From d36f07c6eb2bc43808a49d16942cc43e2ea70d87 Mon Sep 17 00:00:00 2001
From: Alexander Neumann <alexander@bumpern.de>
Date: Sun, 12 Jul 2015 17:09:48 +0200
Subject: [PATCH] checker: add option to remove orphaned packs

---
 checker/checker.go      | 22 ++++++++++++++-------
 cmd/restic/cmd_check.go | 44 +++++++++++++++++++++++++++++++----------
 2 files changed, 49 insertions(+), 17 deletions(-)

diff --git a/checker/checker.go b/checker/checker.go
index 0238a53b3..852c85a43 100644
--- a/checker/checker.go
+++ b/checker/checker.go
@@ -48,7 +48,8 @@ type Checker struct {
 		sync.Mutex
 		M map[mapID]uint
 	}
-	indexes map[mapID]*repository.Index
+	indexes       map[mapID]*repository.Index
+	orphanedPacks backend.IDs
 
 	masterIndex *repository.Index
 
@@ -139,12 +140,13 @@ func (c *Checker) LoadIndex() error {
 
 // PackError describes an error with a specific pack.
 type PackError struct {
-	ID backend.ID
-	error
+	ID       backend.ID
+	Orphaned bool
+	Err      error
 }
 
 func (e PackError) Error() string {
-	return "pack " + e.ID.String() + ": " + e.error.Error()
+	return "pack " + e.ID.String() + ": " + e.Err.Error()
 }
 
 func packIDTester(repo *repository.Repository, inChan <-chan mapID, errChan chan<- error, wg *sync.WaitGroup, done <-chan struct{}) {
@@ -156,10 +158,10 @@ func packIDTester(repo *repository.Repository, inChan <-chan mapID, errChan chan
 	for id := range inChan {
 		ok, err := repo.Backend().Test(backend.Data, map2str(id))
 		if err != nil {
-			err = PackError{map2id(id), err}
+			err = PackError{ID: map2id(id), Err: err}
 		} else {
 			if !ok {
-				err = PackError{map2id(id), errors.New("does not exist")}
+				err = PackError{ID: map2id(id), Err: errors.New("does not exist")}
 			}
 		}
 
@@ -227,10 +229,11 @@ func (c *Checker) Packs(errChan chan<- error, done <-chan struct{}) {
 	for id := range c.repo.List(backend.Data, done) {
 		debug.Log("Checker.Packs", "check data blob %v", id.Str())
 		if _, ok := seenPacks[id2map(id)]; !ok {
+			c.orphanedPacks = append(c.orphanedPacks, id)
 			select {
 			case <-done:
 				return
-			case errChan <- PackError{id, errors.New("not referenced in any index")}:
+			case errChan <- PackError{ID: id, Orphaned: true, Err: errors.New("not referenced in any index")}:
 			}
 		}
 	}
@@ -591,3 +594,8 @@ func (c *Checker) UnusedBlobs() (blobs backend.IDs) {
 
 	return blobs
 }
+
+// OrphanedPacks returns a slice of unused packs (only available after Packs() was run).
+func (c *Checker) OrphanedPacks() backend.IDs {
+	return c.orphanedPacks
+}
diff --git a/cmd/restic/cmd_check.go b/cmd/restic/cmd_check.go
index ccf65d4d7..d47de1c7b 100644
--- a/cmd/restic/cmd_check.go
+++ b/cmd/restic/cmd_check.go
@@ -5,11 +5,13 @@ import (
 	"fmt"
 	"os"
 
+	"github.com/restic/restic/backend"
 	"github.com/restic/restic/checker"
 )
 
 type CmdCheck struct {
-	ReadData bool `          long:"read-data"       description:"Read data blobs" default:"false"`
+	ReadData       bool `long:"read-data"  description:"Read data blobs" default:"false"`
+	RemoveOrphaned bool `long:"remove"     description:"Remove data that isn't used" default:"false"`
 
 	global *GlobalOptions
 }
@@ -45,10 +47,10 @@ func (cmd CmdCheck) Execute(args []string) error {
 		return err
 	}
 
-	checker := checker.New(repo)
+	chkr := checker.New(repo)
 
 	cmd.global.Verbosef("Load indexes\n")
-	if err = checker.LoadIndex(); err != nil {
+	if err = chkr.LoadIndex(); err != nil {
 		return err
 	}
 
@@ -59,20 +61,42 @@ func (cmd CmdCheck) Execute(args []string) error {
 	errChan := make(chan error)
 
 	cmd.global.Verbosef("Check all packs\n")
-	go checker.Packs(errChan, done)
+	go chkr.Packs(errChan, done)
+
+	foundOrphanedPacks := false
+	for err := range errChan {
+		errorsFound = true
+		fmt.Fprintf(os.Stderr, "%v\n", err)
+
+		if e, ok := err.(checker.PackError); ok && e.Orphaned {
+			foundOrphanedPacks = true
+		}
+	}
+
+	cmd.global.Verbosef("Check snapshots, trees and blobs\n")
+	errChan = make(chan error)
+	go chkr.Structure(errChan, done)
 
 	for err := range errChan {
 		errorsFound = true
 		fmt.Fprintf(os.Stderr, "error: %v\n", err)
 	}
 
-	cmd.global.Verbosef("Check snapshots, trees and blobs\n")
-	errChan = make(chan error)
-	go checker.Structure(errChan, done)
+	for _, id := range chkr.UnusedBlobs() {
+		cmd.global.Verbosef("unused blob %v\n", id.Str())
+	}
 
-	for err := range errChan {
-		errorsFound = true
-		fmt.Fprintf(os.Stderr, "error: %v\n", err)
+	if foundOrphanedPacks && cmd.RemoveOrphaned {
+		IDs := chkr.OrphanedPacks()
+		cmd.global.Verbosef("Remove %d orphaned packs... ", len(IDs))
+
+		for _, id := range IDs {
+			if err := repo.Backend().Remove(backend.Data, id.String()); err != nil {
+				fmt.Fprintf(os.Stderr, "%v\n", err)
+			}
+		}
+
+		cmd.global.Verbosef("done\n")
 	}
 
 	if errorsFound {