Fix born digital SIP PREMIS event recording (#19)

Added support for born digital SIPs to PREMIS event recording.

internal/activities/add_premis_objects.go

@@ -8,13 +8,14 @@ import (
 	"github.com/google/uuid"
 
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/premis"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 const AddPREMISObjectsName = "add-premis-objects"
 
 type AddPREMISObjectsParams struct {
+	SIP            sip.SIP
 	PREMISFilePath string
-	ContentPath    string
 }
 
 type AddPREMISObjectsResult struct{}
@@ -30,7 +31,7 @@ func (md *AddPREMISObjectsActivity) Execute(
 	params *AddPREMISObjectsParams,
 ) (*AddPREMISObjectsResult, error) {
 	// Get subpaths of files in transfer.
-	subpaths, err := premis.FilesWithinDirectory(params.ContentPath)
+	subpaths, err := premis.FilesWithinDirectory(params.SIP.ContentPath)
 	if err != nil {
 		return nil, err
 	}
@@ -50,7 +51,7 @@ func (md *AddPREMISObjectsActivity) Execute(
 		object := premis.Object{
 			IdType:       "UUID",
 			IdValue:      uuid.New().String(),
-			OriginalName: premis.OriginalNameForSubpath(params.ContentPath, subpath),
+			OriginalName: premis.OriginalNameForSubpath(params.SIP, subpath),
 		}
 
 		err = premis.AppendObjectXML(doc, object)

internal/activities/add_premis_objects_test.go

@@ -9,7 +9,9 @@ import (
 	"gotest.tools/v3/fs"
 
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/activities"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/enums"
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/premis"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 func TestAddPREMISObjects(t *testing.T) {
@@ -62,24 +64,33 @@ func TestAddPREMISObjects(t *testing.T) {
 		{
 			name: "Add PREMIS objects for normal content",
 			params: activities.AddPREMISObjectsParams{
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: ContentFilesNormal.Path(),
+				},
 				PREMISFilePath: PREMISFilePathNormal,
-				ContentPath:    ContentFilesNormal.Path(),
 			},
 			result: activities.AddPREMISObjectsResult{},
 		},
 		{
 			name: "Add PREMIS objects for no content",
 			params: activities.AddPREMISObjectsParams{
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: ContentNoFiles.Path(),
+				},
 				PREMISFilePath: PREMISFilePathNoFiles,
-				ContentPath:    ContentNoFiles.Path(),
 			},
 			result: activities.AddPREMISObjectsResult{},
 		},
 		{
 			name: "Add PREMIS objects for bad path",
 			params: activities.AddPREMISObjectsParams{
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: ContentNonExistent.Path(),
+				},
 				PREMISFilePath: PREMISFilePathNonExistent,
-				ContentPath:    ContentNonExistent.Path(),
 			},
 			result:  activities.AddPREMISObjectsResult{},
 			wantErr: "no such file or directory",
@@ -116,7 +127,7 @@ func TestAddPREMISObjects(t *testing.T) {
 			assert.DeepEqual(t, res, tt.result)
 
 			// If the content directory has files, make sure that PREMIS file can be parsed.
-			contentFiles, err := premis.FilesWithinDirectory(tt.params.ContentPath)
+			contentFiles, err := premis.FilesWithinDirectory(tt.params.SIP.ContentPath)
 			assert.NilError(t, err)
 
 			if len(contentFiles) > 0 {

internal/activities/validate_file_formats.go

@@ -8,12 +8,13 @@ import (
 
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/fformat"
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/premis"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 const ValidateFileFormatsName = "validate-file-formats"
 
 type ValidateFileFormatsParams struct {
-	ContentPath    string
+	SIP            sip.SIP
 	PREMISFilePath string
 	Agent          premis.Agent
 }
@@ -72,7 +73,7 @@ func (a *ValidateFileFormats) Execute(
 		"fmt/653":   {},
 	}
 
-	err := filepath.WalkDir(params.ContentPath, func(p string, d fs.DirEntry, err error) error {
+	err := filepath.WalkDir(params.SIP.ContentPath, func(p string, d fs.DirEntry, err error) error {
 		if err != nil {
 			return err
 		}
@@ -106,13 +107,13 @@ func (a *ValidateFileFormats) Execute(
 		}
 
 		// Get subpath within content.
-		subpath, err := filepath.Rel(params.ContentPath, p)
+		subpath, err := filepath.Rel(params.SIP.ContentPath, p)
 		if err != nil {
 			return err
 		}
 
 		// Append PREMIS event to XML and write results.
-		originalName := premis.OriginalNameForSubpath(params.ContentPath, subpath)
+		originalName := premis.OriginalNameForSubpath(params.SIP, subpath)
 
 		doc, err := premis.ParseOrInitialize(params.PREMISFilePath)
 		if err != nil {

internal/activities/validate_file_formats_test.go

@@ -11,7 +11,9 @@ import (
 	"gotest.tools/v3/fs"
 
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/activities"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/enums"
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/premis"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 const pngContent = "\x89PNG\r\n\x1a\n\x00\x00\x00\x0DIHDR\x00\x00\x00\x01\x00\x00\x00\x01\x08\x02\x00\x00\x00\x90\x77\x53\xDE\x00\x00\x00\x00IEND\xAE\x42\x60\x82"
@@ -127,7 +129,10 @@ func TestValidateFileFormats(t *testing.T) {
 		{
 			name: "Successes with valid formats",
 			params: activities.ValidateFileFormatsParams{
-				ContentPath: validFormatsContentPath,
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: validFormatsContentPath,
+				},
 				PREMISFilePath: fs.NewFile(t, "premis.xml",
 					fs.WithContent(premisValidFormatsContent),
 				).Path(),
@@ -138,7 +143,10 @@ func TestValidateFileFormats(t *testing.T) {
 		{
 			name: "Fails with invalid formats",
 			params: activities.ValidateFileFormatsParams{
-				ContentPath: invalidFormatsContentPath,
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: invalidFormatsContentPath,
+				},
 				PREMISFilePath: fs.NewFile(t, "premis.xml",
 					fs.WithContent(premisInvalidFormatsContent),
 				).Path(),
@@ -163,7 +171,10 @@ func TestValidateFileFormats(t *testing.T) {
 		{
 			name: "Fails with an invalid content path",
 			params: activities.ValidateFileFormatsParams{
-				ContentPath: "/path/to/missing/dir",
+				SIP: sip.SIP{
+					Type:        enums.SIPTypeDigitizedAIP,
+					ContentPath: "/path/to/missing/dir",
+				},
 				PREMISFilePath: fs.NewFile(t, "premis.xml",
 					fs.WithContent(premis.EmptyXML),
 				).Path(),
@@ -174,9 +185,12 @@ func TestValidateFileFormats(t *testing.T) {
 		{
 			name: "Fails with empty source",
 			params: activities.ValidateFileFormatsParams{
-				ContentPath: fs.NewDir(t, "",
-					fs.WithFile("file.txt", ""),
-				).Path(),
+				SIP: sip.SIP{
+					Type: enums.SIPTypeDigitizedAIP,
+					ContentPath: fs.NewDir(t, "",
+						fs.WithFile("file.txt", ""),
+					).Path(),
+				},
 				PREMISFilePath: fs.NewFile(t, "premis.xml",
 					fs.WithContent(premis.EmptyXML),
 				).Path(),

internal/premis/premis.go

@@ -10,7 +10,9 @@ import (
 	"github.com/beevik/etree"
 	"github.com/google/uuid"
 
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/enums"
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/fsutil"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 const EmptyXML = `<?xml version="1.0" encoding="UTF-8"?>
@@ -419,11 +421,18 @@ func FilesWithinDirectory(contentPath string) ([]string, error) {
 	return subpaths, nil
 }
 
-func OriginalNameForSubpath(contentPath, subpath string) string {
-	transferDirName := filepath.Base(filepath.Dir(filepath.Dir(contentPath)))
+func OriginalNameForSubpath(sip sip.SIP, subpath string) string {
+	// Handle born digital SIP structure separately.
+	var transferDirName string
 
+	if sip.Type == enums.SIPTypeBornDigital {
+		transferDirName = filepath.Base(filepath.Dir(sip.ContentPath))
+	} else {
+		transferDirName = filepath.Base(filepath.Dir(filepath.Dir(sip.ContentPath)))
+	}
+
+	// Handle one file differently (as it gets renamed latest in TransformSIP).
 	if filepath.Base(subpath) == "Prozess_Digitalisierung_PREMIS.xml" {
-		// This file later gets renamed in TransformSIP.
 		parentDirName := filepath.Base(filepath.Dir(subpath))
 		filename := fmt.Sprintf("Prozess_Digitalisierung_PREMIS_%s.xml", parentDirName)
 		return filepath.Join("data", "metadata", filename)

internal/premis/premis_test.go

@@ -8,7 +8,9 @@ import (
 	"gotest.tools/v3/assert"
 	"gotest.tools/v3/fs"
 
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/enums"
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/premis"
+	"github.com/artefactual-sdps/preprocessing-sfa/internal/sip"
 )
 
 const premisObjectAddContent = `<?xml version="1.0" encoding="UTF-8"?>
@@ -219,22 +221,41 @@ func TestFilesWithinDirectory(t *testing.T) {
 }
 
 func TestOriginalNameForSubpath(t *testing.T) {
-	// Check for correct adjustment of file paths in PREMIS.
-	originalName := premis.OriginalNameForSubpath(
-		"test_transfer/content/content",
+	// Check for correct adjustment of AIP file path in PREMIS.
+	aipSIP := sip.SIP{
+		Type:        enums.SIPTypeDigitizedAIP,
+		ContentPath: "test_transfer/content/content",
+	}
+
+	aipOriginalName := premis.OriginalNameForSubpath(
+		aipSIP,
+		"d_0000001/00000001.jp2",
+	)
+
+	assert.Equal(t, aipOriginalName,
+		"data/objects/test_transfer/content/d_0000001/00000001.jp2")
+
+	// Check for correct adjustment of born digital SIP file path in PREMIS.
+	sipSIP := sip.SIP{
+		Type:        enums.SIPTypeBornDigital,
+		ContentPath: "test_transfer/content",
+	}
+
+	sipOriginalName := premis.OriginalNameForSubpath(
+		sipSIP,
 		"d_0000001/00000001.jp2",
 	)
 
-	assert.Equal(t, originalName,
+	assert.Equal(t, sipOriginalName,
 		"data/objects/test_transfer/content/d_0000001/00000001.jp2")
 
 	// Check for special handling of this specific file's path in PREMIS.
-	originalName = premis.OriginalNameForSubpath(
-		"test_transfer/content/content",
+	metadataOriginalName := premis.OriginalNameForSubpath(
+		aipSIP,
 		"content/content/d_0000001/Prozess_Digitalisierung_PREMIS.xml",
 	)
 
-	assert.Equal(t, originalName,
+	assert.Equal(t, metadataOriginalName,
 		"data/metadata/Prozess_Digitalisierung_PREMIS_d_0000001.xml")
 }
 
@@ -257,7 +278,11 @@ func TestAppendPREMISEventAndLinkToObject(t *testing.T) {
 	err = doc.ReadFromString(premisObjectAddContent)
 	assert.NilError(t, err)
 
-	originalName := premis.OriginalNameForSubpath("test_transfer/content/content", "cat.jpg")
+	aipSIP := sip.SIP{
+		Type:        enums.SIPTypeDigitizedAIP,
+		ContentPath: "test_transfer/content/content",
+	}
+	originalName := premis.OriginalNameForSubpath(aipSIP, "cat.jpg")
 
 	err = premis.AppendEventXMLForSingleObject(doc, eventSummary, premis.AgentDefault(), originalName)
 	assert.NilError(t, err)

internal/workflow/preprocessing.go

@@ -172,8 +172,8 @@ func (w *PreprocessingWorkflow) Execute(
 		withLocalActOpts(ctx),
 		activities.AddPREMISObjectsName,
 		&activities.AddPREMISObjectsParams{
+			SIP:            identifySIP.SIP,
 			PREMISFilePath: premisFilePath,
-			ContentPath:    identifySIP.SIP.ContentPath,
 		},
 	).Get(ctx, &addPREMISObjects)
 	if e != nil {
@@ -210,7 +210,7 @@ func (w *PreprocessingWorkflow) Execute(
 		withLocalActOpts(ctx),
 		activities.ValidateFileFormatsName,
 		&activities.ValidateFileFormatsParams{
-			ContentPath:    identifySIP.SIP.ContentPath,
+			SIP:            identifySIP.SIP,
 			PREMISFilePath: premisFilePath,
 			Agent:          premis.AgentDefault(),
 		},

internal/workflow/preprocessing_test.go

@@ -243,8 +243,8 @@ func (s *PreprocessingTestSuite) TestPreprocessingWorkflowSuccess() {
 		activities.AddPREMISObjectsName,
 		sessionCtx,
 		&activities.AddPREMISObjectsParams{
+			SIP:            expectedSIP,
 			PREMISFilePath: premisFilePath,
-			ContentPath:    expectedSIP.ContentPath,
 		},
 	).Return(
 		&activities.AddPREMISObjectsResult{}, nil,
@@ -267,7 +267,7 @@ func (s *PreprocessingTestSuite) TestPreprocessingWorkflowSuccess() {
 		activities.ValidateFileFormatsName,
 		sessionCtx,
 		&activities.ValidateFileFormatsParams{
-			ContentPath:    expectedSIP.ContentPath,
+			SIP:            expectedSIP,
 			PREMISFilePath: premisFilePath,
 			Agent:          premis.AgentDefault(),
 		},
@@ -478,7 +478,7 @@ func (s *PreprocessingTestSuite) TestPreprocessingWorkflowValidationFails() {
 		activities.ValidateFileFormatsName,
 		sessionCtx,
 		&activities.ValidateFileFormatsParams{
-			ContentPath:    expectedSIP.ContentPath,
+			SIP:            expectedSIP,
 			PREMISFilePath: premisFilePath,
 			Agent:          premis.AgentDefault(),
 		},