add both filters into one func

extends testing for filtered details results by
adding a regression test that crates a folder
with zero items, backs it up, and calls the
repo.BackupDetails to ensure that the folder
and its dirmeta files are not present.

CHANGELOG.md

@@ -12,6 +12,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 - Corso-generated .meta files and permissions no longer appear in the backup details.
+- repo.BackupDetails from the SDK now filter out meta and dirmeta files, and empty folders, by standard.
 
 ### Known Issues
 - Folders and Calendars containing zero items or subfolders are not included in the backup.

src/cli/backup/exchange.go

@@ -136,6 +136,8 @@ func addExchangeCommands(cmd *cobra.Command) *cobra.Command {
 		c.Use = c.Use + " " + exchangeServiceCommandDetailsUseSuffix
 		c.Example = exchangeServiceCommandDetailsExamples
 
+		options.AddSkipReduceFlag(c)
+
 		// Flags addition ordering should follow the order we want them to appear in help and docs:
 		// More generic (ex: --user) and more frequently used flags take precedence.
 		fs.StringVar(&backupID,
@@ -212,7 +214,6 @@ func addExchangeCommands(cmd *cobra.Command) *cobra.Command {
 			&contactFolder,
 			utils.ContactFolderFN, nil,
 			"Select backup details for contacts within a folder; accepts '"+utils.Wildcard+"' to select all contact folders.")
-
 		fs.StringVar(
 			&contactName,
 			utils.ContactNameFN, "",
@@ -434,7 +435,8 @@ func exchangeDetailsCmd() *cobra.Command {
 	}
 }
 
-// lists the history of backup operations
+// lists all items in the backup, running the results first through
+// selector reduction as a filtering step.
 func detailsExchangeCmd(cmd *cobra.Command, args []string) error {
 	if utils.HasNoFlagsAndShownHelp(cmd) {
 		return nil
@@ -468,14 +470,16 @@ func detailsExchangeCmd(cmd *cobra.Command, args []string) error {
 		return Only(ctx, err)
 	}
 
-	r, err := repository.Connect(ctx, acct, s, options.Control())
+	ctrlOpts := options.Control()
+
+	r, err := repository.Connect(ctx, acct, s, ctrlOpts)
 	if err != nil {
 		return Only(ctx, errors.Wrapf(err, "Failed to connect to the %s repository", s.Provider))
 	}
 
 	defer utils.CloseRepo(ctx, r)
 
-	ds, err := runDetailsExchangeCmd(ctx, r, backupID, opts)
+	ds, err := runDetailsExchangeCmd(ctx, r, backupID, opts, ctrlOpts.SkipReduce)
 	if err != nil {
 		return Only(ctx, err)
 	}
@@ -498,6 +502,7 @@ func runDetailsExchangeCmd(
 	r repository.BackupGetter,
 	backupID string,
 	opts utils.ExchangeOpts,
+	skipReduce bool,
 ) (*details.Details, error) {
 	if err := utils.ValidateExchangeRestoreFlags(backupID, opts); err != nil {
 		return nil, err
@@ -513,10 +518,13 @@ func runDetailsExchangeCmd(
 		return nil, errors.Wrap(errs.Failure(), "Failed to get backup details in the repository")
 	}
 
-	sel := utils.IncludeExchangeRestoreDataSelectors(opts)
-	utils.FilterExchangeRestoreInfoSelectors(sel, opts)
+	if !skipReduce {
+		sel := utils.IncludeExchangeRestoreDataSelectors(opts)
+		utils.FilterExchangeRestoreInfoSelectors(sel, opts)
+		d = sel.Reduce(ctx, d, errs)
+	}
 
-	return sel.Reduce(ctx, d, errs), nil
+	return d, nil
 }
 
 // ------------------------------------------------------------------------------------------------

src/cli/backup/exchange_test.go

@@ -231,7 +231,8 @@ func (suite *ExchangeSuite) TestExchangeBackupDetailsSelectors() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.NoError(t, err, "failure")
 			assert.ElementsMatch(t, test.Expected, output.Entries)
 		})
@@ -250,7 +251,8 @@ func (suite *ExchangeSuite) TestExchangeBackupDetailsSelectorsBadFormats() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.Error(t, err, "failure")
 			assert.Empty(t, output)
 		})

src/cli/backup/onedrive.go

@@ -103,6 +103,8 @@ func addOneDriveCommands(cmd *cobra.Command) *cobra.Command {
 		c.Use = c.Use + " " + oneDriveServiceCommandDetailsUseSuffix
 		c.Example = oneDriveServiceCommandDetailsExamples
 
+		options.AddSkipReduceFlag(c)
+
 		fs.StringVar(&backupID,
 			utils.BackupFN, "",
 			"ID of the backup to explore. (required)")
@@ -348,7 +350,9 @@ func detailsOneDriveCmd(cmd *cobra.Command, args []string) error {
 		return Only(ctx, err)
 	}
 
-	r, err := repository.Connect(ctx, acct, s, options.Control())
+	ctrlOpts := options.Control()
+
+	r, err := repository.Connect(ctx, acct, s, ctrlOpts)
 	if err != nil {
 		return Only(ctx, errors.Wrapf(err, "Failed to connect to the %s repository", s.Provider))
 	}
@@ -367,7 +371,7 @@ func detailsOneDriveCmd(cmd *cobra.Command, args []string) error {
 		Populated: utils.GetPopulatedFlags(cmd),
 	}
 
-	ds, err := runDetailsOneDriveCmd(ctx, r, backupID, opts)
+	ds, err := runDetailsOneDriveCmd(ctx, r, backupID, opts, ctrlOpts.SkipReduce)
 	if err != nil {
 		return Only(ctx, err)
 	}
@@ -390,6 +394,7 @@ func runDetailsOneDriveCmd(
 	r repository.BackupGetter,
 	backupID string,
 	opts utils.OneDriveOpts,
+	skipReduce bool,
 ) (*details.Details, error) {
 	if err := utils.ValidateOneDriveRestoreFlags(backupID, opts); err != nil {
 		return nil, err
@@ -405,10 +410,13 @@ func runDetailsOneDriveCmd(
 		return nil, errors.Wrap(errs.Failure(), "Failed to get backup details in the repository")
 	}
 
-	sel := utils.IncludeOneDriveRestoreDataSelectors(opts)
-	utils.FilterOneDriveRestoreInfoSelectors(sel, opts)
+	if !skipReduce {
+		sel := utils.IncludeOneDriveRestoreDataSelectors(opts)
+		utils.FilterOneDriveRestoreInfoSelectors(sel, opts)
+		d = sel.Reduce(ctx, d, errs)
+	}
 
-	return sel.Reduce(ctx, d, errs), nil
+	return d, nil
 }
 
 // `corso backup delete onedrive [<flag>...]`

src/cli/backup/onedrive_test.go

@@ -102,7 +102,8 @@ func (suite *OneDriveSuite) TestOneDriveBackupDetailsSelectors() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.NoError(t, err)
 			assert.ElementsMatch(t, test.Expected, output.Entries)
 		})
@@ -121,7 +122,8 @@ func (suite *OneDriveSuite) TestOneDriveBackupDetailsSelectorsBadFormats() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.Error(t, err)
 			assert.Empty(t, output)
 		})

src/cli/backup/sharepoint.go

@@ -114,6 +114,8 @@ func addSharePointCommands(cmd *cobra.Command) *cobra.Command {
 		c.Use = c.Use + " " + sharePointServiceCommandDetailsUseSuffix
 		c.Example = sharePointServiceCommandDetailsExamples
 
+		options.AddSkipReduceFlag(c)
+
 		fs.StringVar(&backupID,
 			utils.BackupFN, "",
 			"ID of the backup to retrieve.")
@@ -479,7 +481,9 @@ func detailsSharePointCmd(cmd *cobra.Command, args []string) error {
 		return Only(ctx, err)
 	}
 
-	r, err := repository.Connect(ctx, acct, s, options.Control())
+	ctrlOpts := options.Control()
+
+	r, err := repository.Connect(ctx, acct, s, ctrlOpts)
 	if err != nil {
 		return Only(ctx, errors.Wrapf(err, "Failed to connect to the %s repository", s.Provider))
 	}
@@ -495,7 +499,7 @@ func detailsSharePointCmd(cmd *cobra.Command, args []string) error {
 		Populated: utils.GetPopulatedFlags(cmd),
 	}
 
-	ds, err := runDetailsSharePointCmd(ctx, r, backupID, opts)
+	ds, err := runDetailsSharePointCmd(ctx, r, backupID, opts, ctrlOpts.SkipReduce)
 	if err != nil {
 		return Only(ctx, err)
 	}
@@ -518,6 +522,7 @@ func runDetailsSharePointCmd(
 	r repository.BackupGetter,
 	backupID string,
 	opts utils.SharePointOpts,
+	skipReduce bool,
 ) (*details.Details, error) {
 	if err := utils.ValidateSharePointRestoreFlags(backupID, opts); err != nil {
 		return nil, err
@@ -533,8 +538,11 @@ func runDetailsSharePointCmd(
 		return nil, errors.Wrap(errs.Failure(), "Failed to get backup details in the repository")
 	}
 
-	sel := utils.IncludeSharePointRestoreDataSelectors(opts)
-	utils.FilterSharePointRestoreInfoSelectors(sel, opts)
+	if !skipReduce {
+		sel := utils.IncludeSharePointRestoreDataSelectors(opts)
+		utils.FilterSharePointRestoreInfoSelectors(sel, opts)
+		d = sel.Reduce(ctx, d, errs)
+	}
 
-	return sel.Reduce(ctx, d, errs), nil
+	return d, nil
 }

src/cli/backup/sharepoint_test.go

@@ -219,7 +219,8 @@ func (suite *SharePointSuite) TestSharePointBackupDetailsSelectors() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.NoError(t, err)
 			assert.ElementsMatch(t, test.Expected, output.Entries)
 		})
@@ -238,7 +239,8 @@ func (suite *SharePointSuite) TestSharePointBackupDetailsSelectorsBadFormats() {
 				ctx,
 				test.BackupGetter,
 				"backup-ID",
-				test.Opts)
+				test.Opts,
+				false)
 			assert.Error(t, err)
 			assert.Empty(t, output)
 		})

src/cli/options/options.go

@@ -14,6 +14,7 @@ func Control() control.Options {
 	opt.FailFast = fastFail
 	opt.DisableMetrics = noStats
 	opt.RestorePermissions = restorePermissions
+	opt.SkipReduce = skipReduce
 	opt.ToggleFeatures.DisableIncrementals = disableIncrementals
 	opt.ToggleFeatures.EnablePermissionsBackup = enablePermissionsBackup
 
@@ -28,6 +29,7 @@ var (
 	fastFail           bool
 	noStats            bool
 	restorePermissions bool
+	skipReduce         bool
 )
 
 // AddOperationFlags adds command-local operation flags
@@ -52,6 +54,14 @@ func AddRestorePermissionsFlag(cmd *cobra.Command) {
 	cobra.CheckErr(fs.MarkHidden("restore-permissions"))
 }
 
+// AddSkipReduceFlag adds a hidden flag that allows callers to skip the selector
+// reduction step.  Currently only intended for details commands, not restore.
+func AddSkipReduceFlag(cmd *cobra.Command) {
+	fs := cmd.Flags()
+	fs.BoolVar(&skipReduce, "skip-reduce", false, "Skip the selector reduce filtering")
+	cobra.CheckErr(fs.MarkHidden("skip-reduce"))
+}
+
 // ---------------------------------------------------------------------------
 // Feature Flags
 // ---------------------------------------------------------------------------

src/internal/common/ptr/pointer.go

@@ -43,3 +43,10 @@ func OrNow(t *time.Time) time.Time {
 
 	return *t
 }
+
+// To makes it easy to get a pointer to ad-hoc primitives
+// without needing to declare additional variables.
+// ex: ptr.To("a string")
+func To[T any](t T) *T {
+	return &t
+}

src/internal/connector/onedrive/collections.go

@@ -272,7 +272,7 @@ func (c *Collections) Get(
 
 	retry := c.source == OneDriveSource
 
-	drives, err := drives(ctx, pager, retry)
+	drives, err := Drives(ctx, pager, retry)
 	if err != nil {
 		return nil, nil, err
 	}

src/internal/connector/onedrive/drive.go

@@ -72,7 +72,7 @@ func PagerForSource(
 	}
 }
 
-func drives(
+func Drives(
 	ctx context.Context,
 	pager drivePager,
 	retry bool,
@@ -362,7 +362,7 @@ func GetAllFolders(
 	prefix string,
 	errs *fault.Bus,
 ) ([]*Displayable, error) {
-	drives, err := drives(ctx, pager, true)
+	drives, err := Drives(ctx, pager, true)
 	if err != nil {
 		return nil, errors.Wrap(err, "getting OneDrive folders")
 	}

src/internal/connector/onedrive/drive_test.go

@@ -313,7 +313,7 @@ func (suite *OneDriveUnitSuite) TestDrives() {
 				toReturn: test.pagerResults,
 			}
 
-			drives, err := drives(ctx, pager, test.retry)
+			drives, err := Drives(ctx, pager, test.retry)
 			test.expectedErr(t, err)
 
 			assert.ElementsMatch(t, test.expectedResults, drives)
@@ -355,7 +355,7 @@ func (suite *OneDriveSuite) TestCreateGetDeleteFolder() {
 	pager, err := PagerForSource(OneDriveSource, gs, suite.userID, nil)
 	require.NoError(t, err)
 
-	drives, err := drives(ctx, pager, true)
+	drives, err := Drives(ctx, pager, true)
 	require.NoError(t, err)
 	require.NotEmpty(t, drives)
 

src/internal/connector/onedrive/item_test.go

@@ -46,7 +46,7 @@ func (suite *ItemIntegrationSuite) SetupSuite() {
 	pager, err := PagerForSource(OneDriveSource, suite.service, suite.user, nil)
 	require.NoError(t, err)
 
-	odDrives, err := drives(ctx, pager, true)
+	odDrives, err := Drives(ctx, pager, true)
 	require.NoError(t, err)
 	// Test Requirement 1: Need a drive
 	require.Greaterf(t, len(odDrives), 0, "user %s does not have a drive", suite.user)

src/pkg/backup/details/details.go

@@ -305,6 +305,158 @@ func (d *Details) addFolder(folder folderEntry) {
 	})
 }
 
+// FilterMetaFiles returns a new Details struct with a copy of the DetailsModel
+// that has had all .meta and .dirmeta files stripped out.
+func (d *Details) FilterMetaFiles() *Details {
+	return &Details{
+		DetailsModel: d.DetailsModel.FilterMetaFiles(),
+	}
+}
+
+// FilterEmptyContainers returns a new Details struct all empty (ie: containing no
+// items) stripped out.  If meta files have not been filtered out already, they
+// will continue to count as a "populated" container.
+func (d *Details) FilterEmptyContainers() *Details {
+	type entCount struct {
+		ent       DetailsEntry
+		itemCount int
+	}
+
+	var (
+		// shortRef: entCount
+		srec  = map[string]entCount{}
+		items = []DetailsEntry{}
+	)
+
+	// split the entries into items and folders.
+	// folders are stored in a map by their shortRef for lookup.
+	for _, ent := range d.Entries {
+		if ent.Folder == nil {
+			items = append(items, ent)
+		} else {
+			srec[ent.ShortRef] = entCount{ent, 0}
+		}
+	}
+
+	// for every item, add a count to the owning folder.
+	// this assumes item parentRef == folder shortRef.
+	for _, ent := range items {
+		if len(ent.ParentRef) == 0 {
+			continue
+		}
+
+		ec := srec[ent.ParentRef]
+		ec.itemCount++
+		srec[ent.ParentRef] = ec
+
+		// to maintain a hierarchical count so that we don't
+		// slice parent folders, this loop walks the tree upward
+		// by parent ref, adding one count to each parent up
+		// to the root.
+		parentRef := ec.ent.ParentRef
+		parentCount := 0
+
+		for len(parentRef) > 0 && parentCount == 0 {
+			ec := srec[parentRef]
+
+			// minor optimization: if the parentCount is already
+			// >zero, then all of its parents are guaranteed >zero.
+			parentCount = ec.itemCount
+
+			ec.itemCount++
+			srec[parentRef] = ec
+
+			parentRef = ec.ent.ParentRef
+		}
+	}
+
+	// walk the map of folder entries; every folder with one or more
+	// items gets added back to the items slice to be returned.
+	for _, ec := range srec {
+		if ec.itemCount > 0 {
+			items = append(items, ec.ent)
+		}
+	}
+
+	return &Details{
+		DetailsModel: DetailsModel{items},
+	}
+}
+
+// FilterEmptyContainers returns a new Details struct all empty (ie: containing no
+// items) stripped out.  If meta files have not been filtered out already, they
+// will continue to count as a "populated" container.
+func (d *Details) FilterBoth() *Details {
+	type entCount struct {
+		ent       DetailsEntry
+		itemCount int
+	}
+
+	var (
+		// shortRef: entCount
+		srec  = map[string]entCount{}
+		items = []DetailsEntry{}
+	)
+
+	// split the entries into items and folders.
+	// folders are stored in a map by their shortRef for lookup.
+	for _, ent := range d.Entries {
+		if ent.isMetaFile() {
+			continue
+		}
+
+		if ent.Folder == nil {
+			items = append(items, ent)
+		} else {
+			srec[ent.ShortRef] = entCount{ent, 0}
+		}
+	}
+
+	// for every item, add a count to the owning folder.
+	// this assumes item parentRef == folder shortRef.
+	for _, ent := range items {
+		if len(ent.ParentRef) == 0 {
+			continue
+		}
+
+		ec := srec[ent.ParentRef]
+		ec.itemCount++
+		srec[ent.ParentRef] = ec
+
+		// to maintain a hierarchical count so that we don't
+		// slice parent folders, this loop walks the tree upward
+		// by parent ref, adding one count to each parent up
+		// to the root.
+		parentRef := ec.ent.ParentRef
+		parentCount := 0
+
+		for len(parentRef) > 0 && parentCount == 0 {
+			ec := srec[parentRef]
+
+			// minor optimization: if the parentCount is already
+			// >zero, then all of its parents are guaranteed >zero.
+			parentCount = ec.itemCount
+
+			ec.itemCount++
+			srec[parentRef] = ec
+
+			parentRef = ec.ent.ParentRef
+		}
+	}
+
+	// walk the map of folder entries; every folder with one or more
+	// items gets added back to the items slice to be returned.
+	for _, ec := range srec {
+		if ec.itemCount > 0 {
+			items = append(items, ec.ent)
+		}
+	}
+
+	return &Details{
+		DetailsModel: DetailsModel{items},
+	}
+}
+
 // --------------------------------------------------------------------------------
 // Entry
 // --------------------------------------------------------------------------------

src/pkg/backup/details/details_benchmark_test.go

@@ -0,0 +1,157 @@
+package details
+
+import (
+	"math/rand"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+)
+
+var (
+	details512k = detailsBuilder(512000)
+	details256k = detailsBuilder(256000)
+	details102k = detailsBuilder(102000)
+)
+
+func detailsBuilder(fileCount int) *Details {
+	alpha := make([]string, 26)
+
+	for i := 0; i < 26; i++ {
+		alpha[i] = string(rune(97 + i))
+	}
+
+	alphaPath := basePath.Append(alpha...)
+	fes := FolderEntriesForPath(alphaPath, alphaPath)
+
+	// populate items
+	for i := 0; i < fileCount; i++ {
+		var (
+			ri       = rand.Int31n(26)
+			itemName = "item_" + uuid.NewString()
+			itemP    = basePath.Append(alpha[:ri]...).Append(itemName)
+		)
+
+		info := ItemInfo{
+			OneDrive: &OneDriveInfo{
+				ItemName: itemName,
+			},
+		}
+
+		ent := folderEntry{
+			RepoRef:   itemP.String(),
+			ShortRef:  itemP.ShortRef(),
+			ParentRef: itemP.Dir().String(),
+			Info:      info,
+		}
+
+		fes = append(fes, ent)
+	}
+
+	// populate 13 empty folders with dirmeta items
+	for i := 0; i < 13; i++ {
+		var (
+			ri       = rand.Int31n(26)
+			fldName  = "empty_" + uuid.NewString()
+			fldP     = basePath.Append(alpha[:ri]...).Append(fldName)
+			itemName = fldName + ".dirmeta"
+			itemP    = fldP.Append(itemName)
+		)
+
+		// folder
+		info := ItemInfo{
+			Folder: &FolderInfo{
+				DisplayName: fldName,
+			},
+		}
+
+		ent := folderEntry{
+			RepoRef:   fldP.String(),
+			ShortRef:  fldP.ShortRef(),
+			ParentRef: fldP.Dir().String(),
+			Info:      info,
+		}
+
+		fes = append(fes, ent)
+
+		// dirmeta
+		info = ItemInfo{
+			OneDrive: &OneDriveInfo{
+				ItemName: itemName,
+			},
+		}
+
+		ent = folderEntry{
+			RepoRef:   itemP.String(),
+			ShortRef:  itemP.ShortRef(),
+			ParentRef: itemP.Dir().String(),
+			Info:      info,
+		}
+
+		fes = append(fes, ent)
+	}
+
+	// shuffle the array, to avoid miscalculations due to ordering
+	rand.Seed(time.Now().UnixNano())
+	rand.Shuffle(len(fes), func(i, j int) {
+		fes[i], fes[j] = fes[j], fes[i]
+	})
+
+	// return it as a details struct
+	return toDetails(fes)
+}
+
+var result *Details
+
+func BenchmarkDetailsFiltering_512_both(b *testing.B)      { benchmarkBoth(details512k, b) }
+func BenchmarkDetailsFiltering_256_both(b *testing.B)      { benchmarkBoth(details256k, b) }
+func BenchmarkDetailsFiltering_102_both(b *testing.B)      { benchmarkBoth(details102k, b) }
+func BenchmarkDetailsFiltering_512_each(b *testing.B)      { benchmarkEach(details512k, b) }
+func BenchmarkDetailsFiltering_256_each(b *testing.B)      { benchmarkEach(details256k, b) }
+func BenchmarkDetailsFiltering_102_each(b *testing.B)      { benchmarkEach(details102k, b) }
+func BenchmarkDetailsFiltering_512_meta(b *testing.B)      { benchmarkMeta(details512k, b) }
+func BenchmarkDetailsFiltering_256_meta(b *testing.B)      { benchmarkMeta(details256k, b) }
+func BenchmarkDetailsFiltering_102_meta(b *testing.B)      { benchmarkMeta(details102k, b) }
+func BenchmarkDetailsFiltering_512_container(b *testing.B) { benchmarkContainer(details512k, b) }
+func BenchmarkDetailsFiltering_256_container(b *testing.B) { benchmarkContainer(details256k, b) }
+func BenchmarkDetailsFiltering_102_container(b *testing.B) { benchmarkContainer(details102k, b) }
+
+func benchmarkBoth(d *Details, b *testing.B) {
+	var d2 *Details
+
+	for n := 0; n < b.N; n++ {
+		d2 = d.FilterBoth()
+	}
+
+	result = d2
+}
+
+func benchmarkEach(d *Details, b *testing.B) {
+	var d2 *Details
+
+	for n := 0; n < b.N; n++ {
+		d2 = d.FilterMetaFiles().FilterEmptyContainers()
+	}
+
+	result = d2
+}
+
+func benchmarkMeta(d *Details, b *testing.B) {
+	var d2 *Details
+
+	for n := 0; n < b.N; n++ {
+		d2 = d.FilterMetaFiles()
+	}
+
+	result = d2
+}
+
+func benchmarkContainer(d *Details, b *testing.B) {
+	var d2 *Details
+
+	for n := 0; n < b.N; n++ {
+		d2 = d.FilterEmptyContainers()
+	}
+
+	result = d2
+}

src/pkg/backup/details/details_test.go

@@ -995,3 +995,103 @@ func (suite *DetailsUnitSuite) TestFolderEntriesForPath() {
 		})
 	}
 }
+
+func (suite *DetailsUnitSuite) TestDetails_FilterMetaFiles() {
+	t := suite.T()
+
+	dm := DetailsModel{
+		Entries: []DetailsEntry{
+			{
+				RepoRef: "a.data",
+				ItemInfo: ItemInfo{
+					OneDrive: &OneDriveInfo{IsMeta: false},
+				},
+			},
+			{
+				RepoRef: "b.meta",
+				ItemInfo: ItemInfo{
+					OneDrive: &OneDriveInfo{IsMeta: false},
+				},
+			},
+			{
+				RepoRef: "c.meta",
+				ItemInfo: ItemInfo{
+					OneDrive: &OneDriveInfo{IsMeta: true},
+				},
+			},
+		},
+	}
+
+	d := &Details{dm}
+
+	d2 := d.FilterMetaFiles()
+
+	assert.Len(t, d2.DetailsModel.Entries, 1)
+	assert.Len(t, d.DetailsModel.Entries, 3)
+}
+
+func toDetails(fes []folderEntry) *Details {
+	d := &Details{
+		DetailsModel: DetailsModel{
+			Entries: make([]DetailsEntry, len(fes)),
+		},
+	}
+
+	for i, fe := range fes {
+		d.DetailsModel.Entries[i] = DetailsEntry{
+			RepoRef:     fe.RepoRef,
+			ShortRef:    fe.ShortRef,
+			ParentRef:   fe.ParentRef,
+			LocationRef: fe.LocationRef,
+			ItemInfo:    fe.Info,
+		}
+	}
+
+	return d
+}
+
+func (suite *DetailsUnitSuite) TestDetails_FilterEmptyContainers() {
+	var (
+		t     = suite.T()
+		empty = basePath.Append("populated", "empty")
+		fes   = FolderEntriesForPath(empty, empty)
+		d     = toDetails(fes)
+		itemP = basePath.Append("populated", "item")
+	)
+
+	item := DetailsEntry{
+		RepoRef:     itemP.String(),
+		ShortRef:    itemP.ShortRef(),
+		ParentRef:   itemP.Dir().ShortRef(),
+		LocationRef: "todo - not currently needed",
+		ItemInfo: ItemInfo{
+			OneDrive: &OneDriveInfo{
+				ItemName: "item",
+			},
+		},
+	}
+
+	d.DetailsModel.Entries = append(d.DetailsModel.Entries, item)
+
+	var (
+		ds     = d.DetailsModel.Entries
+		result = d.FilterEmptyContainers()
+		rs     = result.DetailsModel.Entries
+	)
+
+	assert.Equal(t, len(ds)-1, len(rs), "one empty folder should have been removed")
+
+	for _, r := range rs {
+		assert.NotEqual(t, empty.String(), r.RepoRef, "the empty path should have been removed")
+	}
+
+	dFilt := make([]DetailsEntry, 0, len(ds)-1)
+
+	for _, d := range ds {
+		if d.RepoRef != empty.String() {
+			dFilt = append(dFilt, d)
+		}
+	}
+
+	assert.ElementsMatch(t, dFilt, rs, "all other paths should be present")
+}

src/pkg/control/options.go

@@ -10,6 +10,7 @@ type Options struct {
 	DisableMetrics     bool            `json:"disableMetrics"`
 	FailFast           bool            `json:"failFast"`
 	RestorePermissions bool            `json:"restorePermissions"`
+	SkipReduce         bool            `json:"skipReduce"`
 	ToggleFeatures     Toggles         `json:"ToggleFeatures"`
 }
 

src/pkg/repository/loadtest/repository_load_test.go

@@ -395,7 +395,7 @@ func normalizeCategorySet(t *testing.T, cats map[string]struct{}) []string {
 // multiple users
 
 type RepositoryLoadTestExchangeSuite struct {
-	suite.Suite
+	tester.Suite
 	ctx            context.Context
 	repo           repository.Repository
 	acct           account.Account //lint:ignore U1000 future test use

src/pkg/repository/repository.go

@@ -2,7 +2,6 @@ package repository
 
 import (
 	"context"
-	"strings"
 	"time"
 
 	"github.com/alcionai/clues"
@@ -10,14 +9,12 @@ import (
 	"github.com/pkg/errors"
 
 	"github.com/alcionai/corso/src/internal/common/crash"
-	"github.com/alcionai/corso/src/internal/connector/onedrive"
 	"github.com/alcionai/corso/src/internal/events"
 	"github.com/alcionai/corso/src/internal/kopia"
 	"github.com/alcionai/corso/src/internal/model"
 	"github.com/alcionai/corso/src/internal/observe"
 	"github.com/alcionai/corso/src/internal/operations"
 	"github.com/alcionai/corso/src/internal/streamstore"
-	"github.com/alcionai/corso/src/internal/version"
 	"github.com/alcionai/corso/src/pkg/account"
 	"github.com/alcionai/corso/src/pkg/backup"
 	"github.com/alcionai/corso/src/pkg/backup/details"
@@ -361,21 +358,7 @@ func (r repository) BackupDetails(
 		return nil, nil, errs.Fail(err)
 	}
 
-	// Retroactively fill in isMeta information for items in older
-	// backup versions without that info
-	// version.Restore2 introduces the IsMeta flag, so only v1 needs a check.
-	if b.Version >= version.OneDrive1DataAndMetaFiles && b.Version < version.OneDrive3IsMetaMarker {
-		for _, d := range deets.Entries {
-			if d.OneDrive != nil {
-				if strings.HasSuffix(d.RepoRef, onedrive.MetaFileSuffix) ||
-					strings.HasSuffix(d.RepoRef, onedrive.DirMetaFileSuffix) {
-					d.OneDrive.IsMeta = true
-				}
-			}
-		}
-	}
-
-	return deets, b, errs
+	return deets.FilterMetaFiles().FilterEmptyContainers(), b, errs
 }
 
 // DeleteBackup removes the backup from both the model store and the backup storage.

src/pkg/repository/repository_test.go

@@ -3,10 +3,15 @@ package repository_test
 import (
 	"testing"
 
+	"github.com/microsoftgraph/msgraph-sdk-go/models"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"github.com/stretchr/testify/suite"
 
+	"github.com/alcionai/corso/src/internal/common"
+	"github.com/alcionai/corso/src/internal/common/ptr"
+	"github.com/alcionai/corso/src/internal/connector/graph"
+	"github.com/alcionai/corso/src/internal/connector/onedrive"
 	"github.com/alcionai/corso/src/internal/tester"
 	"github.com/alcionai/corso/src/pkg/account"
 	"github.com/alcionai/corso/src/pkg/control"
@@ -97,6 +102,7 @@ func (suite *RepositorySuite) TestConnect() {
 
 type RepositoryIntegrationSuite struct {
 	tester.Suite
+	userID string
 }
 
 func TestRepositoryIntegrationSuite(t *testing.T) {
@@ -109,6 +115,10 @@ func TestRepositoryIntegrationSuite(t *testing.T) {
 	})
 }
 
+func (suite *RepositoryIntegrationSuite) SetupSuite() {
+	suite.userID = tester.M365UserID(suite.T())
+}
+
 func (suite *RepositoryIntegrationSuite) TestInitialize() {
 	ctx, flush := tester.NewContext()
 	defer flush()
@@ -219,3 +229,67 @@ func (suite *RepositoryIntegrationSuite) TestNewRestore() {
 	require.NoError(t, err)
 	require.NotNil(t, ro)
 }
+
+func (suite *RepositoryIntegrationSuite) TestBackupDetails_regression() {
+	ctx, flush := tester.NewContext()
+	defer flush()
+
+	var (
+		t    = suite.T()
+		acct = tester.NewM365Account(t)
+		st   = tester.NewPrefixedS3Storage(t)
+		dest = "Corso_Restore_empty_" + common.FormatNow(common.SimpleTimeTesting)
+	)
+
+	m365, err := acct.M365Config()
+	require.NoError(t, err)
+
+	adpt, err := graph.CreateAdapter(acct.ID(), m365.AzureClientID, m365.AzureClientSecret)
+	require.NoError(t, err)
+
+	srv := graph.NewService(adpt)
+
+	pager, err := onedrive.PagerForSource(onedrive.OneDriveSource, srv, suite.userID, nil)
+	require.NoError(t, err)
+
+	drives, err := onedrive.Drives(ctx, pager, false)
+	require.NoError(t, err)
+
+	d0 := drives[0]
+	body := models.DriveItem{}
+	body.SetName(&dest)
+
+	fld := models.Folder{}
+	fld.SetChildCount(ptr.To[int32](0))
+	body.SetFolder(&fld)
+
+	_, err = srv.Client().
+		UsersById(suite.userID).
+		DrivesById(*d0.GetId()).
+		Items().
+		Post(ctx, &body, nil)
+	require.NoErrorf(t, err, "%+v", graph.ErrData(err))
+
+	r, err := repository.Initialize(ctx, acct, st, control.Options{})
+	require.NoError(t, err)
+
+	sel := selectors.NewOneDriveBackup([]string{suite.userID})
+	sel.Include(sel.Folders([]string{dest}))
+
+	op, err := r.NewBackup(ctx, sel.Selector)
+	require.NoError(t, err)
+	require.NoError(t, op.Run(ctx))
+	require.NotZero(t, op.Results.ItemsWritten)
+
+	// the actual test.  The backup details, having backed up an empty folder,
+	// should not return the folder within the backup details.  That value
+	// should get filtered out, along with .meta and .dirmeta files.
+	deets, _, ferr := r.BackupDetails(ctx, string(op.Results.BackupID))
+	require.NoError(t, ferr.Failure())
+
+	for _, ent := range deets.Entries {
+		assert.NotContains(t, ent.RepoRef, dest)
+		assert.NotContains(t, ent.RepoRef, onedrive.MetaFileSuffix)
+		assert.NotContains(t, ent.RepoRef, onedrive.DirMetaFileSuffix)
+	}
+}