checker: Add checker and tests

checker/checker.go

@@ -2,26 +2,15 @@ package checker
 
 import (
 	"encoding/hex"
+	"errors"
+	"fmt"
 
+	"github.com/restic/restic"
 	"github.com/restic/restic/backend"
 	"github.com/restic/restic/debug"
 	"github.com/restic/restic/repository"
 )
 
-// Error is an error in the repository detected by the checker.
-type Error struct {
-	Message string
-	Err     error
-}
-
-func (e Error) Error() string {
-	if e.Err != nil {
-		return e.Message + ": " + e.Err.Error()
-	}
-
-	return e.Message
-}
-
 type mapID [backend.IDSize]byte
 
 func id2map(id backend.ID) (mid mapID) {
@@ -38,6 +27,14 @@ func str2map(s string) (mid mapID, err error) {
 	return id2map(data), nil
 }
 
+func map2str(id mapID) string {
+	return hex.EncodeToString(id[:])
+}
+
+func map2id(id mapID) backend.ID {
+	return backend.ID(id[:])
+}
+
 // Checker runs various checks on a repository. It is advisable to create an
 // exclusive Lock in the repository before running any checks.
 //
@@ -127,10 +124,199 @@ func (c *Checker) LoadIndex() error {
 	}
 
 	debug.Log("LoadIndex", "done, error %v", perr)
+
+	c.repo.SetIndex(c.masterIndex)
+
 	return perr
 }
 
-// Packs checks that all packs referenced in the index are still available.
+// PackError describes an error with a specific pack.
-func (c *Checker) Packs() error {
+type PackError struct {
-	return nil
+	ID backend.ID
+	error
+}
+
+func (e PackError) Error() string {
+	return "pack " + e.ID.String() + ": " + e.error.Error()
+}
+
+// Packs checks that all packs referenced in the index are still available and
+// there are no packs that aren't in an index.
+func (c *Checker) Packs() (errs []error) {
+	debug.Log("Checker.Packs", "checking for %d packs", len(c.packs))
+	seenPacks := make(map[mapID]struct{})
+
+	for id := range c.packs {
+		seenPacks[id] = struct{}{}
+		ok, err := c.repo.Backend().Test(backend.Data, map2str(id))
+		if err != nil {
+			debug.Log("Checker.Packs", "error checking for pack %s", map2id(id).Str())
+			errs = append(errs, PackError{map2id(id), err})
+			continue
+		}
+
+		if !ok {
+			debug.Log("Checker.Packs", "pack %s does not exist", map2id(id).Str())
+			errs = append(errs, PackError{map2id(id), errors.New("does not exist")})
+			continue
+		}
+		debug.Log("Checker.Packs", "pack %s exists", map2id(id).Str())
+	}
+
+	done := make(chan struct{})
+	defer close(done)
+
+	for id := range c.repo.List(backend.Data, done) {
+		if _, ok := seenPacks[id2map(id)]; !ok {
+			errs = append(errs, PackError{id, errors.New("not referenced in any index")})
+		}
+	}
+
+	return errs
+}
+
+// Error is an error that occurred while checking a repository.
+type Error struct {
+	TreeID backend.ID
+	BlobID backend.ID
+	Err    error
+}
+
+func (e Error) Error() string {
+	if e.BlobID != nil && e.TreeID != nil {
+		msg := "tree " + e.TreeID.String()
+		msg += ", blob " + e.BlobID.String()
+		msg += ": " + e.Err.Error()
+		return msg
+	}
+
+	if e.TreeID != nil {
+		return "tree " + e.TreeID.String() + ": " + e.Err.Error()
+	}
+
+	return e.Err.Error()
+}
+
+func loadTreeFromSnapshot(repo *repository.Repository, id backend.ID) (backend.ID, error) {
+	sn, err := restic.LoadSnapshot(repo, id)
+	if err != nil {
+		debug.Log("Checker.loadTreeFromSnapshot", "error loading snapshot %v: %v", id.Str(), err)
+		return nil, err
+	}
+
+	if sn.Tree == nil {
+		debug.Log("Checker.loadTreeFromSnapshot", "snapshot %v has no tree", id.Str())
+		return nil, fmt.Errorf("snapshot %v has no tree", id)
+	}
+
+	return sn.Tree, nil
+}
+
+// Structure checks that for all snapshots all referenced blobs are available
+// in the index.
+func (c *Checker) Structure() (errs []error) {
+	done := make(chan struct{})
+	defer close(done)
+
+	var todo backend.IDs
+
+	for id := range c.repo.List(backend.Snapshot, done) {
+		debug.Log("Checker.Snaphots", "check snapshot %v", id.Str())
+
+		treeID, err := loadTreeFromSnapshot(c.repo, id)
+		if err != nil {
+			errs = append(errs, err)
+			continue
+		}
+
+		debug.Log("Checker.Snaphots", "snapshot %v has tree %v", id.Str(), treeID.Str())
+		todo = append(todo, treeID)
+	}
+
+	errs = append(errs, c.trees(todo)...)
+	return errs
+}
+
+func (c *Checker) trees(treeIDs backend.IDs) (errs []error) {
+	treesChecked := make(map[mapID]struct{})
+
+	for len(treeIDs) > 0 {
+		id := treeIDs[0]
+		treeIDs = treeIDs[1:]
+
+		c.blobRefs[id2map(id)]++
+		debug.Log("Checker.trees", "tree %v refcount %d", id.Str(), c.blobRefs[id2map(id)])
+
+		if _, ok := treesChecked[id2map(id)]; ok {
+			debug.Log("Checker.trees", "tree %v already checked", id.Str())
+			continue
+		}
+
+		debug.Log("Checker.trees", "check tree %v", id.Str())
+
+		if _, ok := c.blobs[id2map(id)]; !ok {
+			errs = append(errs, Error{TreeID: id, Err: errors.New("not found in index")})
+			continue
+		}
+
+		blobs, subtrees, treeErrors := c.tree(id)
+		if treeErrors != nil {
+			debug.Log("Checker.trees", "error checking tree %v: %v", id.Str(), treeErrors)
+			errs = append(errs, treeErrors...)
+			continue
+		}
+
+		for _, blobID := range blobs {
+			c.blobRefs[id2map(blobID)]++
+			debug.Log("Checker.trees", "blob %v refcount %d", blobID.Str(), c.blobRefs[id2map(blobID)])
+
+			if _, ok := c.blobs[id2map(blobID)]; !ok {
+				debug.Log("Checker.trees", "tree %v references blob %v which isn't contained in index", id.Str(), blobID.Str())
+
+				errs = append(errs, Error{TreeID: id, BlobID: blobID, Err: errors.New("not found in index")})
+			}
+		}
+
+		treeIDs = append(treeIDs, subtrees...)
+
+		treesChecked[id2map(id)] = struct{}{}
+	}
+
+	return errs
+}
+
+func (c *Checker) tree(id backend.ID) (blobs backend.IDs, subtrees backend.IDs, errs []error) {
+	tree, err := restic.LoadTree(c.repo, id)
+	if err != nil {
+		return nil, nil, []error{Error{TreeID: id, Err: err}}
+	}
+
+	for i, node := range tree.Nodes {
+		switch node.Type {
+		case "file":
+			blobs = append(blobs, node.Content...)
+		case "dir":
+			if node.Subtree == nil {
+				errs = append(errs, Error{TreeID: id, Err: fmt.Errorf("node %d is dir but has no subtree", i)})
+				continue
+			}
+
+			subtrees = append(subtrees, node.Subtree)
+		}
+	}
+
+	return blobs, subtrees, errs
+}
+
+// UnusedBlobs returns all blobs that have never been referenced.
+func (c *Checker) UnusedBlobs() (blobs backend.IDs) {
+	debug.Log("Checker.UnusedBlobs", "checking %d blobs", len(c.blobs))
+	for id := range c.blobs {
+		if c.blobRefs[id] == 0 {
+			debug.Log("Checker.UnusedBlobs", "blob %v not not referenced", map2id(id).Str())
+			blobs = append(blobs, map2id(id))
+		}
+	}
+
+	return blobs
 }

checker/checker_test.go

@@ -2,6 +2,7 @@ package checker_test
 
 import (
 	"path/filepath"
+	"sort"
 	"testing"
 
 	"github.com/restic/restic/backend"
@@ -27,7 +28,85 @@ func TestCheckRepo(t *testing.T) {
 	WithTestEnvironment(t, checkerTestData, func(repodir string) {
 		repo := OpenLocalRepo(t, repodir)
 
-		checker := checker.New(repo)
+		chkr := checker.New(repo)
-		OK(t, checker.LoadIndex())
+		OK(t, chkr.LoadIndex())
+		OKs(t, chkr.Packs())
+		OKs(t, chkr.Structure())
+	})
+}
+
+func TestMissingPack(t *testing.T) {
+	WithTestEnvironment(t, checkerTestData, func(repodir string) {
+		repo := OpenLocalRepo(t, repodir)
+
+		packID := "657f7fb64f6a854fff6fe9279998ee09034901eded4e6db9bcee0e59745bbce6"
+		OK(t, repo.Backend().Remove(backend.Data, packID))
+
+		chkr := checker.New(repo)
+		OK(t, chkr.LoadIndex())
+		errs := chkr.Packs()
+
+		Assert(t, len(errs) == 1,
+			"expected exactly one error, got %v", len(errs))
+
+		if err, ok := errs[0].(checker.PackError); ok {
+			Equals(t, packID, err.ID.String())
+		} else {
+			t.Errorf("expected error returned by checker.Packs() to be PackError, got %v", err)
+		}
+	})
+}
+
+func TestUnreferencedPack(t *testing.T) {
+	WithTestEnvironment(t, checkerTestData, func(repodir string) {
+		repo := OpenLocalRepo(t, repodir)
+
+		// index 8eb5 only references pack 60e0
+		indexID := "8eb5b61062bf8e959f244fba0c971108bc8d4d2a4b236f71a704998e28cc5cf6"
+		packID := "60e0438dcb978ec6860cc1f8c43da648170ee9129af8f650f876bad19f8f788e"
+		OK(t, repo.Backend().Remove(backend.Index, indexID))
+
+		chkr := checker.New(repo)
+		OK(t, chkr.LoadIndex())
+		errs := chkr.Packs()
+
+		Assert(t, len(errs) == 1,
+			"expected exactly one error, got %v", len(errs))
+
+		if err, ok := errs[0].(checker.PackError); ok {
+			Equals(t, packID, err.ID.String())
+		} else {
+			t.Errorf("expected error returned by checker.Packs() to be PackError, got %v", err)
+		}
+	})
+}
+
+func TestUnreferencedBlobs(t *testing.T) {
+	WithTestEnvironment(t, checkerTestData, func(repodir string) {
+		repo := OpenLocalRepo(t, repodir)
+
+		snID := "51d249d28815200d59e4be7b3f21a157b864dc343353df9d8e498220c2499b02"
+		OK(t, repo.Backend().Remove(backend.Snapshot, snID))
+
+		unusedBlobsBySnapshot := backend.IDs{
+			ParseID("58c748bbe2929fdf30c73262bd8313fe828f8925b05d1d4a87fe109082acb849"),
+			ParseID("988a272ab9768182abfd1fe7d7a7b68967825f0b861d3b36156795832c772235"),
+			ParseID("c01952de4d91da1b1b80bc6e06eaa4ec21523f4853b69dc8231708b9b7ec62d8"),
+			ParseID("bec3a53d7dc737f9a9bee68b107ec9e8ad722019f649b34d474b9982c3a3fec7"),
+			ParseID("2a6f01e5e92d8343c4c6b78b51c5a4dc9c39d42c04e26088c7614b13d8d0559d"),
+			ParseID("18b51b327df9391732ba7aaf841a4885f350d8a557b2da8352c9acf8898e3f10"),
+		}
+
+		sort.Sort(unusedBlobsBySnapshot)
+
+		chkr := checker.New(repo)
+		OK(t, chkr.LoadIndex())
+		OKs(t, chkr.Packs())
+		OKs(t, chkr.Structure())
+
+		blobs := chkr.UnusedBlobs()
+		sort.Sort(blobs)
+
+		Equals(t, unusedBlobsBySnapshot, blobs)
 	})
 }

cmd/restic/cmd_check.go

@@ -2,6 +2,8 @@ package main
 
 import (
 	"errors"
+	"fmt"
+	"os"
 
 	"github.com/restic/restic/checker"
 )
@@ -50,5 +52,21 @@ func (cmd CmdCheck) Execute(args []string) error {
 		return err
 	}
 
+	errorsFound := false
+	cmd.global.Verbosef("Check all packs\n")
+	for _, err := range checker.Packs() {
+		errorsFound = true
+		fmt.Fprintf(os.Stderr, "error: %v\n", err)
+	}
+
+	cmd.global.Verbosef("Check snapshots, trees and blobs\n")
+	for _, err := range checker.Structure() {
+		errorsFound = true
+		fmt.Fprintf(os.Stderr, "error: %v\n", err)
+	}
+
+	if errorsFound {
+		return errors.New("repository contains errors")
+	}
 	return nil
 }

test/helpers.go

@@ -35,6 +35,21 @@ func OK(tb testing.TB, err error) {
 	}
 }
 
+// OKs fails the test if any error from errs is not nil.
+func OKs(tb testing.TB, errs []error) {
+	errFound := false
+	for _, err := range errs {
+		if err != nil {
+			errFound = true
+			_, file, line, _ := runtime.Caller(1)
+			fmt.Printf("\033[31m%s:%d: unexpected error: %s\033[39m\n\n", filepath.Base(file), line, err.Error())
+		}
+	}
+	if errFound {
+		tb.FailNow()
+	}
+}
+
 // Equals fails the test if exp is not equal to act.
 func Equals(tb testing.TB, exp, act interface{}) {
 	if !reflect.DeepEqual(exp, act) {