feat(os/rpm): extend os/rpm extractor setting PRETTYNAME and add openEuler accurate mapping

binary/proto/scan_result.proto

@@ -359,6 +359,7 @@ message RPMPackageMetadata {
   string os_name = 7;
   string vendor = 8;
   string architecture = 9;
+  string os_pretty_name = 11;
 
   reserved "license";
 }

extractor/filesystem/os/ecosystem/ecosystem.go

@@ -55,7 +55,7 @@ func MakeEcosystem(metadata any) osvecosystem.Parsed {
 			return osvecosystem.FromEcosystem(osvschema.EcosystemRockyLinux)
 		}
 		if m.OSID == "openEuler" {
-			return osvecosystem.FromEcosystem(osvschema.EcosystemOpenEuler)
+			return osvecosystem.Parsed{Ecosystem: osvschema.EcosystemOpenEuler, Suffix: m.OpenEulerEcosystemSuffix()}
 		}
 
 	case *snapmeta.Metadata:

extractor/filesystem/os/ecosystem/ecosystem_test.go

@@ -223,7 +223,39 @@ func TestEcosystemRPM(t *testing.T) {
 			want: "Rocky Linux",
 		},
 		{
-			desc: "openEuler",
+			desc: "openEuler base version from pretty name",
+			metadata: &rpmmeta.Metadata{
+				OSID:         "openEuler",
+				OSPrettyName: "openEuler 24.03",
+			},
+			want: "openEuler:24.03",
+		},
+		{
+			desc: "openEuler LTS qualifier",
+			metadata: &rpmmeta.Metadata{
+				OSID:         "openEuler",
+				OSPrettyName: "openEuler 24.03 (LTS)",
+			},
+			want: "openEuler:24.03-LTS",
+		},
+		{
+			desc: "openEuler LTS service pack qualifier",
+			metadata: &rpmmeta.Metadata{
+				OSID:         "openEuler",
+				OSPrettyName: "openEuler 24.03 (LTS-SP1)",
+			},
+			want: "openEuler:24.03-LTS-SP1",
+		},
+		{
+			desc: "openEuler fallback to version ID",
+			metadata: &rpmmeta.Metadata{
+				OSID:        "openEuler",
+				OSVersionID: "24.03",
+			},
+			want: "openEuler:24.03",
+		},
+		{
+			desc: "openEuler no version details",
 			metadata: &rpmmeta.Metadata{
 				OSID: "openEuler",
 			},

extractor/filesystem/os/rpm/metadata/metadata.go

@@ -17,18 +17,29 @@ package metadata
 
 import (
 	"fmt"
+	"regexp"
+	"strings"
 
 	"github.com/google/osv-scalibr/log"
 
 	pb "github.com/google/osv-scalibr/binary/proto/scan_result_go_proto"
 )
 
+var (
+	// pattern to match: openEuler version (qualifier)
+	// where (qualifier) is optional. Those without qualifier are considered innovation versions.
+	// version: YY.MM format. e.g. 24.03
+	// qualifier: currently supported are LTS, LTS SPd. e.g. LTS, LTS-SP1, LTS-SP2
+	openEulerPrettyNameRegexp = regexp.MustCompile(`^openEuler\s+([0-9]{2}\.[0-9]{2})(?:\s*\((LTS(?:[- ]SP\d+)?)\))?$`)
+)
+
 // Metadata holds parsing information for an rpm package.
 type Metadata struct {
 	PackageName  string
 	SourceRPM    string
 	Epoch        int
 	OSName       string
+	OSPrettyName string
 	OSID         string
 	OSVersionID  string
 	OSBuildID    string
@@ -80,6 +91,7 @@ func (m *Metadata) SetProto(p *pb.Package) {
 			SourceRpm:    m.SourceRPM,
 			Epoch:        int32(m.Epoch),
 			OsName:       m.OSName,
+			OsPrettyName: m.OSPrettyName,
 			OsId:         m.OSID,
 			OsVersionId:  m.OSVersionID,
 			OsBuildId:    m.OSBuildID,
@@ -100,10 +112,41 @@ func ToStruct(m *pb.RPMPackageMetadata) *Metadata {
 		SourceRPM:    m.GetSourceRpm(),
 		Epoch:        int(m.GetEpoch()),
 		OSName:       m.GetOsName(),
+		OSPrettyName: m.GetOsPrettyName(),
 		OSID:         m.GetOsId(),
 		OSVersionID:  m.GetOsVersionId(),
 		OSBuildID:    m.GetOsBuildId(),
 		Vendor:       m.GetVendor(),
 		Architecture: m.GetArchitecture(),
 	}
 }
+
+// OpenEulerEcosystemSuffix returns the normalized ecosystem suffix for openEuler RPMs with uses of pretty name.
+// e.g. openEuler 24.03 (LTS) -> 24.03-LTS
+func (m *Metadata) OpenEulerEcosystemSuffix() string {
+	if m == nil {
+		return ""
+	}
+
+	prettyName := strings.TrimSpace(m.OSPrettyName)
+	if prettyName != "" {
+		if matches := openEulerPrettyNameRegexp.FindStringSubmatch(prettyName); len(matches) > 0 {
+			version := matches[1]
+			qualifier := strings.TrimSpace(matches[2])
+			if qualifier == "" {
+				return version
+			}
+
+			qualifier = strings.ReplaceAll(qualifier, " ", "-")
+			qualifier = strings.Trim(qualifier, "-")
+			if qualifier == "" {
+				return version
+			}
+
+			return version + "-" + qualifier
+		}
+	}
+
+	versionID := strings.TrimSpace(m.OSVersionID)
+	return versionID
+}

extractor/filesystem/os/rpm/metadata/metadata_test.go

@@ -90,6 +90,7 @@ func TestSetProto(t *testing.T) {
 				SourceRPM:    "source-rpm",
 				Epoch:        1,
 				OSName:       "os-name",
+				OSPrettyName: "os-pretty-name",
 				OSID:         "os-id",
 				OSVersionID:  "os-version-id",
 				OSBuildID:    "os-build-id",
@@ -105,6 +106,7 @@ func TestSetProto(t *testing.T) {
 						SourceRpm:    "source-rpm",
 						Epoch:        1,
 						OsName:       "os-name",
+						OsPrettyName: "os-pretty-name",
 						OsId:         "os-id",
 						OsVersionId:  "os-version-id",
 						OsBuildId:    "os-build-id",
@@ -168,6 +170,7 @@ func TestToStruct(t *testing.T) {
 				SourceRpm:    "source-rpm",
 				Epoch:        1,
 				OsName:       "os-name",
+				OsPrettyName: "os-pretty-name",
 				OsId:         "os-id",
 				OsVersionId:  "os-version-id",
 				OsBuildId:    "os-build-id",
@@ -179,6 +182,7 @@ func TestToStruct(t *testing.T) {
 				SourceRPM:    "source-rpm",
 				Epoch:        1,
 				OSName:       "os-name",
+				OSPrettyName: "os-pretty-name",
 				OSID:         "os-id",
 				OSVersionID:  "os-version-id",
 				OSBuildID:    "os-build-id",
@@ -217,3 +221,68 @@ func TestToStruct(t *testing.T) {
 		})
 	}
 }
+
+func TestOpenEulerEcosystemSuffix(t *testing.T) {
+	testCases := []struct {
+		desc string
+		meta *metadata.Metadata
+		want string
+	}{
+		{
+			desc: "base version from pretty name",
+			meta: &metadata.Metadata{
+				OSPrettyName: "openEuler 24.03",
+			},
+			want: "24.03",
+		},
+		{
+			desc: "lts qualifier",
+			meta: &metadata.Metadata{
+				OSPrettyName: "openEuler 24.03 (LTS)",
+			},
+			want: "24.03-LTS",
+		},
+		{
+			desc: "lts space qualifier",
+			meta: &metadata.Metadata{
+				OSPrettyName: "openEuler 24.03 (LTS SP1)",
+			},
+			want: "24.03-LTS-SP1",
+		},
+		{
+			desc: "lts hyphen qualifier",
+			meta: &metadata.Metadata{
+				OSPrettyName: "openEuler 24.03 (LTS-SP1)",
+			},
+			want: "24.03-LTS-SP1",
+		},
+		{
+			desc: "fallback to version id",
+			meta: &metadata.Metadata{
+				OSVersionID: "24.03",
+			},
+			want: "24.03",
+		},
+		{
+			desc: "non openEuler pretty name",
+			meta: &metadata.Metadata{
+				OSPrettyName: "Fedora Linux 38 (Container Image)",
+				OSVersionID:  "38",
+			},
+			want: "38",
+		},
+		{
+			desc: "no details",
+			meta: &metadata.Metadata{},
+			want: "",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.desc, func(t *testing.T) {
+			if got := tc.meta.OpenEulerEcosystemSuffix(); got != tc.want {
+				t.Fatalf("OpenEulerEcosystemSuffix() = %q, want %q", got, tc.want)
+			}
+		})
+	}
+}

extractor/filesystem/os/rpm/rpm_linux.go

@@ -195,6 +195,7 @@ func (e Extractor) extractFromInput(ctx context.Context, input *filesystem.ScanI
 			SourceRPM:    p.SourceRPM,
 			Epoch:        p.Epoch,
 			OSName:       m["NAME"],
+			OSPrettyName: m["PRETTY_NAME"],
 			OSID:         m["ID"],
 			OSVersionID:  m["VERSION_ID"],
 			OSBuildID:    m["BUILD_ID"],