Add an archive selector to Siegfried

Using the -zs flag will aloow users to specify which archive types
Siegfried cracks open to identify the contents of.

cmd/sf/sf.go

@@ -40,31 +40,32 @@ const maxMulti = 1024
 
 // flags
 var (
-	updateShort  = flag.Bool("u", false, "update or install the default signature file")
-	update       = flag.Bool("update", false, "update or install the default signature file")
-	versionShort = flag.Bool("v", false, "display version information")
-	version      = flag.Bool("version", false, "display version information")
-	logf         = flag.String("log", "error", "log errors, warnings, debug or slow output, knowns or unknowns to stderr or stdout e.g. -log error,warn,unknown,stdout")
-	nr           = flag.Bool("nr", false, "prevent automatic directory recursion")
-	yaml         = flag.Bool("yaml", true, "YAML output format")
-	csvo         = flag.Bool("csv", false, "CSV output format")
-	jsono        = flag.Bool("json", false, "JSON output format")
-	droido       = flag.Bool("droid", false, "DROID CSV output format")
-	sig          = flag.String("sig", config.SignatureBase(), "set the signature file")
-	home         = flag.String("home", config.Home(), "override the default home directory")
-	serve        = flag.String("serve", "", "start siegfried server e.g. -serve localhost:5138")
-	multi        = flag.Int("multi", 1, "set number of parallel file ID processes")
-	archive      = flag.Bool("z", false, fmt.Sprintf("scan archive formats: (%s)", config.ListAllArcTypes()))
-	hashf        = flag.String("hash", "", "calculate file checksum with hash algorithm; options "+checksum.HashChoices)
-	throttlef    = flag.Duration("throttle", 0, "set a time to wait between scanning files e.g. 50ms")
-	utcf         = flag.Bool("utc", false, "report file modified times in UTC, rather than local, TZ")
-	coe          = flag.Bool("coe", false, "continue on fatal errors during directory walks (this may result in directories being skipped)")
-	replay       = flag.Bool("replay", false, "replay one (or more) results files to change output or logging e.g. sf -replay -csv results.yaml")
-	list         = flag.Bool("f", false, "scan one (or more) lists of filenames e.g. sf -f myfiles.txt")
-	name         = flag.String("name", "", "provide a filename when scanning a stream e.g. sf -name myfile.txt -")
-	conff        = flag.String("conf", "", "set the configuration file")
-	setconff     = flag.Bool("setconf", false, "record flags used with this command in configuration file")
-	sourceinline = flag.Bool("sourceinline", false, "display provenance in-line (basis field) when it is available for an identifier, e.g. Wikidata")
+	updateShort    = flag.Bool("u", false, "update or install the default signature file")
+	update         = flag.Bool("update", false, "update or install the default signature file")
+	versionShort   = flag.Bool("v", false, "display version information")
+	version        = flag.Bool("version", false, "display version information")
+	logf           = flag.String("log", "error", "log errors, warnings, debug or slow output, knowns or unknowns to stderr or stdout e.g. -log error,warn,unknown,stdout")
+	nr             = flag.Bool("nr", false, "prevent automatic directory recursion")
+	yaml           = flag.Bool("yaml", true, "YAML output format")
+	csvo           = flag.Bool("csv", false, "CSV output format")
+	jsono          = flag.Bool("json", false, "JSON output format")
+	droido         = flag.Bool("droid", false, "DROID CSV output format")
+	sig            = flag.String("sig", config.SignatureBase(), "set the signature file")
+	home           = flag.String("home", config.Home(), "override the default home directory")
+	serve          = flag.String("serve", "", "start siegfried server e.g. -serve localhost:5138")
+	multi          = flag.Int("multi", 1, "set number of parallel file ID processes")
+	archive        = flag.Bool("z", false, fmt.Sprintf("scan archive formats: (%s)", config.ListAllArcTypes()))
+	selectArchives = flag.String("zs", config.ListAllArcTypes(), "select the archive types to decompress and identify the contents of")
+	hashf          = flag.String("hash", "", "calculate file checksum with hash algorithm; options "+checksum.HashChoices)
+	throttlef      = flag.Duration("throttle", 0, "set a time to wait between scanning files e.g. 50ms")
+	utcf           = flag.Bool("utc", false, "report file modified times in UTC, rather than local, TZ")
+	coe            = flag.Bool("coe", false, "continue on fatal errors during directory walks (this may result in directories being skipped)")
+	replay         = flag.Bool("replay", false, "replay one (or more) results files to change output or logging e.g. sf -replay -csv results.yaml")
+	list           = flag.Bool("f", false, "scan one (or more) lists of filenames e.g. sf -f myfiles.txt")
+	name           = flag.String("name", "", "provide a filename when scanning a stream e.g. sf -name myfile.txt -")
+	conff          = flag.String("conf", "", "set the configuration file")
+	setconff       = flag.Bool("setconf", false, "record flags used with this command in configuration file")
+	sourceinline   = flag.Bool("sourceinline", false, "display provenance in-line (basis field) when it is available for an identifier, e.g. Wikidata")
 )
 
 var (
@@ -372,6 +373,10 @@ func main() {
 		}
 		return
 	}
+	// handle -zs
+	if *selectArchives != "" {
+		config.SetArchiveFilterPermissive(*selectArchives)
+	}
 	// handle -fpr
 	if *fprflag {
 		log.Printf("FPR server started at %s. Use CTRL-C to quit.\n", config.Fpr())

pkg/config/decompress.go

@@ -16,6 +16,7 @@ package config
 
 import (
 	"fmt"
+	"strings"
 )
 
 // Archive is a file format capable of decompression by sf.
@@ -106,8 +107,41 @@ func ListAllArcTypes() string {
 	)
 }
 
-func (a Archive) String() string {
-	switch a {
+var permissiveFilter []string
+
+// SetArchiveFilterPermissive enables a filter to be created on the
+// types of archive that we want to extract from. Anything not in this
+// list is not extracted.
+func SetArchiveFilterPermissive(value string) []string {
+	arr := []string{}
+	arcList := strings.Split(value, ",")
+	for _, arc := range arcList {
+		switch strings.TrimSpace(strings.ToLower(arc)) {
+		case zipArc:
+			arr = append(arr, ArcZipTypes()...)
+		case tarArc:
+			arr = append(arr, ArcTarTypes()...)
+		case gzipArc:
+			arr = append(arr, ArcGzipTypes()...)
+		case warcArc:
+			arr = append(arr, ArcWarcTypes()...)
+		case arcArc:
+			arr = append(arr, ArcArcTypes()...)
+		}
+	}
+	permissiveFilter = arr
+	return arr
+}
+
+// archiveFilterPermissive provides a getter for the configured
+// zip-types we want to extract and identify the contents of with
+// Siegfried.
+func archiveFilterPermissive() []string {
+	return permissiveFilter
+}
+
+func (archive Archive) String() string {
+	switch archive {
 	case Zip:
 		return "zip"
 	case Gzip:

pkg/config/identifier.go

@@ -287,6 +287,9 @@ func contains(v string, s []string) bool {
 
 // IsArchive returns an Archive that corresponds to the provided id (or none if no match).
 func IsArchive(id string) Archive {
+	if !contains(id, archiveFilterPermissive()) {
+		return None
+	}
 	switch {
 	case contains(id, ArcZipTypes()):
 		return Zip

pkg/config/identifier_test.go

@@ -33,26 +33,31 @@ var nonArcUID = "fmt/1000"
 
 // arcTest defines the structure needed for our table driven testing.
 type arcTest struct {
+	filter string  // The set of zip-type files to provide SetArchiveFilterPermissive(...)
 	uid    string  // A UID (PUID, FDD) that identifies a zip-type file.
 	result Archive // The anticipated result from our test.
 }
 
 // isArcTests provide us a slice of tests and results to loop through.
 var isArcTests = []arcTest{
 	// Positive tests should return valid Archive values.
-	arcTest{proZipUID, Zip},
-	arcTest{mimeTarUID, Tar},
-	arcTest{mimeGzipUID, Gzip},
-	arcTest{mimeWarcUID, WARC},
-	arcTest{locArcUID, ARC},
+	arcTest{ListAllArcTypes(), proZipUID, Zip},
+	arcTest{"TAR", mimeTarUID, Tar},
+	arcTest{"gZip", mimeGzipUID, Gzip},
+	arcTest{"warc,zip,tar", mimeWarcUID, WARC},
+	arcTest{"zip,arc", locArcUID, ARC},
 	// Negative tests should all return None.
-	arcTest{nonArcUID, None},
+	arcTest{"zip,arc", mimeWarcUID, None},
+	arcTest{"zip,arc", mimeGzipUID, None},
+	arcTest{ListAllArcTypes(), nonArcUID, None},
+	arcTest{"", nonArcUID, None},
 }
 
-// TestIsArchive tests cases whether we return the correct result when
-// testing whether something is an Archive.
-func TestIsArchive(t *testing.T) {
+// TestIsArchivePositive tests cases where the filter should return a
+// positive match.
+func TestIsArchivePositive(t *testing.T) {
 	for _, test := range isArcTests {
+		SetArchiveFilterPermissive(test.filter)
 		arc := IsArchive(test.uid)
 		if arc != test.result {
 			t.Errorf(