chore: java binary data

Signed-off-by: Keith Zantow <kzantow@gmail.com>

internal/capabilities/generate/README.md

@@ -135,10 +135,8 @@ graph LR
     style E2 fill:#fff9c4
 ```
 
-### Key Data Flows
-
 1. **Cataloger Discovery**: AST parser walks `syft/pkg/cataloger/` to find `generic.NewCataloger()` calls and extract parser information
-2. **Config Discovery**: AST parser finds config structs and extracts fields with `// app-config:` annotations
+2. **Cataloger Config Discovery**: AST parser finds config structs and extracts fields with `// app-config:` annotations
 3. **App Config Discovery**: AST parser extracts ecosystem configurations from options package, including descriptions and defaults
 4. **Metadata Discovery**: JSON reader loads test observations that record what metadata/package types each parser produces
 5. **Linking**: AST analyzer connects catalogers to their config structs by examining constructor parameters

internal/capabilities/generate/merge.go

@@ -478,10 +478,7 @@ func (e *EnrichmentData) EnrichWithBinaryClassifier(catalogerName string, entry
 			// strip @version from PURL
 			purlStr := stripPURLVersion(classifier.PURL.String())
 
-			detectors = append(detectors, capabilities.Detector{
-				Method:   "glob",
-				Criteria: []string{classifier.FileGlob},
-				Packages: []capabilities.DetectorPackageInfo{
+			packages := []capabilities.DetectorPackageInfo{
 				{
 					Class: classifier.Class,
 					Name:  classifier.Package,
@@ -489,7 +486,22 @@ func (e *EnrichmentData) EnrichWithBinaryClassifier(catalogerName string, entry
 					CPEs:  cpeStrings,
 					Type:  "BinaryPkg",
 				},
-				},
+			}
+
+			for _, o := range binaryClassifierOverrides[classifier.Class] {
+				packages = append(packages, capabilities.DetectorPackageInfo{
+					Class: o.Class,
+					Name:  o.Package,
+					PURL:  o.PURL,
+					CPEs:  o.CPEs,
+					Type:  "BinaryPkg",
+				})
+			}
+
+			detectors = append(detectors, capabilities.Detector{
+				Method:   "glob",
+				Criteria: []string{classifier.FileGlob},
+				Packages: packages,
 			})
 		}
 		entry.Detectors = detectors

internal/capabilities/generate/overrides.go

@@ -0,0 +1,107 @@
+package main
+
+import (
+	"github.com/anchore/syft/syft/cpe"
+)
+
+// this is a hack to get some information in the output that is otherwise difficult to extract.
+// it should be removed after we figure out how to extract it properly from the classifiers
+type binaryClassifierOverride struct {
+	Class   string
+	Package string
+	PURL    string
+	CPEs    []string
+}
+
+var binaryClassifierOverrides = map[string][]binaryClassifierOverride{
+	"java-binary": {
+		{
+			Class:   "java-binary-graalvm",
+			Package: "graalvm",
+			PURL:    mustPURL("pkg:generic/oracle/graalvm@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:graalvm:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-openjdk-zulu",
+			Package: "zulu",
+			PURL:    mustPURL("pkg:generic/azul/zulu@version"),
+			CPEs:    singleCPE("cpe:2.3:a:azul:zulu:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-openjdk-with-update",
+			Package: "openjdk",
+			PURL:    mustPURL("pkg:generic/oracle/openjdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:openjdk:{{.primary}}:update{{.update}}:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-openjdk",
+			Package: "openjdk",
+			PURL:    mustPURL("pkg:generic/oracle/openjdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-ibm",
+			Package: "java",
+			PURL:    mustPURL("pkg:generic/ibm/java@version"),
+			CPEs:    singleCPE("cpe:2.3:a:ibm:java:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-openjdk-fallthrough",
+			Package: "jre",
+			PURL:    mustPURL("pkg:generic/oracle/jre@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:jre:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-oracle",
+			Package: "jre",
+			PURL:    mustPURL("pkg:generic/oracle/jre@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:jre:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+	},
+	"java-jdb-binary": {
+		{
+			Class:   "java-binary-graalvm",
+			Package: "graalvm",
+			PURL:    mustPURL("pkg:generic/oracle/graalvm@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:graalvm_for_jdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "jdb-binary-openjdk-zulu",
+			Package: "zulu",
+			PURL:    mustPURL("pkg:generic/azul/zulu@version"),
+			CPEs:    singleCPE("cpe:2.3:a:azul:zulu:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-jdb-binary-openjdk",
+			Package: "openjdk",
+			PURL:    mustPURL("pkg:generic/oracle/openjdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-sdk-binary-ibm",
+			Package: "java_sdk",
+			PURL:    mustPURL("pkg:generic/ibm/java_sdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:ibm:java_sdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-openjdk-fallthrough",
+			Package: "openjdk",
+			PURL:    mustPURL("pkg:generic/oracle/openjdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+		{
+			Class:   "java-binary-jdk",
+			Package: "jdk",
+			PURL:    mustPURL("pkg:generic/oracle/jdk@version"),
+			CPEs:    singleCPE("cpe:2.3:a:oracle:jdk:*:*:*:*:*:*:*:*", cpe.NVDDictionaryLookupSource),
+		},
+	},
+}
+
+func mustPURL(purl string) string {
+	return purl
+}
+
+func singleCPE(cpeString string, _ ...any) []string {
+	return []string{cpeString}
+}

internal/capabilities/packages.yaml

@@ -994,6 +994,48 @@ catalogers:
             purl: pkg:/
             cpes: []
             type: BinaryPkg
+          - class: java-binary-graalvm
+            name: graalvm
+            purl: pkg:generic/oracle/graalvm@version
+            cpes:
+              - cpe:2.3:a:oracle:graalvm:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-openjdk-zulu
+            name: zulu
+            purl: pkg:generic/azul/zulu@version
+            cpes:
+              - cpe:2.3:a:azul:zulu:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-openjdk-with-update
+            name: openjdk
+            purl: pkg:generic/oracle/openjdk@version
+            cpes:
+              - cpe:2.3:a:oracle:openjdk:{{.primary}}:update{{.update}}:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-openjdk
+            name: openjdk
+            purl: pkg:generic/oracle/openjdk@version
+            cpes:
+              - cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-ibm
+            name: java
+            purl: pkg:generic/ibm/java@version
+            cpes:
+              - cpe:2.3:a:ibm:java:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-openjdk-fallthrough
+            name: jre
+            purl: pkg:generic/oracle/jre@version
+            cpes:
+              - cpe:2.3:a:oracle:jre:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-oracle
+            name: jre
+            purl: pkg:generic/oracle/jre@version
+            cpes:
+              - cpe:2.3:a:oracle:jre:*:*:*:*:*:*:*:*
+            type: BinaryPkg
       - method: glob
         criteria:
           - '**/jdb'
@@ -1003,6 +1045,42 @@ catalogers:
             purl: pkg:/
             cpes: []
             type: BinaryPkg
+          - class: java-binary-graalvm
+            name: graalvm
+            purl: pkg:generic/oracle/graalvm@version
+            cpes:
+              - cpe:2.3:a:oracle:graalvm_for_jdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: jdb-binary-openjdk-zulu
+            name: zulu
+            purl: pkg:generic/azul/zulu@version
+            cpes:
+              - cpe:2.3:a:azul:zulu:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-jdb-binary-openjdk
+            name: openjdk
+            purl: pkg:generic/oracle/openjdk@version
+            cpes:
+              - cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-sdk-binary-ibm
+            name: java_sdk
+            purl: pkg:generic/ibm/java_sdk@version
+            cpes:
+              - cpe:2.3:a:ibm:java_sdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-openjdk-fallthrough
+            name: openjdk
+            purl: pkg:generic/oracle/openjdk@version
+            cpes:
+              - cpe:2.3:a:oracle:openjdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
+          - class: java-binary-jdk
+            name: jdk
+            purl: pkg:generic/oracle/jdk@version
+            cpes:
+              - cpe:2.3:a:oracle:jdk:*:*:*:*:*:*:*:*
+            type: BinaryPkg
     metadata_types: # AUTO-GENERATED
       - pkg.BinarySignature
     package_types: # AUTO-GENERATED
@@ -3365,6 +3443,37 @@ catalogers:
       - package
       - python
     parsers: # AUTO-GENERATED structure
+      - function: parsePdmLock # AUTO-GENERATED
+        detector: # AUTO-GENERATED
+          method: glob # AUTO-GENERATED
+          criteria: # AUTO-GENERATED
+            - '**/pdm.lock'
+        metadata_types: # AUTO-GENERATED
+          - pkg.PythonPdmLockEntry
+        package_types: # AUTO-GENERATED
+          - python
+        json_schema_types: # AUTO-GENERATED
+          - PythonPdmLockEntry
+        capabilities: # MANUAL - config-driven capability definitions
+          - name: license
+            default: false
+          - name: dependency.depth
+            default:
+              - direct
+              - indirect
+          - name: dependency.edges
+            default: complete
+          - name: dependency.kinds
+            default:
+              - runtime
+              - dev
+              - optional
+          - name: package_manager.files.listing
+            default: false
+          - name: package_manager.files.digests
+            default: false
+          - name: package_manager.package_integrity_hash
+            default: false
       - function: parseUvLock # AUTO-GENERATED
         detector: # AUTO-GENERATED
           method: glob # AUTO-GENERATED
@@ -3387,6 +3496,7 @@ catalogers:
             default: complete
           - name: dependency.kinds
             default:
+              - runtime
               - dev
               - optional
           - name: package_manager.files.listing
@@ -3471,6 +3581,7 @@ catalogers:
             default: complete
           - name: dependency.kinds
             default:
+              - runtime
               - dev
               - optional
           - name: package_manager.files.listing
@@ -3478,8 +3589,10 @@ catalogers:
           - name: package_manager.files.digests
             default: false
           - name: package_manager.package_integrity_hash
-            default: false
-      - function: parseRequirementsTxt # AUTO-GENERATED
+            default: true
+            evidence:
+              - PythonPoetryLockEntry.PackageHashes
+      - function: parseRequirementsTxt
         detector: # AUTO-GENERATED
           method: glob # AUTO-GENERATED
           criteria: # AUTO-GENERATED
@@ -3490,7 +3603,7 @@ catalogers:
           - python
         json_schema_types: # AUTO-GENERATED
           - PythonPipRequirementsEntry
-        capabilities: # MANUAL - config-driven capability definitions
+        capabilities: # MANUAL - preserved across regeneration
           - name: license
             default: false
           - name: dependency.depth

syft/pkg/cataloger/python/cataloger_test.go

@@ -14,7 +14,7 @@ import (
 	"github.com/anchore/syft/syft/pkg/cataloger/internal/pkgtest"
 )
 
-func Test_PackageCataloger(t *testing.T) {
+func Test_InstalledPackageCataloger(t *testing.T) {
 	ctx := context.TODO()
 	tests := []struct {
 		name             string
@@ -402,10 +402,57 @@ func Test_PackageCataloger(t *testing.T) {
 
 	for _, test := range tests {
 		t.Run(test.name, func(t *testing.T) {
-			(pkgtest.NewCatalogTester().
+			pkgtest.NewCatalogTester().
 				FromDirectory(t, test.fixture).
 				Expects(test.expectedPackages, nil).
-				TestCataloger(t, NewInstalledPackageCataloger()))
+				TestCataloger(t, NewInstalledPackageCataloger())
+		})
+	}
+}
+
+func Test_PackageCataloger(t *testing.T) {
+
+	tests := []struct {
+		name                  string
+		fixture               string
+		expectedPackages      []string
+		expectedRelationships []string
+	}{
+		{
+			name:    "pdm",
+			fixture: "test-fixtures/pdm-lock",
+			expectedPackages: []string{
+				"certifi @ 2025.1.31 (pdm.lock)",
+				"chardet @ 3.0.4 (pdm.lock)",
+				"charset-normalizer @ 2.0.12 (pdm.lock)",
+				"colorama @ 0.3.9 (pdm.lock)",
+				"idna @ 2.7 (pdm.lock)",
+				"py @ 1.4.34 (pdm.lock)",
+				"pytest @ 3.2.5 (pdm.lock)",
+				"requests @ 2.27.1 (pdm.lock)",
+				"setuptools @ 39.2.0 (pdm.lock)",
+				"urllib3 @ 1.26.20 (pdm.lock)",
+			},
+			expectedRelationships: []string{
+				"certifi @ 2025.1.31 (pdm.lock) [dependency-of] requests @ 2.27.1 (pdm.lock)",
+				"chardet @ 3.0.4 (pdm.lock) [dependency-of] requests @ 2.27.1 (pdm.lock)",
+				"charset-normalizer @ 2.0.12 (pdm.lock) [dependency-of] requests @ 2.27.1 (pdm.lock)",
+				"colorama @ 0.3.9 (pdm.lock) [dependency-of] pytest @ 3.2.5 (pdm.lock)",
+				"idna @ 2.7 (pdm.lock) [dependency-of] requests @ 2.27.1 (pdm.lock)",
+				"py @ 1.4.34 (pdm.lock) [dependency-of] pytest @ 3.2.5 (pdm.lock)",
+				"setuptools @ 39.2.0 (pdm.lock) [dependency-of] pytest @ 3.2.5 (pdm.lock)",
+				"urllib3 @ 1.26.20 (pdm.lock) [dependency-of] requests @ 2.27.1 (pdm.lock)",
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			pkgtest.NewCatalogTester().
+				FromDirectory(t, tt.fixture).
+				ExpectsPackageStrings(tt.expectedPackages).
+				ExpectsRelationshipStrings(tt.expectedRelationships).
+				TestCataloger(t, NewPackageCataloger(DefaultCatalogerConfig()))
 		})
 	}
 }