fix(javascript): accept scalar people fields in package.json (#4779)

Signed-off-by: Yoav Alon <yoav@orca.security>

syft/pkg/cataloger/javascript/parse_package_json.go

@@ -215,48 +215,25 @@ func pathContainsNodeModulesDirectory(p string) bool {
 }
 
 func (p *people) UnmarshalJSON(b []byte) error {
-	// Try to unmarshal as an array of strings
+	// Accept either a JSON array of authors, or a single author as a string or
-	var authorStrings []string
+	// object — the latter is used in the wild (e.g. ghost@5.98.1) and dropping
-	if err := json.Unmarshal(b, &authorStrings); err == nil {
+	// the whole package.json on those was https://github.com/anchore/syft/issues/4778.
-		// Successfully parsed as an array of strings
+	var elements []json.RawMessage
-		auths := make([]person, len(authorStrings))
+	if err := json.Unmarshal(b, &elements); err != nil {
-		for i, authorStr := range authorStrings {
+		// not an array — treat the whole payload as a single element
-			// Parse each string into author fields
+		elements = []json.RawMessage{b}
-			fields := internal.MatchNamedCaptureGroups(authorPattern, authorStr)
-			var auth person
-			if err := mapstructure.Decode(fields, &auth); err != nil {
-				return fmt.Errorf("unable to decode package.json author: %w", err)
 	}
-			// Trim whitespace from name if it was parsed
+
-			if auth.Name != "" {
+	auths := make([]person, len(elements))
-				auth.Name = strings.TrimSpace(auth.Name)
+	for i, e := range elements {
+		if err := json.Unmarshal(e, &auths[i]); err != nil {
+			return fmt.Errorf("unable to parse package.json author: %w", err)
 		}
-			auths[i] = auth
 	}
 	*p = auths
 	return nil
 }
 
-	// Try to unmarshal as an array of objects
-	var authorObjs []map[string]any
-	if err := json.Unmarshal(b, &authorObjs); err == nil {
-		// Successfully parsed as an array of objects
-		auths := make([]person, len(authorObjs))
-		for i, fields := range authorObjs {
-			var auth person
-			if err := mapstructure.Decode(fields, &auth); err != nil {
-				return fmt.Errorf("unable to decode package.json author object: %w", err)
-			}
-			auths[i] = auth
-		}
-		*p = auths
-		return nil
-	}
-
-	// If we get here, it means neither format matched
-	return fmt.Errorf("unable to parse package.json authors field: expected array of strings or array of objects")
-}
-
 func (p people) String() string {
 	if len(p) == 0 {
 		return ""

syft/pkg/cataloger/javascript/parse_package_json_test.go

@@ -307,6 +307,71 @@ func TestParsePackageJSON(t *testing.T) {
 				},
 			},
 		},
+		{
+			// see https://github.com/anchore/syft/issues/4778
+			// Ghost CMS publishes package.json with contributors as a single string
+			Fixture: "testdata/pkg-json/package-contributors-string.json",
+			ExpectedPkg: pkg.Package{
+				Name:    "ghost",
+				Version: "5.98.1",
+				PURL:    "pkg:npm/ghost@5.98.1",
+				Type:    pkg.NpmPkg,
+				Licenses: pkg.NewLicenseSet(
+					pkg.NewLicenseFromLocationsWithContext(ctx, "Artistic-2.0", file.NewLocation("testdata/pkg-json/package-contributors-string.json")),
+				),
+				Language: pkg.JavaScript,
+				Metadata: pkg.NpmPackage{
+					Name:        "ghost",
+					Version:     "5.98.1",
+					Author:      "https://github.com/TryGhost/Ghost/graphs/contributors",
+					Homepage:    "https://docs.npmjs.com/",
+					URL:         "https://github.com/npm/cli",
+					Description: "a package manager for JavaScript",
+				},
+			},
+		},
+		{
+			Fixture: "testdata/pkg-json/package-authors-string.json",
+			ExpectedPkg: pkg.Package{
+				Name:    "npm",
+				Version: "6.14.6",
+				PURL:    "pkg:npm/npm@6.14.6",
+				Type:    pkg.NpmPkg,
+				Licenses: pkg.NewLicenseSet(
+					pkg.NewLicenseFromLocationsWithContext(ctx, "Artistic-2.0", file.NewLocation("testdata/pkg-json/package-authors-string.json")),
+				),
+				Language: pkg.JavaScript,
+				Metadata: pkg.NpmPackage{
+					Name:        "npm",
+					Version:     "6.14.6",
+					Author:      "Harry Potter <hp@hogwards.com> (http://youknowwho.com/)",
+					Homepage:    "https://docs.npmjs.com/",
+					URL:         "https://github.com/npm/cli",
+					Description: "a package manager for JavaScript",
+				},
+			},
+		},
+		{
+			Fixture: "testdata/pkg-json/package-maintainers-object.json",
+			ExpectedPkg: pkg.Package{
+				Name:    "npm",
+				Version: "6.14.6",
+				PURL:    "pkg:npm/npm@6.14.6",
+				Type:    pkg.NpmPkg,
+				Licenses: pkg.NewLicenseSet(
+					pkg.NewLicenseFromLocationsWithContext(ctx, "Artistic-2.0", file.NewLocation("testdata/pkg-json/package-maintainers-object.json")),
+				),
+				Language: pkg.JavaScript,
+				Metadata: pkg.NpmPackage{
+					Name:        "npm",
+					Version:     "6.14.6",
+					Author:      "Charlie Maintainer <charlie@example.com>",
+					Homepage:    "https://docs.npmjs.com/",
+					URL:         "https://github.com/npm/cli",
+					Description: "a package manager for JavaScript",
+				},
+			},
+		},
 		{
 			Fixture: "testdata/pkg-json/package-all-author-fields.json",
 			ExpectedPkg: pkg.Package{

syft/pkg/cataloger/javascript/testdata/pkg-json/package-authors-string.json

@@ -0,0 +1,12 @@
+{
+  "version": "6.14.6",
+  "name": "npm",
+  "description": "a package manager for JavaScript",
+  "homepage": "https://docs.npmjs.com/",
+  "authors": "Harry Potter <hp@hogwards.com> (http://youknowwho.com/)",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/npm/cli"
+  },
+  "license": "Artistic-2.0"
+}

syft/pkg/cataloger/javascript/testdata/pkg-json/package-contributors-string.json

@@ -0,0 +1,12 @@
+{
+  "version": "5.98.1",
+  "name": "ghost",
+  "description": "a package manager for JavaScript",
+  "homepage": "https://docs.npmjs.com/",
+  "contributors": "https://github.com/TryGhost/Ghost/graphs/contributors",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/npm/cli"
+  },
+  "license": "Artistic-2.0"
+}

syft/pkg/cataloger/javascript/testdata/pkg-json/package-maintainers-object.json

@@ -0,0 +1,15 @@
+{
+  "version": "6.14.6",
+  "name": "npm",
+  "description": "a package manager for JavaScript",
+  "homepage": "https://docs.npmjs.com/",
+  "maintainers": {
+    "name": "Charlie Maintainer",
+    "email": "charlie@example.com"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/npm/cli"
+  },
+  "license": "Artistic-2.0"
+}