fix(python): skip dev group's deps for poetry (#8106)

Signed-off-by: nikpivkin <nikita.pivkin@smartforce.io>

pkg/dependency/parser/python/poetry/parse.go

@@ -2,13 +2,13 @@ package poetry
 
 import (
 	"sort"
-	"strings"
 
 	"github.com/BurntSushi/toml"
 	"golang.org/x/xerrors"
 
 	version "github.com/aquasecurity/go-pep440-version"
 	"github.com/aquasecurity/trivy/pkg/dependency"
+	"github.com/aquasecurity/trivy/pkg/dependency/parser/python"
 	ftypes "github.com/aquasecurity/trivy/pkg/fanal/types"
 	"github.com/aquasecurity/trivy/pkg/log"
 	xio "github.com/aquasecurity/trivy/pkg/x/io"
@@ -105,7 +105,7 @@ func (p *Parser) parseDependencies(deps map[string]any, pkgVersions map[string][
 }
 
 func (p *Parser) parseDependency(name string, versRange any, pkgVersions map[string][]string) (string, error) {
-	name = NormalizePkgName(name)
+	name = python.NormalizePkgName(name)
 	vers, ok := pkgVersions[name]
 	if !ok {
 		return "", xerrors.Errorf("no version found for %q", name)
@@ -149,17 +149,6 @@ func matchVersion(currentVersion, constraint string) (bool, error) {
 	return c.Check(v), nil
 }
 
-// NormalizePkgName normalizes the package name based on pep-0426
-func NormalizePkgName(name string) string {
-	// The package names don't use `_`, `.` or upper case, but dependency names can contain them.
-	// We need to normalize those names.
-	// cf. https://peps.python.org/pep-0426/#name
-	name = strings.ToLower(name)              // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L819
-	name = strings.ReplaceAll(name, "_", "-") // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L50
-	name = strings.ReplaceAll(name, ".", "-") // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L816
-	return name
-}
-
 func packageID(name, ver string) string {
 	return dependency.ID(ftypes.Poetry, name, ver)
 }

pkg/dependency/parser/python/poetry/parse_testcase.go

@@ -3,9 +3,9 @@ package poetry
 import ftypes "github.com/aquasecurity/trivy/pkg/fanal/types"
 
 var (
-	// docker run --name pipenv --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
+	// docker run --name poetry --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
 	// apk add curl
-	// curl -sSL https://install.python-poetry.org | python3 -
+	// curl -sSL https://install.python-poetry.org | POETRY_VERSION=1.1.7 python3 -
 	// export PATH=/root/.local/bin:$PATH
 	// poetry new normal && cd normal
 	// poetry add pypi@2.1
@@ -14,9 +14,9 @@ var (
 		{ID: "pypi@2.1", Name: "pypi", Version: "2.1"},
 	}
 
-	// docker run --name pipenv --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
+	// docker run --name poetry --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
 	// apk add curl
-	// curl -sSL https://install.python-poetry.org | python3 -
+	// curl -sSL https://install.python-poetry.org | POETRY_VERSION=1.1.7 python3 -
 	// export PATH=/root/.local/bin:$PATH
 	// poetry new many && cd many
 	// curl -o poetry.lock https://raw.githubusercontent.com/python-poetry/poetry/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock
@@ -108,9 +108,9 @@ var (
 		{ID: "xattr@0.10.1", DependsOn: []string{"cffi@1.15.1"}},
 	}
 
-	// docker run --name pipenv --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
+	// docker run --name poetry --rm -it python@sha256:e1141f10176d74d1a0e87a7c0a0a5a98dd98ec5ac12ce867768f40c6feae2fd9 sh
 	// apk add curl
-	// curl -sSL https://install.python-poetry.org | python3 -
+	// curl -sSL https://install.python-poetry.org | POETRY_VERSION=1.1.7 python3 -
 	// export PATH=/root/.local/bin:$PATH
 	// poetry new web && cd web
 	// poetry add flask@1.0.3

pkg/dependency/parser/python/pyproject/pyproject.go

@@ -4,7 +4,10 @@ import (
 	"io"
 
 	"github.com/BurntSushi/toml"
+	"github.com/samber/lo"
 	"golang.org/x/xerrors"
+
+	"github.com/aquasecurity/trivy/pkg/dependency/parser/python"
 )
 
 type PyProject struct {
@@ -16,7 +19,26 @@ type Tool struct {
 }
 
 type Poetry struct {
-	Dependencies map[string]any `toml:"dependencies"`
+	Dependencies dependencies     `toml:"dependencies"`
+	Groups       map[string]Group `toml:"group"`
+}
+
+type Group struct {
+	Dependencies dependencies `toml:"dependencies"`
+}
+
+type dependencies map[string]struct{}
+
+func (d *dependencies) UnmarshalTOML(data any) error {
+	m, ok := data.(map[string]any)
+	if !ok {
+		return xerrors.Errorf("dependencies must be map, but got: %T", data)
+	}
+
+	*d = lo.MapEntries(m, func(pkgName string, _ any) (string, struct{}) {
+		return python.NormalizePkgName(pkgName), struct{}{}
+	})
+	return nil
 }
 
 // Parser parses pyproject.toml defined in PEP518.
@@ -28,10 +50,10 @@ func NewParser() *Parser {
 	return &Parser{}
 }
 
-func (p *Parser) Parse(r io.Reader) (map[string]any, error) {
+func (p *Parser) Parse(r io.Reader) (PyProject, error) {
 	var conf PyProject
 	if _, err := toml.NewDecoder(r).Decode(&conf); err != nil {
-		return nil, xerrors.Errorf("toml decode error: %w", err)
+		return PyProject{}, xerrors.Errorf("toml decode error: %w", err)
 	}
-	return conf.Tool.Poetry.Dependencies, nil
+	return conf, nil
 }

pkg/dependency/parser/python/pyproject/pyproject_test.go

@@ -15,26 +15,33 @@ func TestParser_Parse(t *testing.T) {
 	tests := []struct {
 		name    string
 		file    string
-		want    map[string]any
+		want    pyproject.PyProject
 		wantErr assert.ErrorAssertionFunc
 	}{
 		{
 			name: "happy path",
 			file: "testdata/happy.toml",
-			want: map[string]any{
-				"flask":  "^1.0",
-				"python": "^3.9",
-				"requests": map[string]any{
-					"version":  "2.28.1",
-					"optional": true,
-				},
-				"virtualenv": []any{
-					map[string]any{
-						"version": "^20.4.3,!=20.4.5,!=20.4.6",
-					},
-					map[string]any{
-						"version": "<20.16.6",
-						"markers": "sys_platform == 'win32' and python_version == '3.9'",
+			want: pyproject.PyProject{
+				Tool: pyproject.Tool{
+					Poetry: pyproject.Poetry{
+						Dependencies: map[string]struct{}{
+							"flask":      {},
+							"python":     {},
+							"requests":   {},
+							"virtualenv": {},
+						},
+						Groups: map[string]pyproject.Group{
+							"dev": {
+								Dependencies: map[string]struct{}{
+									"pytest": {},
+								},
+							},
+							"lint": {
+								Dependencies: map[string]struct{}{
+									"ruff": {},
+								},
+							},
+						},
 					},
 				},
 			},

pkg/dependency/parser/python/pyproject/testdata/happy.toml

@@ -14,6 +14,13 @@ virtualenv = [
 
 [tool.poetry.dev-dependencies]
 
+[tool.poetry.group.dev.dependencies]
+pytest = "8.3.4"
+
+
+[tool.poetry.group.lint.dependencies]
+ruff = "0.8.3"
+
 [build-system]
 requires = ["poetry-core>=1.0.0"]
 build-backend = "poetry.core.masonry.api"

pkg/dependency/parser/python/python.go

@@ -0,0 +1,14 @@
+package python
+
+import "strings"
+
+// NormalizePkgName normalizes the package name based on pep-0426
+func NormalizePkgName(name string) string {
+	// The package names don't use `_`, `.` or upper case, but dependency names can contain them.
+	// We need to normalize those names.
+	// cf. https://peps.python.org/pep-0426/#name
+	name = strings.ToLower(name)              // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L819
+	name = strings.ReplaceAll(name, "_", "-") // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L50
+	name = strings.ReplaceAll(name, ".", "-") // e.g. https://github.com/python-poetry/poetry/blob/c8945eb110aeda611cc6721565d7ad0c657d453a/poetry.lock#L816
+	return name
+}

pkg/dependency/parser/python/python_test.go

@@ -0,0 +1,39 @@
+package python_test
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"github.com/aquasecurity/trivy/pkg/dependency/parser/python"
+)
+
+func Test_NormalizePkgName(t *testing.T) {
+	tests := []struct {
+		pkgName  string
+		expected string
+	}{
+		{
+			pkgName:  "SecretStorage",
+			expected: "secretstorage",
+		},
+		{
+			pkgName:  "pywin32-ctypes",
+			expected: "pywin32-ctypes",
+		},
+		{
+			pkgName:  "jaraco.classes",
+			expected: "jaraco-classes",
+		},
+		{
+			pkgName:  "green_gdk",
+			expected: "green-gdk",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.pkgName, func(t *testing.T) {
+			assert.Equal(t, tt.expected, python.NormalizePkgName(tt.pkgName))
+		})
+	}
+}

pkg/fanal/analyzer/language/python/poetry/poetry.go

@@ -94,7 +94,7 @@ func (a poetryAnalyzer) parsePoetryLock(path string, r io.Reader) (*types.Applic
 func (a poetryAnalyzer) mergePyProject(fsys fs.FS, dir string, app *types.Application) error {
 	// Parse pyproject.toml to identify the direct dependencies
 	path := filepath.Join(dir, types.PyProject)
-	p, err := a.parsePyProject(fsys, path)
+	project, err := a.parsePyProject(fsys, path)
 	if errors.Is(err, fs.ErrNotExist) {
 		// Assume all the packages are direct dependencies as it cannot identify them from poetry.lock
 		a.logger.Debug("pyproject.toml not found", log.FilePath(path))
@@ -105,34 +105,68 @@ func (a poetryAnalyzer) mergePyProject(fsys fs.FS, dir string, app *types.Applic
 
 	// Identify the direct/transitive dependencies
 	for i, pkg := range app.Packages {
-		if _, ok := p[pkg.Name]; ok {
+		if _, ok := project.Tool.Poetry.Dependencies[pkg.Name]; ok {
 			app.Packages[i].Relationship = types.RelationshipDirect
 		} else {
 			app.Packages[i].Indirect = true
 			app.Packages[i].Relationship = types.RelationshipIndirect
 		}
 	}
 
+	filterProdPackages(project, app)
 	return nil
 }
 
-func (a poetryAnalyzer) parsePyProject(fsys fs.FS, path string) (map[string]any, error) {
+func filterProdPackages(project pyproject.PyProject, app *types.Application) {
+	packages := lo.SliceToMap(app.Packages, func(pkg types.Package) (string, types.Package) {
+		return pkg.ID, pkg
+	})
+
+	visited := make(map[string]struct{})
+	deps := project.Tool.Poetry.Dependencies
+
+	for group, groupDeps := range project.Tool.Poetry.Groups {
+		if group == "dev" {
+			continue
+		}
+		deps = lo.Assign(deps, groupDeps.Dependencies)
+	}
+
+	for _, pkg := range packages {
+		if _, prodDep := deps[pkg.Name]; !prodDep {
+			continue
+		}
+		walkPackageDeps(pkg.ID, packages, visited)
+	}
+
+	app.Packages = lo.Filter(app.Packages, func(pkg types.Package, _ int) bool {
+		_, ok := visited[pkg.ID]
+		return ok
+	})
+}
+
+func walkPackageDeps(pkgID string, packages map[string]types.Package, visited map[string]struct{}) {
+	if _, ok := visited[pkgID]; ok {
+		return
+	}
+	visited[pkgID] = struct{}{}
+	for _, dep := range packages[pkgID].DependsOn {
+		walkPackageDeps(dep, packages, visited)
+	}
+}
+
+func (a poetryAnalyzer) parsePyProject(fsys fs.FS, path string) (pyproject.PyProject, error) {
 	// Parse pyproject.toml
 	f, err := fsys.Open(path)
 	if err != nil {
-		return nil, xerrors.Errorf("file open error: %w", err)
+		return pyproject.PyProject{}, xerrors.Errorf("file open error: %w", err)
 	}
 	defer f.Close()
 
-	parsed, err := a.pyprojectParser.Parse(f)
+	project, err := a.pyprojectParser.Parse(f)
 	if err != nil {
-		return nil, err
+		return pyproject.PyProject{}, err
 	}
 
-	// Packages from `pyproject.toml` can use uppercase characters, `.` and `_`.
-	parsed = lo.MapKeys(parsed, func(_ any, pkgName string) string {
-		return poetry.NormalizePkgName(pkgName)
-	})
-
-	return parsed, nil
+	return project, nil
 }