restore: New --verify flag to verify restored files content

Signed-off-by: Igor Fedorenko <igor@ifedorenko.com>

changelog/unreleased/pull-1772

@@ -0,0 +1,6 @@
+Enhancement: Add restore --verify to verify restored file content
+
+Restore will print error message if restored file content does not match
+expected SHA256 checksum
+
+https://github.com/restic/restic/pull/1772

cmd/restic/cmd_restore.go

@@ -34,6 +34,7 @@ type RestoreOptions struct {
 	Host    string
 	Paths   []string
 	Tags    restic.TagLists
+	Verify  bool
 }
 
 var restoreOptions RestoreOptions
@@ -49,6 +50,7 @@ func init() {
 	flags.StringVarP(&restoreOptions.Host, "host", "H", "", `only consider snapshots for this host when the snapshot ID is "latest"`)
 	flags.Var(&restoreOptions.Tags, "tag", "only consider snapshots which include this `taglist` for snapshot ID \"latest\"")
 	flags.StringArrayVar(&restoreOptions.Paths, "path", nil, "only consider snapshots which include this (absolute) `path` for snapshot ID \"latest\"")
+	flags.BoolVar(&restoreOptions.Verify, "verify", false, "verify restored files content")
 }
 
 func runRestore(opts RestoreOptions, gopts GlobalOptions, args []string) error {
@@ -154,6 +156,12 @@ func runRestore(opts RestoreOptions, gopts GlobalOptions, args []string) error {
 	Verbosef("restoring %s to %s\n", res.Snapshot(), opts.Target)
 
 	err = res.RestoreTo(ctx, opts.Target)
+	if err == nil && opts.Verify {
+		Verbosef("verifying files in %s\n", opts.Target)
+		var count int
+		count, err = res.VerifyFiles(ctx, opts.Target)
+		Verbosef("finished verifying %d files in %s\n", count, opts.Target)
+	}
 	if totalErrors > 0 {
 		Printf("There were %d errors\n", totalErrors)
 	}

internal/restorer/restorer.go

@@ -2,8 +2,10 @@ package restorer
 
 import (
 	"context"
+	"os"
 	"path/filepath"
 
+	"github.com/restic/restic/internal/crypto"
 	"github.com/restic/restic/internal/errors"
 
 	"github.com/restic/restic/internal/debug"
@@ -218,3 +220,51 @@ func (res *Restorer) RestoreTo(ctx context.Context, dst string) error {
 func (res *Restorer) Snapshot() *restic.Snapshot {
 	return res.sn
 }
+
+// VerifyFiles reads all snapshot files and verifies their contents
+func (res *Restorer) VerifyFiles(ctx context.Context, dst string) (int, error) {
+	// TODO multithreaded?
+
+	count := 0
+	err := res.traverseTree(ctx, dst, string(filepath.Separator), *res.sn.Tree, treeVisitor{
+		enterDir: func(node *restic.Node, target, location string) error { return nil },
+		visitNode: func(node *restic.Node, target, location string) error {
+			if node.Type != "file" {
+				return nil
+			}
+
+			count++
+			stat, err := os.Stat(target)
+			if err != nil {
+				return err
+			}
+			if int64(node.Size) != stat.Size() {
+				return errors.Errorf("Invalid file size: expected %d got %d", node.Size, stat.Size())
+			}
+
+			offset := int64(0)
+			for _, blobID := range node.Content {
+				rd, err := os.Open(target)
+				if err != nil {
+					return err
+				}
+				blobs, _ := res.repo.Index().Lookup(blobID, restic.DataBlob)
+				length := blobs[0].Length - uint(crypto.Extension)
+				buf := make([]byte, length) // TODO do I want to reuse the buffer somehow?
+				_, err = rd.ReadAt(buf, offset)
+				if err != nil {
+					return err
+				}
+				if !blobID.Equal(restic.Hash(buf)) {
+					return errors.Errorf("Unexpected contents starting at offset %d", offset)
+				}
+				offset += int64(length)
+			}
+
+			return nil
+		},
+		leaveDir: func(node *restic.Node, target, location string) error { return nil },
+	})
+
+	return count, err
+}