Fix TestCreateSnapshot, do not generate duplicate data

src/restic/archive_reader_test.go

@@ -63,7 +63,7 @@ func TestArchiveReader(t *testing.T) {
 	size := int64(rand.Intn(50*1024*1024) + 50*1024*1024)
 	t.Logf("seed is 0x%016x, size is %v", seed, size)
 
-	f := fakeFile(t, seed, size)
+	f := fakeFile(seed, size)
 
 	sn, id, err := ArchiveReader(repo, nil, f, "fakefile")
 	if err != nil {
@@ -76,7 +76,7 @@ func TestArchiveReader(t *testing.T) {
 
 	t.Logf("snapshot saved as %v, tree is %v", id.Str(), sn.Tree.Str())
 
-	checkSavedFile(t, repo, *sn.Tree, "fakefile", fakeFile(t, seed, size))
+	checkSavedFile(t, repo, *sn.Tree, "fakefile", fakeFile(seed, size))
 }
 
 func BenchmarkArchiveReader(t *testing.B) {
@@ -86,7 +86,7 @@ func BenchmarkArchiveReader(t *testing.B) {
 	const size = 50 * 1024 * 1024
 
 	buf := make([]byte, size)
-	_, err := io.ReadFull(fakeFile(t, 23, size), buf)
+	_, err := io.ReadFull(fakeFile(23, size), buf)
 	if err != nil {
 		t.Fatal(err)
 	}

src/restic/testing.go

@@ -1,6 +1,7 @@
 package restic
 
 import (
+	"encoding/json"
 	"fmt"
 	"io"
 	"math/rand"
@@ -85,14 +86,21 @@ func (rd *randReader) Read(p []byte) (int, error) {
 }
 
 // fakeFile returns a reader which yields deterministic pseudo-random data.
-func fakeFile(t testing.TB, seed, size int64) io.Reader {
+func fakeFile(seed, size int64) io.Reader {
 	return io.LimitReader(newRandReader(rand.New(rand.NewSource(seed))), size)
 }
 
+type fakeTree struct {
+	t          testing.TB
+	repo       *repository.Repository
+	knownBlobs backend.IDSet
+}
+
 // saveFile reads from rd and saves the blobs in the repository. The list of
 // IDs is returned.
-func saveFile(t testing.TB, repo *repository.Repository, rd io.Reader) (blobs backend.IDs) {
-	ch := chunker.New(rd, repo.Config.ChunkerPolynomial)
+func (f fakeTree) saveFile(rd io.Reader) (blobs backend.IDs) {
+	blobs = backend.IDs{}
+	ch := chunker.New(rd, f.repo.Config.ChunkerPolynomial)
 
 	for {
 		chunk, err := ch.Next(getBuf())
@@ -101,47 +109,92 @@ func saveFile(t testing.TB, repo *repository.Repository, rd io.Reader) (blobs ba
 		}
 
 		if err != nil {
-			t.Fatalf("unabel to save chunk in repo: %v", err)
+			f.t.Fatalf("unable to save chunk in repo: %v", err)
 		}
 
-		id, err := repo.SaveAndEncrypt(pack.Data, chunk.Data, nil)
-		if err != nil {
-			t.Fatalf("error saving chunk: %v", err)
+		id := backend.Hash(chunk.Data)
+		if !f.knownBlobs.Has(id) {
+			_, err := f.repo.SaveAndEncrypt(pack.Data, chunk.Data, &id)
+			if err != nil {
+				f.t.Fatalf("error saving chunk: %v", err)
+			}
+			f.knownBlobs.Insert(id)
 		}
+
 		blobs = append(blobs, id)
 	}
 
 	return blobs
 }
 
-const maxFileSize = 1500000
-const maxSeed = 100
+const (
+	maxFileSize = 1500000
+	maxSeed     = 32
+	maxNodes    = 32
+)
 
-// saveTree saves a tree of fake files in the repo and returns the ID.
-func saveTree(t testing.TB, repo *repository.Repository, seed int64) backend.ID {
+func (f fakeTree) treeIsKnown(tree *Tree) (bool, backend.ID) {
+	data, err := json.Marshal(tree)
+	if err != nil {
+		f.t.Fatalf("json.Marshal(tree) returned error: %v", err)
+		return false, backend.ID{}
+	}
+	data = append(data, '\n')
+
+	// check if tree has been saved before
+	id := backend.Hash(data)
+	if f.knownBlobs.Has(id) {
+		return true, id
+	}
+
+	return false, id
+}
+
+// save stores a tree of fake files in the repo and returns the ID.
+func (f fakeTree) saveTree(seed int64, depth int) backend.ID {
 	rnd := rand.NewSource(seed)
-	numNodes := int(rnd.Int63() % 64)
-	t.Logf("create %v nodes", numNodes)
+	numNodes := int(rnd.Int63() % maxNodes)
 
 	var tree Tree
 	for i := 0; i < numNodes; i++ {
-		seed := rnd.Int63() % maxSeed
-		size := rnd.Int63() % maxFileSize
 
-		node := &Node{
-			Name: fmt.Sprintf("file-%v", seed),
-			Type: "file",
-			Mode: 0644,
-			Size: uint64(size),
+		// randomly select the type of the node, either tree (p = 1/4) or file (p = 3/4).
+		if depth > 1 && rnd.Int63()%4 == 0 {
+			treeSeed := rnd.Int63() % maxSeed
+			id := f.saveTree(treeSeed, depth-1)
+
+			node := &Node{
+				Name:    fmt.Sprintf("dir-%v", treeSeed),
+				Type:    "dir",
+				Mode:    0755,
+				Subtree: &id,
+			}
+
+			tree.Nodes = append(tree.Nodes, node)
+			continue
 		}
 
-		node.Content = saveFile(t, repo, fakeFile(t, seed, size))
+		fileSeed := rnd.Int63() % maxSeed
+		fileSize := (maxFileSize / maxSeed) * fileSeed
+
+		node := &Node{
+			Name: fmt.Sprintf("file-%v", fileSeed),
+			Type: "file",
+			Mode: 0644,
+			Size: uint64(fileSize),
+		}
+
+		node.Content = f.saveFile(fakeFile(fileSeed, fileSize))
 		tree.Nodes = append(tree.Nodes, node)
 	}
 
-	id, err := repo.SaveJSON(pack.Tree, tree)
+	if known, id := f.treeIsKnown(&tree); known {
+		return id
+	}
+
+	id, err := f.repo.SaveJSON(pack.Tree, tree)
 	if err != nil {
-		t.Fatal(err)
+		f.t.Fatal(err)
 	}
 
 	return id
@@ -149,8 +202,12 @@ func saveTree(t testing.TB, repo *repository.Repository, seed int64) backend.ID
 
 // TestCreateSnapshot creates a snapshot filled with fake data. The
 // fake data is generated deterministically from the timestamp `at`, which is
-// also used as the snapshot's timestamp.
-func TestCreateSnapshot(t testing.TB, repo *repository.Repository, at time.Time) backend.ID {
+// also used as the snapshot's timestamp. The tree's depth can be specified
+// with the parameter depth.
+func TestCreateSnapshot(t testing.TB, repo *repository.Repository, at time.Time, depth int) *Snapshot {
+	seed := at.Unix()
+	t.Logf("create fake snapshot at %s with seed %d", at, seed)
+
 	fakedir := fmt.Sprintf("fakedir-at-%v", at.Format("2006-01-02 15:04:05"))
 	snapshot, err := NewSnapshot([]string{fakedir})
 	if err != nil {
@@ -158,7 +215,13 @@ func TestCreateSnapshot(t testing.TB, repo *repository.Repository, at time.Time)
 	}
 	snapshot.Time = at
 
-	treeID := saveTree(t, repo, at.UnixNano())
+	f := fakeTree{
+		t:          t,
+		repo:       repo,
+		knownBlobs: backend.NewIDSet(),
+	}
+
+	treeID := f.saveTree(seed, depth)
 	snapshot.Tree = &treeID
 
 	id, err := repo.SaveJSONUnpacked(backend.Snapshot, snapshot)
@@ -166,6 +229,8 @@ func TestCreateSnapshot(t testing.TB, repo *repository.Repository, at time.Time)
 		t.Fatal(err)
 	}
 
+	snapshot.id = &id
+
 	t.Logf("saved snapshot %v", id.Str())
 
 	err = repo.Flush()
@@ -178,5 +243,5 @@ func TestCreateSnapshot(t testing.TB, repo *repository.Repository, at time.Time)
 		t.Fatal(err)
 	}
 
-	return id
+	return snapshot
 }

src/restic/testing_test.go

@@ -10,14 +10,17 @@ import (
 
 var testSnapshotTime = time.Unix(1460289341, 207401672)
 
-const testCreateSnapshots = 3
+const (
+	testCreateSnapshots = 3
+	testDepth           = 2
+)
 
 func TestCreateSnapshot(t *testing.T) {
 	repo, cleanup := repository.TestRepository(t)
 	defer cleanup()
 
 	for i := 0; i < testCreateSnapshots; i++ {
-		restic.TestCreateSnapshot(t, repo, testSnapshotTime.Add(time.Duration(i)*time.Second))
+		restic.TestCreateSnapshot(t, repo, testSnapshotTime.Add(time.Duration(i)*time.Second), testDepth)
 	}
 
 	snapshots, err := restic.LoadAllSnapshots(repo)