Refactor drive item reader

src/internal/m365/onedrive/collection.go

@@ -21,6 +21,7 @@ import (
 	"github.com/alcionai/corso/src/internal/observe"
 	"github.com/alcionai/corso/src/pkg/backup/details"
 	"github.com/alcionai/corso/src/pkg/control"
+	"github.com/alcionai/corso/src/pkg/extensions"
 	"github.com/alcionai/corso/src/pkg/fault"
 	"github.com/alcionai/corso/src/pkg/logger"
 	"github.com/alcionai/corso/src/pkg/path"
@@ -403,17 +404,21 @@ func readItemContents(
 	return rc, nil
 }
 
+type driveStats struct {
+	dirsRead   int64
+	itemsRead  int64
+	byteCount  int64
+	itemsFound int64
+	dirsFound  int64
+}
+
 // populateItems iterates through items added to the collection
 // and uses the collection `itemReader` to read the item
 func (oc *Collection) populateItems(ctx context.Context, errs *fault.Bus) {
 	var (
-		byteCount  int64
-		itemsRead  int64
-		dirsRead   int64
-		itemsFound int64
-		dirsFound  int64
-		wg         sync.WaitGroup
 		el    = errs.Local()
+		stats driveStats
+		wg    sync.WaitGroup
 	)
 
 	// Retrieve the OneDrive folder path to set later in
@@ -445,12 +450,41 @@ func (oc *Collection) populateItems(ctx context.Context, errs *fault.Bus) {
 
 		wg.Add(1)
 
-		go func(ctx context.Context, item models.DriveItemable) {
+		go func(item models.DriveItemable) {
 			defer wg.Done()
 			defer func() { <-semaphoreCh }()
 
 			// Read the item
+			oc.populateDriveItem(
+				ctx,
+				parentPath,
+				item,
+				&stats,
+				&extensions.ItemExtensionHandler{},
+				oc.ctrl.BackupItemExtensions,
+				errs,
+			)
+
+			folderProgress <- struct{}{}
+		}(item) // TODO: is copy okay here?
+	}
+
+	wg.Wait()
+
+	oc.reportAsCompleted(ctx, int(stats.itemsFound), int(stats.itemsRead), stats.byteCount)
+}
+
+func (oc *Collection) populateDriveItem(
+	ctx context.Context,
+	parentPath *path.Builder,
+	item models.DriveItemable,
+	stats *driveStats,
+	aie extensions.AddItemExtensioner,
+	factories []extensions.CorsoItemExtensionFactory,
+	errs *fault.Bus,
+) {
 	var (
+		el           = errs.Local()
 		itemID       = ptr.Val(item.GetId())
 		itemName     = ptr.Val(item.GetName())
 		itemSize     = ptr.Val(item.GetSize())
@@ -473,12 +507,12 @@ func (oc *Collection) populateItems(ctx context.Context, errs *fault.Bus) {
 	isFile := item.GetFile() != nil
 
 	if isFile {
-				atomic.AddInt64(&itemsFound, 1)
+		atomic.AddInt64(&stats.itemsFound, 1)
 
 		metaFileName = itemID
 		metaSuffix = metadata.MetaFileSuffix
 	} else {
-				atomic.AddInt64(&dirsFound, 1)
+		atomic.AddInt64(&stats.dirsFound, 1)
 
 		// metaFileName not set for directories so we get just ".dirmeta"
 		metaSuffix = metadata.DirMetaFileSuffix
@@ -508,6 +542,32 @@ func (oc *Collection) populateItems(ctx context.Context, errs *fault.Bus) {
 				return nil, err
 			}
 
+			if aie != nil && len(factories) != 0 {
+				logger.Ctx(ctx).Info("enabling drive item extensions")
+
+				extRc, extInfo, err := aie.AddItemExtensions(
+					ctx,
+					itemData,
+					itemInfo,
+					oc.ctrl.BackupItemExtensions)
+				if err != nil {
+					return nil, clues.Wrap(err, "adding item extensions")
+				}
+
+				if extInfo == nil {
+					return nil, clues.New("nil extension info")
+				}
+
+				if extRc == nil {
+					return nil, clues.New("nil extension reader")
+				}
+
+				itemInfo.OneDrive.Extension = extInfo
+				itemData = extRc
+			} else {
+				logger.Ctx(ctx).Info("drive item extensions disabled")
+			}
+
 			// display/log the item download
 			progReader, _ := observe.ItemProgress(
 				ctx,
@@ -546,21 +606,12 @@ func (oc *Collection) populateItems(ctx context.Context, errs *fault.Bus) {
 
 	// Item read successfully, add to collection
 	if isFile {
-				atomic.AddInt64(&itemsRead, 1)
+		atomic.AddInt64(&stats.itemsRead, 1)
 	} else {
-				atomic.AddInt64(&dirsRead, 1)
+		atomic.AddInt64(&stats.dirsRead, 1)
 	}
 
-			// byteCount iteration
-			atomic.AddInt64(&byteCount, itemSize)
-
-			folderProgress <- struct{}{}
-		}(ctx, item)
-	}
-
-	wg.Wait()
-
-	oc.reportAsCompleted(ctx, int(itemsFound), int(itemsRead), byteCount)
+	atomic.AddInt64(&stats.byteCount, itemSize)
 }
 
 func (oc *Collection) reportAsCompleted(ctx context.Context, itemsFound, itemsRead int, byteCount int64) {

src/pkg/backup/details/details.go

@@ -988,6 +988,7 @@ type OneDriveInfo struct {
 	Owner      string         `json:"owner,omitempty"`
 	ParentPath string         `json:"parentPath"`
 	Size       int64          `json:"size,omitempty"`
+	Extension  *ExtensionInfo `json:"extensionData,omitempty"`
 }
 
 // Headers returns the human-readable names of properties in a OneDriveInfo

src/pkg/control/options.go

@@ -8,6 +8,7 @@ import (
 
 	"github.com/alcionai/corso/src/internal/common/dttm"
 	"github.com/alcionai/corso/src/pkg/control/repository"
+	"github.com/alcionai/corso/src/pkg/extensions"
 	"github.com/alcionai/corso/src/pkg/logger"
 )
 
@@ -20,6 +21,7 @@ type Options struct {
 	ToggleFeatures       Toggles                                `json:"toggleFeatures"`
 	Parallelism          Parallelism                            `json:"parallelism"`
 	Repo                 repository.Options                     `json:"repo"`
+	BackupItemExtensions []extensions.CorsoItemExtensionFactory `json:"-"`
 }
 
 type Parallelism struct {
@@ -49,6 +51,7 @@ func Defaults() Options {
 			CollectionBuffer: 4,
 			ItemFetch:        4,
 		},
+		BackupItemExtensions: []extensions.CorsoItemExtensionFactory{},
 	}
 }