Enable reading of format type from a DROID sig

Changes to DROID signature file parsing to enable reading of file format types. Changes assume use of an element for this currently, but can be easily changed with the tests still applicable here.
richardlehane · Nov 5, 2023 · f2f643d · f2f643d
1 parent ccff91e
commit f2f643d
Show file tree

Hide file tree

Showing 6 changed files with 80 additions and 25 deletions.
diff --git a/cmd/sf/pronom_test.go b/cmd/sf/pronom_test.go
@@ -128,17 +128,9 @@ var pronomIDs = []pronomIdentificationTests{
 	},
 }
 
-// TestPronom looks to see if PRONOM identification results for a
-// minimized PRONOM dataset are correct and contain the information we
-// anticipate.
-func TestPronom(t *testing.T) {
-	sf := siegfried.New()
-	config.SetHome(DataPath)
-	identifier, err := pronom.New(config.SetLimit(minimalPronom))
-	if err != nil {
-		t.Errorf("Error creating new PRONOM identifier: %s", err)
-	}
-	sf.Add(identifier)
+// runIdentificationWithSF provides a number of tests that can be run
+// against a Siegfried.
+func runIdentificationWithSF(sf *siegfried.Siegfried, t *testing.T) {
 	makeSkeletons()
 	skeletonFS := fstest.MapFS(skeletons)
 	testDirListing, err := skeletonFS.ReadDir(".")
@@ -182,5 +174,38 @@ func TestPronom(t *testing.T) {
 			t.Errorf("Results not equal for %s; expected %v; got %v", res.puid, pronomIDs[idx], res)
 		}
 	}
+}
+
+// TestPronom looks to see if PRONOM identification results for a
+// minimized PRONOM dataset are correct and contain the information we
+// anticipate.
+func TestPronom(t *testing.T) {
+	sf := siegfried.New()
+	config.SetHome(DataPath)
+	identifier, err := pronom.New(config.SetLimit(minimalPronom))
+	if err != nil {
+		t.Errorf("Error creating new PRONOM identifier: %s", err)
+	}
+	sf.Add(identifier)
+	runIdentificationWithSF(sf, t)
+	config.Clear()()
+}
+
+// TestPronomNoReports performs the same tests as TestPronom, but
+// against a Siegfried created purely from a signature file.
+func TestPronomNoReports(t *testing.T) {
+	sf := siegfried.New()
+	config.SetHome(DataPath)
+	config.SetNoContainer()()
+	config.SetNoReports()()
+	if config.Reports() != "" {
+		t.Errorf("pronon.reports should be unset, not: %s", config.Reports())
+	}
+	identifier, err := pronom.New(config.SetLimit(minimalPronom))
+	if err != nil {
+		t.Errorf("Error creating new PRONOM identifier: %s", err)
+	}
+	sf.Add(identifier)
+	runIdentificationWithSF(sf, t)
 	config.Clear()()
 }
diff --git a/pkg/config/identifier.go b/pkg/config/identifier.go
@@ -316,6 +316,7 @@ func IsArchive(id string) Archive {
 // Clear clears loc and mimeinfo details to avoid pollution when creating multiple identifiers in same session
 func Clear() func() private {
 	return func() private {
+		identifier.noContainer = false
 		identifier.name = ""
 		identifier.extend = nil
 		identifier.limit = nil

diff --git a/pkg/pronom/identifier.go b/pkg/pronom/identifier.go
@@ -119,7 +119,7 @@ func New(opts ...config.Option) (core.Identifier, error) {
 	pronom = identifier.ApplyConfig(pronom)
 	id := &Identifier{
 		Base:     identifier.New(pronom, config.ZipPuid()),
-		hasClass: config.Reports() != "" && !config.NoClass(),
+		hasClass: !config.NoClass(),
 		infos:    infos(pronom.Infos()),
 	}
 	if id.Multi() == config.DROID {

diff --git a/pkg/pronom/internal/mappings/droid.go b/pkg/pronom/internal/mappings/droid.go
@@ -73,6 +73,7 @@ type FileFormat struct {
 	Name       string   `xml:",attr"`
 	Version    string   `xml:",attr"`
 	MIMEType   string   `xml:",attr"`
+	Types      string   `xml:"FormatTypes"`
 	Extensions []string `xml:"Extension"`
 	Signatures []int    `xml:"InternalSignatureID"`
 	Priorities []int    `xml:"HasPriorityOverFileFormatID"`

diff --git a/pkg/pronom/parseable.go b/pkg/pronom/parseable.go
@@ -233,6 +233,7 @@ func (d *droid) Infos() map[string]identifier.FormatInfo {
 			name:     strings.TrimSpace(v.Name),
 			version:  strings.TrimSpace(v.Version),
 			mimeType: strings.TrimSpace(v.MIMEType),
+			class:    strings.TrimSpace(v.Types),
 		}
 	}
 	return infos

diff --git a/pkg/pronom/pronom_test.go b/pkg/pronom/pronom_test.go
@@ -6,6 +6,7 @@ import (
 	"sort"
 	"testing"
 
+	"github.com/richardlehane/siegfried/internal/identifier"
 	"github.com/richardlehane/siegfried/pkg/config"
 )
 
@@ -21,15 +22,9 @@ func TestNew(t *testing.T) {
 	}
 }
 
-// TestFormatInfos inspects the values loaded into a PRONOM identifier
-// from a minimal PRONOM dataset, i.e. fewer than loading all of PRONOM.
-func TestFormatInfos(t *testing.T) {
-	config.SetHome(dataPath)
-	config.SetLimit(minimalPronom)()
-	i, err := NewPronom()
-	if err != nil {
-		t.Error(err)
-	}
+// verifyIdentifier provides a number of tests that can be run against
+// a PRONOM identifier.
+func verifyIdentifier(i identifier.Parseable, t *testing.T) {
 	const minReports int = 5
 	if len(i.Infos()) != minReports {
 		t.Error("Unexpected number of reports for PRONOM minimal tests")
@@ -84,7 +79,7 @@ func TestFormatInfos(t *testing.T) {
 	sort.Strings(puids)
 	sort.Strings(expectedPuids)
 	if !reflect.DeepEqual(puids, expectedPuids) {
-		t.Error("PUIDs from minimal PRONOM set do not match expected values")
+		t.Errorf("PUIDs from minimal PRONOM set do not match expected values;  expected %v; got %v", puids, expectedPuids)
 	}
 	sort.Strings(names)
 	sort.Strings(expectedNames)
@@ -94,17 +89,49 @@ func TestFormatInfos(t *testing.T) {
 	sort.Strings(versions)
 	sort.Strings(expectedVersions)
 	if !reflect.DeepEqual(versions, expectedVersions) {
-		t.Error("Format versions from minimal PRONOM set do not match expected values")
+		t.Errorf("Format versions from minimal PRONOM set do not match expected values;  expected %v; got %v", versions, expectedVersions)
 	}
 	sort.Strings(mimes)
 	sort.Strings(expectedMimes)
 	if !reflect.DeepEqual(mimes, expectedMimes) {
-		t.Error("MIMETypes from minimal PRONOM set do not match expected values")
+		t.Errorf("MIMETypes from minimal PRONOM set do not match expected values; expected %v; got %v", mimes, expectedMimes)
 	}
 	sort.Strings(types)
 	sort.Strings(expectedTypes)
 	if !reflect.DeepEqual(types, expectedTypes) {
-		t.Error("Format types from minimal PRONOM set do not match expected values")
+		t.Errorf("Format types from minimal PRONOM set do not match expected values; expected %v; got %v", types, expectedTypes)
+	}
+}
+
+// TestFormatInfosDefault inspects the values loaded into a PRONOM
+// identifier from a minimal PRONOM dataset, i.e. fewer than loading
+// all of PRONOM.
+func TestFormatInfosDefault(t *testing.T) {
+	config.SetHome(dataPath)
+	config.SetLimit(minimalPronom)()
+	i, err := NewPronom()
+	if err != nil {
+		t.Error(err)
+	}
+	verifyIdentifier(i, t)
+	config.Clear()()
+}
+
+// TestFormatInfosNoReports performs the same tests as TestFormatInfosDefault
+// but does so without loading PRONOM reports, preferring to create an
+// identifier using a signature file only.
+func TestFormatInfosNoReports(t *testing.T) {
+	config.SetHome(dataPath)
+	config.SetLimit(minimalPronom)()
+	config.SetNoContainer()()
+	config.SetNoReports()()
+	if config.Reports() != "" {
+		t.Errorf("pronon.reports should be unset, not: %s", config.Reports())
+	}
+	i, err := NewPronom()
+	if err != nil {
+		t.Error(err)
 	}
+	verifyIdentifier(i, t)
 	config.Clear()()
 }