if distro != nil {
 		namespace = distro.ID
 	}
+
+	qualifiers := map[string]string{
+		purlArchQualifier: m.Architecture,
+	}
+
+	if m.Source != "" {
+		if m.SourceVersion != "" {
+			qualifiers[purlUpstreamQualifier] = fmt.Sprintf("%s@%s", m.Source, m.SourceVersion)
+		} else {
+			qualifiers[purlUpstreamQualifier] = m.Source
+		}
+	}
+
 	return packageurl.NewPackageURL(
 		// TODO: replace with `packageurl.TypeDebian` upon merge of https://github.com/package-url/packageurl-go/pull/21
 		// TODO: or, since we're now using an Anchore fork of this module, we could do this sooner.
@@ -51,9 +65,7 @@ func (m DpkgMetadata) PackageURL(distro *linux.Release) string {
 		m.Package,
 		m.Version,
 		purlQualifiers(
-			map[string]string{
-				purlArchQualifier: m.Architecture,
-			},
+			qualifiers,
 			distro,
 		),
 		"",

syft/pkg/dpkg_metadata_test.go

@@ -25,7 +25,6 @@ func TestDpkgMetadata_pURL(t *testing.T) {
 			},
 			metadata: DpkgMetadata{
 				Package: "p",
-				Source:  "s",
 				Version: "v",
 			},
 			expected: "pkg:deb/debian/p@v?distro=debian-11",
@@ -38,7 +37,6 @@ func TestDpkgMetadata_pURL(t *testing.T) {
 			},
 			metadata: DpkgMetadata{
 				Package:      "p",
-				Source:       "s",
 				Version:      "v",
 				Architecture: "a",
 			},
@@ -48,11 +46,37 @@ func TestDpkgMetadata_pURL(t *testing.T) {
 			name: "missing distro",
 			metadata: DpkgMetadata{
 				Package: "p",
-				Source:  "s",
 				Version: "v",
 			},
 			expected: "pkg:deb/p@v",
 		},
+		{
+			name: "with upstream qualifier with source pkg name info",
+			distro: &linux.Release{
+				ID:        "debian",
+				VersionID: "11",
+			},
+			metadata: DpkgMetadata{
+				Package: "p",
+				Source:  "s",
+				Version: "v",
+			},
+			expected: "pkg:deb/debian/p@v?upstream=s&distro=debian-11",
+		},
+		{
+			name: "with upstream qualifier with source pkg name and version info",
+			distro: &linux.Release{
+				ID:        "debian",
+				VersionID: "11",
+			},
+			metadata: DpkgMetadata{
+				Package:       "p",
+				Source:        "s",
+				Version:       "v",
+				SourceVersion: "2.3",
+			},
+			expected: "pkg:deb/debian/p@v?upstream=s@2.3&distro=debian-11",
+		},
 	}
 
 	for _, test := range tests {

syft/pkg/python_package_metadata.go

@@ -106,7 +106,7 @@ func (p PythonDirectURLOriginInfo) vcsURLQualifier() packageurl.Qualifiers {
 	if p.VCS != "" {
 		// Taken from https://github.com/package-url/purl-spec/blob/master/PURL-SPECIFICATION.rst#known-qualifiers-keyvalue-pairs
 		// packageurl-go still doesn't support all qualifier names
-		return packageurl.Qualifiers{{Key: purlVCSURL, Value: fmt.Sprintf("%s+%s@%s", p.VCS, p.URL, p.CommitID)}}
+		return packageurl.Qualifiers{{Key: purlVCSURLQualifier, Value: fmt.Sprintf("%s+%s@%s", p.VCS, p.URL, p.CommitID)}}
 	}
 	return nil
 }

syft/pkg/rpmdb_metadata.go

@@ -63,6 +63,10 @@ func (m RpmdbMetadata) PackageURL(distro *linux.Release) string {
 		qualifiers[purlEpochQualifier] = strconv.Itoa(*m.Epoch)
 	}
 
+	if m.SourceRpm != "" {
+		qualifiers[purlUpstreamQualifier] = m.SourceRpm
+	}
+
 	return packageurl.NewPackageURL(
 		packageurl.TypeRPM,
 		namespace,

syft/pkg/rpmdb_metadata_test.go

@@ -18,33 +18,33 @@ func TestRpmMetadata_pURL(t *testing.T) {
 		expected string
 	}{
 		{
-			name: "with arch and epoch",
+			name: "go case",
 			distro: &linux.Release{
-				ID:        "centos",
-				VersionID: "7",
+				ID:        "rhel",
+				VersionID: "8.4",
 			},
 			metadata: RpmdbMetadata{
 				Name:    "p",
 				Version: "v",
-				Arch:    "a",
 				Release: "r",
-				Epoch:   intRef(1),
+				Epoch:   nil,
 			},
-			expected: "pkg:rpm/centos/p@v-r?arch=a&epoch=1&distro=centos-7",
+			expected: "pkg:rpm/rhel/p@v-r?distro=rhel-8.4",
 		},
 		{
-			name: "go case",
+			name: "with arch and epoch",
 			distro: &linux.Release{
-				ID:        "rhel",
-				VersionID: "8.4",
+				ID:        "centos",
+				VersionID: "7",
 			},
 			metadata: RpmdbMetadata{
 				Name:    "p",
 				Version: "v",
+				Arch:    "a",
 				Release: "r",
-				Epoch:   nil,
+				Epoch:   intRef(1),
 			},
-			expected: "pkg:rpm/rhel/p@v-r?distro=rhel-8.4",
+			expected: "pkg:rpm/centos/p@v-r?arch=a&epoch=1&distro=centos-7",
 		},
 		{
 			name: "missing distro",
@@ -56,6 +56,20 @@ func TestRpmMetadata_pURL(t *testing.T) {
 			},
 			expected: "pkg:rpm/p@v-r",
 		},
+		{
+			name: "with upstream source rpm info",
+			distro: &linux.Release{
+				ID:        "rhel",
+				VersionID: "8.4",
+			},
+			metadata: RpmdbMetadata{
+				Name:      "p",
+				Version:   "v",
+				Release:   "r",
+				SourceRpm: "sourcerpm",
+			},
+			expected: "pkg:rpm/rhel/p@v-r?upstream=sourcerpm&distro=rhel-8.4",
+		},
 	}
 
 	for _, test := range tests {

syft/pkg/url.go

@@ -14,7 +14,10 @@ const (
 	purlArchQualifier   = "arch"
 	purlDistroQualifier = "distro"
 	purlEpochQualifier  = "epoch"
-	purlVCSURL          = "vcs_url"
+	purlVCSURLQualifier = "vcs_url"
+
+	// this qualifier is not in the pURL spec, but is used by grype to perform indirect matching based on source information
+	purlUpstreamQualifier = "upstream"
 )
 
 type urlIdentifier interface {