Merge branch 'main' into 1577-license-revamp

* main:
  rename pkg.Catalog to pkg.Collection (#1764)

Signed-off-by: Christopher Phillips <christopher.phillips@anchore.com>

syft/formats/common/cyclonedxhelpers/decoder.go

@@ -54,7 +54,7 @@ func ToSyftModel(bom *cyclonedx.BOM) (*sbom.SBOM, error) {
 
 	s := &sbom.SBOM{
 		Artifacts: sbom.Artifacts{
-			PackageCatalog:    pkg.NewCatalog(),
+			PackageCatalog:    pkg.NewCollection(),
 			LinuxDistribution: linuxReleaseFromComponents(*bom.Components),
 		},
 		Source:     extractComponents(bom.Metadata),

syft/formats/common/spdxhelpers/to_format_model.go

@@ -141,10 +141,10 @@ func toSPDXID(identifiable artifact.Identifiable) spdx.ElementID {
 	return spdx.ElementID(id)
 }
 
-// packages populates all Package Information from the package Catalog (see https://spdx.github.io/spdx-spec/3-package-information/)
+// packages populates all Package Information from the package Collection (see https://spdx.github.io/spdx-spec/3-package-information/)
 //
 //nolint:funlen
-func toPackages(catalog *pkg.Catalog, sbom sbom.SBOM) (results []*spdx.Package) {
+func toPackages(catalog *pkg.Collection, sbom sbom.SBOM) (results []*spdx.Package) {
 	for _, p := range catalog.Sorted() {
 		// name should be guaranteed to be unique, but semantically useful and stable
 		id := toSPDXID(p)
@@ -519,7 +519,7 @@ func toFileTypes(metadata *source.FileMetadata) (ty []string) {
 
 // other licenses are for licenses from the pkg.Package that do not have an SPDXExpression
 // field. The spdxexpression field is only filled given a validated Value field.
-func toOtherLicenses(catalog *pkg.Catalog) []*spdx.OtherLicense {
+func toOtherLicenses(catalog *pkg.Collection) []*spdx.OtherLicense {
 	licenses := map[string]bool{}
 	for _, p := range catalog.Sorted() {
 		d, c := parseLicenses(p.Licenses)

syft/formats/common/spdxhelpers/to_format_model_test.go

@@ -422,7 +422,7 @@ func Test_H1Digest(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {
-			catalog := pkg.NewCatalog(test.pkg)
+			catalog := pkg.NewCollection(test.pkg)
 			pkgs := toPackages(catalog, s)
 			require.Len(t, pkgs, 1)
 			for _, p := range pkgs {
@@ -498,7 +498,7 @@ func Test_OtherLicenses(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {
-			catalog := pkg.NewCatalog(test.pkg)
+			catalog := pkg.NewCollection(test.pkg)
 			otherLicenses := toOtherLicenses(catalog)
 			require.Len(t, otherLicenses, len(test.expected))
 			require.Equal(t, test.expected, otherLicenses)

syft/formats/common/spdxhelpers/to_syft_model.go

@@ -34,7 +34,7 @@ func ToSyftModel(doc *spdx.Document) (*sbom.SBOM, error) {
 	s := &sbom.SBOM{
 		Source: src,
 		Artifacts: sbom.Artifacts{
-			PackageCatalog:    pkg.NewCatalog(),
+			PackageCatalog:    pkg.NewCollection(),
 			FileMetadata:      map[source.Coordinates]source.FileMetadata{},
 			FileDigests:       map[source.Coordinates][]file.Digest{},
 			LinuxDistribution: findLinuxReleaseByPURL(doc),

syft/formats/github/encoder_test.go

@@ -28,7 +28,7 @@ func Test_toGithubModel(t *testing.T) {
 				VersionID: "18.04",
 				IDLike:    []string{"debian"},
 			},
-			PackageCatalog: pkg.NewCatalog(),
+			PackageCatalog: pkg.NewCollection(),
 		},
 	}
 	for _, p := range []pkg.Package{

syft/formats/internal/testutils/utils.go

@@ -94,7 +94,7 @@ func AssertEncoderAgainstGoldenSnapshot(t *testing.T, format sbom.Format, sbom s
 
 func ImageInput(t testing.TB, testImage string, options ...ImageOption) sbom.SBOM {
 	t.Helper()
-	catalog := pkg.NewCatalog()
+	catalog := pkg.NewCollection()
 	var cfg imageCfg
 	var img *image.Image
 	for _, opt := range options {
@@ -146,7 +146,7 @@ func carriageRedactor(s []byte) []byte {
 	return []byte(msg)
 }
 
-func populateImageCatalog(catalog *pkg.Catalog, img *image.Image) {
+func populateImageCatalog(catalog *pkg.Collection, img *image.Image) {
 	_, ref1, _ := img.SquashedTree().File("/somefile-1.txt", filetree.FollowBasenameLinks)
 	_, ref2, _ := img.SquashedTree().File("/somefile-2.txt", filetree.FollowBasenameLinks)
 
@@ -260,8 +260,8 @@ func DirectoryInputWithAuthorField(t testing.TB) sbom.SBOM {
 	}
 }
 
-func newDirectoryCatalog() *pkg.Catalog {
-	catalog := pkg.NewCatalog()
+func newDirectoryCatalog() *pkg.Collection {
+	catalog := pkg.NewCollection()
 
 	// populate catalog with test data
 	catalog.Add(pkg.Package{
@@ -318,8 +318,8 @@ func newDirectoryCatalog() *pkg.Catalog {
 	return catalog
 }
 
-func newDirectoryCatalogWithAuthorField() *pkg.Catalog {
-	catalog := pkg.NewCatalog()
+func newDirectoryCatalogWithAuthorField() *pkg.Collection {
+	catalog := pkg.NewCollection()
 
 	// populate catalog with test data
 	catalog.Add(pkg.Package{

syft/formats/spdxtagvalue/encoder_test.go

@@ -49,7 +49,7 @@ func TestSPDXJSONSPDXIDs(t *testing.T) {
 		Format(),
 		sbom.SBOM{
 			Artifacts: sbom.Artifacts{
-				PackageCatalog: pkg.NewCatalog(pkgs...),
+				PackageCatalog: pkg.NewCollection(pkgs...),
 			},
 			Relationships: nil,
 			Source: source.Metadata{

syft/formats/syftjson/encoder_test.go

@@ -48,7 +48,7 @@ func schemaVersionRedactor(s []byte) []byte {
 }
 
 func TestEncodeFullJSONDocument(t *testing.T) {
-	catalog := pkg.NewCatalog()
+	catalog := pkg.NewCollection()
 
 	p1 := pkg.Package{
 		Name:    "package-1",

syft/formats/syftjson/to_format_model.go

@@ -173,7 +173,7 @@ func toFileType(ty stereoscopeFile.Type) string {
 	}
 }
 
-func toPackageModels(catalog *pkg.Catalog) []model.Package {
+func toPackageModels(catalog *pkg.Collection) []model.Package {
 	artifacts := make([]model.Package, 0)
 	if catalog == nil {
 		return artifacts

syft/formats/syftjson/to_syft_model.go

@@ -131,7 +131,7 @@ func toSyftLinuxRelease(d model.LinuxRelease) *linux.Release {
 	}
 }
 
-func toSyftRelationships(doc *model.Document, catalog *pkg.Catalog, relationships []model.Relationship, idAliases map[string]string) []artifact.Relationship {
+func toSyftRelationships(doc *model.Document, catalog *pkg.Collection, relationships []model.Relationship, idAliases map[string]string) []artifact.Relationship {
 	idMap := make(map[string]interface{})
 
 	for _, p := range catalog.Sorted() {
@@ -256,8 +256,8 @@ func toSyftSourceData(s model.Source) *source.Metadata {
 	return nil
 }
 
-func toSyftCatalog(pkgs []model.Package, idAliases map[string]string) *pkg.Catalog {
-	catalog := pkg.NewCatalog()
+func toSyftCatalog(pkgs []model.Package, idAliases map[string]string) *pkg.Collection {
+	catalog := pkg.NewCollection()
 	for _, p := range pkgs {
 		catalog.Add(toSyftPackage(p, idAliases))
 	}

syft/lib.go

@@ -34,7 +34,7 @@ import (
 // CatalogPackages takes an inventory of packages from the given image from a particular perspective
 // (e.g. squashed source, all-layers source). Returns the discovered  set of packages, the identified Linux
 // distribution, and the source object used to wrap the data source.
-func CatalogPackages(src *source.Source, cfg cataloger.Config) (*pkg.Catalog, []artifact.Relationship, *linux.Release, error) {
+func CatalogPackages(src *source.Source, cfg cataloger.Config) (*pkg.Collection, []artifact.Relationship, *linux.Release, error) {
 	resolver, err := src.FileResolver(cfg.Search.Scope)
 	if err != nil {
 		return nil, nil, nil, fmt.Errorf("unable to determine resolver while cataloging packages: %w", err)
@@ -76,7 +76,7 @@ func CatalogPackages(src *source.Source, cfg cataloger.Config) (*pkg.Catalog, []
 	return catalog, relationships, release, err
 }
 
-func newSourceRelationshipsFromCatalog(src *source.Source, c *pkg.Catalog) []artifact.Relationship {
+func newSourceRelationshipsFromCatalog(src *source.Source, c *pkg.Collection) []artifact.Relationship {
 	relationships := make([]artifact.Relationship, 0) // Should we pre-allocate this by giving catalog a Len() method?
 	for p := range c.Enumerate() {
 		relationships = append(relationships, artifact.Relationship{

syft/pkg/catalog.go

@@ -10,43 +10,18 @@ import (
 	"github.com/anchore/syft/syft/artifact"
 )
 
-type orderedIDSet struct {
-	slice []artifact.ID
-}
-
-func (s *orderedIDSet) add(ids ...artifact.ID) {
-loopNewIDs:
-	for _, newID := range ids {
-		for _, existingID := range s.slice {
-			if existingID == newID {
-				continue loopNewIDs
-			}
-		}
-		s.slice = append(s.slice, newID)
-	}
-}
-
-func (s *orderedIDSet) delete(id artifact.ID) {
-	for i, existingID := range s.slice {
-		if existingID == id {
-			s.slice = append(s.slice[:i], s.slice[i+1:]...)
-			return
-		}
-	}
-}
-
-// Catalog represents a collection of Packages.
-type Catalog struct {
+// Collection represents a collection of Packages.
+type Collection struct {
 	byID      map[artifact.ID]Package
 	idsByName map[string]orderedIDSet
 	idsByType map[Type]orderedIDSet
 	idsByPath map[string]orderedIDSet // note: this is real path or virtual path
 	lock      sync.RWMutex
 }
 
-// NewCatalog returns a new empty Catalog
-func NewCatalog(pkgs ...Package) *Catalog {
-	catalog := Catalog{
+// NewCollection returns a new empty Collection
+func NewCollection(pkgs ...Package) *Collection {
+	catalog := Collection{
 		byID:      make(map[artifact.ID]Package),
 		idsByName: make(map[string]orderedIDSet),
 		idsByType: make(map[Type]orderedIDSet),
@@ -61,12 +36,12 @@ func NewCatalog(pkgs ...Package) *Catalog {
 }
 
 // PackageCount returns the total number of packages that have been added.
-func (c *Catalog) PackageCount() int {
+func (c *Collection) PackageCount() int {
 	return len(c.byID)
 }
 
 // Package returns the package with the given ID.
-func (c *Catalog) Package(id artifact.ID) *Package {
+func (c *Collection) Package(id artifact.ID) *Package {
 	v, exists := c.byID[id]
 	if !exists {
 		return nil
@@ -81,17 +56,17 @@ func (c *Catalog) Package(id artifact.ID) *Package {
 }
 
 // PackagesByPath returns all packages that were discovered from the given path.
-func (c *Catalog) PackagesByPath(path string) []Package {
+func (c *Collection) PackagesByPath(path string) []Package {
 	return c.Packages(c.idsByPath[path].slice)
 }
 
 // PackagesByName returns all packages that were discovered with a matching name.
-func (c *Catalog) PackagesByName(name string) []Package {
+func (c *Collection) PackagesByName(name string) []Package {
 	return c.Packages(c.idsByName[name].slice)
 }
 
 // Packages returns all packages for the given ID.
-func (c *Catalog) Packages(ids []artifact.ID) (result []Package) {
+func (c *Collection) Packages(ids []artifact.ID) (result []Package) {
 	for _, i := range ids {
 		p, exists := c.byID[i]
 		if exists {
@@ -102,7 +77,7 @@ func (c *Catalog) Packages(ids []artifact.ID) (result []Package) {
 }
 
 // Add n packages to the catalog.
-func (c *Catalog) Add(pkgs ...Package) {
+func (c *Collection) Add(pkgs ...Package) {
 	c.lock.Lock()
 	defer c.lock.Unlock()
 
@@ -129,26 +104,26 @@ func (c *Catalog) Add(pkgs ...Package) {
 	}
 }
 
-func (c *Catalog) addToIndex(p Package) {
+func (c *Collection) addToIndex(p Package) {
 	c.byID[p.id] = p
 	c.addNameToIndex(p)
 	c.addTypeToIndex(p)
 	c.addPathsToIndex(p)
 }
 
-func (c *Catalog) addNameToIndex(p Package) {
+func (c *Collection) addNameToIndex(p Package) {
 	nameIndex := c.idsByName[p.Name]
 	nameIndex.add(p.id)
 	c.idsByName[p.Name] = nameIndex
 }
 
-func (c *Catalog) addTypeToIndex(p Package) {
+func (c *Collection) addTypeToIndex(p Package) {
 	typeIndex := c.idsByType[p.Type]
 	typeIndex.add(p.id)
 	c.idsByType[p.Type] = typeIndex
 }
 
-func (c *Catalog) addPathsToIndex(p Package) {
+func (c *Collection) addPathsToIndex(p Package) {
 	observedPaths := internal.NewStringSet()
 	for _, l := range p.Locations.ToSlice() {
 		if l.RealPath != "" && !observedPaths.Contains(l.RealPath) {
@@ -162,13 +137,13 @@ func (c *Catalog) addPathsToIndex(p Package) {
 	}
 }
 
-func (c *Catalog) addPathToIndex(id artifact.ID, path string) {
+func (c *Collection) addPathToIndex(id artifact.ID, path string) {
 	pathIndex := c.idsByPath[path]
 	pathIndex.add(id)
 	c.idsByPath[path] = pathIndex
 }
 
-func (c *Catalog) Delete(ids ...artifact.ID) {
+func (c *Collection) Delete(ids ...artifact.ID) {
 	c.lock.Lock()
 	defer c.lock.Unlock()
 
@@ -185,19 +160,19 @@ func (c *Catalog) Delete(ids ...artifact.ID) {
 	}
 }
 
-func (c *Catalog) deleteNameFromIndex(p Package) {
+func (c *Collection) deleteNameFromIndex(p Package) {
 	nameIndex := c.idsByName[p.Name]
 	nameIndex.delete(p.id)
 	c.idsByName[p.Name] = nameIndex
 }
 
-func (c *Catalog) deleteTypeFromIndex(p Package) {
+func (c *Collection) deleteTypeFromIndex(p Package) {
 	typeIndex := c.idsByType[p.Type]
 	typeIndex.delete(p.id)
 	c.idsByType[p.Type] = typeIndex
 }
 
-func (c *Catalog) deletePathsFromIndex(p Package) {
+func (c *Collection) deletePathsFromIndex(p Package) {
 	observedPaths := internal.NewStringSet()
 	for _, l := range p.Locations.ToSlice() {
 		if l.RealPath != "" && !observedPaths.Contains(l.RealPath) {
@@ -211,7 +186,7 @@ func (c *Catalog) deletePathsFromIndex(p Package) {
 	}
 }
 
-func (c *Catalog) deletePathFromIndex(id artifact.ID, path string) {
+func (c *Collection) deletePathFromIndex(id artifact.ID, path string) {
 	pathIndex := c.idsByPath[path]
 	pathIndex.delete(id)
 	if len(pathIndex.slice) == 0 {
@@ -222,7 +197,7 @@ func (c *Catalog) deletePathFromIndex(id artifact.ID, path string) {
 }
 
 // Enumerate all packages for the given type(s), enumerating all packages if no type is specified.
-func (c *Catalog) Enumerate(types ...Type) <-chan Package {
+func (c *Collection) Enumerate(types ...Type) <-chan Package {
 	channel := make(chan Package)
 	go func() {
 		defer close(channel)
@@ -257,7 +232,7 @@ func (c *Catalog) Enumerate(types ...Type) <-chan Package {
 
 // Sorted enumerates all packages for the given types sorted by package name. Enumerates all packages if no type
 // is specified.
-func (c *Catalog) Sorted(types ...Type) (pkgs []Package) {
+func (c *Collection) Sorted(types ...Type) (pkgs []Package) {
 	for p := range c.Enumerate(types...) {
 		pkgs = append(pkgs, p)
 	}
@@ -266,3 +241,28 @@ func (c *Catalog) Sorted(types ...Type) (pkgs []Package) {
 
 	return pkgs
 }
+
+type orderedIDSet struct {
+	slice []artifact.ID
+}
+
+func (s *orderedIDSet) add(ids ...artifact.ID) {
+loopNewIDs:
+	for _, newID := range ids {
+		for _, existingID := range s.slice {
+			if existingID == newID {
+				continue loopNewIDs
+			}
+		}
+		s.slice = append(s.slice, newID)
+	}
+}
+
+func (s *orderedIDSet) delete(id artifact.ID) {
+	for i, existingID := range s.slice {
+		if existingID == id {
+			s.slice = append(s.slice[:i], s.slice[i+1:]...)
+			return
+		}
+	}
+}