fix: separating multiple licenses from one line in dpkg copyright files (#2508)

Co-authored-by: knqyf263 <knqyf263@gmail.com>

.github/workflows/semantic-pr.yaml

@@ -34,6 +34,7 @@ jobs:
             vuln
             misconf
             secret
+            license
 
             image
             fs

docs/community/contribute/pr.md

@@ -42,6 +42,7 @@ checks:
 - vuln
 - misconf
 - secret
+- license
 
 mode:
 

pkg/fanal/analyzer/language/analyze.go

@@ -1,6 +1,8 @@
 package language
 
 import (
+	"strings"
+
 	"golang.org/x/xerrors"
 
 	dio "github.com/aquasecurity/go-dep-parser/pkg/io"
@@ -35,7 +37,10 @@ func ToAnalysisResult(fileType, filePath, libFilePath string, libs []godeptypes.
 	for _, lib := range libs {
 		var licenses []string
 		if lib.License != "" {
-			licenses = []string{licensing.Normalize(lib.License)}
+			licenses = strings.Split(lib.License, ",")
+			for i, license := range licenses {
+				licenses[i] = licensing.Normalize(strings.TrimSpace(license))
+			}
 		}
 		pkgs = append(pkgs, types.Package{
 			ID:        lib.ID,

pkg/fanal/analyzer/language/ruby/gemspec/gemspec_test.go

@@ -31,7 +31,7 @@ func Test_gemspecLibraryAnalyzer_Analyze(t *testing.T) {
 							{
 								Name:     "test-unit",
 								Version:  "3.3.7",
-								Licenses: []string{"Ruby, BSDL, PSFL"},
+								Licenses: []string{"Ruby", "BSDL", "PSFL"},
 								FilePath: "testdata/multiple_licenses.gemspec",
 							},
 						},

pkg/fanal/analyzer/pkg/dpkg/copyright.go

@@ -37,6 +37,7 @@ var (
 	licenseClassifier *classifier.Classifier
 
 	commonLicenseReferenceRegexp = regexp.MustCompile(`/?usr/share/common-licenses/([0-9A-Za-z_.+-]+[0-9A-Za-z+])`)
+	licenseSplitRegexp           = regexp.MustCompile("(,?[_ ]+or[_ ]+)|(,?[_ ]+and[_ ])|(,[ ]*)")
 )
 
 // dpkgLicenseAnalyzer parses copyright files and detect licenses
@@ -52,7 +53,7 @@ func (a dpkgLicenseAnalyzer) Analyze(_ context.Context, input analyzer.AnalysisI
 	}
 
 	findings := lo.Map(licenses, func(license string, _ int) types.LicenseFinding {
-		return types.LicenseFinding{Name: licensing.Normalize(license)}
+		return types.LicenseFinding{Name: license}
 	})
 
 	// e.g. "usr/share/doc/zlib1g/copyright" => "zlib1g"
@@ -82,14 +83,29 @@ func (a dpkgLicenseAnalyzer) parseCopyright(r dio.ReadSeekerAt) ([]string, error
 			// Machine-readable format
 			// cf. https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/#:~:text=The%20debian%2Fcopyright%20file%20must,in%20the%20Debian%20Policy%20Manual.
 			l := strings.TrimSpace(line[8:])
-			if len(l) > 0 && !slices.Contains(licenses, l) {
+			if len(l) > 0 {
-				licenses = append(licenses, l)
+				// Split licenses without considering "and"/"or"
+				// examples:
+				// 'GPL-1+,GPL-2' => {"GPL-1", "GPL-2"}
+				// 'GPL-1+ or Artistic or Artistic-dist' => {"GPL-1", "Artistic", "Artistic-dist"}
+				// 'LGPLv3+_or_GPLv2+' => {"LGPLv3", "GPLv2"}
+				// 'BSD-3-CLAUSE and GPL-2' => {"BSD-3-CLAUSE", "GPL-2"}
+				// 'GPL-1+ or Artistic, and BSD-4-clause-POWERDOG' => {"GPL-1+", "Artistic", "BSD-4-clause-POWERDOG"}
+				for _, lic := range licenseSplitRegexp.Split(l, -1) {
+					lic = licensing.Normalize(lic)
+					if !slices.Contains(licenses, lic) {
+						licenses = append(licenses, lic)
+					}
+				}
 			}
 		case strings.Contains(line, "/usr/share/common-licenses/"):
 			// Common license pattern
 			license := commonLicenseReferenceRegexp.FindStringSubmatch(line)
-			if len(license) == 2 && !slices.Contains(licenses, license[1]) {
+			if len(license) == 2 {
-				licenses = append(licenses, license[1])
+				l := licensing.Normalize(license[1])
+				if !slices.Contains(licenses, l) {
+					licenses = append(licenses, l)
+				}
 			}
 		}
 	}

pkg/fanal/analyzer/pkg/dpkg/copyright_test.go

@@ -29,6 +29,9 @@ func Test_dpkgLicenseAnalyzer_Analyze(t *testing.T) {
 						Type:     types.LicenseTypeDpkg,
 						FilePath: "usr/share/doc/zlib1g/copyright",
 						Findings: []types.LicenseFinding{
+							{Name: "GPL-1.0"},
+							{Name: "Artistic"},
+							{Name: "BSD-4-clause-POWERDOG"},
 							{Name: "Zlib"},
 						},
 						PkgName: "zlib1g",
@@ -64,7 +67,6 @@ func Test_dpkgLicenseAnalyzer_Analyze(t *testing.T) {
 						FilePath: "usr/share/doc/apt/copyright",
 						Findings: []types.LicenseFinding{
 							{Name: "GPL-2.0"},
-							{Name: "GPL-2.0"},
 						},
 						PkgName: "apt",
 					},

pkg/fanal/analyzer/pkg/dpkg/testdata/license-pattern-and-classifier-copyright

@@ -37,7 +37,7 @@ Files-Excluded:
 
 Files: *
 Copyright: 1995-2013 Jean-loup Gailly and Mark Adler
-License: Zlib
+License: GPL-1+ or Artistic, and BSD-4-clause-POWERDOG
 
 Files: amiga/Makefile.pup
 Copyright: 1998 by Andreas R. Kleinert

pkg/fanal/artifact/image/image_test.go

@@ -316,7 +316,6 @@ func TestArtifact_Inspect(t *testing.T) {
 									Type:     types.LicenseTypeDpkg,
 									FilePath: "usr/share/doc/ca-certificates/copyright",
 									Findings: []types.LicenseFinding{
-										{Name: "GPL-2.0"},
 										{Name: "GPL-2.0"},
 										{Name: "MPL-2.0"},
 									},

pkg/licensing/normalize.go

@@ -6,7 +6,11 @@ var mapping = map[string]string{
 	// GPL
 	"GPL-1":                          GPL10,
 	"GPL-1+":                         GPL10,
+	"GPL 1.0":                        GPL10,
+	"GPL 1":                          GPL10,
 	"GPL2":                           GPL20,
+	"GPL 2.0":                        GPL20,
+	"GPL 2":                          GPL20,
 	"GPL-2":                          GPL20,
 	"GPL-2.0-ONLY":                   GPL20,
 	"GPL2+":                          GPL20,
@@ -16,6 +20,8 @@ var mapping = map[string]string{
 	"GPL-2.0-OR-LATER":               GPL20,
 	"GPL-2+ WITH AUTOCONF EXCEPTION": GPL20withautoconfexception,
 	"GPL3":                           GPL30,
+	"GPL 3.0":                        GPL30,
+	"GPL 3":                          GPL30,
 	"GPLV3+":                         GPL30,
 	"GPL-3":                          GPL30,
 	"GPL-3.0-ONLY":                   GPL30,
@@ -27,22 +33,30 @@ var mapping = map[string]string{
 
 	// LGPL
 	"LGPL2":     LGPL20,
+	"LGPL 2":    LGPL20,
+	"LGPL 2.0":  LGPL20,
 	"LGPL-2":    LGPL20,
 	"LGPL2+":    LGPL20,
 	"LGPL-2+":   LGPL20,
 	"LGPL-2.0+": LGPL20,
 	"LGPL-2.1":  LGPL21,
+	"LGPL 2.1":  LGPL21,
 	"LGPL-2.1+": LGPL21,
 	"LGPLV2.1+": LGPL21,
 	"LGPL-3":    LGPL30,
+	"LGPL 3":    LGPL30,
 	"LGPL-3+":   LGPL30,
 	"LGPL":      LGPL30, // 2? 3?
 
 	// MPL
 	"MPL1.0":  MPL10,
 	"MPL1":    MPL10,
+	"MPL 1.0": MPL10,
+	"MPL 1":   MPL10,
 	"MPL2.0":  MPL20,
+	"MPL 2.0": MPL20,
 	"MPL2":    MPL20,
+	"MPL 2":   MPL20,
 
 	// BSD
 	"BSD":          BSD3Clause, // 2? 3?
@@ -51,8 +65,9 @@ var mapping = map[string]string{
 	"BSD-4-CLAUSE": BSD4Clause,
 
 	"APACHE":     Apache20, // 1? 2?
-	"ZLIB":   Zlib,
+	"APACHE 2.0": Apache20,
 	"RUBY":       Ruby,
+	"ZLIB":       Zlib,
 }
 
 func Normalize(name string) string {