feat: Parse pnpm v9 lockfiles (#4256)

Signed-off-by: bernardoamc <bernardo.amc@gmail.com>

syft/pkg/cataloger/javascript/parse_pnpm_lock.go

@@ -4,7 +4,7 @@ import (
 	"context"
 	"fmt"
 	"io"
-	"regexp"
+	"sort"
 	"strconv"
 	"strings"
 
@@ -21,92 +21,186 @@ import (
 // integrity check
 var _ generic.Parser = parsePnpmLock
 
-type pnpmLockYaml struct {
-	Version      string                 `json:"lockfileVersion" yaml:"lockfileVersion"`
-	Dependencies map[string]interface{} `json:"dependencies" yaml:"dependencies"`
-	Packages     map[string]interface{} `json:"packages" yaml:"packages"`
+// pnpmPackage holds the raw name and version extracted from the lockfile.
+type pnpmPackage struct {
+	Name    string
+	Version string
 }
 
+// pnpmLockfileParser defines the interface for parsing different versions of pnpm lockfiles.
+type pnpmLockfileParser interface {
+	Parse(version float64, data []byte) ([]pnpmPackage, error)
+}
+
+// pnpmV6LockYaml represents the structure of pnpm lockfiles for versions < 9.0.
+type pnpmV6LockYaml struct {
+	Dependencies map[string]interface{} `yaml:"dependencies"`
+	Packages     map[string]interface{} `yaml:"packages"`
+}
+
+// pnpmV9LockYaml represents the structure of pnpm lockfiles for versions >= 9.0.
+type pnpmV9LockYaml struct {
+	LockfileVersion string                 `yaml:"lockfileVersion"`
+	Importers       map[string]interface{} `yaml:"importers"` // Using interface{} for forward compatibility
+	Packages        map[string]interface{} `yaml:"packages"`
+}
+
+// Parse implements the pnpmLockfileParser interface for v6-v8 lockfiles.
+func (p *pnpmV6LockYaml) Parse(version float64, data []byte) ([]pnpmPackage, error) {
+	if err := yaml.Unmarshal(data, p); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal pnpm v6 lockfile: %w", err)
+	}
+
+	packages := make(map[string]pnpmPackage)
+
+	// Direct dependencies
+	for name, info := range p.Dependencies {
+		ver, err := parseVersionField(name, info)
+		if err != nil {
+			log.WithFields("package", name, "error", err).Trace("unable to parse pnpm dependency")
+			continue
+		}
+		key := name + "@" + ver
+		packages[key] = pnpmPackage{Name: name, Version: ver}
+	}
+
+	splitChar := "/"
+	if version >= 6.0 {
+		splitChar = "@"
+	}
+
+	// All transitive dependencies
+	for key := range p.Packages {
+		name, ver, ok := parsePnpmPackageKey(key, splitChar)
+		if !ok {
+			log.WithFields("key", key).Trace("unable to parse pnpm package key")
+			continue
+		}
+		pkgKey := name + "@" + ver
+		packages[pkgKey] = pnpmPackage{Name: name, Version: ver}
+	}
+
+	return toSortedSlice(packages), nil
+}
+
+// Parse implements the PnpmLockfileParser interface for v9+ lockfiles.
+func (p *pnpmV9LockYaml) Parse(_ float64, data []byte) ([]pnpmPackage, error) {
+	if err := yaml.Unmarshal(data, p); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal pnpm v9 lockfile: %w", err)
+	}
+
+	packages := make(map[string]pnpmPackage)
+
+	// In v9, all resolved dependencies are listed in the top-level "packages" field.
+	// The key format is like /<name>@<version> or /<name>@<version>(<peer-deps>).
+	for key := range p.Packages {
+		// The separator for name and version is consistently '@' in v9+ keys.
+		name, ver, ok := parsePnpmPackageKey(key, "@")
+		if !ok {
+			log.WithFields("key", key).Trace("unable to parse pnpm v9 package key")
+			continue
+		}
+		pkgKey := name + "@" + ver
+		packages[pkgKey] = pnpmPackage{Name: name, Version: ver}
+	}
+
+	return toSortedSlice(packages), nil
+}
+
+// newPnpmLockfileParser is a factory function that returns the correct parser for the given lockfile version.
+func newPnpmLockfileParser(version float64) pnpmLockfileParser {
+	if version >= 9.0 {
+		return &pnpmV9LockYaml{}
+	}
+	return &pnpmV6LockYaml{}
+}
+
+// parsePnpmLock is the main parser function for pnpm-lock.yaml files.
 func parsePnpmLock(ctx context.Context, resolver file.Resolver, _ *generic.Environment, reader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
-	bytes, err := io.ReadAll(reader)
+	data, err := io.ReadAll(reader)
 	if err != nil {
 		return nil, nil, fmt.Errorf("failed to load pnpm-lock.yaml file: %w", err)
 	}
 
-	var pkgs []pkg.Package
-	var lockFile pnpmLockYaml
+	var lockfile struct {
+		Version string `yaml:"lockfileVersion"`
+	}
+	if err := yaml.Unmarshal(data, &lockfile); err != nil {
+		return nil, nil, fmt.Errorf("failed to parse pnpm-lock.yaml version: %w", err)
+	}
 
-	if err := yaml.Unmarshal(bytes, &lockFile); err != nil {
+	version, err := strconv.ParseFloat(lockfile.Version, 64)
+	if err != nil {
+		return nil, nil, fmt.Errorf("invalid lockfile version %q: %w", lockfile.Version, err)
+	}
+
+	parser := newPnpmLockfileParser(version)
+	pnpmPkgs, err := parser.Parse(version, data)
+	if err != nil {
 		return nil, nil, fmt.Errorf("failed to parse pnpm-lock.yaml file: %w", err)
 	}
 
-	lockVersion, _ := strconv.ParseFloat(lockFile.Version, 64)
+	packages := make([]pkg.Package, len(pnpmPkgs))
+	for i, p := range pnpmPkgs {
+		packages[i] = newPnpmPackage(ctx, resolver, reader.Location, p.Name, p.Version)
+	}
 
-	for name, info := range lockFile.Dependencies {
-		version := ""
+	return packages, nil, unknown.IfEmptyf(packages, "unable to determine packages")
+}
 
-		switch info := info.(type) {
+// parseVersionField extracts the version string from a dependency entry.
+func parseVersionField(name string, info interface{}) (string, error) {
+	switch v := info.(type) {
 	case string:
-			version = info
+		return v, nil
 	case map[string]interface{}:
-			v, ok := info["version"]
-			if !ok {
-				break
-			}
-			ver, ok := v.(string)
-			if ok {
-				version = parseVersion(ver)
+		if ver, ok := v["version"].(string); ok {
+			// e.g., "1.2.3(react@17.0.0)" -> "1.2.3"
+			return strings.SplitN(ver, "(", 2)[0], nil
 		}
+		return "", fmt.Errorf("version field is not a string for %q", name)
 	default:
-			log.Tracef("unsupported pnpm dependency type: %+v", info)
-			continue
-		}
-
-		if hasPkg(pkgs, name, version) {
-			continue
-		}
-
-		pkgs = append(pkgs, newPnpmPackage(ctx, resolver, reader.Location, name, version))
-	}
-
-	packageNameRegex := regexp.MustCompile(`^/?([^(]*)(?:\(.*\))*$`)
-	splitChar := "/"
-	if lockVersion >= 6.0 {
-		splitChar = "@"
-	}
-
-	// parse packages from packages section of pnpm-lock.yaml
-	for nameVersion := range lockFile.Packages {
-		nameVersion = packageNameRegex.ReplaceAllString(nameVersion, "$1")
-		nameVersionSplit := strings.Split(strings.TrimPrefix(nameVersion, "/"), splitChar)
-
-		// last element in split array is version
-		version := nameVersionSplit[len(nameVersionSplit)-1]
-
-		// construct name from all array items other than last item (version)
-		name := strings.Join(nameVersionSplit[:len(nameVersionSplit)-1], splitChar)
-
-		if hasPkg(pkgs, name, version) {
-			continue
-		}
-
-		pkgs = append(pkgs, newPnpmPackage(ctx, resolver, reader.Location, name, version))
-	}
-
-	pkg.Sort(pkgs)
-
-	return pkgs, nil, unknown.IfEmptyf(pkgs, "unable to determine packages")
-}
-
-func hasPkg(pkgs []pkg.Package, name, version string) bool {
-	for _, p := range pkgs {
-		if p.Name == name && p.Version == version {
-			return true
+		return "", fmt.Errorf("unsupported dependency type %T for %q", info, name)
 	}
 }
-	return false
+
+// parsePnpmPackageKey extracts the package name and version from a lockfile package key.
+// Handles formats like:
+// - /@babel/runtime/7.16.7
+// - /@types/node@14.18.12
+// - /is-glob@4.0.3
+// - /@babel/helper-plugin-utils@7.24.7(@babel/core@7.24.7)
+func parsePnpmPackageKey(key, separator string) (name, version string, ok bool) {
+	// Strip peer dependency information, e.g., (...)
+	key = strings.SplitN(key, "(", 2)[0]
+
+	// Strip leading slash
+	key = strings.TrimPrefix(key, "/")
+
+	parts := strings.Split(key, separator)
+	if len(parts) < 2 {
+		return "", "", false
 	}
 
-func parseVersion(version string) string {
-	return strings.SplitN(version, "(", 2)[0]
+	version = parts[len(parts)-1]
+	name = strings.Join(parts[:len(parts)-1], separator)
+
+	return name, version, true
+}
+
+// toSortedSlice converts the map of packages to a sorted slice for deterministic output.
+func toSortedSlice(packages map[string]pnpmPackage) []pnpmPackage {
+	pkgs := make([]pnpmPackage, 0, len(packages))
+	for _, p := range packages {
+		pkgs = append(pkgs, p)
+	}
+
+	sort.Slice(pkgs, func(i, j int) bool {
+		if pkgs[i].Name == pkgs[j].Name {
+			return pkgs[i].Version < pkgs[j].Version
+		}
+		return pkgs[i].Name < pkgs[j].Name
+	})
+
+	return pkgs
 }

syft/pkg/cataloger/javascript/parse_pnpm_lock_test.go

@@ -145,6 +145,50 @@ func TestParsePnpmV6Lock(t *testing.T) {
 	pkgtest.TestFileParser(t, fixture, parsePnpmLock, expectedPkgs, expectedRelationships)
 }
 
+func TestParsePnpmLockV9(t *testing.T) {
+	var expectedRelationships []artifact.Relationship
+	fixture := "test-fixtures/pnpm-v9/pnpm-lock.yaml"
+	locationSet := file.NewLocationSet(file.NewLocation(fixture))
+
+	expected := []pkg.Package{
+		{
+			Name:      "@babel/core",
+			Version:   "7.24.7",
+			PURL:      "pkg:npm/%40babel/core@7.24.7",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "@babel/helper-plugin-utils",
+			Version:   "7.24.7",
+			PURL:      "pkg:npm/%40babel/helper-plugin-utils@7.24.7",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "is-positive",
+			Version:   "3.1.0",
+			PURL:      "pkg:npm/is-positive@3.1.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+		{
+			Name:      "rollup",
+			Version:   "4.18.0",
+			PURL:      "pkg:npm/rollup@4.18.0",
+			Locations: locationSet,
+			Language:  pkg.JavaScript,
+			Type:      pkg.NpmPkg,
+		},
+	}
+
+	// TODO: no relationships are under test
+	pkgtest.TestFileParser(t, fixture, parsePnpmLock, expected, expectedRelationships)
+}
+
 func Test_corruptPnpmLock(t *testing.T) {
 	pkgtest.NewCatalogTester().
 		FromFile(t, "test-fixtures/corrupt/pnpm-lock.yaml").

syft/pkg/cataloger/javascript/test-fixtures/pnpm-v9/pnpm-lock.yaml

@@ -0,0 +1,31 @@
+lockfileVersion: '9.0'
+
+importers:
+  .:
+    dependencies:
+      is-positive:
+        specifier: ^3.1.0
+        version: 3.1.0
+      rollup:
+        specifier: ^4.18.0
+        version: 4.18.0
+
+packages:
+  /@babel/core@7.24.7:
+    resolution: {integrity: sha512-4RjkiFFI42+268iBv2nC+iMLTJGQW3u9P7YvA3x/6MDrJ9IYZ8I/xx5a2GIhY5gBTOcI4iC5S5in2fGjE+P4Yw==}
+    dev: false
+
+  /@babel/helper-plugin-utils@7.24.7(@babel/core@7.24.7):
+    resolution: {integrity: sha512-8A2+zKm53/3w4rwbX11FMW/yFS6c5Vam02P/dw01aK6KbwkKqBaIt3eEATiKtn9I2uS1itk8/aZ2yZ/kURee4Q==}
+    peerDependencies:
+      '@babel/core': ^7.4.0 || ^8.0.0-0 <8.0.0
+    dev: false
+
+  /is-positive@3.1.0:
+    resolution: {integrity: sha512-9ffLCf_f5sopimAhg2g91a7b9Rw5A1aA9eI6S391S3VEzYw99I3iKjcZGxLp25s0cRxNBV5aL2mhn7421SSlA==}
+    dev: false
+
+  /rollup@4.18.0:
+    resolution: {integrity: sha512-QpQY2Q5i0y0Q3RoAvoChE/R5iN2k05N//bNvQbC2XvRjHFT1qWJ2r3n1bNqE+gGRJaeuQf0BxE42D7CyuLh3ZQ==}
+    hasBin: true
+    dev: false