add dynamic package.json parsing of author field

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

internal/parse.go

@@ -0,0 +1,15 @@
+package internal
+
+import "regexp"
+
+// MatchCaptureGroups takes a regular expression and string and returns all of the named capture group results in a map.
+func MatchCaptureGroups(regEx *regexp.Regexp, str string) map[string]string {
+	match := regEx.FindStringSubmatch(str)
+	results := make(map[string]string)
+	for i, name := range regEx.SubexpNames() {
+		if i > 0 && i <= len(match) {
+			results[name] = match[i]
+		}
+	}
+	return results
+}

syft/cataloger/javascript/parse_package_json.go

@@ -4,6 +4,11 @@ import (
 	"encoding/json"
 	"fmt"
 	"io"
+	"regexp"
+
+	"github.com/anchore/syft/internal"
+
+	"github.com/mitchellh/mapstructure"
 
 	"github.com/anchore/syft/syft/cataloger/common"
 	"github.com/anchore/syft/syft/pkg"
@@ -16,7 +21,7 @@ var _ common.ParserFn = parsePackageLock
 type PackageJSON struct {
 	Version      string            `json:"version"`
 	Latest       []string          `json:"latest"`
-	Author       string            `json:"author"`
+	Author       Author            `json:"author"`
 	License      string            `json:"license"`
 	Name         string            `json:"name"`
 	Homepage     string            `json:"homepage"`
@@ -24,6 +29,57 @@ type PackageJSON struct {
 	Dependencies map[string]string `json:"dependencies"`
 }
 
+type Author struct {
+	Name  string `json:"name" mapstruct:"name"`
+	Email string `json:"email" mapstruct:"email"`
+	URL   string `json:"url" mapstruct:"url"`
+}
+
+var authorPattern = regexp.MustCompile(`^\s*(?P<name>[^<(]*)(\s+<(?P<email>.*)>)?(\s\((?P<url>.*)\))?\s*$`)
+
+func (a *Author) UnmarshalJSON(b []byte) error {
+	var authorStr string
+	if err := json.Unmarshal(b, &authorStr); err != nil {
+		// string parsing did not work, assume a map was given
+		// for more information: https://docs.npmjs.com/files/package.json#people-fields-author-contributors
+		var fields map[string]string
+		var author Author
+		if err := json.Unmarshal(b, &fields); err != nil {
+			return fmt.Errorf("unable to parse package.json author: %w", err)
+		}
+		// translate the map into a structure
+		if err := mapstructure.Decode(fields, &author); err != nil {
+			return fmt.Errorf("unable to decode package.json author: %w", err)
+		}
+		*a = author
+	} else {
+		// parse out "name <email> (url)" into an Author struct
+		var fields = internal.MatchCaptureGroups(authorPattern, authorStr)
+		*a = Author{
+			Name:  fields["name"],
+			Email: fields["email"],
+			URL:   fields["url"],
+		}
+	}
+
+	if a.Name == "" {
+		return fmt.Errorf("package.json author name is empty")
+	}
+
+	return nil
+}
+
+func (a *Author) String() string {
+	result := a.Name
+	if a.Email != "" {
+		result += fmt.Sprintf(" <%s>", a.Email)
+	}
+	if a.URL != "" {
+		result += fmt.Sprintf(" (%s)", a.URL)
+	}
+	return result
+}
+
 // parsePackageJson parses a package.json and returns the discovered JavaScript packages.
 func parsePackageJSON(_ string, reader io.Reader) ([]pkg.Package, error) {
 	packages := make([]pkg.Package, 0)
@@ -44,7 +100,7 @@ func parsePackageJSON(_ string, reader io.Reader) ([]pkg.Package, error) {
 			Language: pkg.JavaScript,
 			Type:     pkg.NpmPkg,
 			Metadata: pkg.NpmMetadata{
-				Author:   p.Author,
+				Author:   p.Author.String(),
 				Homepage: p.Homepage,
 			},
 		})

syft/cataloger/javascript/parse_package_json_test.go

@@ -9,35 +9,61 @@ import (
 )
 
 func TestParsePackageJSON(t *testing.T) {
-	expected := pkg.Package{
-		Name:     "npm",
-		Version:  "6.14.6",
-		Type:     pkg.NpmPkg,
-		Licenses: []string{"Artistic-2.0"},
-		Language: pkg.JavaScript,
-		Metadata: pkg.NpmMetadata{
-			Author:   "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
-			Homepage: "https://docs.npmjs.com/",
+	tests := []struct {
+		Fixture     string
+		ExpectedPkg pkg.Package
+	}{
+		{
+			Fixture: "test-fixtures/pkg-json/package.json",
+			ExpectedPkg: pkg.Package{
+				Name:     "npm",
+				Version:  "6.14.6",
+				Type:     pkg.NpmPkg,
+				Licenses: []string{"Artistic-2.0"},
+				Language: pkg.JavaScript,
+				Metadata: pkg.NpmMetadata{
+					Author:   "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
+					Homepage: "https://docs.npmjs.com/",
+				},
+			},
+		},
+		{
+			Fixture: "test-fixtures/pkg-json/package-nested-author.json",
+			ExpectedPkg: pkg.Package{
+				Name:     "npm",
+				Version:  "6.14.6",
+				Type:     pkg.NpmPkg,
+				Licenses: []string{"Artistic-2.0"},
+				Language: pkg.JavaScript,
+				Metadata: pkg.NpmMetadata{
+					Author:   "Isaac Z. Schlueter <i@izs.me> (http://blog.izs.me)",
+					Homepage: "https://docs.npmjs.com/",
+				},
+			},
 		},
 	}
-	fixture, err := os.Open("test-fixtures/pkg-json/package.json")
-	if err != nil {
-		t.Fatalf("failed to open fixture: %+v", err)
-	}
 
-	actual, err := parsePackageJSON(fixture.Name(), fixture)
-	if err != nil {
-		t.Fatalf("failed to parse package-lock.json: %+v", err)
-	}
-	if len(actual) != 1 {
-		for _, a := range actual {
-			t.Log("   ", a)
-		}
-		t.Fatalf("unexpected package count: %d!=1", len(actual))
-	}
+	for _, test := range tests {
+		t.Run(test.Fixture, func(t *testing.T) {
+			fixture, err := os.Open(test.Fixture)
+			if err != nil {
+				t.Fatalf("failed to open fixture: %+v", err)
+			}
 
-	for _, d := range deep.Equal(actual[0], expected) {
-		t.Errorf("diff: %+v", d)
-	}
+			actual, err := parsePackageJSON(fixture.Name(), fixture)
+			if err != nil {
+				t.Fatalf("failed to parse package-lock.json: %+v", err)
+			}
+			if len(actual) != 1 {
+				for _, a := range actual {
+					t.Log("   ", a)
+				}
+				t.Fatalf("unexpected package count: %d!=1", len(actual))
+			}
 
+			for _, d := range deep.Equal(actual[0], test.ExpectedPkg) {
+				t.Errorf("diff: %+v", d)
+			}
+		})
+	}
 }

syft/cataloger/javascript/test-fixtures/pkg-json/package-nested-author.json

@@ -0,0 +1,12 @@
+{
+  "version": "6.14.6",
+  "name": "npm",
+  "description": "a package manager for JavaScript",
+  "homepage": "https://docs.npmjs.com/",
+  "author": {
+    "name": "Isaac Z. Schlueter",
+    "email": "i@izs.me",
+    "url": "http://blog.izs.me"
+  },
+  "license": "Artistic-2.0"
+}

syft/cataloger/ruby/parse_gemspec.go

@@ -8,6 +8,8 @@ import (
 	"regexp"
 	"strings"
 
+	"github.com/anchore/syft/internal"
+
 	"github.com/mitchellh/mapstructure"
 
 	"github.com/anchore/syft/syft/cataloger/common"
@@ -76,7 +78,7 @@ func parseGemSpecEntries(_ string, reader io.Reader) ([]pkg.Package, error) {
 		}
 
 		for field, pattern := range patterns {
-			matchMap := matchCaptureGroups(pattern, sanitizedLine)
+			matchMap := internal.MatchCaptureGroups(pattern, sanitizedLine)
 			if value := matchMap[field]; value != "" {
 				if postProcessor := postProcessors[field]; postProcessor != nil {
 					fields[field] = postProcessor(value)
@@ -121,15 +123,3 @@ func renderUtf8(s string) string {
 	})
 	return fullReplacement
 }
-
-// matchCaptureGroups takes a regular expression and string and returns all of the named capture group results in a map.
-func matchCaptureGroups(regEx *regexp.Regexp, str string) map[string]string {
-	match := regEx.FindStringSubmatch(str)
-	results := make(map[string]string)
-	for i, name := range regEx.SubexpNames() {
-		if i > 0 && i <= len(match) {
-			results[name] = match[i]
-		}
-	}
-	return results
-}

syft/pkg/metadata.go

@@ -0,0 +1,11 @@
+package pkg
+
+type MetadataType string
+
+const (
+	UnknownMetadataType MetadataType = "UnknownMetadata"
+	ApkMetadataType     MetadataType = "apk-metadata"
+	DpkgMetadataType    MetadataType = "dpkg-metadata"
+	GemgMetadataType    MetadataType = "gem-metadata"
+	RpmdbMetadataType   MetadataType = "rpmdb-metadata"
+)