fix: golang PURL should include full module (#4395)

* fixed #4316 go mod with ver purl

Signed-off-by: Rez Moss <hi@rezmoss.com>

* go mod purl fixed, added func to handle go.mod

Signed-off-by: Rez Moss <hi@rezmoss.com>

* fix: use module name in PURL string everywhere

Signed-off-by: Keith Zantow <kzantow@gmail.com>

---------

Signed-off-by: Rez Moss <hi@rezmoss.com>
Signed-off-by: Keith Zantow <kzantow@gmail.com>
Co-authored-by: Keith Zantow <kzantow@gmail.com>

cmd/syft/internal/options/catalog.go

@@ -188,7 +188,8 @@ func (cfg Catalog) ToPackagesConfig() pkgcataloging.Config {
 					WithFromContents(cfg.Golang.MainModuleVersion.FromContents).
 					WithFromBuildSettings(cfg.Golang.MainModuleVersion.FromBuildSettings).
 					WithFromLDFlags(cfg.Golang.MainModuleVersion.FromLDFlags),
-			),
+			).
+			WithUsePackagesLib(*multiLevelOption(true, enrichmentEnabled(cfg.Enrich, task.Go, task.Golang), cfg.Golang.UsePackagesLib)),
 		JavaScript: javascript.DefaultCatalogerConfig().
 			WithIncludeDevDependencies(*multiLevelOption(false, cfg.JavaScript.IncludeDevDependencies)).
 			WithSearchRemoteLicenses(*multiLevelOption(false, enrichmentEnabled(cfg.Enrich, task.JavaScript, task.Node, task.NPM), cfg.JavaScript.SearchRemoteLicenses)).

cmd/syft/internal/options/golang.go

@@ -16,6 +16,7 @@ type golangConfig struct {
 	Proxy                       string                        `json:"proxy" yaml:"proxy" mapstructure:"proxy"`
 	NoProxy                     string                        `json:"no-proxy" yaml:"no-proxy" mapstructure:"no-proxy"`
 	MainModuleVersion           golangMainModuleVersionConfig `json:"main-module-version" yaml:"main-module-version" mapstructure:"main-module-version"`
+	UsePackagesLib              *bool                         `json:"use-packages-lib" yaml:"use-packages-lib" mapstructure:"use-packages-lib"`
 }
 
 var _ interface {
@@ -37,6 +38,7 @@ if unset this defaults to $GONOPROXY`)
 	descriptions.Add(&o.MainModuleVersion, `the go main module version discovered from binaries built with the go compiler will
 always show (devel) as the version. Use these options to control heuristics to guess
 a more accurate version from the binary.`)
+	descriptions.Add(&o.UsePackagesLib, `use the golang.org/x/tools/go/packages library, which executes golang tooling found on the path in addition to potential network access to get the most accurate results`)
 	descriptions.Add(&o.MainModuleVersion.FromLDFlags, `look for LD flags that appear to be setting a version (e.g. -X main.version=1.0.0)`)
 	descriptions.Add(&o.MainModuleVersion.FromBuildSettings, `use the build settings (e.g. vcs.version & vcs.time) to craft a v0 pseudo version 
 (e.g. v0.0.0-20220308212642-53e6d0aaf6fb) when a more accurate version cannot be found otherwise`)
@@ -64,5 +66,6 @@ func defaultGolangConfig() golangConfig {
 			FromContents:      def.MainModuleVersion.FromContents,
 			FromBuildSettings: def.MainModuleVersion.FromBuildSettings,
 		},
+		UsePackagesLib: nil, // this defaults to true, which is the API default
 	}
 }

syft/pkg/cataloger/golang/config.go

@@ -48,6 +48,9 @@ type CatalogerConfig struct {
 	NoProxy []string `yaml:"no-proxy,omitempty" json:"no-proxy,omitempty" mapstructure:"no-proxy"`
 
 	MainModuleVersion MainModuleVersionConfig `yaml:"main-module-version" json:"main-module-version" mapstructure:"main-module-version"`
+
+	// Whether to use the golang.org/x/tools/go/packages, which executes golang tooling found on the path in addition to potential network access
+	UsePackagesLib bool `json:"use-packages-lib" yaml:"use-packages-lib" mapstructure:"use-packages-lib"`
 }
 
 type MainModuleVersionConfig struct {
@@ -70,6 +73,7 @@ type MainModuleVersionConfig struct {
 // - setting the default local module cache dir if none is provided
 func DefaultCatalogerConfig() CatalogerConfig {
 	g := CatalogerConfig{
+		UsePackagesLib:    true,
 		MainModuleVersion: DefaultMainModuleVersionConfig(),
 		LocalModCacheDir:  defaultGoModDir(),
 	}
@@ -180,6 +184,11 @@ func (g CatalogerConfig) WithMainModuleVersion(input MainModuleVersionConfig) Ca
 	return g
 }
 
+func (g CatalogerConfig) WithUsePackagesLib(useLib bool) CatalogerConfig {
+	g.UsePackagesLib = useLib
+	return g
+}
+
 func (g MainModuleVersionConfig) WithFromLDFlags(input bool) MainModuleVersionConfig {
 	g.FromLDFlags = input
 	return g

syft/pkg/cataloger/golang/config_test.go

@@ -57,6 +57,7 @@ func Test_Config(t *testing.T) {
 				Proxies:                     []string{"https://my.proxy"},
 				NoProxy:                     []string{"my.private", "no.proxy"},
 				MainModuleVersion:           DefaultMainModuleVersionConfig(),
+				UsePackagesLib:              true,
 			},
 		},
 		{
@@ -84,6 +85,7 @@ func Test_Config(t *testing.T) {
 				Proxies:                     []string{"https://alt.proxy", "direct"},
 				NoProxy:                     []string{"alt.no.proxy"},
 				MainModuleVersion:           DefaultMainModuleVersionConfig(),
+				UsePackagesLib:              true,
 			},
 		},
 	}

syft/pkg/cataloger/golang/package.go

@@ -64,27 +64,23 @@ func newBinaryMetadata(dep *debug.Module, mainModule, goVersion, architecture st
 func packageURL(moduleName, moduleVersion string) string {
 	// source: https://github.com/package-url/purl-spec/blob/master/PURL-TYPES.rst#golang
 	// note: "The version is often empty when a commit is not specified and should be the commit in most cases when available."
-
+	if moduleName == "" {
-	fields := strings.Split(moduleName, "/")
-	if len(fields) == 0 {
 		return ""
 	}
 
 	namespace := ""
-	name := ""
+	name := moduleName
-	// The subpath is used to point to a subpath inside a package (e.g. pkg:golang/google.golang.org/genproto#googleapis/api/annotations)
-	subpath := ""
 
-	switch len(fields) {
+	// golang PURLs from _modules_ are constructed as pkg:golang/<module>@version, where
-	case 1:
+	// the full module name often includes multiple segments including `/v#`-type versions, for example:
-		name = fields[0]
+	//  pkg:golang/github.com/cli/cli/v2@2.63.0
-	case 2:
+	// see: https://github.com/package-url/purl-spec/issues/63
-		name = fields[1]
+	// and: https://github.com/package-url/purl-spec/blob/main/types-doc/golang-definition.md#subpath-definition
-		namespace = fields[0]
+	// by setting the namespace this way, it does not escape the slashes:
-	default:
+	lastSlash := strings.LastIndex(moduleName, "/")
-		name = fields[2]
+	if lastSlash > 0 && lastSlash < len(moduleName)-1 {
-		namespace = strings.Join(fields[0:2], "/")
+		name = moduleName[lastSlash+1:]
-		subpath = strings.Join(fields[3:], "/")
+		namespace = moduleName[0:lastSlash]
 	}
 
 	return packageurl.NewPackageURL(
@@ -93,6 +89,6 @@ func packageURL(moduleName, moduleVersion string) string {
 		name,
 		moduleVersion,
 		nil,
-		subpath,
+		"", // subpath is used to reference a specific _package_ within the module
 	).ToString()
 }

syft/pkg/cataloger/golang/package_test.go

@@ -38,14 +38,14 @@ func Test_packageURL(t *testing.T) {
 				Name:    "github.com/coreos/go-systemd/v22",
 				Version: "v22.1.0",
 			},
-			expected: "pkg:golang/github.com/coreos/go-systemd@v22.1.0#v22",
+			expected: "pkg:golang/github.com/coreos/go-systemd/v22@v22.1.0",
 		},
 		{
 			name: "golang with subpath deep",
 			pkg: pkg.Package{
 				Name: "google.golang.org/genproto/googleapis/api/annotations",
 			},
-			expected: "pkg:golang/google.golang.org/genproto/googleapis#api/annotations",
+			expected: "pkg:golang/google.golang.org/genproto/googleapis/api/annotations",
 		},
 	}
 

syft/pkg/cataloger/golang/parse_go_binary_test.go

@@ -278,7 +278,7 @@ func TestBuildGoPkgInfo(t *testing.T) {
 				{
 					Name:     "github.com/a/b/c",
 					Version:  "", // this was (devel) but we cleared it explicitly
-					PURL:     "pkg:golang/github.com/a/b#c",
+					PURL:     "pkg:golang/github.com/a/b/c",
 					Language: pkg.Go,
 					Type:     pkg.GoModulePkg,
 					Locations: file.NewLocationSet(

syft/pkg/cataloger/golang/parse_go_mod.go

@@ -8,7 +8,6 @@ import (
 	"io"
 	"path/filepath"
 	"slices"
-	"sort"
 	"strings"
 
 	"github.com/spf13/afero"
@@ -26,17 +25,19 @@ import (
 )
 
 type goModCataloger struct {
+	usePackagesLib  bool
 	licenseResolver goLicenseResolver
 }
 
 func newGoModCataloger(opts CatalogerConfig) *goModCataloger {
 	return &goModCataloger{
+		usePackagesLib:  opts.UsePackagesLib,
 		licenseResolver: newGoLicenseResolver(modFileCatalogerName, opts),
 	}
 }
 
 // parseGoModFile takes a go.mod and tries to resolve and lists all packages discovered.
-func (c *goModCataloger) parseGoModFile(ctx context.Context, resolver file.Resolver, _ *generic.Environment, reader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+func (c *goModCataloger) parseGoModFile(ctx context.Context, resolver file.Resolver, _ *generic.Environment, reader file.LocationReadCloser) (pkgs []pkg.Package, relationships []artifact.Relationship, err error) {
 	modDir := filepath.Dir(string(reader.Location.Reference().RealPath))
 	digests, err := parseGoSumFile(resolver, reader)
 	if err != nil {
@@ -48,24 +49,34 @@ func (c *goModCataloger) parseGoModFile(ctx context.Context, resolver file.Resol
 		scanRoot = dir.Chroot.Base()
 	}
 
-	// source analysis using go toolchain if available
-	syftSourcePackages, sourceModules, sourceDependencies, unknownErr := c.loadPackages(modDir, reader.Location)
-	catalogedModules, sourceModuleToPkg := c.catalogModules(ctx, scanRoot, syftSourcePackages, sourceModules, reader, digests)
-	relationships := buildModuleRelationships(catalogedModules, sourceDependencies, sourceModuleToPkg)
-
 	// base case go.mod file parsing
 	modFile, err := c.parseModFileContents(reader)
 	if err != nil {
 		return nil, nil, err
 	}
 
-	// only use mod packages NOT found in source analysis
+	// source analysis using go toolchain if available
+	var sourceModules map[string]*packages.Module
+	var catalogedModules []pkg.Package
+
+	if c.usePackagesLib {
+		var sourcePackages map[string][]pkgInfo
+		var sourceDependencies map[string][]string
+		var sourceModuleToPkg map[string]artifact.Identifiable
+
+		sourcePackages, sourceModules, sourceDependencies, err = c.loadPackages(modDir, reader.Location)
+		catalogedModules, sourceModuleToPkg = c.catalogModules(ctx, scanRoot, sourcePackages, sourceModules, reader, digests)
+		relationships = buildModuleRelationships(catalogedModules, sourceDependencies, sourceModuleToPkg)
+	}
+
+	// only use go.mod packages NOT found in source analysis
 	goModPackages := c.createGoModPackages(ctx, resolver, modFile, sourceModules, reader, digests)
 	c.applyReplaceDirectives(ctx, resolver, modFile, goModPackages, reader, digests)
 	c.applyExcludeDirectives(modFile, goModPackages)
 
-	finalPkgs := c.assembleResults(catalogedModules, goModPackages)
+	pkgs = c.assembleResults(catalogedModules, goModPackages)
-	return finalPkgs, relationships, unknownErr
+
+	return pkgs, relationships, err
 }
 
 // loadPackages uses golang.org/x/tools/go/packages to get dependency information.
@@ -327,7 +338,7 @@ func (c *goModCataloger) createGoModPackages(ctx context.Context, resolver file.
 	goModPackages := make(map[string]pkg.Package)
 
 	for _, m := range modFile.Require {
-		if _, exists := sourceModules[m.Mod.Path]; !exists {
+		if sourceModules == nil || sourceModules[m.Mod.Path] == nil {
 			lics := c.licenseResolver.getLicenses(ctx, resolver, m.Mod.Path, m.Mod.Version)
 			goModPkg := pkg.Package{
 				Name:      m.Mod.Path,
@@ -392,9 +403,7 @@ func (c *goModCataloger) assembleResults(catalogedPkgs []pkg.Package, goModPacka
 		pkgsSlice = append(pkgsSlice, p)
 	}
 
-	sort.SliceStable(pkgsSlice, func(i, j int) bool {
+	pkg.Sort(pkgsSlice)
-		return pkgsSlice[i].Name < pkgsSlice[j].Name
-	})
 
 	return pkgsSlice
 }

syft/pkg/cataloger/golang/parse_go_mod_test.go

@@ -54,7 +54,7 @@ func TestParseGoMod(t *testing.T) {
 				{
 					Name:      "github.com/anchore/archiver/v3",
 					Version:   "v3.5.2",
-					PURL:      "pkg:golang/github.com/anchore/archiver@v3.5.2#v3",
+					PURL:      "pkg:golang/github.com/anchore/archiver/v3@v3.5.2",
 					Locations: file.NewLocationSet(file.NewLocation("test-fixtures/go-mod-fixtures/many-packages/go.mod")),
 					Language:  pkg.Go,
 					Type:      pkg.GoModulePkg,
@@ -111,7 +111,7 @@ func TestParseGoMod(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.fixture, func(t *testing.T) {
-			c := newGoModCataloger(DefaultCatalogerConfig())
+			c := newGoModCataloger(DefaultCatalogerConfig().WithUsePackagesLib(false))
 			pkgtest.NewCatalogTester().
 				FromFile(t, test.fixture).
 				Expects(test.expected, nil).
@@ -172,7 +172,9 @@ func Test_GoSumHashes(t *testing.T) {
 			pkgtest.NewCatalogTester().
 				FromDirectory(t, test.fixture).
 				Expects(test.expected, nil).
-				TestCataloger(t, NewGoModuleFileCataloger(CatalogerConfig{}))
+				TestCataloger(t, NewGoModuleFileCataloger(CatalogerConfig{
+					UsePackagesLib: false,
+				}))
 		})
 	}
 }
@@ -189,7 +191,6 @@ func Test_parseGoSource_packageResolution(t *testing.T) {
 	tests := []struct {
 		name             string
 		fixturePath      string
-		config           CatalogerConfig
 		expectedPkgs     []string
 		expectedRels     []string
 		expectedLicenses map[string][]string
@@ -333,7 +334,7 @@ func Test_parseGoSource_packageResolution(t *testing.T) {
 						t.Errorf("mismatch in licenses (-want +got):\n%s", diff)
 					}
 				}).
-				TestCataloger(t, NewGoModuleFileCataloger(CatalogerConfig{}))
+				TestCataloger(t, NewGoModuleFileCataloger(DefaultCatalogerConfig().WithUsePackagesLib(true)))
 		})
 	}
 }