Re-introduce linux kernel cataloger (#2526)

* re-add linux kernel cataloger

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* ensure there is at least a directory or image tag on each task

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* fix CLI tests to account for kernel finding (+2)

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

---------

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

internal/task/package_tasks.go

@@ -18,6 +18,7 @@ import (
 	"github.com/anchore/syft/syft/pkg/cataloger/haskell"
 	"github.com/anchore/syft/syft/pkg/cataloger/java"
 	"github.com/anchore/syft/syft/pkg/cataloger/javascript"
+	"github.com/anchore/syft/syft/pkg/cataloger/kernel"
 	"github.com/anchore/syft/syft/pkg/cataloger/nix"
 	"github.com/anchore/syft/syft/pkg/cataloger/php"
 	"github.com/anchore/syft/syft/pkg/cataloger/python"
@@ -112,9 +113,16 @@ func DefaultPackageTaskFactories() PackageTaskFactories {
 			func(cfg CatalogingFactoryConfig) pkg.Cataloger {
 				return binary.NewCataloger(cfg.PackagesConfig.Binary)
 			},
-			pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, pkgcataloging.InstalledTag, pkgcataloging.ImageTag, "binary"),
+			pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, pkgcataloging.InstalledTag, pkgcataloging.ImageTag, "binary",
+		),
 		newSimplePackageTaskFactory(githubactions.NewActionUsageCataloger, pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, "github", "github-actions"),
 		newSimplePackageTaskFactory(githubactions.NewWorkflowUsageCataloger, pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, "github", "github-actions"),
+		newPackageTaskFactory(
+			func(cfg CatalogingFactoryConfig) pkg.Cataloger {
+				return kernel.NewLinuxKernelCataloger(cfg.PackagesConfig.LinuxKernel)
+			},
+			pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, pkgcataloging.InstalledTag, pkgcataloging.ImageTag, "linux", "kernel",
+		),
 		newSimplePackageTaskFactory(sbomCataloger.NewCataloger, pkgcataloging.ImageTag, pkgcataloging.DeclaredTag, pkgcataloging.DirectoryTag, "sbom"), // note: not evidence of installed packages
 	}
 }

test/cli/scan_cmd_test.go

@@ -9,7 +9,7 @@ import (
 const (
 	// this is the number of packages that should be found in the image-pkg-coverage fixture image
 	// when analyzed with the squashed scope.
-	coverageImageSquashedPackageCount = 25
+	coverageImageSquashedPackageCount = 27
 )
 
 func TestPackagesCmdFlags(t *testing.T) {

test/cli/trait_assertions_test.go

@@ -122,6 +122,7 @@ func assertPackageCount(length uint) traitAssertion {
 		type NameAndVersion struct {
 			Name    string `json:"name"`
 			Version string `json:"version"`
+			Type    string `json:"type"`
 		}
 		type partial struct {
 			Artifacts []NameAndVersion `json:"artifacts"`
@@ -136,7 +137,7 @@ func assertPackageCount(length uint) traitAssertion {
 			tb.Errorf("expected package count of %d, but found %d", length, len(data.Artifacts))
 			debugArtifacts := make([]string, len(data.Artifacts))
 			for i, a := range data.Artifacts {
-				debugArtifacts[i] = fmt.Sprintf("%s:%s", a.Name, a.Version)
+				debugArtifacts[i] = fmt.Sprintf("%s@%s (%s)", a.Name, a.Version, a.Type)
 			}
 			sort.Strings(debugArtifacts)
 			for i, a := range debugArtifacts {

test/integration/package_catalogers_represented_test.go

@@ -0,0 +1,201 @@
+package integration
+
+import (
+	"bytes"
+	"go/ast"
+	"go/parser"
+	"go/token"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/scylladb/go-set/strset"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/anchore/syft/internal/task"
+	"github.com/anchore/syft/syft/cataloging/pkgcataloging"
+)
+
+func TestAllPackageCatalogersReachableInTasks(t *testing.T) {
+	// we want to see if we can get a task for all package catalogers. This is a bit tricky since we
+	// don't have a nice way to find all cataloger names in the codebase. Instead, we'll look at the
+	// count of unique task names from the package task factory set and compare that with the known constructors
+	// from a source analysis... they should match.
+
+	// additionally, at this time they should either have a "directory" or "image" tag as well. If there is no tag
+	// on a cataloger task then the test should fail.
+
+	taskFactories := task.DefaultPackageTaskFactories()
+	taskTagsByName := make(map[string][]string)
+	for _, factory := range taskFactories {
+		tsk := factory(task.DefaultCatalogingFactoryConfig())
+		if taskTagsByName[tsk.Name()] != nil {
+			t.Fatalf("duplicate task name: %q", tsk.Name())
+		}
+
+		require.NotNil(t, tsk)
+		if sel, ok := tsk.(task.Selector); ok {
+			taskTagsByName[tsk.Name()] = sel.Selectors()
+		} else {
+			taskTagsByName[tsk.Name()] = []string{}
+		}
+	}
+
+	var constructorCount int
+	constructorsPerPackage := getCatalogerConstructors(t)
+	for _, constructors := range constructorsPerPackage {
+		constructorCount += constructors.Size()
+	}
+
+	assert.Equal(t, len(taskTagsByName), constructorCount, "mismatch in number of cataloger constructors and task names")
+
+	for taskName, tags := range taskTagsByName {
+		if !strset.New(tags...).HasAny(pkgcataloging.ImageTag, pkgcataloging.DirectoryTag) {
+			t.Errorf("task %q is missing 'directory' or 'image' a tag", taskName)
+		}
+	}
+
+}
+
+func TestAllPackageCatalogersRepresentedInSource(t *testing.T) {
+	// find all functions in syft/pkg/cataloger/** that either:
+	// - match the name glob "New*Cataloger"
+	// - are in cataloger.go and match the name glob "New*"
+	//
+	// Then:
+	// - keep track of all packages with cataloger constructors
+	// - keep track of all constructors
+	constructorsPerPackage := getCatalogerConstructors(t)
+
+	// look at the source file in internal/task/package_tasks.go:
+	// - ensure all go packages that have constructors are imported
+	// - ensure there is a reference to all package constructors
+	assertAllPackageCatalogersRepresented(t, constructorsPerPackage)
+}
+
+func getCatalogerConstructors(t *testing.T) map[string]*strset.Set {
+	t.Helper()
+	root := repoRoot(t)
+	catalogerPath := filepath.Join(root, "syft", "pkg", "cataloger")
+
+	constructorsPerPackage := make(map[string]*strset.Set)
+
+	err := filepath.Walk(catalogerPath, func(path string, info os.FileInfo, err error) error {
+		require.NoError(t, err)
+
+		// ignore directories and test files...
+		if info.IsDir() || strings.HasSuffix(info.Name(), "_test.go") {
+			return nil
+		}
+
+		partialResults := getConstructorsFromExpectedFile(t, path, info)
+
+		constructorsPerPackage = mergeConstructors(constructorsPerPackage, partialResults)
+
+		partialResults = getCatalogerConstructorsFromPackage(t, path, info)
+
+		constructorsPerPackage = mergeConstructors(constructorsPerPackage, partialResults)
+
+		return nil
+	})
+
+	require.NoError(t, err)
+
+	// remove some exceptions
+	delete(constructorsPerPackage, "generic") // this is not an actual cataloger
+
+	return constructorsPerPackage
+}
+
+func getConstructorsFromExpectedFile(t *testing.T, path string, info os.FileInfo) map[string][]string {
+	constructorsPerPackage := make(map[string][]string)
+
+	if !strings.HasSuffix(info.Name(), "cataloger.go") && !strings.HasSuffix(info.Name(), "catalogers.go") {
+		return nil
+	}
+
+	fset := token.NewFileSet()
+	node, err := parser.ParseFile(fset, path, nil, parser.ParseComments)
+	require.NoError(t, err)
+
+	for _, f := range node.Decls {
+		fn, ok := f.(*ast.FuncDecl)
+		if !ok || fn.Recv != nil || !strings.HasPrefix(fn.Name.Name, "New") {
+			continue
+		}
+
+		pkg := node.Name.Name
+		constructorsPerPackage[pkg] = append(constructorsPerPackage[pkg], fn.Name.Name)
+	}
+
+	return constructorsPerPackage
+}
+
+func getCatalogerConstructorsFromPackage(t *testing.T, path string, info os.FileInfo) map[string][]string {
+	constructorsPerPackage := make(map[string][]string)
+
+	if info.IsDir() || !strings.HasSuffix(info.Name(), ".go") {
+		return nil
+	}
+
+	fset := token.NewFileSet()
+	node, err := parser.ParseFile(fset, path, nil, parser.ParseComments)
+	require.NoError(t, err)
+
+	for _, f := range node.Decls {
+		fn, ok := f.(*ast.FuncDecl)
+		if !ok || fn.Recv != nil || !strings.HasPrefix(fn.Name.Name, "New") || !strings.HasSuffix(fn.Name.Name, "Cataloger") {
+			continue
+		}
+
+		pkg := node.Name.Name
+		constructorsPerPackage[pkg] = append(constructorsPerPackage[pkg], fn.Name.Name)
+	}
+
+	return constructorsPerPackage
+}
+
+func assertAllPackageCatalogersRepresented(t *testing.T, constructorsPerPackage map[string]*strset.Set) {
+	t.Helper()
+
+	contents, err := os.ReadFile(filepath.Join(repoRoot(t), "internal", "task", "package_tasks.go"))
+	require.NoError(t, err)
+
+	// ensure all packages (keys) are represented in the package_tasks.go file
+	for pkg, constructors := range constructorsPerPackage {
+		if !assert.True(t, bytes.Contains(contents, []byte(pkg)), "missing package %q", pkg) {
+			continue
+		}
+		for _, constructor := range constructors.List() {
+			assert.True(t, bytes.Contains(contents, []byte(constructor)), "missing constructor %q for package %q", constructor, pkg)
+		}
+	}
+
+}
+
+func repoRoot(t testing.TB) string {
+	t.Helper()
+	root, err := exec.Command("git", "rev-parse", "--show-toplevel").Output()
+	if err != nil {
+		t.Fatalf("unable to find repo root dir: %+v", err)
+	}
+	absRepoRoot, err := filepath.Abs(strings.TrimSpace(string(root)))
+	if err != nil {
+		t.Fatal("unable to get abs path to repo root:", err)
+	}
+	return absRepoRoot
+}
+
+func mergeConstructors(constructorsPerPackage map[string]*strset.Set, partialResults map[string][]string) map[string]*strset.Set {
+	for pkg, constructors := range partialResults {
+		if _, ok := constructorsPerPackage[pkg]; !ok {
+			constructorsPerPackage[pkg] = strset.New()
+		}
+		constructorsPerPackage[pkg].Add(constructors...)
+	}
+
+	return constructorsPerPackage
+}