check: enable --read-data-* for specified snapshots - rebase step 1

Add code to cmd/restic/cmd_check.go to detect snapshots
Resolved conflict for cmd/restic/cmd_check.go - runCheck
integrated newCheckCommand(...)

changelog/unreleased/issue-3326

@@ -0,0 +1,8 @@
+check: enable --read-data-subset and --read-data for specified snapshot(s)
+
+When snapshots are specified on the command line, the metadata for these
+snapshots will be read and a set of packfiles will be created representing the data
+parts of these snapshots.
+
+https://github.com/restic/restic/issues/3326
+https://github.com/restic/restic/pull/5213

cmd/restic/cmd_check.go

@@ -73,6 +73,7 @@ type CheckOptions struct {
 	ReadDataSubset string
 	CheckUnused    bool
 	WithCache      bool
+	restic.SnapshotFilter
 }
 
 func (opts *CheckOptions) AddFlags(f *pflag.FlagSet) {
@@ -86,6 +87,7 @@ func (opts *CheckOptions) AddFlags(f *pflag.FlagSet) {
 		panic(err)
 	}
 	f.BoolVar(&opts.WithCache, "with-cache", false, "use existing cache, only read uncached data from repository")
+	initMultiSnapshotFilter(f, &opts.SnapshotFilter, true)
 }
 
 func checkFlags(opts CheckOptions) error {
@@ -222,9 +224,6 @@ func prepareCheckCache(opts CheckOptions, gopts *GlobalOptions, printer progress
 
 func runCheck(ctx context.Context, opts CheckOptions, gopts GlobalOptions, args []string, term *termstatus.Terminal) (checkSummary, error) {
 	summary := checkSummary{MessageType: "summary"}
-	if len(args) != 0 {
-		return summary, errors.Fatal("the check command expects no arguments, only options - please see `restic help check` for usage and flags")
-	}
 
 	var printer progress.Printer
 	if !gopts.JSON {
@@ -258,6 +257,21 @@ func runCheck(ctx context.Context, opts CheckOptions, gopts GlobalOptions, args
 		return summary, ctx.Err()
 	}
 
+	if len(args) > 0 {
+		snapshotLister, err := restic.MemorizeList(ctx, repo, restic.SnapshotFile)
+		if err != nil {
+			return summary, err
+		}
+
+		// run down the tree, take note of the data packfiles involved
+		for sn := range FindFilteredSnapshots(ctx, snapshotLister, repo, &opts.SnapshotFilter, args) {
+			err := chkr.FindDataPackfiles(ctx, repo, sn)
+			if err != nil {
+				return summary, err
+			}
+		}
+	}
+
 	errorsFound := false
 	for _, hint := range hints {
 		switch hint.(type) {

internal/checker/checker.go

@@ -15,6 +15,7 @@ import (
 	"github.com/restic/restic/internal/repository/pack"
 	"github.com/restic/restic/internal/restic"
 	"github.com/restic/restic/internal/ui/progress"
+	"github.com/restic/restic/internal/walker"
 	"golang.org/x/sync/errgroup"
 )
 
@@ -29,6 +30,7 @@ type Checker struct {
 		sync.Mutex
 		M restic.BlobSet
 	}
+	packSet     restic.IDSet
 	trackUnused bool
 
 	masterIndex *index.MasterIndex
@@ -41,6 +43,7 @@ type Checker struct {
 func New(repo restic.Repository, trackUnused bool) *Checker {
 	c := &Checker{
 		packs:       make(map[restic.ID]int64),
+		packSet:     restic.NewIDSet(),
 		masterIndex: index.NewMasterIndex(),
 		repo:        repo,
 		trackUnused: trackUnused,
@@ -431,12 +434,24 @@ func (c *Checker) UnusedBlobs(ctx context.Context) (blobs restic.BlobHandles, er
 
 // CountPacks returns the number of packs in the repository.
 func (c *Checker) CountPacks() uint64 {
-	return uint64(len(c.packs))
+	if len(c.packSet) == 0 {
+		return uint64(len(c.packs))
+	} else {
+		return uint64(len(c.packSet))
+	}
 }
 
 // GetPacks returns IDSet of packs in the repository
 func (c *Checker) GetPacks() map[restic.ID]int64 {
-	return c.packs
+	if len(c.packSet) == 0 {
+		return c.packs
+	} else {
+		result := map[restic.ID]int64{}
+		for packID := range c.packSet {
+			result[packID] = c.packs[packID]
+		}
+		return result
+	}
 }
 
 // ReadData loads all data from the repository and checks the integrity.
@@ -522,3 +537,35 @@ func (c *Checker) ReadPacks(ctx context.Context, packs map[restic.ID]int64, p *p
 		}
 	}
 }
+
+// find data packfiles for checking repository based on snapshots
+func (c *Checker) FindDataPackfiles(ctx context.Context, repo *repository.Repository, sn *restic.Snapshot) error {
+	err := walker.Walk(ctx, repo, *sn.Tree, walker.WalkVisitor{ProcessNode: func(parentTreeID restic.ID, _ string, node *restic.Node, err error) error {
+		if err != nil {
+			fmt.Printf("Unable to load tree %s\n ... which belongs to snapshot %s - reason %v\n", parentTreeID, sn.ID, err)
+			return walker.ErrSkipNode
+		}
+		if node == nil {
+			return nil
+		}
+
+		if node.Type == restic.NodeTypeFile {
+			for _, content := range node.Content {
+				result := repo.LookupBlob(restic.DataBlob, content)
+				if len(result) == 0 {
+					panic("checker.FindDataPackfiles: datablob not mapped!")
+				} else if len(result) > 1 {
+					panic("checker.FindDataPackfiles: datablob found several times!")
+				}
+				c.packSet.Insert(result[0].PackID)
+			}
+		}
+		return nil
+	}})
+
+	if err != nil {
+		return errors.New(fmt.Sprintf("walker.Walk does not want to walk - reason %v\n", err))
+	}
+
+	return nil
+}