Merge pull request #231 from restic/implement-filter-functions

Add filter implementation for files

archiver.go

@@ -34,8 +34,8 @@ type Archiver struct {
 
 	blobToken chan struct{}
 
-	Error  func(dir string, fi os.FileInfo, err error) error
+	Error        func(dir string, fi os.FileInfo, err error) error
-	Filter func(item string, fi os.FileInfo) bool
+	SelectFilter pipe.SelectFunc
 }
 
 // NewArchiver returns a new archiver.
@@ -50,7 +50,7 @@ func NewArchiver(repo *repository.Repository) *Archiver {
 	}
 
 	arch.Error = archiverAbortOnAllErrors
-	arch.Filter = archiverAllowAllFiles
+	arch.SelectFilter = archiverAllowAllFiles
 
 	return arch
 }
@@ -577,7 +577,7 @@ func (arch *Archiver) Snapshot(p *Progress, paths []string, parentID backend.ID)
 	pipeCh := make(chan pipe.Job)
 	resCh := make(chan pipe.Result, 1)
 	go func() {
-		err := pipe.Walk(paths, done, pipeCh, resCh)
+		err := pipe.Walk(paths, arch.SelectFilter, done, pipeCh, resCh)
 		if err != nil {
 			debug.Log("Archiver.Snapshot", "pipe.Walk returned error %v", err)
 			return
@@ -659,7 +659,7 @@ func isRegularFile(fi os.FileInfo) bool {
 
 // Scan traverses the dirs to collect Stat information while emitting progress
 // information with p.
-func Scan(dirs []string, p *Progress) (Stat, error) {
+func Scan(dirs []string, filter pipe.SelectFunc, p *Progress) (Stat, error) {
 	p.Start()
 	defer p.Done()
 
@@ -678,6 +678,15 @@ func Scan(dirs []string, p *Progress) (Stat, error) {
 				fmt.Fprintf(os.Stderr, "error for %v: FileInfo is nil\n", str)
 				return nil
 			}
+
+			if !filter(str, fi) {
+				debug.Log("Scan.Walk", "path %v excluded", str)
+				if fi.IsDir() {
+					return filepath.SkipDir
+				}
+				return nil
+			}
+
 			s := Stat{}
 			if fi.IsDir() {
 				s.Dirs++

cmd/restic/cmd_backup.go

@@ -10,13 +10,15 @@ import (
 
 	"github.com/restic/restic"
 	"github.com/restic/restic/backend"
+	"github.com/restic/restic/filter"
 	"github.com/restic/restic/repository"
 	"golang.org/x/crypto/ssh/terminal"
 )
 
 type CmdBackup struct {
-	Parent string `short:"p" long:"parent"    description:"use this parent snapshot (default: last snapshot in repo that has the same target)"`
+	Parent  string   `short:"p" long:"parent"  description:"use this parent snapshot (default: last snapshot in repo that has the same target)"`
-	Force  bool   `short:"f" long:"force" description:"Force re-reading the target. Overrides the \"parent\" flag"`
+	Force   bool     `short:"f" long:"force"   description:"Force re-reading the target. Overrides the \"parent\" flag"`
+	Exclude []string `short:"e" long:"exclude" description:"Exclude a pattern (can be specified multiple times)"`
 
 	global *GlobalOptions
 }
@@ -282,14 +284,22 @@ func (cmd CmdBackup) Execute(args []string) error {
 
 	cmd.global.Verbosef("scan %v\n", target)
 
-	stat, err := restic.Scan(target, cmd.newScanProgress())
+	selectFilter := func(item string, fi os.FileInfo) bool {
+		matched, err := filter.List(cmd.Exclude, item)
+		if err != nil {
+			cmd.global.Warnf("error for exclude pattern: %v", err)
+		}
 
-	// TODO: add filter
+		return !matched
-	// arch.Filter = func(dir string, fi os.FileInfo) bool {
+	}
-	// 	return true
+
-	// }
+	stat, err := restic.Scan(target, selectFilter, cmd.newScanProgress())
+	if err != nil {
+		return err
+	}
 
 	arch := restic.NewArchiver(repo)
+	arch.SelectFilter = selectFilter
 
 	arch.Error = func(dir string, fi os.FileInfo, err error) error {
 		// TODO: make ignoring errors configurable

cmd/restic/cmd_ls.go

@@ -11,6 +11,8 @@ import (
 )
 
 type CmdLs struct {
+	Long bool `short:"l" long:"long" description:"Use a long listing format showing size and mode"`
+
 	global *GlobalOptions
 }
 
@@ -24,7 +26,11 @@ func init() {
 	}
 }
 
-func printNode(prefix string, n *restic.Node) string {
+func (cmd CmdLs) printNode(prefix string, n *restic.Node) string {
+	if !cmd.Long {
+		return filepath.Join(prefix, n.Name)
+	}
+
 	switch n.Type {
 	case "file":
 		return fmt.Sprintf("%s %5d %5d %6d %s %s",
@@ -40,17 +46,17 @@ func printNode(prefix string, n *restic.Node) string {
 	}
 }
 
-func printTree(prefix string, repo *repository.Repository, id backend.ID) error {
+func (cmd CmdLs) printTree(prefix string, repo *repository.Repository, id backend.ID) error {
 	tree, err := restic.LoadTree(repo, id)
 	if err != nil {
 		return err
 	}
 
 	for _, entry := range tree.Nodes {
-		fmt.Println(printNode(prefix, entry))
+		cmd.global.Printf(cmd.printNode(prefix, entry) + "\n")
 
 		if entry.Type == "dir" && entry.Subtree != nil {
-			err = printTree(filepath.Join(prefix, entry.Name), repo, entry.Subtree)
+			err = cmd.printTree(filepath.Join(prefix, entry.Name), repo, entry.Subtree)
 			if err != nil {
 				return err
 			}
@@ -89,7 +95,7 @@ func (cmd CmdLs) Execute(args []string) error {
 		return err
 	}
 
-	fmt.Printf("snapshot of %v at %s:\n", sn.Paths, sn.Time)
+	cmd.global.Verbosef("snapshot of %v at %s:\n", sn.Paths, sn.Time)
 
-	return printTree("", repo, sn.Tree)
+	return cmd.printTree("", repo, sn.Tree)
 }

cmd/restic/cmd_restore.go

@@ -1,14 +1,19 @@
 package main
 
 import (
+	"errors"
 	"fmt"
-	"path/filepath"
 
 	"github.com/restic/restic"
 	"github.com/restic/restic/debug"
+	"github.com/restic/restic/filter"
 )
 
 type CmdRestore struct {
+	Exclude []string `short:"e" long:"exclude" description:"Exclude a pattern (can be specified multiple times)"`
+	Include []string `short:"i" long:"include" description:"Include a pattern, exclude everything else (can be specified multiple times)"`
+	Target  string   `short:"t" long:"target"  description:"Directory to restore to"`
+
 	global *GlobalOptions
 }
 
@@ -23,14 +28,26 @@ func init() {
 }
 
 func (cmd CmdRestore) Usage() string {
-	return "snapshot-ID TARGETDIR [PATTERN]"
+	return "snapshot-ID"
 }
 
 func (cmd CmdRestore) Execute(args []string) error {
-	if len(args) < 2 || len(args) > 3 {
+	if len(args) != 1 {
 		return fmt.Errorf("wrong number of arguments, Usage: %s", cmd.Usage())
 	}
 
+	if cmd.Target == "" {
+		return errors.New("please specify a directory to restore to (--target)")
+	}
+
+	if len(cmd.Exclude) > 0 && len(cmd.Include) > 0 {
+		return errors.New("exclude and include patterns are mutually exclusive")
+	}
+
+	snapshotIDString := args[0]
+
+	debug.Log("restore", "restore %v to %v", snapshotIDString, cmd.Target)
+
 	repo, err := cmd.global.OpenRepository()
 	if err != nil {
 		return err
@@ -47,14 +64,11 @@ func (cmd CmdRestore) Execute(args []string) error {
 		return err
 	}
 
-	id, err := restic.FindSnapshot(repo, args[0])
+	id, err := restic.FindSnapshot(repo, snapshotIDString)
 	if err != nil {
-		cmd.global.Exitf(1, "invalid id %q: %v", args[0], err)
+		cmd.global.Exitf(1, "invalid id %q: %v", snapshotIDString, err)
 	}
 
-	target := args[1]
-
-	// create restorer
 	res, err := restic.NewRestorer(repo, id)
 	if err != nil {
 		cmd.global.Exitf(2, "creating restorer failed: %v\n", err)
@@ -62,41 +76,36 @@ func (cmd CmdRestore) Execute(args []string) error {
 
 	res.Error = func(dir string, node *restic.Node, err error) error {
 		cmd.global.Warnf("error for %s: %+v\n", dir, err)
-
-		// if node.Type == "dir" {
-		// 	if e, ok := err.(*os.PathError); ok {
-		// 		if errn, ok := e.Err.(syscall.Errno); ok {
-		// 			if errn == syscall.EEXIST {
-		// 				fmt.Printf("ignoring already existing directory %s\n", dir)
-		// 				return nil
-		// 			}
-		// 		}
-		// 	}
-		// }
 		return err
 	}
 
-	// TODO: a filter against the full path sucks as filepath.Match doesn't match
+	selectExcludeFilter := func(item string, dstpath string, node *restic.Node) bool {
-	// directory separators on '*'. still, it's better than nothing.
+		matched, err := filter.List(cmd.Exclude, item)
-	if len(args) > 2 {
+		if err != nil {
-		pattern := args[2]
+			cmd.global.Warnf("error for exclude pattern: %v", err)
-		cmd.global.Verbosef("filter pattern %q\n", pattern)
-
-		res.SelectForRestore = func(item string, dstpath string, node *restic.Node) bool {
-			matched, err := filepath.Match(pattern, node.Name)
-			if err != nil {
-				panic(err)
-			}
-			if !matched {
-				debug.Log("restic.restore", "item %v doesn't match pattern %q", item, pattern)
-			}
-			return matched
 		}
+
+		return !matched
 	}
 
-	cmd.global.Verbosef("restoring %s to %s\n", res.Snapshot(), target)
+	selectIncludeFilter := func(item string, dstpath string, node *restic.Node) bool {
+		matched, err := filter.List(cmd.Include, item)
+		if err != nil {
+			cmd.global.Warnf("error for include pattern: %v", err)
+		}
 
-	err = res.RestoreTo(target)
+		return matched
+	}
+
+	if len(cmd.Exclude) > 0 {
+		res.SelectFilter = selectExcludeFilter
+	} else if len(cmd.Include) > 0 {
+		res.SelectFilter = selectIncludeFilter
+	}
+
+	cmd.global.Verbosef("restoring %s to %s\n", res.Snapshot(), cmd.Target)
+
+	err = res.RestoreTo(cmd.Target)
 	if err != nil {
 		return err
 	}

cmd/restic/integration_test.go

@@ -10,12 +10,14 @@ import (
 	"os"
 	"path/filepath"
 	"regexp"
+	"strings"
 	"syscall"
 	"testing"
 	"time"
 
 	"github.com/restic/restic/backend"
 	"github.com/restic/restic/debug"
+	"github.com/restic/restic/filter"
 	. "github.com/restic/restic/test"
 )
 
@@ -44,7 +46,11 @@ func cmdInit(t testing.TB, global GlobalOptions) {
 }
 
 func cmdBackup(t testing.TB, global GlobalOptions, target []string, parentID backend.ID) {
-	cmd := &CmdBackup{global: &global}
+	cmdBackupExcludes(t, global, target, parentID, nil)
+}
+
+func cmdBackupExcludes(t testing.TB, global GlobalOptions, target []string, parentID backend.ID, excludes []string) {
+	cmd := &CmdBackup{global: &global, Exclude: excludes}
 	cmd.Parent = parentID.String()
 
 	t.Logf("backing up %v", target)
@@ -63,9 +69,18 @@ func cmdList(t testing.TB, global GlobalOptions, tpe string) []backend.ID {
 	return IDs
 }
 
-func cmdRestore(t testing.TB, global GlobalOptions, dir string, snapshotID backend.ID, args ...string) {
+func cmdRestore(t testing.TB, global GlobalOptions, dir string, snapshotID backend.ID) {
-	cmd := &CmdRestore{global: &global}
+	cmdRestoreExcludes(t, global, dir, snapshotID, nil)
-	cmd.Execute(append([]string{snapshotID.String(), dir}, args...))
+}
+
+func cmdRestoreExcludes(t testing.TB, global GlobalOptions, dir string, snapshotID backend.ID, excludes []string) {
+	cmd := &CmdRestore{global: &global, Target: dir, Exclude: excludes}
+	OK(t, cmd.Execute([]string{snapshotID.String()}))
+}
+
+func cmdRestoreIncludes(t testing.TB, global GlobalOptions, dir string, snapshotID backend.ID, includes []string) {
+	cmd := &CmdRestore{global: &global, Target: dir, Include: includes}
+	OK(t, cmd.Execute([]string{snapshotID.String()}))
 }
 
 func cmdCheck(t testing.TB, global GlobalOptions) {
@@ -73,6 +88,16 @@ func cmdCheck(t testing.TB, global GlobalOptions) {
 	OK(t, cmd.Execute(nil))
 }
 
+func cmdLs(t testing.TB, global GlobalOptions, snapshotID string) []string {
+	var buf bytes.Buffer
+	global.stdout = &buf
+
+	cmd := &CmdLs{global: &global}
+	OK(t, cmd.Execute([]string{snapshotID}))
+
+	return strings.Split(string(buf.Bytes()), "\n")
+}
+
 func TestBackup(t *testing.T) {
 	withTestEnvironment(t, func(env *testEnvironment, global GlobalOptions) {
 		datafile := filepath.Join("testdata", "backup-data.tar.gz")
@@ -237,6 +262,86 @@ func TestBackupMissingFile2(t *testing.T) {
 	})
 }
 
+func includes(haystack []string, needle string) bool {
+	for _, s := range haystack {
+		if s == needle {
+			return true
+		}
+	}
+
+	return false
+}
+
+func loadSnapshotMap(t testing.TB, global GlobalOptions) map[string]struct{} {
+	snapshotIDs := cmdList(t, global, "snapshots")
+
+	m := make(map[string]struct{})
+	for _, id := range snapshotIDs {
+		m[id.String()] = struct{}{}
+	}
+
+	return m
+}
+
+func lastSnapshot(old, new map[string]struct{}) (map[string]struct{}, string) {
+	for k := range new {
+		if _, ok := old[k]; !ok {
+			old[k] = struct{}{}
+			return old, k
+		}
+	}
+
+	return old, ""
+}
+
+var backupExcludeFilenames = []string{
+	"testfile1",
+	"foo.tar.gz",
+	"private/secret/passwords.txt",
+	"work/source/test.c",
+}
+
+func TestBackupExclude(t *testing.T) {
+	withTestEnvironment(t, func(env *testEnvironment, global GlobalOptions) {
+		cmdInit(t, global)
+
+		datadir := filepath.Join(env.base, "testdata")
+
+		for _, filename := range backupExcludeFilenames {
+			fp := filepath.Join(datadir, filename)
+			OK(t, os.MkdirAll(filepath.Dir(fp), 0755))
+
+			f, err := os.Create(fp)
+			OK(t, err)
+
+			fmt.Fprintf(f, filename)
+			OK(t, f.Close())
+		}
+
+		snapshots := make(map[string]struct{})
+
+		cmdBackup(t, global, []string{datadir}, nil)
+		snapshots, snapshotID := lastSnapshot(snapshots, loadSnapshotMap(t, global))
+		files := cmdLs(t, global, snapshotID)
+		Assert(t, includes(files, filepath.Join("testdata", "foo.tar.gz")),
+			"expected file %q in first snapshot, but it's not included", "foo.tar.gz")
+
+		cmdBackupExcludes(t, global, []string{datadir}, nil, []string{"*.tar.gz"})
+		snapshots, snapshotID = lastSnapshot(snapshots, loadSnapshotMap(t, global))
+		files = cmdLs(t, global, snapshotID)
+		Assert(t, !includes(files, filepath.Join("testdata", "foo.tar.gz")),
+			"expected file %q not in first snapshot, but it's included", "foo.tar.gz")
+
+		cmdBackupExcludes(t, global, []string{datadir}, nil, []string{"*.tar.gz", "private/secret"})
+		snapshots, snapshotID = lastSnapshot(snapshots, loadSnapshotMap(t, global))
+		files = cmdLs(t, global, snapshotID)
+		Assert(t, !includes(files, filepath.Join("testdata", "foo.tar.gz")),
+			"expected file %q not in first snapshot, but it's included", "foo.tar.gz")
+		Assert(t, !includes(files, filepath.Join("testdata", "private", "secret", "passwords.txt")),
+			"expected file %q not in first snapshot, but it's included", "passwords.txt")
+	})
+}
+
 const (
 	incrementalFirstWrite  = 20 * 1042 * 1024
 	incrementalSecondWrite = 12 * 1042 * 1024
@@ -422,10 +527,10 @@ func TestRestoreFilter(t *testing.T) {
 
 		for i, pat := range []string{"*.c", "*.exe", "*", "*file3*"} {
 			base := filepath.Join(env.base, fmt.Sprintf("restore%d", i+1))
-			cmdRestore(t, global, base, snapshotID, pat)
+			cmdRestoreExcludes(t, global, base, snapshotID, []string{pat})
 			for _, test := range testfiles {
 				err := testFileSize(filepath.Join(base, "testdata", test.name), int64(test.size))
-				if ok, _ := filepath.Match(pat, filepath.Base(test.name)); ok {
+				if ok, _ := filter.Match(pat, filepath.Base(test.name)); !ok {
 					OK(t, err)
 				} else {
 					Assert(t, os.IsNotExist(err),
@@ -463,7 +568,7 @@ func TestRestoreNoMetadataOnIgnoredIntermediateDirs(t *testing.T) {
 		// restore with filter "*.ext", this should restore "file.ext", but
 		// since the directories are ignored and only created because of
 		// "file.ext", no meta data should be restored for them.
-		cmdRestore(t, global, filepath.Join(env.base, "restore0"), snapshotID, "*.ext")
+		cmdRestoreIncludes(t, global, filepath.Join(env.base, "restore0"), snapshotID, []string{"*.ext"})
 
 		f1 := filepath.Join(env.base, "restore0", "testdata", "subdir1", "subdir2")
 		fi, err := os.Stat(f1)
@@ -473,7 +578,7 @@ func TestRestoreNoMetadataOnIgnoredIntermediateDirs(t *testing.T) {
 			"meta data of intermediate directory has been restore although it was ignored")
 
 		// restore with filter "*", this should restore meta data on everything.
-		cmdRestore(t, global, filepath.Join(env.base, "restore1"), snapshotID, "*")
+		cmdRestoreIncludes(t, global, filepath.Join(env.base, "restore1"), snapshotID, []string{"*"})
 
 		f2 := filepath.Join(env.base, "restore1", "testdata", "subdir1", "subdir2")
 		fi, err = os.Stat(f2)

filter/doc.go

@@ -0,0 +1,5 @@
+// Package filter implements filters for files similar to filepath.Glob, but
+// in contrast to filepath.Glob a pattern may specify directories.
+//
+// For a list of valid patterns please see the documentation on filepath.Glob.
+package filter

filter/filter.go

@@ -0,0 +1,108 @@
+package filter
+
+import (
+	"errors"
+	"path/filepath"
+	"strings"
+)
+
+// ErrBadString is returned when Match is called with the empty string as the
+// second argument.
+var ErrBadString = errors.New("filter.Match: string is empty")
+
+// Match returns true if str matches the pattern. When the pattern is
+// malformed, filepath.ErrBadPattern is returned. The empty pattern matches
+// everything, when str is the empty string ErrBadString is returned.
+//
+// Pattern can be a combination of patterns suitable for filepath.Match, joined
+// by filepath.Separator.
+func Match(pattern, str string) (matched bool, err error) {
+	if pattern == "" {
+		return true, nil
+	}
+
+	if str == "" {
+		return false, ErrBadString
+	}
+
+	patterns := strings.Split(pattern, string(filepath.Separator))
+	strs := strings.Split(str, string(filepath.Separator))
+
+	return match(patterns, strs)
+}
+
+func hasDoubleWildcard(list []string) (ok bool, pos int) {
+	for i, item := range list {
+		if item == "**" {
+			return true, i
+		}
+	}
+
+	return false, 0
+}
+
+func match(patterns, strs []string) (matched bool, err error) {
+	if ok, pos := hasDoubleWildcard(patterns); ok {
+		// gradually expand '**' into separate wildcards
+		for i := 0; i <= len(strs)-len(patterns)+1; i++ {
+			newPat := make([]string, pos)
+			copy(newPat, patterns[:pos])
+			for k := 0; k < i; k++ {
+				newPat = append(newPat, "*")
+			}
+			newPat = append(newPat, patterns[pos+1:]...)
+
+			matched, err := match(newPat, strs)
+			if err != nil {
+				return false, err
+			}
+
+			if matched {
+				return true, nil
+			}
+		}
+
+		return false, nil
+	}
+
+	if len(patterns) == 0 && len(strs) == 0 {
+		return true, nil
+	}
+
+	if len(patterns) <= len(strs) {
+	outer:
+		for offset := len(strs) - len(patterns); offset >= 0; offset-- {
+
+			for i := len(patterns) - 1; i >= 0; i-- {
+				ok, err := filepath.Match(patterns[i], strs[offset+i])
+				if err != nil {
+					return false, err
+				}
+
+				if !ok {
+					continue outer
+				}
+			}
+
+			return true, nil
+		}
+	}
+
+	return false, nil
+}
+
+// List returns true if str matches one of the patterns.
+func List(patterns []string, str string) (matched bool, err error) {
+	for _, pat := range patterns {
+		matched, err = Match(pat, str)
+		if err != nil {
+			return false, err
+		}
+
+		if matched {
+			return true, nil
+		}
+	}
+
+	return false, nil
+}

filter/filter_test.go

@@ -0,0 +1,252 @@
+package filter_test
+
+import (
+	"bufio"
+	"compress/bzip2"
+	"fmt"
+	"os"
+	"testing"
+
+	"github.com/restic/restic/filter"
+)
+
+var matchTests = []struct {
+	pattern string
+	path    string
+	match   bool
+}{
+	{"", "", true},
+	{"", "foo", true},
+	{"", "/x/y/z/foo", true},
+	{"*.go", "/foo/bar/test.go", true},
+	{"*.c", "/foo/bar/test.go", false},
+	{"*", "/foo/bar/test.go", true},
+	{"foo*", "/foo/bar/test.go", true},
+	{"bar*", "/foo/bar/test.go", true},
+	{"/bar*", "/foo/bar/test.go", false},
+	{"bar/*", "/foo/bar/test.go", true},
+	{"baz/*", "/foo/bar/test.go", false},
+	{"bar/test.go", "/foo/bar/test.go", true},
+	{"bar/*.go", "/foo/bar/test.go", true},
+	{"ba*/*.go", "/foo/bar/test.go", true},
+	{"bb*/*.go", "/foo/bar/test.go", false},
+	{"test.*", "/foo/bar/test.go", true},
+	{"tesT.*", "/foo/bar/test.go", false},
+	{"bar/*", "/foo/bar/baz", true},
+	{"bar", "/foo/bar", true},
+	{"bar", "/foo/bar/baz", true},
+	{"bar", "/foo/bar/test.go", true},
+	{"/foo/*test.*", "/foo/bar/test.go", false},
+	{"/foo/*/test.*", "/foo/bar/test.go", true},
+	{"/foo/*/bar/test.*", "/foo/bar/test.go", false},
+	{"/*/*/bar/test.*", "/foo/bar/test.go", false},
+	{"/*/*/bar/test.*", "/foo/bar/baz/test.go", false},
+	{"/*/*/baz/test.*", "/foo/bar/baz/test.go", true},
+	{"/*/foo/bar/test.*", "/foo/bar/baz/test.go", false},
+	{"/*/foo/bar/test.*", "/foo/bar/baz/test.go", false},
+	{"/foo/bar/test.*", "bar/baz/test.go", false},
+	{"/x/y/bar/baz/test.*", "bar/baz/test.go", false},
+	{"/x/y/bar/baz/test.c", "bar/baz/test.go", false},
+	{"baz/test.*", "bar/baz/test.go", true},
+	{"baz/tesT.*", "bar/baz/test.go", false},
+	{"test.go", "bar/baz/test.go", true},
+	{"*.go", "bar/baz/test.go", true},
+	{"*.c", "bar/baz/test.go", false},
+	{"sdk", "/foo/bar/sdk", true},
+	{"sdk", "/foo/bar/sdk/test/sdk_foo.go", true},
+	{
+		"sdk/*/cpp/*/*vars*.html",
+		"/usr/share/doc/libreoffice/sdk/docs/cpp/ref/a00517.html",
+		false,
+	},
+	{"foo/**/bar/*.go", "/home/user/foo/work/special/project/bar/test.go", true},
+	{"foo/**/bar/*.go", "/home/user/foo/bar/test.go", true},
+	{"foo/**/bar/*.go", "x/foo/bar/test.go", true},
+	{"foo/**/bar/*.go", "foo/bar/test.go", true},
+	{"foo/**/bar/*.go", "/home/user/foo/test.c", false},
+	{"foo/**/bar/*.go", "bar/foo/main.go", false},
+	{"foo/**/bar/*.go", "/foo/bar/main.go", true},
+	{"foo/**/bar/*.go", "bar/main.go", false},
+	{"foo/**/bar", "/home/user/foo/x/y/bar", true},
+	{"foo/**/bar", "/home/user/foo/x/y/bar/main.go", true},
+	{"user/**/important*", "/home/user/work/x/y/hidden/x", false},
+	{"user/**/hidden*/**/c", "/home/user/work/x/y/hidden/z/a/b/c", true},
+}
+
+func TestMatch(t *testing.T) {
+	for i, test := range matchTests {
+		match, err := filter.Match(test.pattern, test.path)
+		if err != nil {
+			t.Errorf("test %d failed: expected no error for pattern %q, but error returned: %v",
+				i, test.pattern, err)
+			continue
+		}
+
+		if match != test.match {
+			t.Errorf("test %d: filter.Match(%q, %q): expected %v, got %v",
+				i, test.pattern, test.path, test.match, match)
+		}
+	}
+}
+
+func ExampleMatch() {
+	match, _ := filter.Match("*.go", "/home/user/file.go")
+	fmt.Printf("match: %v\n", match)
+	// Output:
+	// match: true
+}
+
+func ExampleMatch_wildcards() {
+	match, _ := filter.Match("/home/[uU]ser/?.go", "/home/user/F.go")
+	fmt.Printf("match: %v\n", match)
+	// Output:
+	// match: true
+}
+
+var filterListTests = []struct {
+	patterns []string
+	path     string
+	match    bool
+}{
+	{[]string{"*.go"}, "/foo/bar/test.go", true},
+	{[]string{"*.c"}, "/foo/bar/test.go", false},
+	{[]string{"*.go", "*.c"}, "/foo/bar/test.go", true},
+	{[]string{"*"}, "/foo/bar/test.go", true},
+	{[]string{"x"}, "/foo/bar/test.go", false},
+	{[]string{"?"}, "/foo/bar/test.go", false},
+	{[]string{"?", "x"}, "/foo/bar/x", true},
+	{[]string{"/*/*/bar/test.*"}, "/foo/bar/test.go", false},
+	{[]string{"/*/*/bar/test.*", "*.go"}, "/foo/bar/test.go", true},
+}
+
+func TestMatchList(t *testing.T) {
+	for i, test := range filterListTests {
+		match, err := filter.List(test.patterns, test.path)
+		if err != nil {
+			t.Errorf("test %d failed: expected no error for patterns %q, but error returned: %v",
+				i, test.patterns, err)
+			continue
+		}
+
+		if match != test.match {
+			t.Errorf("test %d: filter.MatchList(%q, %q): expected %v, got %v",
+				i, test.patterns, test.path, test.match, match)
+		}
+	}
+}
+
+func ExampleMatchList() {
+	match, _ := filter.List([]string{"*.c", "*.go"}, "/home/user/file.go")
+	fmt.Printf("match: %v\n", match)
+	// Output:
+	// match: true
+}
+
+func extractTestLines(t testing.TB) (lines []string) {
+	f, err := os.Open("testdata/libreoffice.txt.bz2")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	defer func() {
+		if err := f.Close(); err != nil {
+			t.Fatal(err)
+		}
+	}()
+
+	sc := bufio.NewScanner(bzip2.NewReader(f))
+	for sc.Scan() {
+		lines = append(lines, sc.Text())
+	}
+
+	return lines
+}
+
+func TestFilterPatternsFile(t *testing.T) {
+	lines := extractTestLines(t)
+
+	var testPatterns = []struct {
+		pattern string
+		hits    uint
+	}{
+		{"*.html", 18249},
+		{"sdk", 22186},
+		{"sdk/*/cpp/*/*vars.html", 3},
+	}
+
+	for _, test := range testPatterns {
+		var c uint
+		for _, line := range lines {
+			match, err := filter.Match(test.pattern, line)
+			if err != nil {
+				t.Error(err)
+				continue
+			}
+
+			if match {
+				c++
+				// fmt.Printf("pattern %q, line %q\n", test.pattern, line)
+			}
+		}
+
+		if c != test.hits {
+			t.Errorf("wrong number of hits for pattern %q: want %d, got %d",
+				test.pattern, test.hits, c)
+		}
+	}
+}
+
+func BenchmarkFilterLines(b *testing.B) {
+	pattern := "sdk/*/cpp/*/*vars.html"
+	lines := extractTestLines(b)
+	var c uint
+
+	b.ResetTimer()
+
+	for i := 0; i < b.N; i++ {
+		c = 0
+		for _, line := range lines {
+			match, err := filter.Match(pattern, line)
+			if err != nil {
+				b.Fatal(err)
+			}
+
+			if match {
+				c++
+			}
+		}
+
+		if c != 3 {
+			b.Fatalf("wrong number of matches: expected 3, got %d", c)
+		}
+	}
+}
+
+func BenchmarkFilterPatterns(b *testing.B) {
+	patterns := []string{
+		"sdk/*",
+		"*.html",
+	}
+	lines := extractTestLines(b)
+	var c uint
+
+	b.ResetTimer()
+
+	for i := 0; i < b.N; i++ {
+		c = 0
+		for _, line := range lines {
+			match, err := filter.List(patterns, line)
+			if err != nil {
+				b.Fatal(err)
+			}
+
+			if match {
+				c++
+			}
+		}
+
+		if c != 22185 {
+			b.Fatalf("wrong number of matches: expected 22185, got %d", c)
+		}
+	}
+}

pipe/pipe.go

@@ -82,13 +82,22 @@ func isFile(fi os.FileInfo) bool {
 
 var errCancelled = errors.New("walk cancelled")
 
-func walk(basedir, dir string, done chan struct{}, jobs chan<- Job, res chan<- Result) error {
+// SelectFunc returns true for all items that should be included (files and
+// dirs). If false is returned, files are ignored and dirs are not even walked.
+type SelectFunc func(item string, fi os.FileInfo) bool
+
+func walk(basedir, dir string, selectFunc SelectFunc, done chan struct{}, jobs chan<- Job, res chan<- Result) error {
 	info, err := os.Lstat(dir)
 	if err != nil {
 		debug.Log("pipe.walk", "error for %v: %v", dir, err)
 		return err
 	}
 
+	if !selectFunc(dir, info) {
+		debug.Log("pipe.walk", "file %v excluded by filter", dir)
+		return nil
+	}
+
 	relpath, _ := filepath.Rel(basedir, dir)
 
 	if !info.IsDir() {
@@ -114,13 +123,18 @@ func walk(basedir, dir string, done chan struct{}, jobs chan<- Job, res chan<- R
 	for _, name := range names {
 		subpath := filepath.Join(dir, name)
 
+		fi, statErr := os.Lstat(subpath)
+		if !selectFunc(subpath, fi) {
+			debug.Log("pipe.walk", "file %v excluded by filter", subpath)
+			continue
+		}
+
 		ch := make(chan Result, 1)
 		entries = append(entries, ch)
 
-		fi, err := os.Lstat(subpath)
+		if statErr != nil {
-		if err != nil {
 			select {
-			case jobs <- Entry{info: fi, error: err, basedir: basedir, path: filepath.Join(relpath, name), result: ch}:
+			case jobs <- Entry{info: fi, error: statErr, basedir: basedir, path: filepath.Join(relpath, name), result: ch}:
 			case <-done:
 				return errCancelled
 			}
@@ -132,7 +146,7 @@ func walk(basedir, dir string, done chan struct{}, jobs chan<- Job, res chan<- R
 		debug.RunHook("pipe.walk2", filepath.Join(relpath, name))
 
 		if isDir(fi) {
-			err = walk(basedir, subpath, done, jobs, ch)
+			err = walk(basedir, subpath, selectFunc, done, jobs, ch)
 			if err != nil {
 				return err
 			}
@@ -156,7 +170,7 @@ func walk(basedir, dir string, done chan struct{}, jobs chan<- Job, res chan<- R
 
 // Walk sends a Job for each file and directory it finds below the paths. When
 // the channel done is closed, processing stops.
-func Walk(paths []string, done chan struct{}, jobs chan<- Job, res chan<- Result) error {
+func Walk(paths []string, selectFunc SelectFunc, done chan struct{}, jobs chan<- Job, res chan<- Result) error {
 	defer func() {
 		debug.Log("pipe.Walk", "output channel closed")
 		close(jobs)
@@ -166,7 +180,7 @@ func Walk(paths []string, done chan struct{}, jobs chan<- Job, res chan<- Result
 	for _, path := range paths {
 		debug.Log("pipe.Walk", "start walker for %v", path)
 		ch := make(chan Result, 1)
-		err := walk(filepath.Dir(path), path, done, jobs, ch)
+		err := walk(filepath.Dir(path), path, selectFunc, done, jobs, ch)
 		if err != nil {
 			debug.Log("pipe.Walk", "error for %v: %v", path, err)
 			continue

pipe/pipe_test.go

@@ -19,6 +19,10 @@ type stats struct {
 	dirs, files int
 }
 
+func acceptAll(string, os.FileInfo) bool {
+	return true
+}
+
 func statPath(path string) (stats, error) {
 	var s stats
 
@@ -118,7 +122,7 @@ func TestPipelineWalkerWithSplit(t *testing.T) {
 	}()
 
 	resCh := make(chan pipe.Result, 1)
-	err = pipe.Walk([]string{TestWalkerPath}, done, jobs, resCh)
+	err = pipe.Walk([]string{TestWalkerPath}, acceptAll, done, jobs, resCh)
 	OK(t, err)
 
 	// wait for all workers to terminate
@@ -198,7 +202,7 @@ func TestPipelineWalker(t *testing.T) {
 	}
 
 	resCh := make(chan pipe.Result, 1)
-	err = pipe.Walk([]string{TestWalkerPath}, done, jobs, resCh)
+	err = pipe.Walk([]string{TestWalkerPath}, acceptAll, done, jobs, resCh)
 	OK(t, err)
 
 	// wait for all workers to terminate
@@ -298,7 +302,7 @@ func BenchmarkPipelineWalker(b *testing.B) {
 		}()
 
 		resCh := make(chan pipe.Result, 1)
-		err := pipe.Walk([]string{TestWalkerPath}, done, jobs, resCh)
+		err := pipe.Walk([]string{TestWalkerPath}, acceptAll, done, jobs, resCh)
 		OK(b, err)
 
 		// wait for all workers to terminate
@@ -375,7 +379,7 @@ func TestPipelineWalkerMultiple(t *testing.T) {
 	}
 
 	resCh := make(chan pipe.Result, 1)
-	err = pipe.Walk(paths, done, jobs, resCh)
+	err = pipe.Walk(paths, acceptAll, done, jobs, resCh)
 	OK(t, err)
 
 	// wait for all workers to terminate

restorer.go

@@ -18,8 +18,8 @@ type Restorer struct {
 	repo *repository.Repository
 	sn   *Snapshot
 
-	Error            func(dir string, node *Node, err error) error
+	Error        func(dir string, node *Node, err error) error
-	SelectForRestore func(item string, dstpath string, node *Node) bool
+	SelectFilter func(item string, dstpath string, node *Node) bool
 }
 
 var restorerAbortOnAllErrors = func(str string, node *Node, err error) error { return err }
@@ -28,7 +28,7 @@ var restorerAbortOnAllErrors = func(str string, node *Node, err error) error { r
 func NewRestorer(repo *repository.Repository, id backend.ID) (*Restorer, error) {
 	r := &Restorer{
 		repo: repo, Error: restorerAbortOnAllErrors,
-		SelectForRestore: func(string, string, *Node) bool { return true },
+		SelectFilter: func(string, string, *Node) bool { return true },
 	}
 
 	var err error
@@ -48,7 +48,7 @@ func (res *Restorer) restoreTo(dst string, dir string, treeID backend.ID) error
 	}
 
 	for _, node := range tree.Nodes {
-		selectedForRestore := res.SelectForRestore(filepath.Join(dir, node.Name),
+		selectedForRestore := res.SelectFilter(filepath.Join(dir, node.Name),
 			filepath.Join(dst, dir, node.Name), node)
 		debug.Log("Restorer.restoreNodeTo", "SelectForRestore returned %v", selectedForRestore)
 

walk_test.go

@@ -1,6 +1,7 @@
 package restic_test
 
 import (
+	"os"
 	"path/filepath"
 	"testing"
 
@@ -33,7 +34,11 @@ func TestWalkTree(t *testing.T) {
 	// start filesystem walker
 	fsJobs := make(chan pipe.Job)
 	resCh := make(chan pipe.Result, 1)
-	go pipe.Walk(dirs, done, fsJobs, resCh)
+
+	f := func(string, os.FileInfo) bool {
+		return true
+	}
+	go pipe.Walk(dirs, f, done, fsJobs, resCh)
 
 	for {
 		// receive fs job