// This will deduplicate a buffer of zeros to an non-indexed stream
// written to a file.
// It is not recommended to use a single stream when you are writing to
// a stream.
func ExampleNewStreamWriter_file() {
// We will write to this
data, err := os.Create("outputstream.data")
if err != nil {
panic(err)
}
// Close, print stats and remove it
defer func() {
data.Close()
stat, _ := os.Stat("outputstream.data")
fmt.Println("Stream size:", stat.Size())
os.Remove("outputstream.data")
}()
// This is our input:
input := bytes.NewBuffer(make([]byte, 500000))
// Create a new writer, with each block being 1000 bytes,
// And allow it to use 10000 bytes of memory
w, err := dedup.NewStreamWriter(data, dedup.ModeFixed, 1000, 10000)
if err != nil {
panic(err)
}
defer w.Close()
// Copy our input to the writer.
io.Copy(w, input)
// Print the number of blocks written
fmt.Println("Blocks:", w.Blocks())
// OUTPUT: Blocks: 500
// Stream size: 1518
}
func BenchmarkFixedStreamWriter4K(t *testing.B) {
const totalinput = 10 << 20
input := getBufferSize(totalinput)
const size = 4 << 10
b := input.Bytes()
// Create some duplicates
for i := 0; i < 500; i++ {
// Read from 10 first blocks
src := b[(i%10)*size : (i%10)*size+size]
// Write into the following ones
dst := b[(10+i)*size : (i+10)*size+size]
copy(dst, src)
}
t.ResetTimer()
t.SetBytes(totalinput)
for i := 0; i < t.N; i++ {
input = bytes.NewBuffer(b)
w, _ := dedup.NewStreamWriter(ioutil.Discard, dedup.ModeFixed, size, 10*size)
io.Copy(w, input)
err := w.Close()
if err != nil {
t.Fatal(err)
}
}
}
// This will deduplicate a buffer of zeros to an non-indexed stream
func ExampleNewStreamWriter() {
// We will write to this
data := bytes.Buffer{}
// This is our input:
input := bytes.NewBuffer(make([]byte, 50000))
// Create a new writer, with each block being 1000 bytes,
// And allow it to use 10000 bytes of memory
w, err := dedup.NewStreamWriter(&data, dedup.ModeFixed, 1000, 10000)
if err != nil {
panic(err)
}
// Copy our input to the writer.
io.Copy(w, input)
// Close the writer
err = w.Close()
if err != nil {
panic(err)
}
// Let us inspect what was written:
fmt.Println("Blocks:", w.Blocks())
fmt.Println("Data size:", data.Len())
// OUTPUT: Blocks: 50
// Data size: 1068
}
func TestDynamicStreamWriter(t *testing.T) {
data := bytes.Buffer{}
const totalinput = 10 << 20
input := getBufferSize(totalinput)
const size = 64 << 10
b := input.Bytes()
// Create some duplicates
for i := 0; i < 50; i++ {
// Read from 10 first blocks
src := b[(i%10)*size : (i%10)*size+size]
// Write into the following ones
dst := b[(10+i)*size : (i+10)*size+size]
copy(dst, src)
}
input = bytes.NewBuffer(b)
w, err := dedup.NewStreamWriter(&data, dedup.ModeDynamic, size, 10*8*size)
if err != nil {
t.Fatal(err)
}
io.Copy(w, input)
err = w.Close()
if err != nil {
t.Fatal(err)
}
removed := ((totalinput) - data.Len()) / size
t.Log("Dynamic Data size:", data.Len())
t.Log("Removed", removed, "blocks")
// We don't know how many, but it should remove some blocks
if removed < 40 {
t.Fatal("didn't remove at least 40 blocks")
}
}
// This will deduplicate a buffer of zeros to an indexed stream
func ExampleNewStreamReader() {
// Create data we can read.
var data bytes.Buffer
input := bytes.NewBuffer(make([]byte, 50000))
// Set the memory limit to 10000 bytes
w, _ := dedup.NewStreamWriter(&data, dedup.ModeFixed, 1000, 10000)
_, _ = io.Copy(w, input)
_ = w.Close()
// Create a new stream reader:
r, err := dedup.NewStreamReader(&data)
if err != nil {
panic(err)
}
// Inspect how much memory it will use.
// Since this is a stream, it will print the worst possible scenario
fmt.Println("Memory use:", r.MaxMem())
var dst bytes.Buffer
// Read everything
_, err = io.Copy(&dst, r)
if err != nil && err != io.EOF {
panic(err)
}
// Let us inspect what was written:
fmt.Println("Returned data length:", dst.Len())
fmt.Println("Everything zero:", 0 == bytes.Compare(dst.Bytes(), make([]byte, 50000)))
// OUTPUT: Memory use: 10000
// Returned data length: 50000
// Everything zero: true
}
func TestReaderStream(t *testing.T) {
data := bytes.Buffer{}
const totalinput = 10<<20 + 65
input := getBufferSize(totalinput)
const size = 64 << 10
b := input.Bytes()
// Create some duplicates
for i := 0; i < 50; i++ {
// Read from 10 first blocks
src := b[(i%10)*size : (i%10)*size+size]
// Write into the following ones
dst := b[(10+i)*size : (i+10)*size+size]
copy(dst, src)
}
input = bytes.NewBuffer(b)
w, err := dedup.NewStreamWriter(&data, dedup.ModeFixed, size, 10*size)
if err != nil {
t.Fatal(err)
}
io.Copy(w, input)
err = w.Close()
if err != nil {
t.Fatal(err)
}
t.Log("Fixed Data size:", data.Len(), "-", data.Len()*100/totalinput, "%")
r, err := dedup.NewStreamReader(&data)
if err != nil {
t.Fatal(err)
}
t.Log("Maximum estimated memory:", r.MaxMem(), "bytes")
out, err := ioutil.ReadAll(r)
if err != io.EOF && err != nil {
t.Fatal(err)
}
if len(b) != len(out) {
t.Fatalf("Expected len %d, got %d", len(b), len(out))
}
if bytes.Compare(b, out) != 0 {
t.Fatal("Output mismatch")
}
err = r.Close()
if err != nil {
t.Fatal(err)
}
}
// Stream, 1K blocks on 10MB data.
func BenchmarkReaderStream1K(t *testing.B) {
data := &bytes.Buffer{}
const totalinput = 10 << 20
input := getBufferSize(totalinput)
const size = 1 << 10
b := input.Bytes()
// Create some duplicates
for i := 0; i < 500; i++ {
// Read from 10 first blocks
src := b[(i%10)*size : (i%10)*size+size]
// Write into the following ones
dst := b[(10+i)*size : (i+10)*size+size]
copy(dst, src)
}
input = bytes.NewBuffer(b)
w, err := dedup.NewStreamWriter(data, dedup.ModeFixed, size, 100*size)
if err != nil {
t.Fatal(err)
}
io.Copy(w, input)
err = w.Close()
if err != nil {
t.Fatal(err)
}
alldata := data.Bytes()
t.ResetTimer()
t.SetBytes(totalinput)
for i := 0; i < t.N; i++ {
input := bytes.NewBuffer(alldata)
r, err := dedup.NewStreamReader(input)
if err != nil {
t.Fatal(err)
}
n, err := io.Copy(ioutil.Discard, r)
if err != io.EOF && err != nil {
t.Fatal(err)
}
if len(b) != int(n) {
t.Fatalf("Expected len %d, got %d", len(b), n)
}
err = r.Close()
if err != nil {
t.Fatal(err)
}
}
}
func TestFixedStreamWriter(t *testing.T) {
data := bytes.Buffer{}
const totalinput = 10 << 20
input := getBufferSize(totalinput)
const size = 64 << 10
b := input.Bytes()
// Create some duplicates
for i := 0; i < 50; i++ {
// Read from 10 first blocks
src := b[(i%10)*size : (i%10)*size+size]
// Write into the following ones
dst := b[(10+i)*size : (i+10)*size+size]
copy(dst, src)
}
input = bytes.NewBuffer(b)
w, err := dedup.NewStreamWriter(&data, dedup.ModeFixed, size, 10*size)
if err != nil {
t.Fatal(err)
}
io.Copy(w, input)
err = w.Close()
if err != nil {
t.Fatal(err)
}
removed := ((totalinput) - data.Len()) / size
t.Log("Data size:", data.Len())
t.Log("Removed", removed, "blocks")
// We should get at least 50 blocks, but there is a little overhead
if removed < 49 {
t.Fatal("didn't remove at least 49 blocks")
}
if removed > 60 {
t.Fatal("removed unreasonable high amount of blocks")
}
}
