Avoid multiple file tree traversals on bundle deploy

bundle/bundle.go

@@ -16,6 +16,7 @@ import (
 	"github.com/databricks/cli/bundle/config"
 	"github.com/databricks/cli/bundle/env"
 	"github.com/databricks/cli/bundle/metadata"
+	"github.com/databricks/cli/libs/fileset"
 	"github.com/databricks/cli/libs/folders"
 	"github.com/databricks/cli/libs/git"
 	"github.com/databricks/cli/libs/locker"
@@ -50,6 +51,9 @@ type Bundle struct {
 	clientOnce sync.Once
 	client     *databricks.WorkspaceClient
 
+	// Files that are synced to the workspace.file_path
+	SyncedFiles []fileset.File
+
 	// Stores an initialized copy of this bundle's Terraform wrapper.
 	Terraform *tfexec.Terraform
 

bundle/deploy/files/upload.go

@@ -23,7 +23,7 @@ func (m *upload) Apply(ctx context.Context, b *bundle.Bundle) diag.Diagnostics {
 		return diag.FromErr(err)
 	}
 
-	err = sync.RunOnce(ctx)
+	b.SyncedFiles, err = sync.RunOnce(ctx)
 	if err != nil {
 		return diag.FromErr(err)
 	}

bundle/deploy/state_update.go

@@ -11,7 +11,6 @@ import (
 	"time"
 
 	"github.com/databricks/cli/bundle"
-	"github.com/databricks/cli/bundle/deploy/files"
 	"github.com/databricks/cli/internal/build"
 	"github.com/databricks/cli/libs/diag"
 	"github.com/databricks/cli/libs/log"
@@ -40,19 +39,8 @@ func (s *stateUpdate) Apply(ctx context.Context, b *bundle.Bundle) diag.Diagnost
 	state.CliVersion = build.GetInfo().Version
 	state.Version = DeploymentStateVersion
 
-	// Get the current file list.
-	sync, err := files.GetSync(ctx, bundle.ReadOnly(b))
-	if err != nil {
-		return diag.FromErr(err)
-	}
-
-	files, err := sync.GetFileList(ctx)
-	if err != nil {
-		return diag.FromErr(err)
-	}
-
-	// Update the state with the current file list.
-	fl, err := FromSlice(files)
+	// Update the state with the current list of synced files.
+	fl, err := FromSlice(b.SyncedFiles)
 	if err != nil {
 		return diag.FromErr(err)
 	}

cmd/bundle/sync.go

@@ -72,7 +72,8 @@ func newSyncCommand() *cobra.Command {
 			return s.RunContinuous(ctx)
 		}
 
-		return s.RunOnce(ctx)
+		_, err = s.RunOnce(ctx)
+		return err
 	}
 
 	return cmd

cmd/sync/sync.go

@@ -135,7 +135,7 @@ func New() *cobra.Command {
 		if f.watch {
 			err = s.RunContinuous(ctx)
 		} else {
-			err = s.RunOnce(ctx)
+			_, err = s.RunOnce(ctx)
 		}
 
 		s.Close()

libs/sync/sync.go

@@ -152,36 +152,41 @@ func (s *Sync) notifyComplete(ctx context.Context, d diff) {
 	s.seq++
 }
 
-func (s *Sync) RunOnce(ctx context.Context) error {
+// Upload all files in the file tree rooted at the local path configured in the
+// SyncOptions to the remote path configured in the SyncOptions.
+//
+// Returns the list of files tracked (and synchronized) by the syncer during the run,
+// and an error if any occurred.
+func (s *Sync) RunOnce(ctx context.Context) ([]fileset.File, error) {
 	files, err := s.GetFileList(ctx)
 	if err != nil {
-		return err
+		return files, err
 	}
 
 	change, err := s.snapshot.diff(ctx, files)
 	if err != nil {
-		return err
+		return files, err
 	}
 
 	s.notifyStart(ctx, change)
 	if change.IsEmpty() {
 		s.notifyComplete(ctx, change)
-		return nil
+		return files, nil
 	}
 
 	err = s.applyDiff(ctx, change)
 	if err != nil {
-		return err
+		return files, err
 	}
 
 	err = s.snapshot.Save(ctx)
 	if err != nil {
 		log.Errorf(ctx, "cannot store snapshot: %s", err)
-		return err
+		return files, err
 	}
 
 	s.notifyComplete(ctx, change)
-	return nil
+	return files, nil
 }
 
 func (s *Sync) GetFileList(ctx context.Context) ([]fileset.File, error) {
@@ -231,7 +236,7 @@ func (s *Sync) RunContinuous(ctx context.Context) error {
 		case <-ctx.Done():
 			return ctx.Err()
 		case <-ticker.C:
-			err := s.RunOnce(ctx)
+			_, err := s.RunOnce(ctx)
 			if err != nil {
 				return err
 			}