improve java name and version extraction as well as parent pkg pairing

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

syft/cataloger/java/archive_filename.go

@@ -10,11 +10,20 @@ import (
 	"github.com/anchore/syft/syft/pkg"
 )
 
+// match on versions and anything after the version. This is used to isolate the name from the version.
+// match examples:
+// wagon-webdav-1.0.2-rc1-hudson.jar      --->     -1.0.2-rc1-hudson.jar
+// windows-remote-command-1.0.jar         --->     -1.0.jar
+// wstx-asl-1-2.jar                       --->     -1-2.jar
+// guava-rc0.jar                          --->     -rc0.jar
+var versionAreaPattern = regexp.MustCompile(`-(?P<version>(\d+\.)?(\d+\.)?(r?c?\d+)(-[a-zA-Z0-9\-.]+)*)(?P<remaining>.*)$`)
+
+// match on explicit versions. This is used for extracting version information.
 // match examples:
 //	pkg-extra-field-4.3.2-rc1 --> match(name=pkg-extra-field version=4.3.2-rc1)
 //	pkg-extra-field-4.3-rc1   --> match(name=pkg-extra-field version=4.3-rc1)
 //	pkg-extra-field-4.3       --> match(name=pkg-extra-field version=4.3)
-var versionPattern = regexp.MustCompile(`(?P<name>.+)-(?P<version>(\d+\.)?(\d+\.)?(\*|\d+)(-[a-zA-Z0-9\-\.]+)*)`)
+var versionPattern = regexp.MustCompile(`-(?P<version>(\d+\.)?(\d+\.)?(r?c?\d+)(-[a-zA-Z0-9\-.]+)*)`)
 
 type archiveFilename struct {
 	raw    string
@@ -70,14 +79,8 @@ func (a archiveFilename) version() string {
 }
 
 func (a archiveFilename) name() string {
-	for _, fieldSet := range a.fields {
-		if name, ok := fieldSet["name"]; ok {
-			// return the first name
-			return name
-		}
-	}
-
-	// derive the name from the archive name (no path or extension)
+	// derive the name from the archive name (no path or extension) and remove any versions found
 	basename := filepath.Base(a.raw)
-	return strings.TrimSuffix(basename, filepath.Ext(basename))
+	cleaned := strings.TrimSuffix(basename, filepath.Ext(basename))
+	return versionAreaPattern.ReplaceAllString(cleaned, "")
 }

syft/cataloger/java/archive_filename_test.go

@@ -1,9 +1,10 @@
 package java
 
 import (
+	"testing"
+
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/sergi/go-diff/diffmatchpatch"
-	"testing"
 )
 
 func TestExtractInfoFromJavaArchiveFilename(t *testing.T) {
@@ -56,12 +57,78 @@ func TestExtractInfoFromJavaArchiveFilename(t *testing.T) {
 			name:      "pkg-extra-field-maven",
 			ty:        pkg.JenkinsPluginPkg,
 		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-webdav-1.0.2-beta-2.2.3a-hudson.jar",
+			version:   "1.0.2-beta-2.2.3a-hudson",
+			extension: "jar",
+			name:      "wagon-webdav",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-webdav-1.0.2-beta-2.2.3-hudson.jar",
+			version:   "1.0.2-beta-2.2.3-hudson",
+			extension: "jar",
+			name:      "wagon-webdav",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/windows-remote-command-1.0.jar",
+			version:   "1.0",
+			extension: "jar",
+			name:      "windows-remote-command",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/some/path-with-version-5.4.3/wagon-http-lightweight-1.0.5-beta-2.jar",
+			version:   "1.0.5-beta-2",
+			extension: "jar",
+			name:      "wagon-http-lightweight",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename:  "/hudson.war:WEB-INF/lib/commons-jelly-1.1-hudson-20100305.jar",
+			version:   "1.1-hudson-20100305",
+			extension: "jar",
+			name:      "commons-jelly",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename: "/hudson.war:WEB-INF/lib/jtidy-4aug2000r7-dev-hudson-1.jar",
+			// I don't see how we can reliably account for this case
+			//version:   "4aug2000r7-dev-hudson-1",
+			version:   "",
+			extension: "jar",
+			name:      "jtidy",
+			ty:        pkg.JavaPkg,
+		},
+		{
+			filename: "/hudson.war:WEB-INF/lib/trilead-ssh2-build212-hudson-5.jar",
+			// I don't see how we can reliably account for this case
+			//version:   "build212-hudson-5",
+			version:   "5",
+			extension: "jar",
+			// name: "trilead-ssh2",
+			name: "trilead-ssh2-build212-hudson",
+			ty:   pkg.JavaPkg,
+		},
+		{
+			filename:  "/hudson.war:WEB-INF/lib/guava-r06.jar",
+			version:   "r06",
+			extension: "jar",
+			name:      "guava",
+			ty:        pkg.JavaPkg,
+		},
 	}
 
 	for _, test := range tests {
 		t.Run(test.filename, func(t *testing.T) {
 			obj := newJavaArchiveFilename(test.filename)
 
+			ty := obj.pkgType()
+			if ty != test.ty {
+				t.Errorf("mismatched type: %+v != %v", ty, test.ty)
+			}
+
 			version := obj.version()
 			if version != test.version {
 				dmp := diffmatchpatch.New()

syft/cataloger/java/archive_parser.go

@@ -66,13 +66,17 @@ func newJavaArchiveParser(virtualPath string, reader io.Reader, detectNested boo
 		return nil, cleanupFn, fmt.Errorf("unable to read files from java archive: %w", err)
 	}
 
+	// fetch the last element of the virtual path
+	virtualElements := strings.Split(virtualPath, ":")
+	currentFilepath := virtualElements[len(virtualElements)-1]
+
 	return &archiveParser{
 		discoveredPkgs: internal.NewStringSet(),
 		fileManifest:   fileManifest,
 		virtualPath:    virtualPath,
 		archivePath:    archivePath,
 		contentPath:    contentPath,
-		fileInfo:       newJavaArchiveFilename(virtualPath),
+		fileInfo:       newJavaArchiveFilename(currentFilepath),
 		detectNested:   detectNested,
 	}, cleanupFn, nil
 }
@@ -182,6 +186,7 @@ func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([
 				if !strings.HasPrefix(propsObj.ArtifactID, parentPkg.Name) {
 					vPathSuffix += ":" + propsObj.ArtifactID
 				}
+				virtualPath := j.virtualPath + vPathSuffix
 
 				// discovered props = new package
 				p := pkg.Package{
@@ -191,7 +196,7 @@ func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([
 					Type:         pkg.JavaPkg,
 					MetadataType: pkg.JavaMetadataType,
 					Metadata: pkg.JavaMetadata{
-						VirtualPath:   j.virtualPath + vPathSuffix,
+						VirtualPath:   virtualPath,
 						PomProperties: propsObj,
 						Parent:        parentPkg,
 					},
@@ -199,16 +204,24 @@ func (j *archiveParser) discoverPkgsFromPomProperties(parentPkg *pkg.Package) ([
 
 				pkgKey := uniquePkgKey(&p)
 
-				if !j.discoveredPkgs.Contains(pkgKey) {
-					// only keep packages we haven't seen yet
-					pkgs = append(pkgs, p)
-				} else if pkgKey == parentKey {
+				if pkgKey == parentKey || parentPkg.Metadata.(pkg.JavaMetadata).VirtualPath == virtualPath || len(contents) == 1 {
 					// we've run across more information about our parent package, add this info to the parent package metadata
+					// the pom properties is typically a better source of information for name and version than the manifest
+					if p.Name != parentPkg.Name {
+						parentPkg.Name = p.Name
+					}
+					if p.Version != parentPkg.Version {
+						parentPkg.Version = p.Version
+					}
+
 					parentMetadata, ok := parentPkg.Metadata.(pkg.JavaMetadata)
 					if ok {
 						parentMetadata.PomProperties = propsObj
 						parentPkg.Metadata = parentMetadata
 					}
+				} else if !j.discoveredPkgs.Contains(pkgKey) {
+					// only keep packages we haven't seen yet (and are not related to the parent package)
+					pkgs = append(pkgs, p)
 				}
 			}
 		}

syft/cataloger/java/java_manifest.go

@@ -70,7 +70,7 @@ func parseJavaManifest(path string, reader io.Reader) (*pkg.JavaManifest, error)
 	if len(sections) > 0 {
 		manifest.Main = sections[0]
 		if len(sections) > 1 {
-			manifest.Sections = make(map[string]map[string]string)
+			manifest.NamedSections = make(map[string]map[string]string)
 			for i, s := range sections[1:] {
 				name, ok := s["Name"]
 				if !ok {
@@ -82,7 +82,7 @@ func parseJavaManifest(path string, reader io.Reader) (*pkg.JavaManifest, error)
 				} else {
 					delete(s, "Name")
 				}
-				manifest.Sections[name] = s
+				manifest.NamedSections[name] = s
 			}
 		}
 	}
@@ -117,10 +117,10 @@ func selectName(manifest *pkg.JavaManifest, filenameObj archiveFilename) string
 func selectVersion(manifest *pkg.JavaManifest, filenameObj archiveFilename) string {
 	var version string
 	switch {
-	case manifest.Main["Implementation-Version"] != "":
-		version = manifest.Main["Implementation-Version"]
 	case filenameObj.version() != "":
 		version = filenameObj.version()
+	case manifest.Main["Implementation-Version"] != "":
+		version = manifest.Main["Implementation-Version"]
 	case manifest.Main["Specification-Version"] != "":
 		version = manifest.Main["Specification-Version"]
 	case manifest.Main["Plugin-Version"] != "":

syft/cataloger/java/java_manifest_test.go

@@ -45,7 +45,7 @@ func TestParseJavaManifest(t *testing.T) {
 					"Archiver-Version": "Plexus Archiver",
 					"Created-By":       "Apache Maven 3.6.3",
 				},
-				Sections: map[string]map[string]string{
+				NamedSections: map[string]map[string]string{
 					"thing-1": {
 						"Built-By": "?",
 					},

syft/pkg/java_metadata.go

@@ -22,8 +22,8 @@ type PomProperties struct {
 
 // JavaManifest represents the fields of interest extracted from a Java archive's META-INF/MANIFEST.MF file.
 type JavaManifest struct {
-	Main     map[string]string            `json:"main,omitempty"`
-	Sections map[string]map[string]string `json:"sections,omitempty"`
+	Main          map[string]string            `json:"main,omitempty"`
+	NamedSections map[string]map[string]string `json:"namedSections,omitempty"`
 }
 
 func (m JavaMetadata) PackageURL() string {