add helper to extract regular files

Signed-off-by: Alex Goodman <[email protected]>

internal/err_helper.go

@@ -16,12 +16,13 @@ func CloseAndLogError(closer io.Closer, location string) {
 }
 
 type ErrPath struct {
-	Path string
-	Err  error
+	Cataloger string
+	Path      string
+	Err       error
 }
 
 func (e ErrPath) Error() string {
-	return fmt.Sprintf("unable to observe contents of %+v: %v", e.Path, e.Err)
+	return fmt.Sprintf("%s unable to observe contents of %+v: %v", e.Cataloger, e.Path, e.Err)
 }
 
 func IsErrPath(err error) bool {

syft/file/all_regular_files.go

@@ -0,0 +1,30 @@
+package file
+
+import (
+	"github.com/anchore/syft/internal/log"
+	"github.com/anchore/syft/syft/source"
+)
+
+func allRegularFiles(resolver source.FileResolver) (locations []source.Location) {
+	for location := range resolver.AllLocations() {
+		resolvedLocations, err := resolver.FilesByPath(location.RealPath)
+		if err != nil {
+			log.Warnf("unable to resolve %+v: %+v", location, err)
+			continue
+		}
+
+		for _, resolvedLocation := range resolvedLocations {
+			metadata, err := resolver.FileMetadataByLocation(resolvedLocation)
+			if err != nil {
+				log.Warnf("unable to get metadata for %+v: %+v", location, err)
+				continue
+			}
+
+			if metadata.Type != source.RegularFile {
+				continue
+			}
+			locations = append(locations, resolvedLocation)
+		}
+	}
+	return locations
+}

syft/file/classification_cataloger.go

@@ -19,7 +19,7 @@ func (i *ClassificationCataloger) Catalog(resolver source.FileResolver) (map[sou
 	results := make(map[source.Coordinates][]Classification)
 
 	numResults := 0
-	for location := range resolver.AllLocations() {
+	for _, location := range allRegularFiles(resolver) {
 		for _, classifier := range i.classifiers {
 			result, err := classifier.Classify(resolver, location)
 			if err != nil {

syft/file/contents_cataloger.go

@@ -65,7 +65,7 @@ func (i *ContentsCataloger) catalogLocation(resolver source.FileResolver, locati
 
 	buf := &bytes.Buffer{}
 	if _, err = io.Copy(base64.NewEncoder(base64.StdEncoding, buf), contentReader); err != nil {
-		return "", internal.ErrPath{Path: location.RealPath, Err: err}
+		return "", internal.ErrPath{Cataloger: "contents-cataloger", Path: location.RealPath, Err: err}
 	}
 
 	return buf.String(), nil

syft/file/digest_cataloger.go

@@ -34,10 +34,7 @@ func NewDigestsCataloger(hashes []crypto.Hash) (*DigestsCataloger, error) {
 
 func (i *DigestsCataloger) Catalog(resolver source.FileResolver) (map[source.Coordinates][]Digest, error) {
 	results := make(map[source.Coordinates][]Digest)
-	var locations []source.Location
-	for location := range resolver.AllLocations() {
-		locations = append(locations, location)
-	}
+	locations := allRegularFiles(resolver)
 	stage, prog := digestsCatalogingProgress(int64(len(locations)))
 	for _, location := range locations {
 		stage.Current = location.RealPath
@@ -90,7 +87,7 @@ func (i *DigestsCataloger) catalogLocation(resolver source.FileResolver, locatio
 
 	size, err := io.Copy(io.MultiWriter(writers...), contentReader)
 	if err != nil {
-		return nil, internal.ErrPath{Path: location.RealPath, Err: err}
+		return nil, internal.ErrPath{Cataloger: "digests-cataloger", Path: location.RealPath, Err: err}
 	}
 
 	if size == 0 {

syft/file/secrets_cataloger.go

@@ -42,10 +42,7 @@ func NewSecretsCataloger(patterns map[string]*regexp.Regexp, revealValues bool,
 
 func (i *SecretsCataloger) Catalog(resolver source.FileResolver) (map[source.Coordinates][]SearchResult, error) {
 	results := make(map[source.Coordinates][]SearchResult)
-	var locations []source.Location
-	for location := range resolver.AllLocations() {
-		locations = append(locations, location)
-	}
+	locations := allRegularFiles(resolver)
 	stage, prog, secretsDiscovered := secretsCatalogingProgress(int64(len(locations)))
 	for _, location := range locations {
 		stage.Current = location.RealPath
@@ -86,7 +83,7 @@ func (i *SecretsCataloger) catalogLocation(resolver source.FileResolver, locatio
 	// TODO: in the future we can swap out search strategies here
 	secrets, err := catalogLocationByLine(resolver, location, i.patterns)
 	if err != nil {
-		return nil, internal.ErrPath{Path: location.RealPath, Err: err}
+		return nil, internal.ErrPath{Cataloger: "secrets-cataloger", Path: location.RealPath, Err: err}
 	}
 
 	if i.revealValues {

syft/source/directory_resolver.go

@@ -233,7 +233,9 @@ func (r directoryResolver) addSymlinkToIndex(p string, info os.FileInfo) (string
 	}
 
 	location := NewLocationFromDirectory(p, *ref)
+	location.VirtualPath = p
 	metadata := fileMetadataFromPath(p, usedInfo, r.isInIndex(location))
+	metadata.LinkDestination = linkTarget
 	r.addFileMetadataToIndex(ref, metadata)
 
 	return targetAbsPath, nil

syft/source/file_metadata.go

@@ -63,6 +63,7 @@ func fileMetadataFromPath(path string, info os.FileInfo, withMIMEType bool) File
 		// unsupported across platforms
 		UserID:   uid,
 		GroupID:  gid,
+		Size:     info.Size(),
 		MIMEType: mimeType,
 	}
 }

test/cli/packages_cmd_test.go

@@ -209,55 +209,55 @@ func TestRegistryAuth(t *testing.T) {
 	}{
 		{
 			name: "fallback to keychain",
-			args: []string{"packages", "-vv", "registry:localhost:5000/something:latest"},
+			args: []string{"packages", "-vv", "registry:localhost:17/something:latest"},
 			assertions: []traitAssertion{
 				assertInOutput("source=OciRegistry"),
-				assertInOutput("localhost:5000/something:latest"),
+				assertInOutput("localhost:17/something:latest"),
 				assertInOutput("no registry credentials configured, using the default keychain"),
 			},
 		},
 		{
 			name: "use creds",
-			args: []string{"packages", "-vv", "registry:localhost:5000/something:latest"},
+			args: []string{"packages", "-vv", "registry:localhost:17/something:latest"},
 			env: map[string]string{
-				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:5000",
+				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:17",
 				"SYFT_REGISTRY_AUTH_USERNAME":  "username",
 				"SYFT_REGISTRY_AUTH_PASSWORD":  "password",
 			},
 			assertions: []traitAssertion{
 				assertInOutput("source=OciRegistry"),
-				assertInOutput("localhost:5000/something:latest"),
-				assertInOutput(`using basic auth for registry "localhost:5000"`),
+				assertInOutput("localhost:17/something:latest"),
+				assertInOutput(`using basic auth for registry "localhost:17"`),
 			},
 		},
 		{
 			name: "use token",
-			args: []string{"packages", "-vv", "registry:localhost:5000/something:latest"},
+			args: []string{"packages", "-vv", "registry:localhost:17/something:latest"},
 			env: map[string]string{
-				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:5000",
+				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:17",
 				"SYFT_REGISTRY_AUTH_TOKEN":     "token",
 			},
 			assertions: []traitAssertion{
 				assertInOutput("source=OciRegistry"),
-				assertInOutput("localhost:5000/something:latest"),
-				assertInOutput(`using token for registry "localhost:5000"`),
+				assertInOutput("localhost:17/something:latest"),
+				assertInOutput(`using token for registry "localhost:17"`),
 			},
 		},
 		{
 			name: "not enough info fallsback to keychain",
-			args: []string{"packages", "-vv", "registry:localhost:5000/something:latest"},
+			args: []string{"packages", "-vv", "registry:localhost:17/something:latest"},
 			env: map[string]string{
-				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:5000",
+				"SYFT_REGISTRY_AUTH_AUTHORITY": "localhost:17",
 			},
 			assertions: []traitAssertion{
 				assertInOutput("source=OciRegistry"),
-				assertInOutput("localhost:5000/something:latest"),
+				assertInOutput("localhost:17/something:latest"),
 				assertInOutput(`no registry credentials configured, using the default keychain`),
 			},
 		},
 		{
 			name: "allows insecure http flag",
-			args: []string{"packages", "-vv", "registry:localhost:5000/something:latest"},
+			args: []string{"packages", "-vv", "registry:localhost:17/something:latest"},
 			env: map[string]string{
 				"SYFT_REGISTRY_INSECURE_USE_HTTP": "true",
 			},

test/cli/power_user_cmd_test.go

@@ -83,7 +83,7 @@ func TestPowerUserCmdFlags(t *testing.T) {
 			},
 		},
 		{
-			name: "defaut-secrets-dir-results-w-reveal-values",
+			name: "default-secrets-dir-results-w-reveal-values",
 			env: map[string]string{
 				"SYFT_SECRETS_REVEAL_VALUES": "true",
 			},