Merge pull request #374 from anchore/add-binary-classifier

Add file classification cataloger

README.md

@@ -108,6 +108,17 @@ package:
     # same as -s ; SYFT_PACKAGE_CATALOGER_SCOPE env var
     scope: "squashed"
 
+# cataloging file classifications is exposed through the power-user subcommand
+file-classification:
+  cataloger:
+    # enable/disable cataloging of file classifications
+    # SYFT_FILE_CLASSIFICATION_CATALOGER_ENABLED env var
+    enabled: true
+
+    # the search space to look for file classifications (options: all-layers, squashed)
+    # SYFT_FILE_CLASSIFICATION_CATALOGER_SCOPE env var
+    scope: "squashed"
+
 # cataloging file metadata is exposed through the power-user subcommand
 file-metadata:
   cataloger:

cmd/power_user_tasks.go

@@ -20,6 +20,7 @@ func powerUserTasks() ([]powerUserTask, error) {
 		catalogFileMetadataTask,
 		catalogFileDigestsTask,
 		catalogSecretsTask,
+		catalogFileClassificationsTask,
 	}
 
 	for _, generator := range generators {
@@ -156,3 +157,31 @@ func catalogSecretsTask() (powerUserTask, error) {
 
 	return task, nil
 }
+
+func catalogFileClassificationsTask() (powerUserTask, error) {
+	if !appConfig.FileClassification.Cataloger.Enabled {
+		return nil, nil
+	}
+
+	// TODO: in the future we could expose out the classifiers via configuration
+	classifierCataloger, err := file.NewClassificationCataloger(file.DefaultClassifiers)
+	if err != nil {
+		return nil, err
+	}
+
+	task := func(results *poweruser.JSONDocumentConfig, src source.Source) error {
+		resolver, err := src.FileResolver(appConfig.FileClassification.Cataloger.ScopeOpt)
+		if err != nil {
+			return err
+		}
+
+		result, err := classifierCataloger.Catalog(resolver)
+		if err != nil {
+			return err
+		}
+		results.FileClassifications = result
+		return nil
+	}
+
+	return task, nil
+}

internal/config/application.go

@@ -27,17 +27,18 @@ type parser interface {
 
 // Application is the main syft application configuration.
 type Application struct {
-	ConfigPath        string         `yaml:",omitempty" json:"configPath"`               // the location where the application config was read from (either from -c or discovered while loading)
-	Output            string         `yaml:"output" json:"output" mapstructure:"output"` // -o, the Presenter hint string to use for report formatting
-	Quiet             bool           `yaml:"quiet" json:"quiet" mapstructure:"quiet"`    // -q, indicates to not show any status output to stderr (ETUI or logging UI)
-	Log               logging        `yaml:"log" json:"log" mapstructure:"log"`          // all logging-related options
-	CliOptions        CliOnlyOptions `yaml:"-" json:"-"`                                 // all options only available through the CLI (not via env vars or config)
-	Dev               Development    `yaml:"dev" json:"dev" mapstructure:"dev"`
-	CheckForAppUpdate bool           `yaml:"check-for-app-update" json:"check-for-app-update" mapstructure:"check-for-app-update"` // whether to check for an application update on start up or not
-	Anchore           anchore        `yaml:"anchore" json:"anchore" mapstructure:"anchore"`                                        // options for interacting with Anchore Engine/Enterprise
-	Package           Packages       `yaml:"package" json:"package" mapstructure:"package"`
-	FileMetadata      FileMetadata   `yaml:"file-metadata" json:"file-metadata" mapstructure:"file-metadata"`
-	Secrets           Secrets        `yaml:"secrets" json:"secrets" mapstructure:"secrets"`
+	ConfigPath         string             `yaml:",omitempty" json:"configPath"`                                                         // the location where the application config was read from (either from -c or discovered while loading)
+	Output             string             `yaml:"output" json:"output" mapstructure:"output"`                                           // -o, the Presenter hint string to use for report formatting
+	Quiet              bool               `yaml:"quiet" json:"quiet" mapstructure:"quiet"`                                              // -q, indicates to not show any status output to stderr (ETUI or logging UI)
+	CheckForAppUpdate  bool               `yaml:"check-for-app-update" json:"check-for-app-update" mapstructure:"check-for-app-update"` // whether to check for an application update on start up or not
+	Anchore            anchore            `yaml:"anchore" json:"anchore" mapstructure:"anchore"`                                        // options for interacting with Anchore Engine/Enterprise
+	CliOptions         CliOnlyOptions     `yaml:"-" json:"-"`                                                                           // all options only available through the CLI (not via env vars or config)
+	Dev                development        `yaml:"dev" json:"dev" mapstructure:"dev"`
+	Log                logging            `yaml:"log" json:"log" mapstructure:"log"` // all logging-related options
+	Package            packages           `yaml:"package" json:"package" mapstructure:"package"`
+	FileMetadata       FileMetadata       `yaml:"file-metadata" json:"file-metadata" mapstructure:"file-metadata"`
+	FileClassification fileClassification `yaml:"file-classification" json:"file-classification" mapstructure:"file-classification"`
+	Secrets            secrets            `yaml:"secrets" json:"secrets" mapstructure:"secrets"`
 }
 
 func newApplicationConfig(v *viper.Viper, cliOpts CliOnlyOptions) *Application {

internal/config/development.go

@@ -2,12 +2,12 @@ package config
 
 import "github.com/spf13/viper"
 
-type Development struct {
+type development struct {
 	ProfileCPU bool `yaml:"profile-cpu" json:"profile-cpu" mapstructure:"profile-cpu"`
 	ProfileMem bool `yaml:"profile-mem" json:"profile-mem" mapstructure:"profile-mem"`
 }
 
-func (cfg Development) loadDefaultValues(v *viper.Viper) {
+func (cfg development) loadDefaultValues(v *viper.Viper) {
 	v.SetDefault("dev.profile-cpu", false)
 	v.SetDefault("dev.profile-mem", false)
 }

internal/config/file_classification.go

@@ -0,0 +1,19 @@
+package config
+
+import (
+	"github.com/anchore/syft/syft/source"
+	"github.com/spf13/viper"
+)
+
+type fileClassification struct {
+	Cataloger catalogerOptions `yaml:"cataloger" json:"cataloger" mapstructure:"cataloger"`
+}
+
+func (cfg fileClassification) loadDefaultValues(v *viper.Viper) {
+	v.SetDefault("file-classification.cataloger.enabled", true)
+	v.SetDefault("file-classification.cataloger.scope", source.SquashedScope)
+}
+
+func (cfg *fileClassification) parseConfigValues() error {
+	return cfg.Cataloger.parseConfigValues()
+}

internal/config/packages.go

@@ -2,14 +2,14 @@ package config
 
 import "github.com/spf13/viper"
 
-type Packages struct {
+type packages struct {
 	Cataloger catalogerOptions `yaml:"cataloger" json:"cataloger" mapstructure:"cataloger"`
 }
 
-func (cfg Packages) loadDefaultValues(v *viper.Viper) {
+func (cfg packages) loadDefaultValues(v *viper.Viper) {
 	v.SetDefault("package.cataloger.enabled", true)
 }
 
-func (cfg *Packages) parseConfigValues() error {
+func (cfg *packages) parseConfigValues() error {
 	return cfg.Cataloger.parseConfigValues()
 }

internal/config/secrets.go

@@ -6,7 +6,7 @@ import (
 	"github.com/spf13/viper"
 )
 
-type Secrets struct {
+type secrets struct {
 	Cataloger           catalogerOptions  `yaml:"cataloger" json:"cataloger" mapstructure:"cataloger"`
 	AdditionalPatterns  map[string]string `yaml:"additional-patterns" json:"additional-patterns" mapstructure:"additional-patterns"`
 	ExcludePatternNames []string          `yaml:"exclude-pattern-names" json:"exclude-pattern-names" mapstructure:"exclude-pattern-names"`
@@ -14,7 +14,7 @@ type Secrets struct {
 	SkipFilesAboveSize  int64             `yaml:"skip-files-above-size" json:"skip-files-above-size" mapstructure:"skip-files-above-size"`
 }
 
-func (cfg Secrets) loadDefaultValues(v *viper.Viper) {
+func (cfg secrets) loadDefaultValues(v *viper.Viper) {
 	v.SetDefault("secrets.cataloger.enabled", true)
 	v.SetDefault("secrets.cataloger.scope", source.AllLayersScope)
 	v.SetDefault("secrets.reveal-values", false)
@@ -23,6 +23,6 @@ func (cfg Secrets) loadDefaultValues(v *viper.Viper) {
 	v.SetDefault("secrets.exclude-pattern-names", []string{})
 }
 
-func (cfg *Secrets) parseConfigValues() error {
+func (cfg *secrets) parseConfigValues() error {
 	return cfg.Cataloger.parseConfigValues()
 }

internal/parse.go

@@ -1,15 +0,0 @@
-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
-}

internal/presenter/poweruser/json_document.go

@@ -9,8 +9,9 @@ type JSONDocument struct {
 	// here should be optional by supplying "omitempty" on these fields hint to the jsonschema generator to not
 	// require these fields. As an accepted rule in this repo all collections should still be initialized in the
 	// context of being used in a JSON document.
-	FileMetadata []JSONFileMetadata `json:"fileMetadata,omitempty"` // note: must have omitempty
-	Secrets      []JSONSecrets      `json:"secrets,omitempty"`      // note: must have omitempty
+	FileClassifications []JSONFileClassifications `json:"fileClassifications,omitempty"` // note: must have omitempty
+	FileMetadata        []JSONFileMetadata        `json:"fileMetadata,omitempty"`        // note: must have omitempty
+	Secrets             []JSONSecrets             `json:"secrets,omitempty"`             // note: must have omitempty
 	packages.JSONDocument
 }
 
@@ -27,8 +28,9 @@ func NewJSONDocument(config JSONDocumentConfig) (JSONDocument, error) {
 	}
 
 	return JSONDocument{
-		FileMetadata: fileMetadata,
-		Secrets:      NewJSONSecrets(config.Secrets),
-		JSONDocument: pkgsDoc,
+		FileClassifications: NewJSONFileClassifications(config.FileClassifications),
+		FileMetadata:        fileMetadata,
+		Secrets:             NewJSONSecrets(config.Secrets),
+		JSONDocument:        pkgsDoc,
 	}, nil
 }

internal/presenter/poweruser/json_document_config.go

@@ -9,11 +9,12 @@ import (
 )
 
 type JSONDocumentConfig struct {
-	ApplicationConfig config.Application
-	PackageCatalog    *pkg.Catalog
-	FileMetadata      map[source.Location]source.FileMetadata
-	FileDigests       map[source.Location][]file.Digest
-	Secrets           map[source.Location][]file.SearchResult
-	Distro            *distro.Distro
-	SourceMetadata    source.Metadata
+	ApplicationConfig   config.Application
+	PackageCatalog      *pkg.Catalog
+	FileMetadata        map[source.Location]source.FileMetadata
+	FileDigests         map[source.Location][]file.Digest
+	FileClassifications map[source.Location][]file.Classification
+	Secrets             map[source.Location][]file.SearchResult
+	Distro              *distro.Distro
+	SourceMetadata      source.Metadata
 }

internal/presenter/poweruser/json_file_classifications.go

@@ -0,0 +1,34 @@
+package poweruser
+
+import (
+	"sort"
+
+	"github.com/anchore/syft/syft/file"
+	"github.com/anchore/syft/syft/source"
+)
+
+type JSONFileClassifications struct {
+	Location       source.Location     `json:"location"`
+	Classification file.Classification `json:"classification"`
+}
+
+func NewJSONFileClassifications(data map[source.Location][]file.Classification) []JSONFileClassifications {
+	results := make([]JSONFileClassifications, 0)
+	for location, classifications := range data {
+		for _, classification := range classifications {
+			results = append(results, JSONFileClassifications{
+				Location:       location,
+				Classification: classification,
+			})
+		}
+	}
+
+	// sort by real path then virtual path to ensure the result is stable across multiple runs
+	sort.SliceStable(results, func(i, j int) bool {
+		if results[i].Location.RealPath == results[j].Location.RealPath {
+			return results[i].Location.VirtualPath < results[j].Location.VirtualPath
+		}
+		return results[i].Location.RealPath < results[j].Location.RealPath
+	})
+	return results
+}

internal/presenter/poweruser/test-fixtures/snapshot/TestJSONPresenter.golden

@@ -161,15 +161,6 @@
    "configPath": "",
    "output": "",
    "quiet": false,
-   "log": {
-    "structured": false,
-    "level": "",
-    "file-location": ""
-   },
-   "dev": {
-    "profile-cpu": false,
-    "profile-mem": false
-   },
    "check-for-app-update": false,
    "anchore": {
     "host": "",
@@ -177,6 +168,15 @@
     "dockerfile": "",
     "overwrite-existing-image": false
    },
+   "dev": {
+    "profile-cpu": false,
+    "profile-mem": false
+   },
+   "log": {
+    "structured": false,
+    "level": "",
+    "file-location": ""
+   },
    "package": {
     "cataloger": {
      "enabled": false,
@@ -192,6 +192,12 @@
      "sha256"
     ]
    },
+   "file-classification": {
+    "cataloger": {
+     "enabled": false,
+     "scope": ""
+    }
+   },
    "secrets": {
     "cataloger": {
      "enabled": false,

internal/regex_helpers.go

@@ -0,0 +1,45 @@
+package internal
+
+import "regexp"
+
+// MatchNamedCaptureGroups takes a regular expression and string and returns all of the named capture group results in a map.
+// This is only for the first match in the regex. Callers shouldn't be providing regexes with multiple capture groups with the same name.
+func MatchNamedCaptureGroups(regEx *regexp.Regexp, content string) map[string]string {
+	// note: we are looking across all matches and stopping on the first non-empty match. Why? Take the following example:
+	// input: "cool something to match against" pattern: `((?P<name>match) (?P<version>against))?`. Since the pattern is
+	// encapsulated in an optional capture group, there will be results for each character, but the results will match
+	// on nothing. The only "true" match will be at the end ("match against").
+	allMatches := regEx.FindAllStringSubmatch(content, -1)
+	var results map[string]string
+	for _, match := range allMatches {
+		// fill a candidate results map with named capture group results, accepting empty values, but not groups with
+		// no names
+		for nameIdx, name := range regEx.SubexpNames() {
+			if nameIdx > len(match) || len(name) == 0 {
+				continue
+			}
+			if results == nil {
+				results = make(map[string]string)
+			}
+			results[name] = match[nameIdx]
+		}
+		// note: since we are looking for the first best potential match we should stop when we find the first one
+		// with non-empty results.
+		if !isEmptyMap(results) {
+			break
+		}
+	}
+	return results
+}
+
+func isEmptyMap(m map[string]string) bool {
+	if len(m) == 0 {
+		return true
+	}
+	for _, value := range m {
+		if value != "" {
+			return false
+		}
+	}
+	return true
+}

internal/regex_helpers_test.go

@@ -0,0 +1,70 @@
+package internal
+
+import (
+	"regexp"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestMatchCaptureGroups(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		pattern  string
+		expected map[string]string
+	}{
+		{
+			name:    "go-case",
+			input:   "match this thing",
+			pattern: `(?P<name>match).*(?P<version>thing)`,
+			expected: map[string]string{
+				"name":    "match",
+				"version": "thing",
+			},
+		},
+		{
+			name:    "only matches the first instance",
+			input:   "match this thing batch another think",
+			pattern: `(?P<name>[mb]atch).*?(?P<version>thin[gk])`,
+			expected: map[string]string{
+				"name":    "match",
+				"version": "thing",
+			},
+		},
+		{
+			name:    "nested capture groups",
+			input:   "cool something to match against",
+			pattern: `((?P<name>match) (?P<version>against))`,
+			expected: map[string]string{
+				"name":    "match",
+				"version": "against",
+			},
+		},
+		{
+			name:    "nested optional capture groups",
+			input:   "cool something to match against",
+			pattern: `((?P<name>match) (?P<version>against))?`,
+			expected: map[string]string{
+				"name":    "match",
+				"version": "against",
+			},
+		},
+		{
+			name:    "nested optional capture groups with larger match",
+			input:   "cool something to match against match never",
+			pattern: `.*?((?P<name>match) (?P<version>(against|never)))?`,
+			expected: map[string]string{
+				"name":    "match",
+				"version": "against",
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			actual := MatchNamedCaptureGroups(regexp.MustCompile(test.pattern), test.input)
+			assert.Equal(t, test.expected, actual)
+		})
+	}
+}

schema/json/schema-1.0.6.json

@@ -0,0 +1,935 @@
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "$ref": "#/definitions/Document",
+  "definitions": {
+    "ApkFileRecord": {
+      "required": [
+        "path"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "ownerUid": {
+          "type": "string"
+        },
+        "ownerGid": {
+          "type": "string"
+        },
+        "permissions": {
+          "type": "string"
+        },
+        "checksum": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "ApkMetadata": {
+      "required": [
+        "package",
+        "originPackage",
+        "maintainer",
+        "version",
+        "license",
+        "architecture",
+        "url",
+        "description",
+        "size",
+        "installedSize",
+        "pullDependencies",
+        "pullChecksum",
+        "gitCommitOfApkPort",
+        "files"
+      ],
+      "properties": {
+        "package": {
+          "type": "string"
+        },
+        "originPackage": {
+          "type": "string"
+        },
+        "maintainer": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "license": {
+          "type": "string"
+        },
+        "architecture": {
+          "type": "string"
+        },
+        "url": {
+          "type": "string"
+        },
+        "description": {
+          "type": "string"
+        },
+        "size": {
+          "type": "integer"
+        },
+        "installedSize": {
+          "type": "integer"
+        },
+        "pullDependencies": {
+          "type": "string"
+        },
+        "pullChecksum": {
+          "type": "string"
+        },
+        "gitCommitOfApkPort": {
+          "type": "string"
+        },
+        "files": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/ApkFileRecord"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "CargoPackageMetadata": {
+      "required": [
+        "name",
+        "version",
+        "source",
+        "checksum",
+        "dependencies"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "source": {
+          "type": "string"
+        },
+        "checksum": {
+          "type": "string"
+        },
+        "dependencies": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Classification": {
+      "required": [
+        "class",
+        "metadata"
+      ],
+      "properties": {
+        "class": {
+          "type": "string"
+        },
+        "metadata": {
+          "patternProperties": {
+            ".*": {
+              "type": "string"
+            }
+          },
+          "type": "object"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Descriptor": {
+      "required": [
+        "name",
+        "version"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "configuration": {
+          "additionalProperties": true
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Digest": {
+      "required": [
+        "algorithm",
+        "value"
+      ],
+      "properties": {
+        "algorithm": {
+          "type": "string"
+        },
+        "value": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Distribution": {
+      "required": [
+        "name",
+        "version",
+        "idLike"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "idLike": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Document": {
+      "required": [
+        "artifacts",
+        "artifactRelationships",
+        "source",
+        "distro",
+        "descriptor",
+        "schema"
+      ],
+      "properties": {
+        "fileClassifications": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/FileClassifications"
+          },
+          "type": "array"
+        },
+        "fileMetadata": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/FileMetadata"
+          },
+          "type": "array"
+        },
+        "secrets": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/Secrets"
+          },
+          "type": "array"
+        },
+        "artifacts": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/Package"
+          },
+          "type": "array"
+        },
+        "artifactRelationships": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/Relationship"
+          },
+          "type": "array"
+        },
+        "source": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Source"
+        },
+        "distro": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Distribution"
+        },
+        "descriptor": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Descriptor"
+        },
+        "schema": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Schema"
+        },
+        "artifacts.metadata": {
+          "anyOf": [
+            {
+              "type": "null"
+            },
+            {
+              "$ref": "#/definitions/ApkMetadata"
+            },
+            {
+              "$ref": "#/definitions/CargoPackageMetadata"
+            },
+            {
+              "$ref": "#/definitions/DpkgMetadata"
+            },
+            {
+              "$ref": "#/definitions/GemMetadata"
+            },
+            {
+              "$ref": "#/definitions/JavaMetadata"
+            },
+            {
+              "$ref": "#/definitions/NpmPackageJSONMetadata"
+            },
+            {
+              "$ref": "#/definitions/PythonPackageMetadata"
+            },
+            {
+              "$ref": "#/definitions/RpmdbMetadata"
+            }
+          ]
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "DpkgFileRecord": {
+      "required": [
+        "path",
+        "md5"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "md5": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "DpkgMetadata": {
+      "required": [
+        "package",
+        "source",
+        "version",
+        "sourceVersion",
+        "architecture",
+        "maintainer",
+        "installedSize",
+        "files"
+      ],
+      "properties": {
+        "package": {
+          "type": "string"
+        },
+        "source": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "sourceVersion": {
+          "type": "string"
+        },
+        "architecture": {
+          "type": "string"
+        },
+        "maintainer": {
+          "type": "string"
+        },
+        "installedSize": {
+          "type": "integer"
+        },
+        "files": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/DpkgFileRecord"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "FileClassifications": {
+      "required": [
+        "location",
+        "classification"
+      ],
+      "properties": {
+        "location": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Location"
+        },
+        "classification": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/Classification"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "FileMetadata": {
+      "required": [
+        "location",
+        "metadata"
+      ],
+      "properties": {
+        "location": {
+          "$ref": "#/definitions/Location"
+        },
+        "metadata": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/FileMetadataEntry"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "FileMetadataEntry": {
+      "required": [
+        "mode",
+        "type",
+        "userID",
+        "groupID"
+      ],
+      "properties": {
+        "mode": {
+          "type": "integer"
+        },
+        "type": {
+          "type": "string"
+        },
+        "linkDestination": {
+          "type": "string"
+        },
+        "userID": {
+          "type": "integer"
+        },
+        "groupID": {
+          "type": "integer"
+        },
+        "digests": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/Digest"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "GemMetadata": {
+      "required": [
+        "name",
+        "version"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "files": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "authors": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "licenses": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "homepage": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "JavaManifest": {
+      "properties": {
+        "main": {
+          "patternProperties": {
+            ".*": {
+              "type": "string"
+            }
+          },
+          "type": "object"
+        },
+        "namedSections": {
+          "patternProperties": {
+            ".*": {
+              "patternProperties": {
+                ".*": {
+                  "type": "string"
+                }
+              },
+              "type": "object"
+            }
+          },
+          "type": "object"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "JavaMetadata": {
+      "required": [
+        "virtualPath"
+      ],
+      "properties": {
+        "virtualPath": {
+          "type": "string"
+        },
+        "manifest": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/JavaManifest"
+        },
+        "pomProperties": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/PomProperties"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Location": {
+      "required": [
+        "path"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "layerID": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "NpmPackageJSONMetadata": {
+      "required": [
+        "author",
+        "licenses",
+        "homepage",
+        "description",
+        "url"
+      ],
+      "properties": {
+        "files": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "author": {
+          "type": "string"
+        },
+        "licenses": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "homepage": {
+          "type": "string"
+        },
+        "description": {
+          "type": "string"
+        },
+        "url": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Package": {
+      "required": [
+        "id",
+        "name",
+        "version",
+        "type",
+        "foundBy",
+        "locations",
+        "licenses",
+        "language",
+        "cpes",
+        "purl",
+        "metadataType",
+        "metadata"
+      ],
+      "properties": {
+        "id": {
+          "type": "string"
+        },
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "type": {
+          "type": "string"
+        },
+        "foundBy": {
+          "type": "string"
+        },
+        "locations": {
+          "items": {
+            "$ref": "#/definitions/Location"
+          },
+          "type": "array"
+        },
+        "licenses": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "language": {
+          "type": "string"
+        },
+        "cpes": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        },
+        "purl": {
+          "type": "string"
+        },
+        "metadataType": {
+          "type": "string"
+        },
+        "metadata": {
+          "additionalProperties": true
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "PomProperties": {
+      "required": [
+        "path",
+        "name",
+        "groupId",
+        "artifactId",
+        "version",
+        "extraFields"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "name": {
+          "type": "string"
+        },
+        "groupId": {
+          "type": "string"
+        },
+        "artifactId": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "extraFields": {
+          "patternProperties": {
+            ".*": {
+              "type": "string"
+            }
+          },
+          "type": "object"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "PythonFileDigest": {
+      "required": [
+        "algorithm",
+        "value"
+      ],
+      "properties": {
+        "algorithm": {
+          "type": "string"
+        },
+        "value": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "PythonFileRecord": {
+      "required": [
+        "path"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "digest": {
+          "$schema": "http://json-schema.org/draft-04/schema#",
+          "$ref": "#/definitions/PythonFileDigest"
+        },
+        "size": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "PythonPackageMetadata": {
+      "required": [
+        "name",
+        "version",
+        "license",
+        "author",
+        "authorEmail",
+        "platform",
+        "sitePackagesRootPath"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "license": {
+          "type": "string"
+        },
+        "author": {
+          "type": "string"
+        },
+        "authorEmail": {
+          "type": "string"
+        },
+        "platform": {
+          "type": "string"
+        },
+        "files": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/PythonFileRecord"
+          },
+          "type": "array"
+        },
+        "sitePackagesRootPath": {
+          "type": "string"
+        },
+        "topLevelPackages": {
+          "items": {
+            "type": "string"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Relationship": {
+      "required": [
+        "parent",
+        "child",
+        "type",
+        "metadata"
+      ],
+      "properties": {
+        "parent": {
+          "type": "string"
+        },
+        "child": {
+          "type": "string"
+        },
+        "type": {
+          "type": "string"
+        },
+        "metadata": {
+          "additionalProperties": true
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "RpmdbFileRecord": {
+      "required": [
+        "path",
+        "mode",
+        "size",
+        "sha256"
+      ],
+      "properties": {
+        "path": {
+          "type": "string"
+        },
+        "mode": {
+          "type": "integer"
+        },
+        "size": {
+          "type": "integer"
+        },
+        "sha256": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "RpmdbMetadata": {
+      "required": [
+        "name",
+        "version",
+        "epoch",
+        "architecture",
+        "release",
+        "sourceRpm",
+        "size",
+        "license",
+        "vendor",
+        "files"
+      ],
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "version": {
+          "type": "string"
+        },
+        "epoch": {
+          "type": "integer"
+        },
+        "architecture": {
+          "type": "string"
+        },
+        "release": {
+          "type": "string"
+        },
+        "sourceRpm": {
+          "type": "string"
+        },
+        "size": {
+          "type": "integer"
+        },
+        "license": {
+          "type": "string"
+        },
+        "vendor": {
+          "type": "string"
+        },
+        "files": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/RpmdbFileRecord"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Schema": {
+      "required": [
+        "version",
+        "url"
+      ],
+      "properties": {
+        "version": {
+          "type": "string"
+        },
+        "url": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "SearchResult": {
+      "required": [
+        "classification",
+        "lineNumber",
+        "lineOffset",
+        "seekPosition",
+        "length"
+      ],
+      "properties": {
+        "classification": {
+          "type": "string"
+        },
+        "lineNumber": {
+          "type": "integer"
+        },
+        "lineOffset": {
+          "type": "integer"
+        },
+        "seekPosition": {
+          "type": "integer"
+        },
+        "length": {
+          "type": "integer"
+        },
+        "value": {
+          "type": "string"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Secrets": {
+      "required": [
+        "location",
+        "secrets"
+      ],
+      "properties": {
+        "location": {
+          "$ref": "#/definitions/Location"
+        },
+        "secrets": {
+          "items": {
+            "$schema": "http://json-schema.org/draft-04/schema#",
+            "$ref": "#/definitions/SearchResult"
+          },
+          "type": "array"
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "Source": {
+      "required": [
+        "type",
+        "target"
+      ],
+      "properties": {
+        "type": {
+          "type": "string"
+        },
+        "target": {
+          "additionalProperties": true
+        }
+      },
+      "additionalProperties": true,
+      "type": "object"
+    }
+  }
+}

syft/file/classification_cataloger.go

@@ -0,0 +1,37 @@
+package file
+
+import (
+	"github.com/anchore/syft/internal/log"
+	"github.com/anchore/syft/syft/source"
+)
+
+type ClassificationCataloger struct {
+	classifiers []Classifier
+}
+
+func NewClassificationCataloger(classifiers []Classifier) (*ClassificationCataloger, error) {
+	return &ClassificationCataloger{
+		classifiers: classifiers,
+	}, nil
+}
+
+func (i *ClassificationCataloger) Catalog(resolver source.FileResolver) (map[source.Location][]Classification, error) {
+	results := make(map[source.Location][]Classification)
+
+	numResults := 0
+	for location := range resolver.AllLocations() {
+		for _, classifier := range i.classifiers {
+			result, err := classifier.Classify(resolver, location)
+			if err != nil {
+				return nil, err
+			}
+			if result != nil {
+				results[location] = append(results[location], *result)
+				numResults++
+			}
+		}
+	}
+	log.Debugf("classification cataloger discovered %d results", numResults)
+
+	return results, nil
+}

syft/file/classification_cataloger_test.go

@@ -0,0 +1,145 @@
+package file
+
+import (
+	"testing"
+
+	"github.com/anchore/syft/syft/source"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestClassifierCataloger_DefaultClassifiers_PositiveCases(t *testing.T) {
+	tests := []struct {
+		name        string
+		fixtureDir  string
+		location    string
+		expected    []Classification
+		expectedErr func(assert.TestingT, error, ...interface{}) bool
+	}{
+		{
+			name:       "positive-libpython3.7.so",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/libpython3.7.so",
+			expected: []Classification{
+				{
+					Class: "python-binary",
+					Metadata: map[string]string{
+						"version": "3.7.4a-vZ9",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+		{
+			name:       "positive-python3.6",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/python3.6",
+			expected: []Classification{
+				{
+					Class: "python-binary",
+					Metadata: map[string]string{
+						"version": "3.6.3a-vZ9",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+		{
+			name:       "positive-patchlevel.h",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/patchlevel.h",
+			expected: []Classification{
+				{
+					Class: "cpython-source",
+					Metadata: map[string]string{
+						"version": "3.9-aZ5",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+		{
+			name:       "positive-go",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/go",
+			expected: []Classification{
+				{
+					Class: "go-binary",
+					Metadata: map[string]string{
+						"version": "1.14",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+		{
+			name:       "positive-go-hint",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/VERSION",
+			expected: []Classification{
+				{
+					Class: "go-binary-hint",
+					Metadata: map[string]string{
+						"version": "1.15",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+		{
+			name:       "positive-busybox",
+			fixtureDir: "test-fixtures/classifiers/positive",
+			location:   "test-fixtures/classifiers/positive/busybox",
+			expected: []Classification{
+				{
+					Class: "busybox-binary",
+					Metadata: map[string]string{
+						"version": "3.33.3",
+					},
+				},
+			},
+			expectedErr: assert.NoError,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+
+			c, err := NewClassificationCataloger(DefaultClassifiers)
+			test.expectedErr(t, err)
+
+			src, err := source.NewFromDirectory(test.fixtureDir)
+			test.expectedErr(t, err)
+
+			resolver, err := src.FileResolver(source.SquashedScope)
+			test.expectedErr(t, err)
+
+			actualResults, err := c.Catalog(resolver)
+			test.expectedErr(t, err)
+
+			loc := source.NewLocation(test.location)
+
+			if _, ok := actualResults[loc]; !ok {
+				t.Fatalf("could not find test location=%q", test.location)
+			}
+
+			assert.Equal(t, test.expected, actualResults[loc])
+		})
+	}
+}
+
+func TestClassifierCataloger_DefaultClassifiers_NegativeCases(t *testing.T) {
+
+	c, err := NewClassificationCataloger(DefaultClassifiers)
+	assert.NoError(t, err)
+
+	src, err := source.NewFromDirectory("test-fixtures/classifiers/negative")
+	assert.NoError(t, err)
+
+	resolver, err := src.FileResolver(source.SquashedScope)
+	assert.NoError(t, err)
+
+	actualResults, err := c.Catalog(resolver)
+	assert.NoError(t, err)
+	assert.Equal(t, 0, len(actualResults))
+
+}

syft/file/classifier.go

@@ -0,0 +1,141 @@
+package file
+
+import (
+	"bytes"
+	"fmt"
+	"io/ioutil"
+	"regexp"
+	"text/template"
+
+	"github.com/anchore/syft/internal"
+	"github.com/anchore/syft/syft/source"
+)
+
+var DefaultClassifiers = []Classifier{
+	{
+		Class: "python-binary",
+		FilepathPatterns: []*regexp.Regexp{
+			regexp.MustCompile(`(.*/|^)python(?P<version>[0-9]+\.[0-9]+)$`),
+			regexp.MustCompile(`(.*/|^)libpython(?P<version>[0-9]+\.[0-9]+).so.*$`),
+		},
+		EvidencePatternTemplates: []string{
+			`(?m)(?P<version>{{ .version }}\.[0-9]+[-_a-zA-Z0-9]*)`,
+		},
+	},
+	{
+		Class: "cpython-source",
+		FilepathPatterns: []*regexp.Regexp{
+			regexp.MustCompile(`(.*/|^)patchlevel.h$`),
+		},
+		EvidencePatternTemplates: []string{
+			`(?m)#define\s+PY_VERSION\s+"?(?P<version>[0-9\.\-_a-zA-Z]+)"?`,
+		},
+	},
+	{
+		Class: "go-binary",
+		FilepathPatterns: []*regexp.Regexp{
+			regexp.MustCompile(`(.*/|^)go$`),
+		},
+		EvidencePatternTemplates: []string{
+			`(?m)go(?P<version>[0-9]+\.[0-9]+(\.[0-9]+|beta[0-9]+|alpha[0-9]+|rc[0-9]+)?)`,
+		},
+	},
+	{
+		Class: "go-binary-hint",
+		FilepathPatterns: []*regexp.Regexp{
+			regexp.MustCompile(`(.*/|^)VERSION$`),
+		},
+		EvidencePatternTemplates: []string{
+			`(?m)go(?P<version>[0-9]+\.[0-9]+(\.[0-9]+|beta[0-9]+|alpha[0-9]+|rc[0-9]+)?)`,
+		},
+	},
+	{
+		Class: "busybox-binary",
+		FilepathPatterns: []*regexp.Regexp{
+			regexp.MustCompile(`(.*/|^)busybox$`),
+		},
+		EvidencePatternTemplates: []string{
+			`(?m)BusyBox\s+v(?P<version>[0-9]+\.[0-9]+\.[0-9]+)`,
+		},
+	},
+}
+
+type Classifier struct {
+	Class                    string
+	FilepathPatterns         []*regexp.Regexp
+	EvidencePatternTemplates []string
+}
+
+type Classification struct {
+	Class    string            `json:"class"`
+	Metadata map[string]string `json:"metadata"`
+}
+
+func (c Classifier) Classify(resolver source.FileResolver, location source.Location) (*Classification, error) {
+	doesFilepathMatch, filepathNamedGroupValues := filepathMatches(c.FilepathPatterns, location)
+	if !doesFilepathMatch {
+		return nil, nil
+	}
+
+	contentReader, err := resolver.FileContentsByLocation(location)
+	if err != nil {
+		return nil, err
+	}
+	defer contentReader.Close()
+
+	// TODO: there is room for improvement here, as this may use an excessive amount of memory. Alternate approach is to leverage a RuneReader.
+	contents, err := ioutil.ReadAll(contentReader)
+	if err != nil {
+		return nil, err
+	}
+
+	var result *Classification
+	for _, patternTemplate := range c.EvidencePatternTemplates {
+		tmpl, err := template.New("").Parse(patternTemplate)
+		if err != nil {
+			return nil, fmt.Errorf("unable to parse classifier template=%q : %w", patternTemplate, err)
+		}
+
+		patternBuf := &bytes.Buffer{}
+		err = tmpl.Execute(patternBuf, filepathNamedGroupValues)
+		if err != nil {
+			return nil, fmt.Errorf("unable to render template: %w", err)
+		}
+
+		pattern, err := regexp.Compile(patternBuf.String())
+		if err != nil {
+			return nil, fmt.Errorf("unable to compile rendered regex=%q: %w", patternBuf.String(), err)
+		}
+
+		if !pattern.Match(contents) {
+			continue
+		}
+
+		matchMetadata := internal.MatchNamedCaptureGroups(pattern, string(contents))
+		if result == nil {
+			result = &Classification{
+				Class:    c.Class,
+				Metadata: matchMetadata,
+			}
+		} else {
+			for key, value := range matchMetadata {
+				result.Metadata[key] = value
+			}
+		}
+	}
+	return result, nil
+}
+
+func filepathMatches(patterns []*regexp.Regexp, location source.Location) (bool, map[string]string) {
+	for _, path := range []string{location.RealPath, location.VirtualPath} {
+		if path == "" {
+			continue
+		}
+		for _, pattern := range patterns {
+			if pattern.MatchString(path) {
+				return true, internal.MatchNamedCaptureGroups(pattern, path)
+			}
+		}
+	}
+	return false, nil
+}

syft/file/classifier_test.go

@@ -0,0 +1,90 @@
+package file
+
+import (
+	"regexp"
+	"testing"
+
+	"github.com/anchore/syft/syft/source"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestFilepathMatches(t *testing.T) {
+	tests := []struct {
+		name                string
+		location            source.Location
+		patterns            []string
+		expectedMatches     bool
+		expectedNamedGroups map[string]string
+	}{
+		{
+			name: "simple-filename-match",
+			location: source.Location{
+				RealPath: "python2.7",
+			},
+			patterns: []string{
+				`python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: true,
+		},
+		{
+			name: "filepath-match",
+			location: source.Location{
+				RealPath: "/usr/bin/python2.7",
+			},
+			patterns: []string{
+				`python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: true,
+		},
+		{
+			name: "virtual-filepath-match",
+			location: source.Location{
+				VirtualPath: "/usr/bin/python2.7",
+			},
+			patterns: []string{
+				`python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: true,
+		},
+		{
+			name: "full-filepath-match",
+			location: source.Location{
+				VirtualPath: "/usr/bin/python2.7",
+			},
+			patterns: []string{
+				`.*/bin/python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: true,
+		},
+		{
+			name: "anchored-filename-match-FAILS",
+			location: source.Location{
+				RealPath: "/usr/bin/python2.7",
+			},
+			patterns: []string{
+				`^python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: false,
+		},
+		{
+			name:     "empty-filename-match-FAILS",
+			location: source.Location{},
+			patterns: []string{
+				`^python([0-9]+\.[0-9]+)$`,
+			},
+			expectedMatches: false,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			var patterns []*regexp.Regexp
+			for _, p := range test.patterns {
+				patterns = append(patterns, regexp.MustCompile(p))
+			}
+			actualMatches, actualNamedGroups := filepathMatches(patterns, test.location)
+			assert.Equal(t, test.expectedMatches, actualMatches)
+			assert.Equal(t, test.expectedNamedGroups, actualNamedGroups)
+		})
+	}
+}

syft/file/test-fixtures/classifiers/negative/.gitignore

@@ -0,0 +1 @@
+!libpython2.7.so

syft/file/test-fixtures/classifiers/negative/busybox

@@ -0,0 +1 @@
+another bad binary

syft/file/test-fixtures/classifiers/negative/go

@@ -0,0 +1 @@
+a bad go binary

syft/file/test-fixtures/classifiers/negative/libpython2.7.so

@@ -0,0 +1,3 @@
+# note: this should NOT match
+
+DO NOT DETECT

syft/file/test-fixtures/classifiers/negative/python2.6

@@ -0,0 +1,3 @@
+# note: this should NOT match
+
+just some noise

syft/file/test-fixtures/classifiers/positive/.gitignore

@@ -0,0 +1 @@
+!libpython3.7.so

syft/file/test-fixtures/classifiers/positive/VERSION

@@ -0,0 +1 @@
+go1.15-beta2

syft/file/test-fixtures/classifiers/positive/busybox

@@ -0,0 +1,3 @@
+# note: this SHOULD match as busybox 3.33.3
+
+noise!BusyBox v3.33.3!noise

syft/file/test-fixtures/classifiers/positive/go

@@ -0,0 +1 @@
+go1.14

syft/file/test-fixtures/classifiers/positive/libpython3.7.so

@@ -0,0 +1,2 @@
+# note: this SHOULD match as python 3.7
+noise3.7.4a-vZ9!morenoise

syft/file/test-fixtures/classifiers/positive/patchlevel.h

@@ -0,0 +1,7 @@
+# note: this SHOULD match as python 3.9
+
+some source code...
+
+#define    PY_VERSION   3.9-aZ5
+
+more source!

syft/file/test-fixtures/classifiers/positive/python3.6

@@ -0,0 +1,3 @@
+# note: this SHOULD match as python 3.6
+
+noise3.6.3a-vZ9!morenoise

syft/pkg/cataloger/deb/parse_copyright.go

@@ -21,7 +21,7 @@ func parseLicensesFromCopyright(reader io.Reader) []string {
 	for scanner.Scan() {
 		line := scanner.Text()
 
-		matchesByGroup := internal.MatchCaptureGroups(licensePattern, line)
+		matchesByGroup := internal.MatchNamedCaptureGroups(licensePattern, line)
 		if len(matchesByGroup) > 0 {
 			candidate, ok := matchesByGroup["license"]
 			if !ok {

syft/pkg/cataloger/deb/parse_dpkg_status.go

@@ -145,7 +145,7 @@ func extractAllFields(reader *bufio.Reader) (map[string]interface{}, error) {
 // of the "<name>" form, then return name and nil
 func extractSourceVersion(source string) (string, string) {
 	// special handling for the Source field since it has formatted data
-	match := internal.MatchCaptureGroups(sourceRegexp, source)
+	match := internal.MatchNamedCaptureGroups(sourceRegexp, source)
 	return match["name"], match["version"]
 }
 

syft/pkg/cataloger/javascript/parse_package_json.go

@@ -63,7 +63,7 @@ func (a *Author) UnmarshalJSON(b []byte) error {
 		}
 	} else {
 		// parse out "name <email> (url)" into an Author struct
-		fields = internal.MatchCaptureGroups(authorPattern, authorStr)
+		fields = internal.MatchNamedCaptureGroups(authorPattern, authorStr)
 	}
 
 	// translate the map into a structure

syft/pkg/cataloger/ruby/parse_gemspec.go

@@ -77,7 +77,7 @@ func parseGemSpecEntries(_ string, reader io.Reader) ([]pkg.Package, error) {
 		}
 
 		for field, pattern := range patterns {
-			matchMap := internal.MatchCaptureGroups(pattern, sanitizedLine)
+			matchMap := internal.MatchNamedCaptureGroups(pattern, sanitizedLine)
 			if value := matchMap[field]; value != "" {
 				if postProcessor := postProcessors[field]; postProcessor != nil {
 					fields[field] = postProcessor(value)