mirror
/
brotli
mirror of https://github.com/andybalholm/brotli.git
1
0
Fork 1
Code Issues Packages Projects Releases Wiki Activity
brotli/brotli_test.go

720 lines
19 KiB
Go
Raw Normal View History

Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
// Copyright 2016 Google Inc. All Rights Reserved.
//
// Distributed under MIT license.
// See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
package brotli
import (
"bytes"
Add test for issue 22.
2021-07-15 22:01:31 +03:00
"compress/gzip"
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
"fmt"
"io"
"io/ioutil"
"math"
"math/rand"
Add test for issue 22.
2021-07-15 22:01:31 +03:00
"os"
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
"testing"
"time"
Add matchfinder package. I've been experimenting for a while with a new brotli compressor. Instead of being a translation of the C implementation, it's a rewrite in Go, with a modular structure thanks to interfaces. (A few low-level functions still come from the C version, though.) The performance is getting to the point where it seems to be worth adding to the brotli repository.
2023-12-29 03:09:32 +03:00
"github.com/andybalholm/brotli/matchfinder"
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
)
func checkCompressedData(compressedData, wantOriginalData []byte) error {
uncompressed, err := Decode(compressedData)
if err != nil {
return fmt.Errorf("brotli decompress failed: %v", err)
}
if !bytes.Equal(uncompressed, wantOriginalData) {
if len(wantOriginalData) != len(uncompressed) {
return fmt.Errorf(""+
"Data doesn't uncompress to the original value.\n"+
"Length of original: %v\n"+
"Length of uncompressed: %v",
len(wantOriginalData), len(uncompressed))
}
for i := range wantOriginalData {
if wantOriginalData[i] != uncompressed[i] {
return fmt.Errorf(""+
"Data doesn't uncompress to the original value.\n"+
"Original at %v is %v\n"+
"Uncompressed at %v is %v",
i, wantOriginalData[i], i, uncompressed[i])
}
}
}
return nil
}
func TestEncoderNoWrite(t *testing.T) {
out := bytes.Buffer{}
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
e := NewWriterOptions(&out, WriterOptions{Quality: 5})
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
if err := e.Close(); err != nil {
t.Errorf("Close()=%v, want nil", err)
}
// Check Write after close.
if _, err := e.Write([]byte("hi")); err == nil {
t.Errorf("No error after Close() + Write()")
}
}
func TestEncoderEmptyWrite(t *testing.T) {
out := bytes.Buffer{}
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
e := NewWriterOptions(&out, WriterOptions{Quality: 5})
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
n, err := e.Write([]byte(""))
if n != 0 || err != nil {
t.Errorf("Write()=%v,%v, want 0, nil", n, err)
}
if err := e.Close(); err != nil {
t.Errorf("Close()=%v, want nil", err)
}
}
func TestWriter(t *testing.T) {
Reuse more memory when a Writer is Reset.
2020-05-06 03:18:33 +03:00
for level := BestSpeed; level <= BestCompression; level++ {
// Test basic encoder usage.
input := []byte("<html><body><H1>Hello world</H1></body></html>")
out := bytes.Buffer{}
e := NewWriterOptions(&out, WriterOptions{Quality: level})
in := bytes.NewReader([]byte(input))
n, err := io.Copy(e, in)
if err != nil {
t.Errorf("Copy Error: %v", err)
}
if int(n) != len(input) {
t.Errorf("Copy() n=%v, want %v", n, len(input))
}
if err := e.Close(); err != nil {
t.Errorf("Close Error after copied %d bytes: %v", n, err)
}
if err := checkCompressedData(out.Bytes(), input); err != nil {
t.Error(err)
}
out2 := bytes.Buffer{}
e.Reset(&out2)
n2, err := e.Write(input)
if err != nil {
t.Errorf("Write error after Reset: %v", err)
}
if n2 != len(input) {
t.Errorf("Write() after Reset n=%d, want %d", n2, len(input))
}
if err := e.Close(); err != nil {
t.Errorf("Close error after Reset (copied %d) bytes: %v", n2, err)
}
if !bytes.Equal(out.Bytes(), out2.Bytes()) {
t.Error("Compressed data after Reset doesn't equal first time")
}
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
}
}
Add test for issue 22.
2021-07-15 22:01:31 +03:00
func TestIssue22(t *testing.T) {
f, err := os.Open("testdata/issue22.gz")
if err != nil {
t.Fatalf("Error opening test data file: %v", err)
}
defer f.Close()
zr, err := gzip.NewReader(f)
if err != nil {
t.Fatalf("Error creating gzip reader: %v", err)
}
data, err := io.ReadAll(zr)
if err != nil {
t.Fatalf("Error reading test data: %v", err)
}
if len(data) != 2851073 {
t.Fatalf("Wrong length for test data: got %d, want 2851073", len(data))
}
for level := BestSpeed; level <= BestCompression; level++ {
out := bytes.Buffer{}
e := NewWriterOptions(&out, WriterOptions{Quality: level})
n, err := e.Write(data)
if err != nil {
t.Errorf("Error compressing data: %v", err)
}
if int(n) != len(data) {
t.Errorf("Write() n=%v, want %v", n, len(data))
}
if err := e.Close(); err != nil {
t.Errorf("Close Error after writing %d bytes: %v", n, err)
}
if err := checkCompressedData(out.Bytes(), data); err != nil {
t.Errorf("Error decompressing data at level %d: %v", level, err)
}
}
}
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
func TestEncoderStreams(t *testing.T) {
// Test that output is streamed.
// Adjust window size to ensure the encoder outputs at least enough bytes
// to fill the window.
const lgWin = 16
windowSize := int(math.Pow(2, lgWin))
input := make([]byte, 8*windowSize)
rand.Read(input)
out := bytes.Buffer{}
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
e := NewWriterOptions(&out, WriterOptions{Quality: 11, LGWin: lgWin})
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
halfInput := input[:len(input)/2]
in := bytes.NewReader(halfInput)
n, err := io.Copy(e, in)
if err != nil {
t.Errorf("Copy Error: %v", err)
}
// We've fed more data than the sliding window size. Check that some
// compressed data has been output.
if out.Len() == 0 {
t.Errorf("Output length is 0 after %d bytes written", n)
}
if err := e.Close(); err != nil {
t.Errorf("Close Error after copied %d bytes: %v", n, err)
}
if err := checkCompressedData(out.Bytes(), halfInput); err != nil {
t.Error(err)
}
}
func TestEncoderLargeInput(t *testing.T) {
Reuse more memory when a Writer is Reset.
2020-05-06 03:18:33 +03:00
for level := BestSpeed; level <= BestCompression; level++ {
input := make([]byte, 1000000)
rand.Read(input)
out := bytes.Buffer{}
e := NewWriterOptions(&out, WriterOptions{Quality: level})
in := bytes.NewReader(input)
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
Reuse more memory when a Writer is Reset.
2020-05-06 03:18:33 +03:00
n, err := io.Copy(e, in)
if err != nil {
t.Errorf("Copy Error: %v", err)
}
if int(n) != len(input) {
t.Errorf("Copy() n=%v, want %v", n, len(input))
}
if err := e.Close(); err != nil {
t.Errorf("Close Error after copied %d bytes: %v", n, err)
}
if err := checkCompressedData(out.Bytes(), input); err != nil {
t.Error(err)
}
out2 := bytes.Buffer{}
e.Reset(&out2)
n2, err := e.Write(input)
if err != nil {
t.Errorf("Write error after Reset: %v", err)
}
if n2 != len(input) {
t.Errorf("Write() after Reset n=%d, want %d", n2, len(input))
}
if err := e.Close(); err != nil {
t.Errorf("Close error after Reset (copied %d) bytes: %v", n2, err)
}
if !bytes.Equal(out.Bytes(), out2.Bytes()) {
t.Error("Compressed data after Reset doesn't equal first time")
}
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
}
}
func TestEncoderFlush(t *testing.T) {
input := make([]byte, 1000)
rand.Read(input)
out := bytes.Buffer{}
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
e := NewWriterOptions(&out, WriterOptions{Quality: 5})
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
in := bytes.NewReader(input)
_, err := io.Copy(e, in)
if err != nil {
t.Fatalf("Copy Error: %v", err)
}
if err := e.Flush(); err != nil {
t.Fatalf("Flush(): %v", err)
}
if out.Len() == 0 {
t.Fatalf("0 bytes written after Flush()")
}
decompressed := make([]byte, 1000)
reader := NewReader(bytes.NewReader(out.Bytes()))
n, err := reader.Read(decompressed)
if n != len(decompressed) || err != nil {
t.Errorf("Expected <%v, nil>, but <%v, %v>", len(decompressed), n, err)
}
if !bytes.Equal(decompressed, input) {
t.Errorf(""+
"Decompress after flush: %v\n"+
"%q\n"+
"want:\n%q",
err, decompressed, input)
}
if err := e.Close(); err != nil {
t.Errorf("Close(): %v", err)
}
}
type readerWithTimeout struct {
io.Reader
}
func (r readerWithTimeout) Read(p []byte) (int, error) {
type result struct {
n int
err error
}
ch := make(chan result)
go func() {
n, err := r.Reader.Read(p)
ch <- result{n, err}
}()
select {
case result := <-ch:
return result.n, result.err
case <-time.After(5 * time.Second):
return 0, fmt.Errorf("read timed out")
}
}
func TestDecoderStreaming(t *testing.T) {
pr, pw := io.Pipe()
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
writer := NewWriterOptions(pw, WriterOptions{Quality: 5, LGWin: 20})
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
reader := readerWithTimeout{NewReader(pr)}
defer func() {
go ioutil.ReadAll(pr) // swallow the "EOF" token from writer.Close
if err := writer.Close(); err != nil {
t.Errorf("writer.Close: %v", err)
}
}()
ch := make(chan []byte)
errch := make(chan error)
go func() {
for {
segment, ok := <-ch
if !ok {
return
}
if n, err := writer.Write(segment); err != nil || n != len(segment) {
errch <- fmt.Errorf("write=%v,%v, want %v,%v", n, err, len(segment), nil)
return
}
if err := writer.Flush(); err != nil {
errch <- fmt.Errorf("flush: %v", err)
return
}
}
}()
defer close(ch)
segments := [...][]byte{
[]byte("first"),
[]byte("second"),
[]byte("third"),
}
for k, segment := range segments {
t.Run(fmt.Sprintf("Segment%d", k), func(t *testing.T) {
select {
case ch <- segment:
case err := <-errch:
t.Fatalf("write: %v", err)
case <-time.After(5 * time.Second):
t.Fatalf("timed out")
}
wantLen := len(segment)
got := make([]byte, wantLen)
if n, err := reader.Read(got); err != nil || n != wantLen || !bytes.Equal(got, segment) {
t.Fatalf("read[%d]=%q,%v,%v, want %q,%v,%v", k, got, n, err, segment, wantLen, nil)
}
})
}
}
func TestReader(t *testing.T) {
content := bytes.Repeat([]byte("hello world!"), 10000)
encoded, _ := Encode(content, WriterOptions{Quality: 5})
r := NewReader(bytes.NewReader(encoded))
var decodedOutput bytes.Buffer
n, err := io.Copy(&decodedOutput, r)
if err != nil {
t.Fatalf("Copy(): n=%v, err=%v", n, err)
}
if got := decodedOutput.Bytes(); !bytes.Equal(got, content) {
t.Errorf(""+
"Reader output:\n"+
"%q\n"+
"want:\n"+
"<%d bytes>",
got, len(content))
}
Reuse ringbuffer in Reader. Fixes #33
2022-05-18 22:05:34 +03:00
r.Reset(bytes.NewReader(encoded))
decodedOutput.Reset()
n, err = io.Copy(&decodedOutput, r)
if err != nil {
t.Fatalf("After Reset: Copy(): n=%v, err=%v", n, err)
}
if got := decodedOutput.Bytes(); !bytes.Equal(got, content) {
t.Errorf("After Reset: "+
"Reader output:\n"+
"%q\n"+
"want:\n"+
"<%d bytes>",
got, len(content))
}
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
}
func TestDecode(t *testing.T) {
content := bytes.Repeat([]byte("hello world!"), 10000)
encoded, _ := Encode(content, WriterOptions{Quality: 5})
decoded, err := Decode(encoded)
if err != nil {
t.Errorf("Decode: %v", err)
}
if !bytes.Equal(decoded, content) {
t.Errorf(""+
"Decode content:\n"+
"%q\n"+
"want:\n"+
"<%d bytes>",
decoded, len(content))
}
}
func TestQuality(t *testing.T) {
content := bytes.Repeat([]byte("hello world!"), 10000)
for q := 0; q < 12; q++ {
encoded, _ := Encode(content, WriterOptions{Quality: q})
decoded, err := Decode(encoded)
if err != nil {
t.Errorf("Decode: %v", err)
}
if !bytes.Equal(decoded, content) {
t.Errorf(""+
"Decode content:\n"+
"%q\n"+
"want:\n"+
"<%d bytes>",
decoded, len(content))
}
}
}
func TestDecodeFuzz(t *testing.T) {
// Test that the decoder terminates with corrupted input.
content := bytes.Repeat([]byte("hello world!"), 100)
Fix int overflow in test on 32-bit. On 32-bit systems, the random array indexes were sometimes negative as a result of converting an int64 to int. Fixes #7.
2019-07-25 18:50:51 +03:00
rnd := rand.New(rand.NewSource(0))
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
encoded, err := Encode(content, WriterOptions{Quality: 5})
if err != nil {
t.Fatalf("Encode(<%d bytes>, _) = _, %s", len(content), err)
}
if len(encoded) == 0 {
t.Fatalf("Encode(<%d bytes>, _) produced empty output", len(content))
}
for i := 0; i < 100; i++ {
enc := append([]byte{}, encoded...)
for j := 0; j < 5; j++ {
Fix int overflow in test on 32-bit. On 32-bit systems, the random array indexes were sometimes negative as a result of converting an int64 to int. Fixes #7.
2019-07-25 18:50:51 +03:00
enc[rnd.Intn(len(enc))] = byte(rnd.Intn(256))
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
}
Decode(enc)
}
}
func TestDecodeTrailingData(t *testing.T) {
content := bytes.Repeat([]byte("hello world!"), 100)
encoded, _ := Encode(content, WriterOptions{Quality: 5})
_, err := Decode(append(encoded, 0))
if err == nil {
t.Errorf("Expected 'excessive input' error")
}
}
func TestEncodeDecode(t *testing.T) {
for _, test := range []struct {
data []byte
repeats int
}{
{nil, 0},
{[]byte("A"), 1},
{[]byte("<html><body><H1>Hello world</H1></body></html>"), 10},
{[]byte("<html><body><H1>Hello world</H1></body></html>"), 1000},
} {
t.Logf("case %q x %d", test.data, test.repeats)
input := bytes.Repeat(test.data, test.repeats)
encoded, err := Encode(input, WriterOptions{Quality: 5})
if err != nil {
t.Errorf("Encode: %v", err)
}
// Inputs are compressible, but may be too small to compress.
if maxSize := len(input)/2 + 20; len(encoded) >= maxSize {
t.Errorf(""+
"Encode returned %d bytes, want <%d\n"+
"Encoded=%q",
len(encoded), maxSize, encoded)
}
decoded, err := Decode(encoded)
if err != nil {
t.Errorf("Decode: %v", err)
}
if !bytes.Equal(decoded, input) {
var want string
if len(input) > 320 {
want = fmt.Sprintf("<%d bytes>", len(input))
} else {
want = fmt.Sprintf("%q", input)
}
t.Errorf(""+
"Decode content:\n"+
"%q\n"+
"want:\n"+
"%s",
decoded, want)
}
}
}
Reader.Reset: recover from errors. When a Reader encounters an error, its internal state may be corrupted. So Reset needs to be extra thorough, to avoid cascading errors. Based on PR 37 by Sovianum.
2022-09-24 05:39:05 +03:00
func TestErrorReset(t *testing.T) {
compress := func(input []byte) []byte {
var buf bytes.Buffer
writer := new(Writer)
writer.Reset(&buf)
writer.Write(input)
writer.Close()
return buf.Bytes()
}
corruptReader := func(reader *Reader) {
buf := bytes.NewBuffer([]byte("trash"))
reader.Reset(buf)
_, err := io.ReadAll(reader)
if err == nil {
t.Fatalf("successively decompressed invalid input")
}
}
decompress := func(input []byte, reader *Reader) []byte {
buf := bytes.NewBuffer(input)
reader.Reset(buf)
output, err := io.ReadAll(reader)
if err != nil {
t.Fatalf("failed to decompress data %s", err.Error())
}
return output
}
source := []byte("text")
compressed := compress(source)
reader := new(Reader)
corruptReader(reader)
decompressed := decompress(compressed, reader)
if string(source) != string(decompressed) {
t.Fatalf("decompressed data does not match original state")
}
}
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
// Encode returns content encoded with Brotli.
func Encode(content []byte, options WriterOptions) ([]byte, error) {
var buf bytes.Buffer
Fix tests broken by API change.
2019-04-25 19:03:32 +03:00
writer := NewWriterOptions(&buf, options)
Make BrotliDecoderState into a Reader.
2019-03-07 04:08:24 +03:00
_, err := writer.Write(content)
if closeErr := writer.Close(); err == nil {
err = closeErr
}
return buf.Bytes(), err
}
// Decode decodes Brotli encoded data.
func Decode(encodedData []byte) ([]byte, error) {
r := NewReader(bytes.NewReader(encodedData))
return ioutil.ReadAll(r)
}
Benchmark all compression levels.
2020-05-07 02:08:01 +03:00
func BenchmarkEncodeLevels(b *testing.B) {
opticks, err := ioutil.ReadFile("testdata/Isaac.Newton-Opticks.txt")
if err != nil {
b.Fatal(err)
}
for level := BestSpeed; level <= BestCompression; level++ {
b.Run(fmt.Sprintf("%d", level), func(b *testing.B) {
b.ReportAllocs()
b.SetBytes(int64(len(opticks)))
for i := 0; i < b.N; i++ {
w := NewWriterLevel(ioutil.Discard, level)
w.Write(opticks)
w.Close()
}
})
}
}
func BenchmarkEncodeLevelsReset(b *testing.B) {
opticks, err := ioutil.ReadFile("testdata/Isaac.Newton-Opticks.txt")
if err != nil {
b.Fatal(err)
}
for level := BestSpeed; level <= BestCompression; level++ {
buf := new(bytes.Buffer)
w := NewWriterLevel(buf, level)
w.Write(opticks)
w.Close()
Read multiple bytes in findMatchLengthWithLimit Use 64- or 32-bit loads instead of reading a byte at a time. The original C source did something like this, in a very C-ish way. It needed to be simplified to translate it to Go. The exact way this works was suggested by the assembly code in github.com/golang/snappy.
2020-05-15 20:43:19 +03:00
b.Run(fmt.Sprintf("%d", level), func(b *testing.B) {
Benchmark all compression levels.
2020-05-07 02:08:01 +03:00
b.ReportAllocs()
Read multiple bytes in findMatchLengthWithLimit Use 64- or 32-bit loads instead of reading a byte at a time. The original C source did something like this, in a very C-ish way. It needed to be simplified to translate it to Go. The exact way this works was suggested by the assembly code in github.com/golang/snappy.
2020-05-15 20:43:19 +03:00
b.ReportMetric(float64(len(opticks))/float64(buf.Len()), "ratio")
Benchmark all compression levels.
2020-05-07 02:08:01 +03:00
b.SetBytes(int64(len(opticks)))
for i := 0; i < b.N; i++ {
w.Reset(ioutil.Discard)
w.Write(opticks)
w.Close()
}
})
}
}
func BenchmarkDecodeLevels(b *testing.B) {
opticks, err := ioutil.ReadFile("testdata/Isaac.Newton-Opticks.txt")
if err != nil {
b.Fatal(err)
}
for level := BestSpeed; level <= BestCompression; level++ {
buf := new(bytes.Buffer)
w := NewWriterLevel(buf, level)
w.Write(opticks)
w.Close()
compressed := buf.Bytes()
b.Run(fmt.Sprintf("%d", level), func(b *testing.B) {
b.ReportAllocs()
b.SetBytes(int64(len(opticks)))
for i := 0; i < b.N; i++ {
io.Copy(ioutil.Discard, NewReader(bytes.NewReader(compressed)))
}
})
}
}
Add matchfinder package. I've been experimenting for a while with a new brotli compressor. Instead of being a translation of the C implementation, it's a rewrite in Go, with a modular structure thanks to interfaces. (A few low-level functions still come from the C version, though.) The performance is getting to the point where it seems to be worth adding to the brotli repository.
2023-12-29 03:09:32 +03:00
func test(t *testing.T, filename string, m matchfinder.MatchFinder, blockSize int) {
data, err := ioutil.ReadFile(filename)
if err != nil {
t.Fatal(err)
}
b := new(bytes.Buffer)
w := &matchfinder.Writer{
Dest: b,
MatchFinder: m,
Encoder: &Encoder{},
BlockSize: blockSize,
}
w.Write(data)
w.Close()
compressed := b.Bytes()
sr := NewReader(bytes.NewReader(compressed))
decompressed, err := ioutil.ReadAll(sr)
if err != nil {
t.Fatal(err)
}
if !bytes.Equal(decompressed, data) {
t.Fatal("decompressed output doesn't match")
}
}
func benchmark(b *testing.B, filename string, m matchfinder.MatchFinder, blockSize int) {
b.StopTimer()
b.ReportAllocs()
data, err := ioutil.ReadFile(filename)
if err != nil {
b.Fatal(err)
}
b.SetBytes(int64(len(data)))
buf := new(bytes.Buffer)
w := &matchfinder.Writer{
Dest: buf,
MatchFinder: m,
Encoder: &Encoder{},
BlockSize: blockSize,
}
w.Write(data)
w.Close()
b.ReportMetric(float64(len(data))/float64(buf.Len()), "ratio")
b.StartTimer()
for i := 0; i < b.N; i++ {
w.Reset(ioutil.Discard)
w.Write(data)
w.Close()
}
}
func TestEncodeM4(t *testing.T) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
test(t, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 18, DistanceBitCost: 57}, 1<<16)
Add matchfinder package. I've been experimenting for a while with a new brotli compressor. Instead of being a translation of the C implementation, it's a rewrite in Go, with a modular structure thanks to interfaces. (A few low-level functions still come from the C version, though.) The performance is getting to the point where it seems to be worth adding to the brotli repository.
2023-12-29 03:09:32 +03:00
}
func BenchmarkEncodeM4(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, DistanceBitCost: 57}, 1<<16)
Add matchfinder package. I've been experimenting for a while with a new brotli compressor. Instead of being a translation of the C implementation, it's a rewrite in Go, with a modular structure thanks to interfaces. (A few low-level functions still come from the C version, though.) The performance is getting to the point where it seems to be worth adding to the brotli repository.
2023-12-29 03:09:32 +03:00
}
matchfinder.M4: add LimitedSearch option Using LimitedSearch, it only checks for overlapping matches in one place instead of checking at each byte. This gains about 50% in compression speed while only losing about 2% in compression ratio.
2023-12-31 02:56:13 +03:00
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
func TestEncodeM4Chain1(t *testing.T) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
test(t, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 18, ChainLength: 1, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add LimitedSearch option Using LimitedSearch, it only checks for overlapping matches in one place instead of checking at each byte. This gains about 50% in compression speed while only losing about 2% in compression ratio.
2023-12-31 02:56:13 +03:00
}
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
func BenchmarkEncodeM4Chain1(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 1, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain2(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 2, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain4(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 4, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain8(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 8, HashLen: 5, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain16(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 16, HashLen: 5, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain32(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 32, HashLen: 5, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain64(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 64, HashLen: 5, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add match chain
2024-01-02 03:13:22 +03:00
}
func BenchmarkEncodeM4Chain128(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.M4{MaxDistance: 1 << 20, ChainLength: 128, HashLen: 5, DistanceBitCost: 57}, 1<<16)
matchfinder.M4: add LimitedSearch option Using LimitedSearch, it only checks for overlapping matches in one place instead of checking at each byte. This gains about 50% in compression speed while only losing about 2% in compression ratio.
2023-12-31 02:56:13 +03:00
}
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
func TestEncodeMultiHash6(t *testing.T) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
test(t, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 18, DistanceBitCost: 57, HashLengths: []int{6}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}
func TestEncodeMultiHash6_8(t *testing.T) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
test(t, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 18, DistanceBitCost: 57, HashLengths: []int{6, 8}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}
func BenchmarkEncodeMultiHash6(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 20, DistanceBitCost: 57, HashLengths: []int{6}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}
func BenchmarkEncodeMultiHash5_8(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 20, DistanceBitCost: 57, HashLengths: []int{5, 8}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}
func BenchmarkEncodeMultiHash5_7_9(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 20, DistanceBitCost: 57, HashLengths: []int{5, 7, 9}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}
func BenchmarkEncodeMultiHash5_6_7_9(b *testing.B) {
matchfinder: replace Score function with DistanceBitCost
2024-01-09 16:40:40 +03:00
benchmark(b, "testdata/Isaac.Newton-Opticks.txt", &matchfinder.MultiHash{MaxDistance: 1 << 20, DistanceBitCost: 57, HashLengths: []int{5, 6, 7, 9}}, 1<<16)
matchfinder: add MultiHash
2024-01-09 16:06:39 +03:00
}