fix: improve Jenkins plugin known CPE index construction

Improves the logic for adding known CPEs for Jenkins plugins and
prefers non-deprecated CPEs

Signed-off-by: Weston Steimel <commits@weston.slmail.me>

syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/generate.go

@@ -72,9 +72,14 @@ func filterCpeList(cpeList CpeList) CpeList {
 		normalizedName := normalizeCPE(parsedName).BindToURI()
 		cpeItem.Name = normalizedName
 
-		parsedCPE, err := wfn.Parse(cpeItem.Cpe23Item.Name)
+		cpeName := cpeItem.Cpe23Item.Name
+		if cpeItem.Cpe23Item.Deprecation.DeprecatedBy.Name != "" {
+			cpeName = cpeItem.Cpe23Item.Deprecation.DeprecatedBy.Name
+		}
+
+		parsedCPE, err := wfn.Parse(cpeName)
 		if err != nil {
-			log.Printf("unable to parse CPE value %q: %s", cpeItem.Cpe23Item.Name, err)
+			log.Printf("unable to parse CPE value %q: %s", cpeName, err)
 			continue
 		}
 
@@ -97,18 +102,19 @@ func normalizeCPE(cpe *wfn.Attributes) *wfn.Attributes {
 }
 
 const (
-	prefixForNPMPackages    = "https://www.npmjs.com/package/"
-	prefixForRubyGems       = "https://rubygems.org/gems/"
-	prefixForRubyGemsHTTP   = "http://rubygems.org/gems/"
-	prefixForNativeRubyGems = "https://github.com/ruby/"
-	prefixForPyPIPackages   = "https://pypi.org/project/"
-	prefixForJenkinsPlugins = "https://github.com/jenkinsci/"
-	prefixForRustCrates     = "https://crates.io/crates/"
-	prefixForPHPPear        = "https://pear.php.net/"
-	prefixForPHPPearHTTP    = "http://pear.php.net/"
-	prefixForPHPPecl        = "https://pecl.php.net/"
-	prefixForPHPPeclHTTP    = "http://pecl.php.net/"
-	prefixForPHPComposer    = "https://packagist.org/packages/"
+	prefixForNPMPackages          = "https://www.npmjs.com/package/"
+	prefixForRubyGems             = "https://rubygems.org/gems/"
+	prefixForRubyGemsHTTP         = "http://rubygems.org/gems/"
+	prefixForNativeRubyGems       = "https://github.com/ruby/"
+	prefixForPyPIPackages         = "https://pypi.org/project/"
+	prefixForJenkinsPlugins       = "https://plugins.jenkins.io/"
+	prefixForJenkinsPluginsGitHub = "https://github.com/jenkinsci/"
+	prefixForRustCrates           = "https://crates.io/crates/"
+	prefixForPHPPear              = "https://pear.php.net/"
+	prefixForPHPPearHTTP          = "http://pear.php.net/"
+	prefixForPHPPecl              = "https://pecl.php.net/"
+	prefixForPHPPeclHTTP          = "http://pecl.php.net/"
+	prefixForPHPComposer          = "https://packagist.org/packages/"
 )
 
 // indexCPEList creates an index of CPEs by ecosystem.
@@ -136,8 +142,11 @@ func indexCPEList(list CpeList) *dictionary.Indexed {
 			case strings.HasPrefix(ref, prefixForPyPIPackages):
 				addEntryForPyPIPackage(indexed, ref, cpeItemName)
 
-			case strings.HasPrefix(ref, prefixForJenkinsPlugins):
+			case strings.HasPrefix(ref, prefixForJenkinsPluginsGitHub):
 				// It _might_ be a jenkins plugin!
+				addEntryForJenkinsPluginGitHub(indexed, ref, cpeItemName)
+
+			case strings.HasPrefix(ref, prefixForJenkinsPlugins):
 				addEntryForJenkinsPlugin(indexed, ref, cpeItemName)
 
 			case strings.HasPrefix(ref, prefixForRustCrates):
@@ -170,9 +179,9 @@ func addEntryForRustCrate(indexed *dictionary.Indexed, ref string, cpeItemName s
 	indexed.EcosystemPackages[dictionary.EcosystemRustCrates][ref] = cpeItemName
 }
 
-func addEntryForJenkinsPlugin(indexed *dictionary.Indexed, ref string, cpeItemName string) {
+func addEntryForJenkinsPluginGitHub(indexed *dictionary.Indexed, ref string, cpeItemName string) {
 	// Prune off the non-package-name parts of the URL
-	ref = strings.TrimPrefix(ref, prefixForJenkinsPlugins)
+	ref = strings.TrimPrefix(ref, prefixForJenkinsPluginsGitHub)
 	ref = strings.Split(ref, "/")[0]
 
 	if !strings.HasSuffix(ref, "-plugin") {
@@ -189,6 +198,22 @@ func addEntryForJenkinsPlugin(indexed *dictionary.Indexed, ref string, cpeItemNa
 	indexed.EcosystemPackages[dictionary.EcosystemJenkinsPlugins][ref] = cpeItemName
 }
 
+func addEntryForJenkinsPlugin(indexed *dictionary.Indexed, ref string, cpeItemName string) {
+	// Prune off the non-package-name parts of the URL
+	ref = strings.TrimPrefix(ref, prefixForJenkinsPlugins)
+	ref = strings.Split(ref, "/")[0]
+
+	if ref == "" {
+		return
+	}
+
+	if _, ok := indexed.EcosystemPackages[dictionary.EcosystemJenkinsPlugins]; !ok {
+		indexed.EcosystemPackages[dictionary.EcosystemJenkinsPlugins] = make(dictionary.Packages)
+	}
+
+	indexed.EcosystemPackages[dictionary.EcosystemJenkinsPlugins][ref] = cpeItemName
+}
+
 func addEntryForPyPIPackage(indexed *dictionary.Indexed, ref string, cpeItemName string) {
 	// Prune off the non-package-name parts of the URL
 	ref = strings.TrimPrefix(ref, prefixForPyPIPackages)

syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/generate_test.go

@@ -65,8 +65,8 @@ func Test_addEntryFuncs(t *testing.T) {
 			},
 		},
 		{
-			name:             "addEntryForJenkinsPlugin",
-			addEntryFunc:     addEntryForJenkinsPlugin,
+			name:             "addEntryForJenkinsPluginGitHub",
+			addEntryFunc:     addEntryForJenkinsPluginGitHub,
 			inputRef:         "https://github.com/jenkinsci/sonarqube-plugin",
 			inputCpeItemName: "cpe:2.3:a:sonarsource:sonarqube_scanner:2.7:*:*:*:*:jenkins:*:*",
 			expectedIndexed: dictionary.Indexed{
@@ -78,14 +78,27 @@ func Test_addEntryFuncs(t *testing.T) {
 			},
 		},
 		{
-			name:             "addEntryForJenkinsPlugin: not actually a plugin",
-			addEntryFunc:     addEntryForJenkinsPlugin,
+			name:             "addEntryForJenkinsPluginGitHub: not actually a plugin",
+			addEntryFunc:     addEntryForJenkinsPluginGitHub,
 			inputRef:         "https://github.com/jenkinsci/jenkins",
 			inputCpeItemName: "cpe:2.3:a:jenkins:jenkinsci:2.7:*:*:*:*:*:*:*",
 			expectedIndexed: dictionary.Indexed{
 				EcosystemPackages: map[string]dictionary.Packages{},
 			},
 		},
+		{
+			name:             "addEntryForJenkinsPlugin",
+			addEntryFunc:     addEntryForJenkinsPlugin,
+			inputRef:         "https://plugins.jenkins.io/svn-partial-release-mgr/release",
+			inputCpeItemName: "cpe:2.3:a:jenkins:subversion_partial_release_manager:1.0.1:*:*:*:*:jenkins:*:*",
+			expectedIndexed: dictionary.Indexed{
+				EcosystemPackages: map[string]dictionary.Packages{
+					dictionary.EcosystemJenkinsPlugins: {
+						"svn-partial-release-mgr": "cpe:2.3:a:jenkins:subversion_partial_release_manager:1.0.1:*:*:*:*:jenkins:*:*",
+					},
+				},
+			},
+		},
 		{
 			name:             "addEntryForPyPIPackage",
 			addEntryFunc:     addEntryForPyPIPackage,

syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/nvd.go

@@ -10,7 +10,12 @@ type CpeItem struct {
 		} `xml:"reference"`
 	} `xml:"references"`
 	Cpe23Item struct {
-		Name string `xml:"name,attr"`
+		Name        string `xml:"name,attr"`
+		Deprecation struct {
+			DeprecatedBy struct {
+				Name string `xml:"name,attr"`
+			} `xml:"deprecated-by"`
+		} `xml:"deprecation"`
 	} `xml:"cpe23-item"`
 }
 

syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/testdata/expected-cpe-index.json

@@ -1,7 +1,9 @@
 {
   "ecosystems": {
     "jenkins_plugins": {
-      "sonarqube": "cpe:2.3:a:sonarsource:sonarqube_scanner:*:*:*:*:*:jenkins:*:*"
+      "fireline": "cpe:2.3:a:jenkins:360_fireline:*:*:*:*:*:jenkins:*:*",
+      "sonarqube": "cpe:2.3:a:sonarsource:sonarqube_scanner:*:*:*:*:*:jenkins:*:*",
+      "svn-partial-release-mgr": "cpe:2.3:a:jenkins:subversion_partial_release_manager:*:*:*:*:*:jenkins:*:*"
     },
     "npm": {
       "merge-recursive": "cpe:2.3:a:umbraengineering:merge-recursive:*:*:*:*:*:node.js:*:*",
@@ -30,7 +32,7 @@
       "xhprof": "cpe:2.3:a:php:xhprof:*:*:*:*:*:*:*:*"
     },
     "pypi": {
-      "vault-cli": "cpe:2.3:a:ukg:vault-cli:*:*:*:*:*:python:*:*"
+      "vault-cli": "cpe:2.3:a:vault-cli_project:vault-cli:*:*:*:*:*:python:*:*"
     },
     "rubygems": {
       "openssl": "cpe:2.3:a:ruby-lang:openssl:*:*:*:*:*:*:*:*"

syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/testdata/official-cpe-dictionary_v2.3.xml

@@ -24994,4 +24994,26 @@
     </references>
     <cpe-23:cpe23-item name="cpe:2.3:a:frappant:forms_export:-:*:*:*:*:typo3:*:*"/>
   </cpe-item>
+  <cpe-item name="cpe:/a:jenkins:_subversion_partial_release_manager:1.0.1::~~~jenkins~~" deprecated="true" deprecation_date="2020-06-04T11:50:37.430Z">
+    <title xml:lang="en-US">Jenkins Subversion Partial Release Manager 1.0.1 for Jenkins</title>
+    <references>
+      <reference href="https://github.com/jenkinsci/svn-partial-release-mgr-plugin">Version</reference>
+      <reference href="https://plugins.jenkins.io/svn-partial-release-mgr">Product</reference>
+    </references>
+    <cpe-23:cpe23-item name="cpe:2.3:a:jenkins:_subversion_partial_release_manager:1.0.1:*:*:*:*:jenkins:*:*">
+      <cpe-23:deprecation date="2020-06-04T07:50:37.430-04:00">
+        <cpe-23:deprecated-by name="cpe:2.3:a:jenkins:subversion_partial_release_manager:1.0.1:*:*:*:*:jenkins:*:*" type="NAME_CORRECTION"/>
+      </cpe-23:deprecation>
+    </cpe-23:cpe23-item>
+  </cpe-item>
+  <cpe-item name="cpe:/a:jenkins:360_fireline:1.0::~~~jenkins~~">
+    <title xml:lang="en-US">Jenkins 360 FireLine 1.0 for Jenkins</title>
+    <references>
+      <reference href="https://plugins.jenkins.io/fireline">Product</reference>
+      <reference href="https://github.com/jenkinsci/fireline-plugin">Version</reference>
+    </references>
+    <cpe-23:cpe23-item name="cpe:2.3:a:jenkins:360_fireline:1.0:*:*:*:*:jenkins:*:*"/>
+  </cpe-item>
+</cpe-list>
+</cpe-list>
 </cpe-list>