fix tests and linting

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

internal/capabilities/generate/cataloger_config_linking_test.go

@@ -10,128 +10,6 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-func TestLinkCatalogersToConfigs(t *testing.T) {
-	if testing.Short() {
-		t.Skip("skipping integration test in short mode")
-	}
-
-	repoRoot, err := RepoRoot()
-	require.NoError(t, err)
-
-	linkages, err := LinkCatalogersToConfigs(repoRoot)
-	require.NoError(t, err)
-
-	// verify we discovered multiple catalogers
-	require.NotEmpty(t, linkages, "should discover at least one cataloger linkage")
-
-	// test cases for known catalogers with configs
-	// NOTE: Some catalogers may not be detected if their Name() method is in a different file
-	// than the constructor function. This is a known limitation.
-	tests := []struct {
-		catalogerName string
-		wantConfig    string
-		optional      bool // set to true if detection may not work due to cross-file Name() methods
-	}{
-		{
-			catalogerName: "go-module-binary-cataloger",
-			wantConfig:    "golang.CatalogerConfig",
-		},
-		{
-			catalogerName: "go-module-file-cataloger",
-			wantConfig:    "golang.CatalogerConfig",
-		},
-		{
-			catalogerName: "python-package-cataloger",
-			wantConfig:    "python.CatalogerConfig",
-		},
-		{
-			catalogerName: "java-archive-cataloger",
-			wantConfig:    "java.ArchiveCatalogerConfig",
-		},
-		{
-			catalogerName: "java-pom-cataloger",
-			wantConfig:    "java.ArchiveCatalogerConfig",
-			optional:      true, // Name() method in different file
-		},
-		{
-			catalogerName: "dotnet-deps-binary-cataloger",
-			wantConfig:    "dotnet.CatalogerConfig",
-			optional:      true, // Name() method in different file
-		},
-		{
-			catalogerName: "javascript-lock-cataloger",
-			wantConfig:    "javascript.CatalogerConfig",
-		},
-		{
-			catalogerName: "linux-kernel-cataloger",
-			wantConfig:    "kernel.LinuxKernelCatalogerConfig",
-		},
-		{
-			catalogerName: "nix-cataloger",
-			wantConfig:    "nix.Config",
-			optional:      true, // Name() method in different file
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.catalogerName, func(t *testing.T) {
-			config, ok := linkages[tt.catalogerName]
-			if tt.optional && !ok {
-				t.Skipf("cataloger %s not detected (expected due to cross-file Name() method)", tt.catalogerName)
-				return
-			}
-			require.True(t, ok, "should find linkage for cataloger: %s", tt.catalogerName)
-			require.Equal(t, tt.wantConfig, config, "config type should match for cataloger: %s", tt.catalogerName)
-		})
-	}
-
-	// test catalogers without configs (should have empty string)
-	catalogersWithoutConfig := []string{
-		"python-installed-package-cataloger",
-		"java-gradle-lockfile-cataloger",
-		"java-jvm-cataloger",
-		"dotnet-packages-lock-cataloger",
-		"javascript-package-cataloger",
-	}
-
-	for _, catalogerName := range catalogersWithoutConfig {
-		t.Run(catalogerName+"_no_config", func(t *testing.T) {
-			config, ok := linkages[catalogerName]
-			if ok {
-				require.Empty(t, config, "cataloger %s should have empty config", catalogerName)
-			}
-		})
-	}
-
-	// print summary for manual inspection
-	t.Logf("Discovered %d cataloger-to-config linkages:", len(linkages))
-
-	// separate into catalogers with and without configs
-	withConfig := make(map[string]string)
-	withoutConfig := make([]string, 0)
-
-	for name, config := range linkages {
-		if config != "" {
-			withConfig[name] = config
-		} else {
-			withoutConfig = append(withoutConfig, name)
-		}
-	}
-
-	t.Logf("Catalogers with configs (%d):", len(withConfig))
-	for name, config := range withConfig {
-		t.Logf("  %s -> %s", name, config)
-	}
-
-	t.Logf("Catalogers without configs (%d):", len(withoutConfig))
-	for _, name := range withoutConfig {
-		t.Logf("  %s", name)
-	}
-
-	// ensure we found at least some catalogers with configs
-	require.GreaterOrEqual(t, len(withConfig), 6, "should find at least 6 catalogers with configs")
-}
-
 func TestLinkCatalogersToConfigsFromPath(t *testing.T) {
 	tests := []struct {
 		name             string
@@ -204,7 +82,7 @@ func TestLinkCatalogersToConfigsFromPath(t *testing.T) {
 			name:        "selector expression config",
 			fixturePath: "selector-expression-config",
 			expectedLinkages: map[string]string{
-				"rust-cataloger": "cargo.CatalogerConfig",
+				"rust-cataloger": "rust.CatalogerConfig",
 			},
 		},
 	}
@@ -215,8 +93,9 @@ func TestLinkCatalogersToConfigsFromPath(t *testing.T) {
 				tt.wantErr = require.NoError
 			}
 
-			fixtureDir := filepath.Join("test-fixtures", "config-linking", tt.fixturePath)
+			fixtureDir := filepath.Join("testdata", "cataloger", tt.fixturePath)
-			linkages, err := LinkCatalogersToConfigsFromPath(fixtureDir, fixtureDir)
+			catalogerRoot := filepath.Join(fixtureDir, "cataloger")
+			linkages, err := LinkCatalogersToConfigsFromPath(catalogerRoot, fixtureDir)
 			tt.wantErr(t, err)
 
 			if err != nil {
@@ -229,51 +108,64 @@ func TestLinkCatalogersToConfigsFromPath(t *testing.T) {
 }
 
 func TestExtractConfigTypeName(t *testing.T) {
-	if testing.Short() {
-		t.Skip("skipping integration test in short mode")
-	}
-
 	tests := []struct {
 		name             string
+		fixturePath      string
 		catalogerName    string
 		expectedConfig   string
 		expectedNoConfig bool
 	}{
 		{
 			name:           "golang config",
-			catalogerName:  "go-module-binary-cataloger",
+			fixturePath:    "simple-generic-cataloger",
+			catalogerName:  "go-module-cataloger",
 			expectedConfig: "golang.CatalogerConfig",
 		},
 		{
-			name:           "python config",
+			name:           "python config with constant",
+			fixturePath:    "cataloger-with-constant",
 			catalogerName:  "python-package-cataloger",
 			expectedConfig: "python.CatalogerConfig",
 		},
 		{
-			name:           "java archive config",
+			name:           "java archive config same file",
-			catalogerName:  "java-archive-cataloger",
+			fixturePath:    "custom-cataloger-same-file",
+			catalogerName:  "java-pom-cataloger",
 			expectedConfig: "java.ArchiveCatalogerConfig",
 		},
 		{
-			name:           "kernel config",
+			name:           "kernel config imported type",
+			fixturePath:    "imported-config-type",
 			catalogerName:  "linux-kernel-cataloger",
 			expectedConfig: "kernel.LinuxKernelCatalogerConfig",
 		},
 		{
-			name:             "python installed - no config",
+			name:             "javascript - no config",
-			catalogerName:    "python-installed-package-cataloger",
+			fixturePath:      "no-config-cataloger",
+			catalogerName:    "javascript-cataloger",
 			expectedNoConfig: true,
 		},
+		{
+			name:           "ruby with mixed naming",
+			fixturePath:    "mixed-naming-patterns",
+			catalogerName:  "ruby-cataloger",
+			expectedConfig: "ruby.Config",
+		},
+		{
+			name:           "rust with selector expression",
+			fixturePath:    "selector-expression-config",
+			catalogerName:  "rust-cataloger",
+			expectedConfig: "rust.CatalogerConfig",
+		},
 	}
 
-	repoRoot, err := RepoRoot()
-	require.NoError(t, err)
-
-	linkages, err := LinkCatalogersToConfigs(repoRoot)
-	require.NoError(t, err)
-
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
+			fixtureDir := filepath.Join("testdata", "cataloger", tt.fixturePath)
+			catalogerRoot := filepath.Join(fixtureDir, "cataloger")
+			linkages, err := LinkCatalogersToConfigsFromPath(catalogerRoot, fixtureDir)
+			require.NoError(t, err)
+
 			config, ok := linkages[tt.catalogerName]
 
 			if tt.expectedNoConfig {

internal/capabilities/generate/completeness_test.go

@@ -58,6 +58,8 @@ var observationExceptions = map[string]*strset.Set{
 	"linux-kernel-cataloger": strset.New("relationships"),
 }
 
+// TestCatalogersInSync ensures that all catalogers from the syft binary are documented in packages.yaml
+// and vice versa, and that all capability fields are properly filled without TODOs or null values.
 func TestCatalogersInSync(t *testing.T) {
 	// get canonical list from syft binary
 	catalogersInBinary := getCatalogerNamesFromBinary(t)
@@ -141,6 +143,8 @@ func validateCapabilitiesFilled(t *testing.T, catalogers []capabilities.Cataloge
 	}
 }
 
+// TestPackageTypeCoverage ensures that every package type defined in pkg.AllPkgs is represented in at least
+// one cataloger's capabilities, preventing orphaned package types that are defined but never documented.
 func TestPackageTypeCoverage(t *testing.T) {
 	// load catalogers from embedded YAML
 	catalogerEntries, err := capabilities.Packages()
@@ -184,6 +188,8 @@ func TestPackageTypeCoverage(t *testing.T) {
 		missingTypesWithoutExceptions)
 }
 
+// TestMetadataTypeCoverage ensures that every metadata type defined in packagemetadata.AllTypes() is represented
+// in at least one cataloger's capabilities, preventing orphaned metadata types that are defined but never produced.
 func TestMetadataTypeCoverage(t *testing.T) {
 	// load catalogers from embedded YAML
 	catalogerEntries, err := capabilities.Packages()
@@ -231,6 +237,9 @@ func TestMetadataTypeCoverage(t *testing.T) {
 		missingTypesWithoutExceptions)
 }
 
+// TestCatalogerStructure validates that catalogers follow structural conventions: generic catalogers must have
+// parsers and parser-level capabilities, custom catalogers must have detectors and cataloger-level capabilities,
+// and all catalogers must have an ecosystem set.
 func TestCatalogerStructure(t *testing.T) {
 	// load catalogers from embedded YAML
 	catalogerEntries, err := capabilities.Packages()
@@ -269,6 +278,8 @@ func TestCatalogerStructure(t *testing.T) {
 	}
 }
 
+// TestCatalogerDataQuality checks for data integrity issues in packages.yaml, including duplicate cataloger
+// names, duplicate parser functions within catalogers, and validates that detector definitions are well-formed.
 func TestCatalogerDataQuality(t *testing.T) {
 	// load catalogers from embedded YAML
 	catalogerEntries, err := capabilities.Packages()
@@ -345,7 +356,8 @@ func TestCatalogerDataQuality(t *testing.T) {
 	})
 }
 
-// TestCapabilitiesAreUpToDate verifies that regeneration runs successfully
+// TestCapabilitiesAreUpToDate verifies that packages.yaml is up to date by running regeneration and checking
+// for uncommitted changes. This test only runs in CI to catch cases where code changed but capabilities weren't regenerated.
 func TestCapabilitiesAreUpToDate(t *testing.T) {
 	if os.Getenv("CI") == "" {
 		t.Skip("skipping regeneration test in local environment")
@@ -367,8 +379,9 @@ func TestCapabilitiesAreUpToDate(t *testing.T) {
 	require.NoError(t, err, "packages.yaml has uncommitted changes after regeneration. Run 'go generate ./internal/capabilities' locally and commit the changes.")
 }
 
-// TestCatalogersHaveTestObservations verifies that all catalogers have test observations,
+// TestCatalogersHaveTestObservations ensures that all custom catalogers (and optionally parsers) have
-// ensuring they are using the pkgtest helpers
+// test observations recorded in test-fixtures/test-observations.json, which proves they are using the
+// pkgtest.CatalogTester helpers and have test coverage.
 func TestCatalogersHaveTestObservations(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -486,6 +499,9 @@ func extractPackageName(catalogerName string) string {
 	return catalogerName
 }
 
+// TestConfigCompleteness validates the integrity of config references in packages.yaml, ensuring that all
+// configs in the configs section are referenced by at least one cataloger, all cataloger config references exist,
+// and all app-key references in config fields exist in the application section.
 func TestConfigCompleteness(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -543,6 +559,8 @@ func TestConfigCompleteness(t *testing.T) {
 	}
 }
 
+// TestAppConfigFieldsHaveDescriptions ensures that all application config fields discovered from the
+// options package have descriptions, which are required for user-facing documentation.
 func TestAppConfigFieldsHaveDescriptions(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -561,6 +579,8 @@ func TestAppConfigFieldsHaveDescriptions(t *testing.T) {
 	require.Empty(t, missingDescriptions, "the following configs are missing descriptions: %v", missingDescriptions)
 }
 
+// TestAppConfigKeyFormat validates that all application config keys follow the expected naming convention
+// of "ecosystem.field-name" using kebab-case (lowercase with hyphens, no underscores or spaces).
 func TestAppConfigKeyFormat(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -579,8 +599,9 @@ func TestAppConfigKeyFormat(t *testing.T) {
 	}
 }
 
-// TestCapabilityConfigFieldReferences validates that config field names referenced in CapabilitiesV2
+// TestCapabilityConfigFieldReferences validates that config field names referenced in capability conditions
-// conditions actually exist in the cataloger's config struct
+// actually exist in the cataloger's config struct, preventing typos and ensuring capability conditions can
+// be properly evaluated at runtime.
 func TestCapabilityConfigFieldReferences(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -697,7 +718,9 @@ func TestCapabilityConfigFieldReferences(t *testing.T) {
 	}
 }
 
-// TestCapabilityFieldNaming validates that capability field names follow known patterns
+// TestCapabilityFieldNaming validates that all capability field names follow known patterns
+// (e.g., "license", "dependency.depth", "package_manager.files.listing"), catching typos and ensuring
+// consistency across catalogers.
 func TestCapabilityFieldNaming(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -754,7 +777,9 @@ func TestCapabilityFieldNaming(t *testing.T) {
 	}
 }
 
-// TestCapabilityValueTypes validates that capability field values match expected types
+// TestCapabilityValueTypes validates that capability field values match their expected types based on the
+// field name (e.g., boolean fields like "license" must have bool values, array fields like "dependency.depth"
+// must have []string values), preventing type mismatches that would cause runtime errors.
 func TestCapabilityValueTypes(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -870,8 +895,9 @@ func validateCapabilityValueType(fieldPath string, value interface{}) error {
 	return nil
 }
 
-// TestMetadataTypesHaveJSONSchemaTypes validates that metadata_types and json_schema_types are synchronized
+// TestMetadataTypesHaveJSONSchemaTypes validates that metadata_types and json_schema_types arrays are synchronized
-// in packages.yaml - every metadata type should have a corresponding json_schema_type with correct conversion
+// in packages.yaml, ensuring every metadata type (e.g., "pkg.AlpmDBEntry") has a corresponding json_schema_type
+// (e.g., "AlpmDbEntry") with correct conversion, which is required for JSON schema generation.
 func TestMetadataTypesHaveJSONSchemaTypes(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -1230,7 +1256,8 @@ func validateFieldPath(repoRoot, structName string, fieldPath []string) error {
 }
 
 // TestCapabilityEvidenceFieldReferences validates that evidence field references in capabilities
-// actually exist on their corresponding metadata structs
+// (e.g., "AlpmDBEntry.Files[].Digests") actually exist on their corresponding metadata structs by using
+// AST parsing to verify the field paths, preventing broken references when structs are refactored.
 func TestCapabilityEvidenceFieldReferences(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)
@@ -1305,8 +1332,9 @@ func TestCapabilityEvidenceFieldReferences(t *testing.T) {
 	}
 }
 
-// TestDetectorConfigFieldReferences validates that config field names referenced in detector
+// TestDetectorConfigFieldReferences validates that config field names referenced in detector conditions
-// conditions actually exist in the cataloger's config struct
+// actually exist in the cataloger's config struct, ensuring that conditional detectors can properly
+// evaluate their activation conditions based on configuration.
 func TestDetectorConfigFieldReferences(t *testing.T) {
 	repoRoot, err := RepoRoot()
 	require.NoError(t, err)

internal/capabilities/generate/discover_app_config.go

@@ -156,25 +156,25 @@ func extractConfigStructTypes(filePath string) ([]string, error) {
 // 1. Finding files with cataloger imports in options directory
 // 2. Extracting ecosystem config fields from Catalog struct
 // 3. Matching file structs against Catalog fields
-// Returns a map of file path to top-level YAML key
+// Returns a map of file path to top-level YAML key and a reverse lookup map of YAML key to struct name
-func discoverCatalogerConfigs(repoRoot string) (map[string]string, error) {
+func discoverCatalogerConfigs(repoRoot string) (map[string]string, map[string]string, error) {
 	optionsDir := filepath.Join(repoRoot, "cmd", "syft", "internal", "options")
 	catalogFilePath := filepath.Join(optionsDir, "catalog.go")
 
 	// get ecosystem config fields from Catalog struct
 	ecosystemConfigs, err := extractEcosystemConfigFieldsFromCatalog(catalogFilePath)
 	if err != nil {
-		return nil, err
+		return nil, nil, err
 	}
 
 	if len(ecosystemConfigs) == 0 {
-		return nil, fmt.Errorf("no ecosystem config fields found in Catalog struct")
+		return nil, nil, fmt.Errorf("no ecosystem config fields found in Catalog struct")
 	}
 
 	// find files with cataloger imports
 	candidateFiles, err := findFilesWithCatalogerImports(optionsDir)
 	if err != nil {
-		return nil, err
+		return nil, nil, err
 	}
 
 	// match candidate files against Catalog ecosystem fields
@@ -184,7 +184,7 @@ func discoverCatalogerConfigs(repoRoot string) (map[string]string, error) {
 	for _, filePath := range candidateFiles {
 		structTypes, err := extractConfigStructTypes(filePath)
 		if err != nil {
-			return nil, err
+			return nil, nil, err
 		}
 
 		// check if any struct type matches an ecosystem config
@@ -207,17 +207,23 @@ func discoverCatalogerConfigs(repoRoot string) (map[string]string, error) {
 
 	if len(missingConfigs) > 0 {
 		sort.Strings(missingConfigs)
-		return nil, fmt.Errorf("could not find files for ecosystem configs: %s", strings.Join(missingConfigs, ", "))
+		return nil, nil, fmt.Errorf("could not find files for ecosystem configs: %s", strings.Join(missingConfigs, ", "))
 	}
 
-	return fileToKey, nil
+	// build reverse lookup map (yamlKey -> structName)
+	keyToStruct := make(map[string]string)
+	for structName, yamlKey := range ecosystemConfigs {
+		keyToStruct[yamlKey] = structName
+	}
+
+	return fileToKey, keyToStruct, nil
 }
 
 // DiscoverAppConfigs discovers all application-level cataloger configuration fields
 // from the options package
 func DiscoverAppConfigs(repoRoot string) ([]AppConfigField, error) {
 	// discover cataloger config files dynamically
-	configFiles, err := discoverCatalogerConfigs(repoRoot)
+	configFiles, keyToStruct, err := discoverCatalogerConfigs(repoRoot)
 	if err != nil {
 		return nil, fmt.Errorf("failed to discover cataloger configs: %w", err)
 	}
@@ -225,7 +231,7 @@ func DiscoverAppConfigs(repoRoot string) ([]AppConfigField, error) {
 	// extract configuration fields from each discovered file
 	var configs []AppConfigField
 	for filePath, topLevelKey := range configFiles {
-		fields, err := extractAppConfigFields(filePath, topLevelKey)
+		fields, err := extractAppConfigFields(filePath, topLevelKey, keyToStruct)
 		if err != nil {
 			return nil, fmt.Errorf("failed to extract config from %s: %w", filePath, err)
 		}
@@ -241,7 +247,7 @@ func DiscoverAppConfigs(repoRoot string) ([]AppConfigField, error) {
 }
 
 // extractAppConfigFields extracts config fields from an options file
-func extractAppConfigFields(filePath, topLevelKey string) ([]AppConfigField, error) {
+func extractAppConfigFields(filePath, topLevelKey string, keyToStruct map[string]string) ([]AppConfigField, error) {
 	fset := token.NewFileSet()
 	f, err := parser.ParseFile(fset, filePath, nil, parser.ParseComments)
 	if err != nil {
@@ -251,7 +257,7 @@ func extractAppConfigFields(filePath, topLevelKey string) ([]AppConfigField, err
 	var configs []AppConfigField
 
 	// find the main config struct (not nested ones)
-	configStruct, descriptions := findAppConfigStructAndDescriptions(f, topLevelKey)
+	configStruct, descriptions := findAppConfigStructAndDescriptions(f, topLevelKey, keyToStruct)
 	if configStruct == nil {
 		return nil, fmt.Errorf("no config struct found in %s", filePath)
 	}
@@ -302,26 +308,13 @@ func extractAppConfigFields(filePath, topLevelKey string) ([]AppConfigField, err
 
 // findAppConfigStructAndDescriptions finds the main config struct and extracts field descriptions
 // from the DescribeFields method
-func findAppConfigStructAndDescriptions(f *ast.File, topLevelKey string) (*ast.StructType, map[string]string) {
+func findAppConfigStructAndDescriptions(f *ast.File, topLevelKey string, keyToStruct map[string]string) (*ast.StructType, map[string]string) {
-	expectedName := determineExpectedConfigName(topLevelKey)
+	structName := keyToStruct[topLevelKey]
-	configStruct := findConfigStruct(f, expectedName)
+	configStruct := findConfigStruct(f, structName)
 	descriptions := extractDescriptionsFromDescribeFields(f)
 	return configStruct, descriptions
 }
 
-// determineExpectedConfigName maps the top-level key to the expected config struct name
-func determineExpectedConfigName(topLevelKey string) string {
-	// handle special cases first
-	switch topLevelKey {
-	case "linux-kernel":
-		return "linuxKernelConfig"
-	case "javascript":
-		return "javaScriptConfig"
-	default:
-		return topLevelKey + "Config"
-	}
-}
-
 // findConfigStruct searches for the config struct with the expected name in the AST
 func findConfigStruct(f *ast.File, expectedName string) *ast.StructType {
 	for _, decl := range f.Decls {

internal/capabilities/generate/discover_app_config_test.go

@@ -9,52 +9,6 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-func TestDetermineExpectedConfigName(t *testing.T) {
-	tests := []struct {
-		name        string
-		topLevelKey string
-		wantName    string
-	}{
-		{
-			name:        "linux-kernel special case",
-			topLevelKey: "linux-kernel",
-			wantName:    "linuxKernelConfig",
-		},
-		{
-			name:        "javascript special case",
-			topLevelKey: "javascript",
-			wantName:    "javaScriptConfig",
-		},
-		{
-			name:        "standard config golang",
-			topLevelKey: "golang",
-			wantName:    "golangConfig",
-		},
-		{
-			name:        "standard config python",
-			topLevelKey: "python",
-			wantName:    "pythonConfig",
-		},
-		{
-			name:        "standard config java",
-			topLevelKey: "java",
-			wantName:    "javaConfig",
-		},
-		{
-			name:        "standard config dotnet",
-			topLevelKey: "dotnet",
-			wantName:    "dotnetConfig",
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			got := determineExpectedConfigName(tt.topLevelKey)
-			require.Equal(t, tt.wantName, got)
-		})
-	}
-}
-
 func TestCleanDescription(t *testing.T) {
 	tests := []struct {
 		name string

internal/capabilities/generate/discover_cataloger_configs.go

@@ -33,8 +33,13 @@ var appConfigAnnotationPattern = regexp.MustCompile(`^//\s*app-config:\s*(.+)$`)
 // Returns map where key is "packageName.StructName" (e.g., "golang.CatalogerConfig")
 func DiscoverConfigs(repoRoot string) (map[string]ConfigInfo, error) {
 	catalogerRoot := filepath.Join(repoRoot, "syft", "pkg", "cataloger")
+	return DiscoverConfigsFromPath(catalogerRoot)
+}
 
-	// find all .go files under syft/pkg/cataloger/ recursively
+// DiscoverConfigsFromPath walks the given directory and discovers all configuration structs
+// Returns map where key is "packageName.StructName" (e.g., "golang.CatalogerConfig")
+func DiscoverConfigsFromPath(catalogerRoot string) (map[string]ConfigInfo, error) {
+	// find all .go files under the directory recursively
 	var files []string
 	err := filepath.Walk(catalogerRoot, func(path string, info os.FileInfo, err error) error {
 		if err != nil {
@@ -52,7 +57,7 @@ func DiscoverConfigs(repoRoot string) (map[string]ConfigInfo, error) {
 	discovered := make(map[string]ConfigInfo)
 
 	for _, file := range files {
-		configs, err := discoverConfigsInFile(file, repoRoot)
+		configs, err := discoverConfigsInFile(file)
 		if err != nil {
 			return nil, fmt.Errorf("failed to parse %s: %w", file, err)
 		}
@@ -68,19 +73,15 @@ func DiscoverConfigs(repoRoot string) (map[string]ConfigInfo, error) {
 	return discovered, nil
 }
 
-func discoverConfigsInFile(path, repoRoot string) (map[string]ConfigInfo, error) {
+func discoverConfigsInFile(path string) (map[string]ConfigInfo, error) {
 	fset := token.NewFileSet()
 	f, err := parser.ParseFile(fset, path, nil, parser.ParseComments)
 	if err != nil {
 		return nil, err
 	}
 
-	// extract package name from file path
+	// extract package name from file path (use absolute path, not relative)
-	relPath, err := filepath.Rel(repoRoot, path)
+	packageName := extractPackageNameFromPath(path)
-	if err != nil {
-		relPath = path
-	}
-	packageName := extractPackageNameFromPath(relPath)
 	if packageName == "" {
 		return nil, nil
 	}

internal/capabilities/generate/discover_cataloger_configs_test.go

@@ -2,129 +2,131 @@ package main
 
 import (
 	"go/ast"
+	"path/filepath"
 	"testing"
 
-	"github.com/google/go-cmp/cmp"
 	"github.com/stretchr/testify/require"
 )
 
-// expected config structs that should be discovered with app-config annotations
+func TestDiscoverConfigs(t *testing.T) {
-var expectedCatalogConfigs = []string{
+	tests := []struct {
+		name            string
+		fixturePath     string
+		expectedConfigs []string
+		verifyConfig    func(t *testing.T, configs map[string]ConfigInfo)
+	}{
+		{
+			name:        "simple config with annotations",
+			fixturePath: "simple-config",
+			expectedConfigs: []string{
+				"golang.CatalogerConfig",
+			},
+			verifyConfig: func(t *testing.T, configs map[string]ConfigInfo) {
+				golangConfig := configs["golang.CatalogerConfig"]
+				require.Equal(t, "golang", golangConfig.PackageName)
+				require.Equal(t, "CatalogerConfig", golangConfig.StructName)
+				require.Len(t, golangConfig.Fields, 3, "should have 3 annotated fields")
+
+				// verify specific field
+				var foundSearchLocalModCache bool
+				for _, field := range golangConfig.Fields {
+					if field.Name == "SearchLocalModCacheLicenses" {
+						foundSearchLocalModCache = true
+						require.Equal(t, "bool", field.Type)
+						require.Equal(t, "golang.search-local-mod-cache-licenses", field.AppKey)
+						require.Contains(t, field.Description, "searching for go package licenses")
+					}
+				}
+				require.True(t, foundSearchLocalModCache, "should find SearchLocalModCacheLicenses field")
+			},
+		},
+		{
+			name:        "nested config struct",
+			fixturePath: "nested-config",
+			expectedConfigs: []string{
 				"golang.CatalogerConfig",
 				"golang.MainModuleVersionConfig",
-	"java.ArchiveCatalogerConfig",
+			},
+			verifyConfig: func(t *testing.T, configs map[string]ConfigInfo) {
+				// verify main config
+				golangConfig := configs["golang.CatalogerConfig"]
+				require.Equal(t, "golang", golangConfig.PackageName)
+				require.Equal(t, "CatalogerConfig", golangConfig.StructName)
+
+				// verify nested config
+				mainModuleConfig := configs["golang.MainModuleVersionConfig"]
+				require.Equal(t, "golang", mainModuleConfig.PackageName)
+				require.Equal(t, "MainModuleVersionConfig", mainModuleConfig.StructName)
+				require.Len(t, mainModuleConfig.Fields, 2, "should have 2 annotated fields")
+
+				// check for specific nested field
+				var foundFromLDFlags bool
+				for _, field := range mainModuleConfig.Fields {
+					if field.Name == "FromLDFlags" {
+						foundFromLDFlags = true
+						require.Equal(t, "bool", field.Type)
+						require.Equal(t, "golang.main-module-version.from-ld-flags", field.AppKey)
+						require.Contains(t, field.Description, "extract version from LD flags")
+					}
+				}
+				require.True(t, foundFromLDFlags, "should find FromLDFlags field")
+			},
+		},
+		{
+			name:        "multiple configs in different packages",
+			fixturePath: "multiple-configs",
+			expectedConfigs: []string{
 				"python.CatalogerConfig",
-	"dotnet.CatalogerConfig",
+				"java.ArchiveCatalogerConfig",
-	"kernel.LinuxKernelCatalogerConfig",
+			},
-	"javascript.CatalogerConfig",
+			verifyConfig: func(t *testing.T, configs map[string]ConfigInfo) {
-	"nix.Config",
+				// verify python config
-}
+				pythonConfig := configs["python.CatalogerConfig"]
+				require.Equal(t, "python", pythonConfig.PackageName)
+				require.Len(t, pythonConfig.Fields, 1)
 
-func TestDiscoverConfigs(t *testing.T) {
+				// verify java config
-	repoRoot, err := RepoRoot()
+				javaConfig := configs["java.ArchiveCatalogerConfig"]
+				require.Equal(t, "java", javaConfig.PackageName)
+				require.Len(t, javaConfig.Fields, 1)
+			},
+		},
+		{
+			name:            "config without annotations",
+			fixturePath:     "no-annotations",
+			expectedConfigs: []string{},
+			verifyConfig: func(t *testing.T, configs map[string]ConfigInfo) {
+				// should not discover any configs without annotations
+				require.Empty(t, configs, "should not discover configs without app-config annotations")
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			fixtureDir := filepath.Join("testdata", "config-discovery", tt.fixturePath, "cataloger")
+			configs, err := DiscoverConfigsFromPath(fixtureDir)
 			require.NoError(t, err)
 
-	configs, err := DiscoverConfigs(repoRoot)
+			// Debug: log what was discovered
-	require.NoError(t, err)
+			t.Logf("Discovered %d configs:", len(configs))
+			for key, config := range configs {
+				t.Logf("  %s: %d fields", key, len(config.Fields))
+			}
 
-	// verify we discovered multiple config structs
+			// verify expected configs were discovered
-	require.NotEmpty(t, configs, "should discover at least one config struct")
+			for _, expected := range tt.expectedConfigs {
-
-	// check for known config structs that have app-config annotations
-	for _, expected := range expectedCatalogConfigs {
 				config, ok := configs[expected]
 				require.True(t, ok, "should discover config: %s", expected)
 				require.NotEmpty(t, config.Fields, "config %s should have fields", expected)
 				require.Equal(t, expected, config.PackageName+"."+config.StructName)
 			}
 
-	// verify golang.CatalogerConfig structure
+			// run custom verification
-	golangConfig := configs["golang.CatalogerConfig"]
+			if tt.verifyConfig != nil {
-	wantGolangConfig := ConfigInfo{
+				tt.verifyConfig(t, configs)
-		PackageName: "golang",
-		StructName:  "CatalogerConfig",
-	}
-	if diff := cmp.Diff(wantGolangConfig.PackageName, golangConfig.PackageName); diff != "" {
-		t.Errorf("golang.CatalogerConfig.PackageName mismatch (-want +got):\n%s", diff)
-	}
-	if diff := cmp.Diff(wantGolangConfig.StructName, golangConfig.StructName); diff != "" {
-		t.Errorf("golang.CatalogerConfig.StructName mismatch (-want +got):\n%s", diff)
-	}
-	require.NotEmpty(t, golangConfig.Fields)
-
-	// check for specific field
-	var foundSearchLocalModCache bool
-	for _, field := range golangConfig.Fields {
-		if field.Name == "SearchLocalModCacheLicenses" {
-			foundSearchLocalModCache = true
-			wantField := ConfigField{
-				Name:   "SearchLocalModCacheLicenses",
-				Type:   "bool",
-				AppKey: "golang.search-local-mod-cache-licenses",
-			}
-			if diff := cmp.Diff(wantField.Name, field.Name); diff != "" {
-				t.Errorf("SearchLocalModCacheLicenses field Name mismatch (-want +got):\n%s", diff)
-			}
-			if diff := cmp.Diff(wantField.Type, field.Type); diff != "" {
-				t.Errorf("SearchLocalModCacheLicenses field Type mismatch (-want +got):\n%s", diff)
-			}
-			if diff := cmp.Diff(wantField.AppKey, field.AppKey); diff != "" {
-				t.Errorf("SearchLocalModCacheLicenses field AppKey mismatch (-want +got):\n%s", diff)
-			}
-			require.NotEmpty(t, field.Description)
-			require.Contains(t, field.Description, "searching for go package licenses")
-		}
-	}
-	require.True(t, foundSearchLocalModCache, "should find SearchLocalModCacheLicenses field")
-
-	// verify nested config struct
-	golangMainModuleConfig := configs["golang.MainModuleVersionConfig"]
-	wantMainModuleConfig := ConfigInfo{
-		PackageName: "golang",
-		StructName:  "MainModuleVersionConfig",
-	}
-	if diff := cmp.Diff(wantMainModuleConfig.PackageName, golangMainModuleConfig.PackageName); diff != "" {
-		t.Errorf("golang.MainModuleVersionConfig.PackageName mismatch (-want +got):\n%s", diff)
-	}
-	if diff := cmp.Diff(wantMainModuleConfig.StructName, golangMainModuleConfig.StructName); diff != "" {
-		t.Errorf("golang.MainModuleVersionConfig.StructName mismatch (-want +got):\n%s", diff)
-	}
-	require.NotEmpty(t, golangMainModuleConfig.Fields)
-
-	// check for specific nested field
-	var foundFromLDFlags bool
-	for _, field := range golangMainModuleConfig.Fields {
-		if field.Name == "FromLDFlags" {
-			foundFromLDFlags = true
-			wantField := ConfigField{
-				Name:   "FromLDFlags",
-				Type:   "bool",
-				AppKey: "golang.main-module-version.from-ld-flags",
-			}
-			if diff := cmp.Diff(wantField.Name, field.Name); diff != "" {
-				t.Errorf("FromLDFlags field Name mismatch (-want +got):\n%s", diff)
-			}
-			if diff := cmp.Diff(wantField.Type, field.Type); diff != "" {
-				t.Errorf("FromLDFlags field Type mismatch (-want +got):\n%s", diff)
-			}
-			if diff := cmp.Diff(wantField.AppKey, field.AppKey); diff != "" {
-				t.Errorf("FromLDFlags field AppKey mismatch (-want +got):\n%s", diff)
-			}
-			require.NotEmpty(t, field.Description)
-		}
-	}
-	require.True(t, foundFromLDFlags, "should find FromLDFlags field in MainModuleVersionConfig")
-
-	// print summary for manual inspection
-	t.Logf("Discovered %d config structs:", len(configs))
-	for key, config := range configs {
-		t.Logf("  %s: %d fields", key, len(config.Fields))
-		for _, field := range config.Fields {
-			t.Logf("    - %s (%s): %s", field.Name, field.Type, field.AppKey)
-			if diff := cmp.Diff("", field.Description); diff == "" {
-				t.Logf("      WARNING: field %s has no description", field.Name)
-			}
 			}
+		})
 	}
 }
 

internal/capabilities/generate/discover_catalogers.go

@@ -215,14 +215,20 @@ func parseGenericCatalogerFunction(funcDecl *ast.FuncDecl, filePath, repoRoot st
 func extractPackageNameFromPath(filePath string) string {
 	parts := strings.Split(filePath, string(filepath.Separator))
 
-	// find the index of "cataloger" in the path
+	// find the LAST occurrence of "cataloger" in the path
+	// (to handle test fixtures with multiple "cataloger" segments)
+	lastCatalogerIndex := -1
 	for i, part := range parts {
-		if part == "cataloger" && i+1 < len(parts) {
+		if part == "cataloger" {
-			// return the next segment after "cataloger"
+			lastCatalogerIndex = i
-			return parts[i+1]
 		}
 	}
 
+	if lastCatalogerIndex != -1 && lastCatalogerIndex+1 < len(parts) {
+		// return the next segment after the last "cataloger"
+		return parts[lastCatalogerIndex+1]
+	}
+
 	return ""
 }
 

internal/capabilities/generate/discover_catalogers_test.go

@@ -9,52 +9,6 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-// test helper functions
-
-// parseFuncDecl parses a function declaration from a code string
-func parseFuncDecl(t *testing.T, code string) *ast.FuncDecl {
-	t.Helper()
-	fset := token.NewFileSet()
-	file, err := parser.ParseFile(fset, "", "package test\n"+code, 0)
-	require.NoError(t, err)
-	require.Len(t, file.Decls, 1, "expected exactly one declaration")
-	funcDecl, ok := file.Decls[0].(*ast.FuncDecl)
-	require.True(t, ok, "expected declaration to be a function")
-	return funcDecl
-}
-
-// parseCallExpr parses a call expression from a code string
-func parseCallExpr(t *testing.T, code string) *ast.CallExpr {
-	t.Helper()
-	expr, err := parser.ParseExpr(code)
-	require.NoError(t, err)
-	callExpr, ok := expr.(*ast.CallExpr)
-	require.True(t, ok, "expected expression to be a call expression")
-	return callExpr
-}
-
-// parseCompositeLit parses a composite literal from a code string
-func parseCompositeLit(t *testing.T, code string) *ast.CompositeLit {
-	t.Helper()
-	expr, err := parser.ParseExpr(code)
-	require.NoError(t, err)
-	lit, ok := expr.(*ast.CompositeLit)
-	require.True(t, ok, "expected expression to be a composite literal")
-	return lit
-}
-
-// parseConstDecl parses a const declaration from a code string and returns the GenDecl
-func parseConstDecl(t *testing.T, code string) *ast.GenDecl {
-	t.Helper()
-	fset := token.NewFileSet()
-	file, err := parser.ParseFile(fset, "", "package test\n"+code, 0)
-	require.NoError(t, err)
-	require.Len(t, file.Decls, 1, "expected exactly one declaration")
-	genDecl, ok := file.Decls[0].(*ast.GenDecl)
-	require.True(t, ok, "expected declaration to be a general declaration")
-	return genDecl
-}
-
 func TestReturnsPackageCataloger(t *testing.T) {
 	tests := []struct {
 		name string
@@ -387,3 +341,49 @@ func TestResolveImportPath(t *testing.T) {
 		})
 	}
 }
+
+// test helper functions
+
+// parseFuncDecl parses a function declaration from a code string
+func parseFuncDecl(t *testing.T, code string) *ast.FuncDecl {
+	t.Helper()
+	fset := token.NewFileSet()
+	file, err := parser.ParseFile(fset, "", "package test\n"+code, 0)
+	require.NoError(t, err)
+	require.Len(t, file.Decls, 1, "expected exactly one declaration")
+	funcDecl, ok := file.Decls[0].(*ast.FuncDecl)
+	require.True(t, ok, "expected declaration to be a function")
+	return funcDecl
+}
+
+// parseCallExpr parses a call expression from a code string
+func parseCallExpr(t *testing.T, code string) *ast.CallExpr {
+	t.Helper()
+	expr, err := parser.ParseExpr(code)
+	require.NoError(t, err)
+	callExpr, ok := expr.(*ast.CallExpr)
+	require.True(t, ok, "expected expression to be a call expression")
+	return callExpr
+}
+
+// parseCompositeLit parses a composite literal from a code string
+func parseCompositeLit(t *testing.T, code string) *ast.CompositeLit {
+	t.Helper()
+	expr, err := parser.ParseExpr(code)
+	require.NoError(t, err)
+	lit, ok := expr.(*ast.CompositeLit)
+	require.True(t, ok, "expected expression to be a composite literal")
+	return lit
+}
+
+// parseConstDecl parses a const declaration from a code string and returns the GenDecl
+func parseConstDecl(t *testing.T, code string) *ast.GenDecl {
+	t.Helper()
+	fset := token.NewFileSet()
+	file, err := parser.ParseFile(fset, "", "package test\n"+code, 0)
+	require.NoError(t, err)
+	require.Len(t, file.Decls, 1, "expected exactly one declaration")
+	genDecl, ok := file.Decls[0].(*ast.GenDecl)
+	require.True(t, ok, "expected declaration to be a general declaration")
+	return genDecl
+}

internal/capabilities/generate/test-fixtures/config-linking/conflicting-names/duplicate/cataloger1.go

@@ -1,19 +0,0 @@
-package duplicate
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-type Config1 struct {
-	Option1 bool
-}
-
-func NewDuplicateCataloger1(cfg Config1) pkg.Cataloger {
-	return generic.NewCataloger("duplicate-cataloger").
-		WithParserByGlobs(parse1, "**/*.txt")
-}
-
-func parse1(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/conflicting-names/duplicate/cataloger2.go

@@ -1,19 +0,0 @@
-package duplicate
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-type Config2 struct {
-	Option2 string
-}
-
-func NewDuplicateCataloger2(cfg Config2) pkg.Cataloger {
-	return generic.NewCataloger("duplicate-cataloger").
-		WithParserByGlobs(parse2, "**/*.json")
-}
-
-func parse2(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/imported-config-type/kernel/cataloger.go

@@ -1,15 +0,0 @@
-package kernel
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-func NewLinuxKernelCataloger(cfg LinuxKernelCatalogerConfig) pkg.Cataloger {
-	return generic.NewCataloger("linux-kernel-cataloger").
-		WithParserByGlobs(parse, "**/vmlinuz")
-}
-
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/mixed-naming-patterns/ruby/cataloger.go

@@ -1,19 +0,0 @@
-package ruby
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-type Config struct {
-	Setting bool
-}
-
-func NewRubyCataloger(opts Config) pkg.Cataloger {
-	return generic.NewCataloger("ruby-cataloger").
-		WithParserByGlobs(parse, "**/Gemfile")
-}
-
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/non-config-first-param/binary/cataloger.go

@@ -1,17 +0,0 @@
-package binary
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-type Parser struct{}
-
-func NewBinaryCataloger(parser Parser) pkg.Cataloger {
-	return generic.NewCataloger("binary-cataloger").
-		WithParserByGlobs(parse, "**/*")
-}
-
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/selector-expression-config/rust/cataloger.go

@@ -1,16 +0,0 @@
-package rust
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-	"github.com/test/cargo"
-)
-
-func NewRustCataloger(cfg cargo.CatalogerConfig) pkg.Cataloger {
-	return generic.NewCataloger("rust-cataloger").
-		WithParserByGlobs(parse, "**/Cargo.toml")
-}
-
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/test-fixtures/config-linking/simple-generic-cataloger/golang/cataloger.go

@@ -1,19 +0,0 @@
-package golang
-
-import (
-	"github.com/anchore/syft/syft/pkg"
-	"github.com/anchore/syft/syft/pkg/cataloger/generic"
-)
-
-type CatalogerConfig struct {
-	SomeOption bool
-}
-
-func NewGoModuleCataloger(cfg CatalogerConfig) pkg.Cataloger {
-	return generic.NewCataloger("go-module-cataloger").
-		WithParserByGlobs(parseGoMod, "**/go.mod")
-}
-
-func parseGoMod(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
-	return nil, nil, nil
-}

internal/capabilities/generate/testdata/cataloger/cataloger-with-constant/cataloger/python/cataloger.go

@@ -1,6 +1,10 @@
 package python
 
 import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/anchore/syft/syft/pkg/cataloger/generic"
 )
@@ -11,11 +15,11 @@ type CatalogerConfig struct {
 	Setting string
 }
 
-func NewPythonCataloger(cfg CatalogerConfig) pkg.Cataloger {
+func NewPythonCataloger(_ CatalogerConfig) pkg.Cataloger {
 	return generic.NewCataloger(catalogerName).
 		WithParserByGlobs(parse, "**/*.py")
 }
 
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 	return nil, nil, nil
 }

internal/capabilities/generate/testdata/cataloger/conflicting-names/cataloger/duplicate1/cataloger.go

@@ -0,0 +1,23 @@
+package duplicate1
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+type Config struct {
+	Option1 bool
+}
+
+func NewDuplicateCataloger(_ Config) pkg.Cataloger {
+	return generic.NewCataloger("duplicate-cataloger").
+		WithParserByGlobs(parse, "**/*.txt")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/conflicting-names/cataloger/duplicate2/cataloger.go

@@ -0,0 +1,23 @@
+package duplicate2
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+type Config struct {
+	Option2 string
+}
+
+func NewDuplicateCataloger(_ Config) pkg.Cataloger {
+	return generic.NewCataloger("duplicate-cataloger").
+		WithParserByGlobs(parse, "**/*.json")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/custom-cataloger-different-file/cataloger/dotnet/types.go

@@ -1,6 +1,10 @@
 package dotnet
 
 import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/pkg"
 )
 
@@ -18,6 +22,6 @@ func (d dotnetCataloger) Name() string {
 	return catalogerName
 }
 
-func (d dotnetCataloger) Catalog(resolver any) ([]pkg.Package, []pkg.Relationship, error) {
+func (d dotnetCataloger) Catalog(_ context.Context, _ file.Resolver) ([]pkg.Package, []artifact.Relationship, error) {
 	return nil, nil, nil
 }

internal/capabilities/generate/testdata/cataloger/custom-cataloger-same-file/cataloger/java/cataloger.go

@@ -1,6 +1,10 @@
 package java
 
 import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/pkg"
 )
 
@@ -18,7 +22,7 @@ func (p pomXMLCataloger) Name() string {
 	return pomCatalogerName
 }
 
-func (p pomXMLCataloger) Catalog(resolver any) ([]pkg.Package, []pkg.Relationship, error) {
+func (p pomXMLCataloger) Catalog(_ context.Context, _ file.Resolver) ([]pkg.Package, []artifact.Relationship, error) {
 	return nil, nil, nil
 }
 

internal/capabilities/generate/testdata/cataloger/imported-config-type/cataloger/kernel/cataloger.go

@@ -0,0 +1,19 @@
+package kernel
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+func NewLinuxKernelCataloger(_ LinuxKernelCatalogerConfig) pkg.Cataloger {
+	return generic.NewCataloger("linux-kernel-cataloger").
+		WithParserByGlobs(parse, "**/vmlinuz")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/mixed-naming-patterns/cataloger/ruby/cataloger.go

@@ -0,0 +1,23 @@
+package ruby
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+type Config struct {
+	Setting bool
+}
+
+func NewRubyCataloger(_ Config) pkg.Cataloger {
+	return generic.NewCataloger("ruby-cataloger").
+		WithParserByGlobs(parse, "**/Gemfile")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/no-config-cataloger/cataloger/javascript/cataloger.go

@@ -1,6 +1,10 @@
 package javascript
 
 import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/pkg"
 	"github.com/anchore/syft/syft/pkg/cataloger/generic"
 )
@@ -10,6 +14,6 @@ func NewJavaScriptCataloger() pkg.Cataloger {
 		WithParserByGlobs(parse, "**/*.js")
 }
 
-func parse(path string, reader any) ([]pkg.Package, []pkg.Relationship, error) {
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
 	return nil, nil, nil
 }

internal/capabilities/generate/testdata/cataloger/non-config-first-param/cataloger/binary/cataloger.go

@@ -0,0 +1,21 @@
+package binary
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+type Parser struct{}
+
+func NewBinaryCataloger(_ Parser) pkg.Cataloger {
+	return generic.NewCataloger("binary-cataloger").
+		WithParserByGlobs(parse, "**/*")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/selector-expression-config/cataloger/rust/cataloger.go

@@ -0,0 +1,24 @@
+package rust
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+// CatalogerConfig is imported from a selector expression in the real code
+type CatalogerConfig struct {
+	SomeOption bool
+}
+
+func NewRustCataloger(_ CatalogerConfig) pkg.Cataloger {
+	return generic.NewCataloger("rust-cataloger").
+		WithParserByGlobs(parse, "**/Cargo.toml")
+}
+
+func parse(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/cataloger/simple-generic-cataloger/cataloger/golang/cataloger.go

@@ -0,0 +1,23 @@
+package golang
+
+import (
+	"context"
+
+	"github.com/anchore/syft/syft/artifact"
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/pkg/cataloger/generic"
+)
+
+type CatalogerConfig struct {
+	SomeOption bool
+}
+
+func NewGoModuleCataloger(_ CatalogerConfig) pkg.Cataloger {
+	return generic.NewCataloger("go-module-cataloger").
+		WithParserByGlobs(parseGoMod, "**/go.mod")
+}
+
+func parseGoMod(_ context.Context, _ file.Resolver, _ *generic.Environment, _ file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
+	return nil, nil, nil
+}

internal/capabilities/generate/testdata/config-discovery/multiple-configs/cataloger/java/config.go

@@ -0,0 +1,8 @@
+package java
+
+// ArchiveCatalogerConfig contains configuration for the java archive cataloger
+type ArchiveCatalogerConfig struct {
+	// include archive contents in catalog
+	// app-config: java.use-network
+	UseNetwork bool
+}

internal/capabilities/generate/testdata/config-discovery/multiple-configs/cataloger/python/config.go

@@ -0,0 +1,8 @@
+package python
+
+// CatalogerConfig contains configuration for the python cataloger
+type CatalogerConfig struct {
+	// guess unpinned python package requirements
+	// app-config: python.guess-unpinned-requirements
+	GuessUnpinnedRequirements bool
+}

internal/capabilities/generate/testdata/config-discovery/nested-config/cataloger/golang/config.go

@@ -0,0 +1,22 @@
+package golang
+
+// MainModuleVersionConfig contains nested configuration for main module version detection
+type MainModuleVersionConfig struct {
+	// extract version from LD flags
+	// app-config: golang.main-module-version.from-ld-flags
+	FromLDFlags bool
+
+	// extract version from build info
+	// app-config: golang.main-module-version.from-build-info
+	FromBuildInfo bool
+}
+
+// CatalogerConfig contains configuration for the golang cataloger
+type CatalogerConfig struct {
+	// enable searching for go package licenses in the local mod cache
+	// app-config: golang.search-local-mod-cache-licenses
+	SearchLocalModCacheLicenses bool
+
+	// main module version configuration
+	MainModuleVersion MainModuleVersionConfig
+}

internal/capabilities/generate/testdata/config-discovery/no-annotations/cataloger/javascript/config.go

@@ -0,0 +1,10 @@
+package javascript
+
+// CatalogerConfig contains configuration for the javascript cataloger (no annotations)
+type CatalogerConfig struct {
+	// this field has no app-config annotation
+	SearchRemoteLicenses bool
+
+	// this field also has no annotation
+	IncludeDevDependencies bool
+}

internal/capabilities/generate/testdata/config-discovery/simple-config/cataloger/golang/config.go

@@ -0,0 +1,16 @@
+package golang
+
+// CatalogerConfig contains configuration for the golang cataloger
+type CatalogerConfig struct {
+	// enable searching for go package licenses in the local mod cache
+	// app-config: golang.search-local-mod-cache-licenses
+	SearchLocalModCacheLicenses bool
+
+	// base URL for npm registry
+	// app-config: golang.npm-base-url
+	NpmBaseURL string
+
+	// list of globs to search for go.mod files
+	// app-config: golang.search-patterns
+	SearchPatterns []string
+}