refactor: implement package-manager-specific extractors & replace osv-scanner with scalibr

This commit is contained in:
Sahilb315
2025-06-11 19:01:44 +05:30
parent 8b8a98b13d
commit 6c446a7ad7
9 changed files with 258 additions and 204 deletions
+26 -77
View File
@@ -1,101 +1,50 @@
package extractor
import (
"fmt"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/google/osv-scanner/pkg/lockfile"
)
// EcosystemExtractor defines the interface for ecosystem-specific extractors
type EcosystemExtractor interface {
// PackageManagerExtractor defines the interface for package-manager-specific extractors
type PackageManagerExtractor interface {
// Returns the list of supported lockfiles by the package manager
GetSupportedFiles() []string
// Returns the package manager ecosysetm
GetEcosystem() packagev1.Ecosystem
Extract(depFile lockfile.DepFile, filename string) ([]lockfile.PackageDetails, error)
// Returns the package manager name
GetPackageManager() PackageManagerName
// Extracts the packages from the lockfile
Extract(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error)
}
// NpmExtractor handles NPM ecosystem lockfiles
type NpmExtractor struct{}
type PackageManagerName string
var NpmExtractors []string = []string{"package-lock.json", "pnpm-lock.yaml"}
const (
Npm PackageManagerName = "npm"
Pnpm PackageManagerName = "pnpm"
Pip PackageManagerName = "pip"
)
func (n *NpmExtractor) GetSupportedFiles() []string {
return NpmExtractors
}
func (n *NpmExtractor) GetEcosystem() packagev1.Ecosystem {
return packagev1.Ecosystem_ECOSYSTEM_NPM
}
func (n *NpmExtractor) Extract(depFile lockfile.DepFile, filename string) ([]lockfile.PackageDetails, error) {
switch filename {
case "package-lock.json":
extractor := lockfile.NpmLockExtractor{}
return extractor.Extract(depFile)
case "pnpm-lock.yaml":
extractor := lockfile.PnpmLockExtractor{}
return extractor.Extract(depFile)
default:
return nil, fmt.Errorf("unsupported NPM lockfile: %s", filename)
}
}
// PyPiExtractor handles PyPI ecosystem lockfiles
type PyPiExtractor struct{}
var PyPiExtractors []string = []string{"requirements.txt", "Pipfile.lock"}
func (p *PyPiExtractor) GetSupportedFiles() []string {
return PyPiExtractors
}
func (p *PyPiExtractor) GetEcosystem() packagev1.Ecosystem {
return packagev1.Ecosystem_ECOSYSTEM_PYPI
}
func (p *PyPiExtractor) Extract(depFile lockfile.DepFile, filename string) ([]lockfile.PackageDetails, error) {
switch filename {
case "requirements.txt":
extractor := lockfile.RequirementsTxtExtractor{}
return extractor.Extract(depFile)
case "Pipfile.lock":
extractor := lockfile.PipenvLockExtractor{}
return extractor.Extract(depFile)
default:
return nil, fmt.Errorf("unsupported PyPI lockfile: %s", filename)
}
}
// ExtractorManager manages all ecosystem extractors
type ExtractorManager struct {
extractors map[packagev1.Ecosystem]EcosystemExtractor
extractors map[PackageManagerName]PackageManagerExtractor
}
func NewExtractorManager() *ExtractorManager {
return &ExtractorManager{
extractors: map[packagev1.Ecosystem]EcosystemExtractor{
packagev1.Ecosystem_ECOSYSTEM_NPM: &NpmExtractor{},
packagev1.Ecosystem_ECOSYSTEM_PYPI: &PyPiExtractor{},
extractors: map[PackageManagerName]PackageManagerExtractor{
Npm: &NpmExtractor{},
Pnpm: &PnpmExtractor{},
Pip: &PipExtractor{},
},
}
}
func (em *ExtractorManager) GetExtractorForEcosystem(ecosystem packagev1.Ecosystem) (EcosystemExtractor, bool) {
extractor, exists := em.extractors[ecosystem]
return extractor, exists
func (e *ExtractorManager) GetExtractorForPackageManager(pmn PackageManagerName) PackageManagerExtractor {
return e.extractors[pmn]
}
func (em *ExtractorManager) GetSupportedFilesForEcosystem(ecosystem packagev1.Ecosystem) []string {
if extractor, exists := em.extractors[ecosystem]; exists {
return extractor.GetSupportedFiles()
}
return nil
}
func (em *ExtractorManager) GetAllSupportedFiles() []string {
var allFiles []string
for _, extractor := range em.extractors {
allFiles = append(allFiles, extractor.GetSupportedFiles()...)
}
return allFiles
func (e *ExtractorManager) GetSupportedFilesForPackageManager(pmn PackageManagerName) []string {
return e.extractors[pmn].GetSupportedFiles()
}
+17 -99
View File
@@ -4,23 +4,19 @@ import (
"fmt"
"os"
"path/filepath"
"strings"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/google/osv-scanner/pkg/lockfile"
"github.com/safedep/dry/log"
)
type ExtractorConfig struct {
// List of extractors used by an ecosystem
ExtractorsName []string
ExtractorEcosystem packagev1.Ecosystem
ScanDir string
ExtractorPackageManager PackageManagerName
ScanDir string
}
type extractor struct {
Config ExtractorConfig
extractorManager *ExtractorManager
extractorManager ExtractorManager
}
func NewDefaultExtractorConfig() *ExtractorConfig {
@@ -32,7 +28,7 @@ func NewDefaultExtractorConfig() *ExtractorConfig {
func New(config ExtractorConfig) *extractor {
return &extractor{
Config: config,
extractorManager: NewExtractorManager(),
extractorManager: *NewExtractorManager(),
}
}
@@ -50,8 +46,14 @@ func (e *extractor) ExtractManifest() ([]*packagev1.PackageVersion, error) {
continue
}
extractor, err := e.getExtractorForFile()
if err != nil {
log.Warnf("failed to get extractor\n")
continue
}
// Extract packages from this lockfile
packages, err := e.extractFromLockfile(filePath)
packages, err := extractor.Extract(filePath, e.Config.ScanDir)
if err != nil {
log.Warnf("failed to extract from %s: %v\n", filePath, err)
continue
@@ -64,97 +66,13 @@ func (e *extractor) ExtractManifest() ([]*packagev1.PackageVersion, error) {
}
func (e *extractor) getFilesToCheck() []string {
// If specific extractors are configured, use those
if len(e.Config.ExtractorsName) > 0 {
return e.Config.ExtractorsName
}
// If ecosystem is specified, return files for that ecosystem
if e.Config.ExtractorEcosystem != packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED {
return e.extractorManager.GetSupportedFilesForEcosystem(e.Config.ExtractorEcosystem)
}
// If no ecosystem specified, check all known extractors
return e.extractorManager.GetAllSupportedFiles()
return e.extractorManager.GetSupportedFilesForPackageManager(e.Config.ExtractorPackageManager)
}
func (e *extractor) extractFromLockfile(path string) ([]*packagev1.PackageVersion, error) {
filename := filepath.Base(path)
// Find the appropriate extractor for this file
ecosystemExtractor, ecosystem, err := e.getExtractorForFile(filename)
if err != nil {
return nil, err
func (e *extractor) getExtractorForFile() (PackageManagerExtractor, error) {
extractor := e.extractorManager.GetExtractorForPackageManager(e.Config.ExtractorPackageManager)
if extractor == nil {
return nil, fmt.Errorf("no extractor found for package manager: %s", e.Config.ExtractorPackageManager)
}
// Open the file as a DepFile for extraction
depFile, err := lockfile.OpenLocalDepFile(path)
if err != nil {
return nil, fmt.Errorf("failed to open file %s: %w", path, err)
}
defer depFile.Close()
// Extract dependencies using the appropriate extractor
deps, err := ecosystemExtractor.Extract(depFile, filename)
if err != nil {
return nil, fmt.Errorf("failed to extract from %s: %w", path, err)
}
packages := e.convertToPackageVersions(deps, ecosystem)
return packages, nil
}
func (e *extractor) getExtractorForFile(filename string) (EcosystemExtractor, packagev1.Ecosystem, error) {
// If ecosystem is configured, use that directly
if e.Config.ExtractorEcosystem != packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED {
extractor, exists := e.extractorManager.GetExtractorForEcosystem(e.Config.ExtractorEcosystem)
if !exists {
return nil, packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED, fmt.Errorf("no extractor found for ecosystem: %v", e.Config.ExtractorEcosystem)
}
// Verify the file is supported by this ecosystem
supportedFiles := extractor.GetSupportedFiles()
for _, supportedFile := range supportedFiles {
if filename == supportedFile || strings.Contains(filename, supportedFile) {
return extractor, e.Config.ExtractorEcosystem, nil
}
}
return nil, packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED, fmt.Errorf("file %s not supported by ecosystem %v", filename, e.Config.ExtractorEcosystem)
}
// Try to find extractor by filename
for ecosystem, extractor := range e.extractorManager.extractors {
supportedFiles := extractor.GetSupportedFiles()
for _, supportedFile := range supportedFiles {
if filename == supportedFile || strings.Contains(filename, supportedFile) {
return extractor, ecosystem, nil
}
}
}
return nil, packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED, fmt.Errorf("unsupported lockfile type: %s", filename)
}
func (e *extractor) convertToPackageVersions(deps []lockfile.PackageDetails, ecosystem packagev1.Ecosystem) []*packagev1.PackageVersion {
var packages []*packagev1.PackageVersion
for _, dep := range deps {
// Use the configured ecosystem if provided, otherwise use the detected one
targetEcosystem := ecosystem
if e.Config.ExtractorEcosystem != packagev1.Ecosystem_ECOSYSTEM_UNSPECIFIED {
targetEcosystem = e.Config.ExtractorEcosystem
}
packageVer := &packagev1.PackageVersion{
Package: &packagev1.Package{
Name: dep.Name,
Ecosystem: targetEcosystem,
},
Version: dep.Version,
}
packages = append(packages, packageVer)
}
return packages
return extractor, nil
}
+125
View File
@@ -0,0 +1,125 @@
package extractor
import (
"context"
"fmt"
"os"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/google/osv-scalibr/extractor/filesystem"
"github.com/google/osv-scalibr/extractor/filesystem/language/javascript/packagelockjson"
"github.com/google/osv-scalibr/extractor/filesystem/language/javascript/pnpmlock"
"github.com/google/osv-scalibr/fs"
)
// NpmExtractor handles package-lock.json files
type NpmExtractor struct{}
func (n *NpmExtractor) GetSupportedFiles() []string {
return []string{"package-lock.json"}
}
func (n *NpmExtractor) GetEcosystem() packagev1.Ecosystem {
return packagev1.Ecosystem_ECOSYSTEM_NPM
}
func (n *NpmExtractor) GetPackageManager() PackageManagerName {
return Npm
}
func (n *NpmExtractor) Extract(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
return parseNpmPackageLockFile(lockfilePath, scanDir)
}
func parseNpmPackageLockFile(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
packagelockExtractor := packagelockjson.NewDefault()
file, err := os.Open(lockfilePath)
if err != nil {
return nil, fmt.Errorf("failed to open lockfile: %s", err)
}
defer file.Close()
inputConfig := &filesystem.ScanInput{
FS: fs.DirFS(scanDir),
Path: lockfilePath,
Reader: file,
}
inventory, err := packagelockExtractor.Extract(context.Background(), inputConfig)
if err != nil {
return nil, fmt.Errorf("failed to extract packages: %s", err)
}
var packages []*packagev1.PackageVersion
for _, invPkg := range inventory.Packages {
pkg := &packagev1.PackageVersion{
Package: &packagev1.Package{
Name: invPkg.Name,
Ecosystem: packagev1.Ecosystem_ECOSYSTEM_NPM,
},
Version: invPkg.Version,
}
packages = append(packages, pkg)
}
return packages, nil
}
// PnpmExtractor handles pnpm-lock.yaml files
type PnpmExtractor struct{}
func (p *PnpmExtractor) GetSupportedFiles() []string {
return []string{"pnpm-lock.yaml"}
}
func (p *PnpmExtractor) GetEcosystem() packagev1.Ecosystem {
return packagev1.Ecosystem_ECOSYSTEM_NPM
}
func (p *PnpmExtractor) GetPackageManager() PackageManagerName {
return Pnpm
}
func (p *PnpmExtractor) Extract(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
return parsePnpmLockFile(lockfilePath, scanDir)
}
func parsePnpmLockFile(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
pnpmLockExtractor := pnpmlock.New()
file, err := os.Open(lockfilePath)
if err != nil {
return nil, fmt.Errorf("failed to open lockfile: %s", err)
}
defer file.Close()
inputConfig := &filesystem.ScanInput{
FS: fs.DirFS(scanDir),
Path: lockfilePath,
Reader: file,
}
inventory, err := pnpmLockExtractor.Extract(context.Background(), inputConfig)
if err != nil {
return nil, fmt.Errorf("failed to extract packages: %s", err)
}
var packages []*packagev1.PackageVersion
for _, invPkg := range inventory.Packages {
pkg := &packagev1.PackageVersion{
Package: &packagev1.Package{
Name: invPkg.Name,
Ecosystem: packagev1.Ecosystem_ECOSYSTEM_NPM,
},
Version: invPkg.Version,
}
packages = append(packages, pkg)
}
return packages, nil
}
+68
View File
@@ -0,0 +1,68 @@
package extractor
import (
"context"
"fmt"
"os"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/google/osv-scalibr/extractor/filesystem"
"github.com/google/osv-scalibr/extractor/filesystem/language/python/requirements"
"github.com/google/osv-scalibr/fs"
)
// PipExtractor handles requirements.txt files
type PipExtractor struct{}
func (p *PipExtractor) GetSupportedFiles() []string {
return []string{"requirements.txt"}
}
func (p *PipExtractor) GetEcosystem() packagev1.Ecosystem {
return packagev1.Ecosystem_ECOSYSTEM_PYPI
}
func (p *PipExtractor) GetPackageManager() PackageManagerName {
return Pip
}
func (n *PipExtractor) Extract(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
return parseRequirementsTxtFile(lockfilePath, scanDir)
}
func parseRequirementsTxtFile(lockfilePath, scanDir string) ([]*packagev1.PackageVersion, error) {
requirementsExtractor := requirements.NewDefault()
file, err := os.Open(lockfilePath)
if err != nil {
return nil, fmt.Errorf("failed to open lockfile: %s", err)
}
defer file.Close()
inputConfig := &filesystem.ScanInput{
FS: fs.DirFS(scanDir),
Path: lockfilePath,
Reader: file,
}
inventory, err := requirementsExtractor.Extract(context.Background(), inputConfig)
if err != nil {
return nil, fmt.Errorf("failed to extract packages: %s", err)
}
var packages []*packagev1.PackageVersion
for _, invPkg := range inventory.Packages {
pkg := &packagev1.PackageVersion{
Package: &packagev1.Package{
Name: invPkg.Name,
Ecosystem: packagev1.Ecosystem_ECOSYSTEM_PYPI,
},
Version: invPkg.Version,
}
packages = append(packages, pkg)
}
return packages, nil
}