EZPPFarm/hanayo
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

replace zxq.co/ripple/hanayo

Browse Source
This commit is contained in:
Alicia
2019-02-23 13:29:15 +00:00
commit c3d206c173
5871 changed files with 1353715 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
vendor/github.com/klauspost/compress/.gitignore generated vendored Normal file
View File

@@ -0,0 +1,24 @@
# Compiled Object files, Static and Dynamic libs (Shared Objects)
*.o
*.a
*.so
# Folders
_obj
_test
# Architecture specific extensions/prefixes
*.[568vq]
[568vq].out
*.cgo1.go
*.cgo2.c
_cgo_defun.c
_cgo_gotypes.go
_cgo_export.*
_testmain.go
*.exe
*.test
*.prof

24
vendor/github.com/klauspost/compress/.travis.yml generated vendored Normal file
View File

@@ -0,0 +1,24 @@
language: go
sudo: false
os:
- linux
- osx
go:
- 1.4
- 1.5
- 1.6
- 1.7
- tip
install:
- go get -t ./...
script:
- diff <(gofmt -d .) <(printf "")
- go test -v -cpu=2 ./...
- go test -cpu=2 -tags=noasm ./...
- go test -cpu=1,2,4 -short -race ./...
- go test -cpu=2,4 -short -race -tags=noasm ./...

27
vendor/github.com/klauspost/compress/LICENSE generated vendored Normal file
View File

@@ -0,0 +1,27 @@
Copyright (c) 2012 The Go Authors. All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above
copyright notice, this list of conditions and the following disclaimer
in the documentation and/or other materials provided with the
distribution.
* Neither the name of Google Inc. nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

289
vendor/github.com/klauspost/compress/README.md generated vendored Normal file
View File

@@ -0,0 +1,289 @@
# compress
This package is based on an optimized Deflate function, which is used by gzip/zip/zlib packages.
It offers slightly better compression at lower compression settings, and up to 3x faster encoding at highest compression level.
* [High Throughput Benchmark](http://blog.klauspost.com/go-gzipdeflate-benchmarks/).
* [Small Payload/Webserver Benchmarks](http://blog.klauspost.com/gzip-performance-for-go-webservers/).
* [Linear Time Compression](http://blog.klauspost.com/constant-time-gzipzip-compression/).
* [Re-balancing Deflate Compression Levels](https://blog.klauspost.com/rebalancing-deflate-compression-levels/)
[![Build Status](https://travis-ci.org/klauspost/compress.svg?branch=master)](https://travis-ci.org/klauspost/compress)
# changelog
* Jan 14, 2017: Reduce stack pressure due to array copies. See [Issue #18625(https://github.com/golang/go/issues/18625).
* Oct 25, 2016: Level 2-4 have been rewritten and now offers significantly better performance than before.
* Oct 20, 2016: Port zlib changes from Go 1.7 to fix zlib writer issue. Please update.
* Oct 16, 2016: Go 1.7 changes merged. Apples to apples this package is a few percent faster, but has a significantly better balance between speed and compression per level.
* Mar 24, 2016: Always attempt Huffman encoding on level 4-7. This improves base 64 encoded data compression.
* Mar 24, 2016: Small speedup for level 1-3.
* Feb 19, 2016: Faster bit writer, level -2 is 15% faster, level 1 is 4% faster.
* Feb 19, 2016: Handle small payloads faster in level 1-3.
* Feb 19, 2016: Added faster level 2 + 3 compression modes.
* Feb 19, 2016: [Rebalanced compression levels](https://blog.klauspost.com/rebalancing-deflate-compression-levels/), so there is a more even progresssion in terms of compression. New default level is 5.
* Feb 14, 2016: Snappy: Merge upstream changes.
* Feb 14, 2016: Snappy: Fix aggressive skipping.
* Feb 14, 2016: Snappy: Update benchmark.
* Feb 13, 2016: Deflate: Fixed assembler problem that could lead to sub-optimal compression.
* Feb 12, 2016: Snappy: Added AMD64 SSE 4.2 optimizations to matching, which makes easy to compress material run faster. Typical speedup is around 25%.
* Feb 9, 2016: Added Snappy package fork. This version is 5-7% faster, much more on hard to compress content.
* Jan 30, 2016: Optimize level 1 to 3 by not considering static dictionary or storing uncompressed. ~4-5% speedup.
* Jan 16, 2016: Optimization on deflate level 1,2,3 compression.
* Jan 8 2016: Merge [CL 18317](https://go-review.googlesource.com/#/c/18317): fix reading, writing of zip64 archives.
* Dec 8 2015: Make level 1 and -2 deterministic even if write size differs.
* Dec 8 2015: Split encoding functions, so hashing and matching can potentially be inlined. 1-3% faster on AMD64. 5% faster on other platforms.
* Dec 8 2015: Fixed rare [one byte out-of bounds read](https://github.com/klauspost/compress/issues/20). Please update!
* Nov 23 2015: Optimization on token writer. ~2-4% faster. Contributed by [@dsnet](https://github.com/dsnet).
* Nov 20 2015: Small optimization to bit writer on 64 bit systems.
* Nov 17 2015: Fixed out-of-bound errors if the underlying Writer returned an error. See [#15](https://github.com/klauspost/compress/issues/15).
* Nov 12 2015: Added [io.WriterTo](https://golang.org/pkg/io/#WriterTo) support to gzip/inflate.
* Nov 11 2015: Merged [CL 16669](https://go-review.googlesource.com/#/c/16669/4): archive/zip: enable overriding (de)compressors per file
* Oct 15 2015: Added skipping on uncompressible data. Random data speed up >5x.
# usage
The packages are drop-in replacements for standard libraries. Simply replace the import path to use them:
| old import | new import |
|--------------------|-----------------------------------------|
| `compress/gzip` | `github.com/klauspost/compress/gzip` |
| `compress/zlib` | `github.com/klauspost/compress/zlib` |
| `archive/zip` | `github.com/klauspost/compress/zip` |
| `compress/deflate` | `github.com/klauspost/compress/deflate` |
You may also be interested in [pgzip](https://github.com/klauspost/pgzip), which is a drop in replacement for gzip, which support multithreaded compression on big files and the optimized [crc32](https://github.com/klauspost/crc32) package used by these packages.
The packages contains the same as the standard library, so you can use the godoc for that: [gzip](http://golang.org/pkg/compress/gzip/), [zip](http://golang.org/pkg/archive/zip/), [zlib](http://golang.org/pkg/compress/zlib/), [flate](http://golang.org/pkg/compress/flate/).
Currently there is only minor speedup on decompression (mostly CRC32 calculation).
# deflate optimizations
* Minimum matches are 4 bytes, this leads to fewer searches and better compression. (In Go 1.7)
* Stronger hash (iSCSI CRC32) for matches on x64 with SSE 4.2 support. This leads to fewer hash collisions. (Go 1.7 also has improved hashes)
* Literal byte matching using SSE 4.2 for faster match comparisons. (not in Go)
* Bulk hashing on matches. (In Go 1.7)
* Much faster dictionary indexing with `NewWriterDict()`/`Reset()`. (In Go 1.7)
* Make Bit Coder faster by assuming we are on a 64 bit CPU. (In Go 1.7)
* Level 1 compression replaced by converted "Snappy" algorithm. (In Go 1.7)
* Uncompressible content is detected and skipped faster. (Only in BestSpeed in Go)
* A lot of branching eliminated by having two encoders for levels 4-6 and 7-9. (not in Go)
* All heap memory allocations eliminated. (In Go 1.7)
```
benchmark old ns/op new ns/op delta
BenchmarkEncodeDigitsSpeed1e4-4 554029 265175 -52.14%
BenchmarkEncodeDigitsSpeed1e5-4 3908558 2416595 -38.17%
BenchmarkEncodeDigitsSpeed1e6-4 37546692 24875330 -33.75%
BenchmarkEncodeDigitsDefault1e4-4 781510 486322 -37.77%
BenchmarkEncodeDigitsDefault1e5-4 15530248 6740175 -56.60%
BenchmarkEncodeDigitsDefault1e6-4 174915710 76498625 -56.27%
BenchmarkEncodeDigitsCompress1e4-4 769995 485652 -36.93%
BenchmarkEncodeDigitsCompress1e5-4 15450113 6929589 -55.15%
BenchmarkEncodeDigitsCompress1e6-4 175114660 73348495 -58.11%
BenchmarkEncodeTwainSpeed1e4-4 560122 275977 -50.73%
BenchmarkEncodeTwainSpeed1e5-4 3740978 2506095 -33.01%
BenchmarkEncodeTwainSpeed1e6-4 35542802 21904440 -38.37%
BenchmarkEncodeTwainDefault1e4-4 828534 549026 -33.74%
BenchmarkEncodeTwainDefault1e5-4 13667153 7528455 -44.92%
BenchmarkEncodeTwainDefault1e6-4 141191770 79952170 -43.37%
BenchmarkEncodeTwainCompress1e4-4 830050 545694 -34.26%
BenchmarkEncodeTwainCompress1e5-4 16620852 8460600 -49.10%
BenchmarkEncodeTwainCompress1e6-4 193326820 90808750 -53.03%
benchmark old MB/s new MB/s speedup
BenchmarkEncodeDigitsSpeed1e4-4 18.05 37.71 2.09x
BenchmarkEncodeDigitsSpeed1e5-4 25.58 41.38 1.62x
BenchmarkEncodeDigitsSpeed1e6-4 26.63 40.20 1.51x
BenchmarkEncodeDigitsDefault1e4-4 12.80 20.56 1.61x
BenchmarkEncodeDigitsDefault1e5-4 6.44 14.84 2.30x
BenchmarkEncodeDigitsDefault1e6-4 5.72 13.07 2.28x
BenchmarkEncodeDigitsCompress1e4-4 12.99 20.59 1.59x
BenchmarkEncodeDigitsCompress1e5-4 6.47 14.43 2.23x
BenchmarkEncodeDigitsCompress1e6-4 5.71 13.63 2.39x
BenchmarkEncodeTwainSpeed1e4-4 17.85 36.23 2.03x
BenchmarkEncodeTwainSpeed1e5-4 26.73 39.90 1.49x
BenchmarkEncodeTwainSpeed1e6-4 28.14 45.65 1.62x
BenchmarkEncodeTwainDefault1e4-4 12.07 18.21 1.51x
BenchmarkEncodeTwainDefault1e5-4 7.32 13.28 1.81x
BenchmarkEncodeTwainDefault1e6-4 7.08 12.51 1.77x
BenchmarkEncodeTwainCompress1e4-4 12.05 18.33 1.52x
BenchmarkEncodeTwainCompress1e5-4 6.02 11.82 1.96x
BenchmarkEncodeTwainCompress1e6-4 5.17 11.01 2.13x
```
* "Speed" is compression level 1
* "Default" is compression level 6
* "Compress" is compression level 9
* Test files are [Digits](https://github.com/klauspost/compress/blob/master/testdata/e.txt) (no matches) and [Twain](https://github.com/klauspost/compress/blob/master/testdata/Mark.Twain-Tom.Sawyer.txt) (plain text) .
As can be seen it shows a very good speedup all across the line.
`Twain` is a much more realistic benchmark, and will be closer to JSON/HTML performance. Here speed is equivalent or faster, up to 2 times.
**Without assembly**. This is what you can expect on systems that does not have amd64 and SSE 4:
```
benchmark old ns/op new ns/op delta
BenchmarkEncodeDigitsSpeed1e4-4 554029 249558 -54.96%
BenchmarkEncodeDigitsSpeed1e5-4 3908558 2295216 -41.28%
BenchmarkEncodeDigitsSpeed1e6-4 37546692 22594905 -39.82%
BenchmarkEncodeDigitsDefault1e4-4 781510 579850 -25.80%
BenchmarkEncodeDigitsDefault1e5-4 15530248 10096561 -34.99%
BenchmarkEncodeDigitsDefault1e6-4 174915710 111470780 -36.27%
BenchmarkEncodeDigitsCompress1e4-4 769995 579708 -24.71%
BenchmarkEncodeDigitsCompress1e5-4 15450113 10266373 -33.55%
BenchmarkEncodeDigitsCompress1e6-4 175114660 110170120 -37.09%
BenchmarkEncodeTwainSpeed1e4-4 560122 260679 -53.46%
BenchmarkEncodeTwainSpeed1e5-4 3740978 2097372 -43.94%
BenchmarkEncodeTwainSpeed1e6-4 35542802 20353449 -42.74%
BenchmarkEncodeTwainDefault1e4-4 828534 646016 -22.03%
BenchmarkEncodeTwainDefault1e5-4 13667153 10056369 -26.42%
BenchmarkEncodeTwainDefault1e6-4 141191770 105268770 -25.44%
BenchmarkEncodeTwainCompress1e4-4 830050 642401 -22.61%
BenchmarkEncodeTwainCompress1e5-4 16620852 11157081 -32.87%
BenchmarkEncodeTwainCompress1e6-4 193326820 121780770 -37.01%
benchmark old MB/s new MB/s speedup
BenchmarkEncodeDigitsSpeed1e4-4 18.05 40.07 2.22x
BenchmarkEncodeDigitsSpeed1e5-4 25.58 43.57 1.70x
BenchmarkEncodeDigitsSpeed1e6-4 26.63 44.26 1.66x
BenchmarkEncodeDigitsDefault1e4-4 12.80 17.25 1.35x
BenchmarkEncodeDigitsDefault1e5-4 6.44 9.90 1.54x
BenchmarkEncodeDigitsDefault1e6-4 5.72 8.97 1.57x
BenchmarkEncodeDigitsCompress1e4-4 12.99 17.25 1.33x
BenchmarkEncodeDigitsCompress1e5-4 6.47 9.74 1.51x
BenchmarkEncodeDigitsCompress1e6-4 5.71 9.08 1.59x
BenchmarkEncodeTwainSpeed1e4-4 17.85 38.36 2.15x
BenchmarkEncodeTwainSpeed1e5-4 26.73 47.68 1.78x
BenchmarkEncodeTwainSpeed1e6-4 28.14 49.13 1.75x
BenchmarkEncodeTwainDefault1e4-4 12.07 15.48 1.28x
BenchmarkEncodeTwainDefault1e5-4 7.32 9.94 1.36x
BenchmarkEncodeTwainDefault1e6-4 7.08 9.50 1.34x
BenchmarkEncodeTwainCompress1e4-4 12.05 15.57 1.29x
BenchmarkEncodeTwainCompress1e5-4 6.02 8.96 1.49x
BenchmarkEncodeTwainCompress1e6-4 5.17 8.21 1.59x
```
So even without the assembly optimizations there is a general speedup across the board.
## level 1-3 "snappy" compression
Levels 1 "Best Speed", 2 and 3 are completely replaced by a converted version of the algorithm found in Snappy, modified to be fully
compatible with the deflate bitstream (and thus still compatible with all existing zlib/gzip libraries and tools).
This version is considerably faster than the "old" deflate at level 1. It does however come at a compression loss, usually in the order of 3-4% compared to the old level 1. However, the speed is usually 1.75 times that of the fastest deflate mode.
In my previous experiments the most common case for "level 1" was that it provided no significant speedup, only lower compression compared to level 2 and sometimes even 3. However, the modified Snappy algorithm provides a very good sweet spot. Usually about 75% faster and with only little compression loss. Therefore I decided to *replace* level 1 with this mode entirely.
Input is split into blocks of 64kb of, and they are encoded independently (no backreferences across blocks) for the best speed. Contrary to Snappy the output is entropy-encoded, so you will almost always see better compression than Snappy. But Snappy is still about twice as fast as Snappy in deflate mode.
Level 2 and 3 have also been replaced. Level 2 is capable is matching between blocks and level 3 checks up to two hashes for matches it will try.
## compression levels
This table shows the compression at each level, and the percentage of the output size compared to output
at the similar level with the standard library. Compression data is `Twain`, see above.
(Not up-to-date after rebalancing)
| Level | Bytes | % size |
|-------|--------|--------|
| 1 | 194622 | 103.7% |
| 2 | 174684 | 96.85% |
| 3 | 170301 | 98.45% |
| 4 | 165253 | 97.69% |
| 5 | 161274 | 98.65% |
| 6 | 160464 | 99.71% |
| 7 | 160304 | 99.87% |
| 8 | 160279 | 99.99% |
| 9 | 160279 | 99.99% |
To interpret and example, this version of deflate compresses input of 407287 bytes to 161274 bytes at level 5, which is 98.6% of the size of what the standard library produces; 161274 bytes.
This means that from level 4 you can expect a compression level increase of a few percent. Level 1 is about 3% worse, as descibed above.
# linear time compression (huffman only)
This compression library adds a special compression level, named `ConstantCompression`, which allows near linear time compression. This is done by completely disabling matching of previous data, and only reduce the number of bits to represent each character.
This means that often used characters, like 'e' and ' ' (space) in text use the fewest bits to represent, and rare characters like '¤' takes more bits to represent. For more information see [wikipedia](https://en.wikipedia.org/wiki/Huffman_coding) or this nice [video](https://youtu.be/ZdooBTdW5bM).
Since this type of compression has much less variance, the compression speed is mostly unaffected by the input data, and is usually more than *180MB/s* for a single core.
The downside is that the compression ratio is usually considerably worse than even the fastest conventional compression. The compression raio can never be better than 8:1 (12.5%).
The linear time compression can be used as a "better than nothing" mode, where you cannot risk the encoder to slow down on some content. For comparison, the size of the "Twain" text is *233460 bytes* (+29% vs. level 1) and encode speed is 144MB/s (4.5x level 1). So in this case you trade a 30% size increase for a 4 times speedup.
For more information see my blog post on [Fast Linear Time Compression](http://blog.klauspost.com/constant-time-gzipzip-compression/).
This is implemented on Go 1.7 as "Huffman Only" mode, though not exposed for gzip.
# gzip/zip optimizations
* Uses the faster deflate
* Uses SSE 4.2 CRC32 calculations.
Speed increase is up to 3x of the standard library, but usually around 2x.
This is close to a real world benchmark as you will get. A 2.3MB JSON file. (NOTE: not up-to-date)
```
benchmark old ns/op new ns/op delta
BenchmarkGzipL1-4 95212470 59938275 -37.05%
BenchmarkGzipL2-4 102069730 76349195 -25.20%
BenchmarkGzipL3-4 115472770 82492215 -28.56%
BenchmarkGzipL4-4 153197780 107570890 -29.78%
BenchmarkGzipL5-4 203930260 134387930 -34.10%
BenchmarkGzipL6-4 233172100 145495400 -37.60%
BenchmarkGzipL7-4 297190260 197926950 -33.40%
BenchmarkGzipL8-4 512819750 376244733 -26.63%
BenchmarkGzipL9-4 563366800 403266833 -28.42%
benchmark old MB/s new MB/s speedup
BenchmarkGzipL1-4 52.11 82.78 1.59x
BenchmarkGzipL2-4 48.61 64.99 1.34x
BenchmarkGzipL3-4 42.97 60.15 1.40x
BenchmarkGzipL4-4 32.39 46.13 1.42x
BenchmarkGzipL5-4 24.33 36.92 1.52x
BenchmarkGzipL6-4 21.28 34.10 1.60x
BenchmarkGzipL7-4 16.70 25.07 1.50x
BenchmarkGzipL8-4 9.68 13.19 1.36x
BenchmarkGzipL9-4 8.81 12.30 1.40x
```
Multithreaded compression using [pgzip](https://github.com/klauspost/pgzip) comparison, Quadcore, CPU = 8:
(Not updated, old numbers)
```
benchmark old ns/op new ns/op delta
BenchmarkGzipL1 96155500 25981486 -72.98%
BenchmarkGzipL2 101905830 24601408 -75.86%
BenchmarkGzipL3 113506490 26321506 -76.81%
BenchmarkGzipL4 143708220 31761818 -77.90%
BenchmarkGzipL5 188210770 39602266 -78.96%
BenchmarkGzipL6 209812000 40402313 -80.74%
BenchmarkGzipL7 270015440 56103210 -79.22%
BenchmarkGzipL8 461359700 91255220 -80.22%
BenchmarkGzipL9 498361833 88755075 -82.19%
benchmark old MB/s new MB/s speedup
BenchmarkGzipL1 51.60 190.97 3.70x
BenchmarkGzipL2 48.69 201.69 4.14x
BenchmarkGzipL3 43.71 188.51 4.31x
BenchmarkGzipL4 34.53 156.22 4.52x
BenchmarkGzipL5 26.36 125.29 4.75x
BenchmarkGzipL6 23.65 122.81 5.19x
BenchmarkGzipL7 18.38 88.44 4.81x
BenchmarkGzipL8 10.75 54.37 5.06x
BenchmarkGzipL9 9.96 55.90 5.61x
```
# snappy package
The standard snappy package has now been improved. This repo contains a copy of the snappy repo.
I would advise to use the standard package: https://github.com/golang/snappy
# license
This code is licensed under the same conditions as the original Go code. See LICENSE file.

193
vendor/github.com/klauspost/compress/flate/asm_test.go generated vendored Normal file
View File

@@ -0,0 +1,193 @@
// Copyright 2015, Klaus Post, see LICENSE for details.
//+build amd64
package flate
import (
"math/rand"
"testing"
)
func TestCRC(t *testing.T) {
if !useSSE42 {
t.Skip("Skipping CRC test, no SSE 4.2 available")
}
for _, x := range deflateTests {
y := x.out
if len(y) >= minMatchLength {
t.Logf("In: %v, Out:0x%08x", y[0:minMatchLength], crc32sse(y[0:minMatchLength]))
}
}
}
func TestCRCBulk(t *testing.T) {
if !useSSE42 {
t.Skip("Skipping CRC test, no SSE 4.2 available")
}
for _, x := range deflateTests {
y := x.out
y = append(y, y...)
y = append(y, y...)
y = append(y, y...)
y = append(y, y...)
y = append(y, y...)
y = append(y, y...)
if !testing.Short() {
y = append(y, y...)
y = append(y, y...)
}
y = append(y, 1)
if len(y) >= minMatchLength {
for j := len(y) - 1; j >= 4; j-- {
// Create copy, so we easier detect of-of-bound reads
test := make([]byte, j)
test2 := make([]byte, j)
copy(test, y[:j])
copy(test2, y[:j])
// We allocate one more than we need to test for unintentional overwrites
dst := make([]uint32, j-3+1)
ref := make([]uint32, j-3+1)
for i := range dst {
dst[i] = uint32(i + 100)
ref[i] = uint32(i + 101)
}
// Last entry must NOT be overwritten.
dst[j-3] = 0x1234
ref[j-3] = 0x1234
// Do two encodes we can compare
crc32sseAll(test, dst)
crc32sseAll(test2, ref)
// Check all values
for i, got := range dst {
if i == j-3 {
if dst[i] != 0x1234 {
t.Fatalf("end of expected dst overwritten, was %08x", uint32(dst[i]))
}
continue
}
expect := crc32sse(y[i : i+4])
if got != expect && got == uint32(i)+100 {
t.Errorf("Len:%d Index:%d, expected 0x%08x but not modified", len(y), i, uint32(expect))
} else if got != expect {
t.Errorf("Len:%d Index:%d, got 0x%08x expected:0x%08x", len(y), i, uint32(got), uint32(expect))
}
expect = ref[i]
if got != expect {
t.Errorf("Len:%d Index:%d, got 0x%08x expected:0x%08x", len(y), i, got, expect)
}
}
}
}
}
}
func TestMatchLen(t *testing.T) {
if !useSSE42 {
t.Skip("Skipping Matchlen test, no SSE 4.2 available")
}
// Maximum length tested
var maxLen = 512
// Skips per iteration
is, js, ks := 3, 2, 1
if testing.Short() {
is, js, ks = 7, 5, 3
}
a := make([]byte, maxLen)
b := make([]byte, maxLen)
bb := make([]byte, maxLen)
rand.Seed(1)
for i := range a {
a[i] = byte(rand.Int63())
b[i] = byte(rand.Int63())
}
// Test different lengths
for i := 0; i < maxLen; i += is {
// Test different dst offsets.
for j := 0; j < maxLen-1; j += js {
copy(bb, b)
// Test different src offsets
for k := i - 1; k >= 0; k -= ks {
copy(bb[j:], a[k:i])
maxTest := maxLen - j
if maxTest > maxLen-k {
maxTest = maxLen - k
}
got := matchLenSSE4(a[k:], bb[j:], maxTest)
expect := matchLenReference(a[k:], bb[j:], maxTest)
if got > maxTest || got < 0 {
t.Fatalf("unexpected result %d (len:%d, src offset: %d, dst offset:%d)", got, maxTest, k, j)
}
if got != expect {
t.Fatalf("Mismatch, expected %d, got %d", expect, got)
}
}
}
}
}
// matchLenReference is a reference matcher.
func matchLenReference(a, b []byte, max int) int {
for i := 0; i < max; i++ {
if a[i] != b[i] {
return i
}
}
return max
}
func TestHistogram(t *testing.T) {
if !useSSE42 {
t.Skip("Skipping Matchlen test, no SSE 4.2 available")
}
// Maximum length tested
const maxLen = 65536
var maxOff = 8
// Skips per iteration
is, js := 5, 3
if testing.Short() {
is, js = 9, 1
maxOff = 1
}
a := make([]byte, maxLen+maxOff)
rand.Seed(1)
for i := range a {
a[i] = byte(rand.Int63())
}
// Test different lengths
for i := 0; i <= maxLen; i += is {
// Test different offsets
for j := 0; j < maxOff; j += js {
var got [256]int32
var reference [256]int32
histogram(a[j:i+j], got[:])
histogramReference(a[j:i+j], reference[:])
for k := range got {
if got[k] != reference[k] {
t.Fatalf("mismatch at len:%d, offset:%d, value %d: (got) %d != %d (expected)", i, j, k, got[k], reference[k])
}
}
}
}
}
// histogramReference is a reference
func histogramReference(b []byte, h []int32) {
if len(h) < 256 {
panic("Histogram too small")
}
for _, t := range b {
h[t]++
}
}

32
vendor/github.com/klauspost/compress/flate/copy.go generated vendored Normal file
View File

@@ -0,0 +1,32 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
// forwardCopy is like the built-in copy function except that it always goes
// forward from the start, even if the dst and src overlap.
// It is equivalent to:
// for i := 0; i < n; i++ {
// mem[dst+i] = mem[src+i]
// }
func forwardCopy(mem []byte, dst, src, n int) {
if dst <= src {
copy(mem[dst:dst+n], mem[src:src+n])
return
}
for {
if dst >= src+n {
copy(mem[dst:dst+n], mem[src:src+n])
return
}
// There is some forward overlap. The destination
// will be filled with a repeated pattern of mem[src:src+k].
// We copy one instance of the pattern here, then repeat.
// Each time around this loop k will double.
k := dst - src
copy(mem[dst:dst+k], mem[src:src+k])
n -= k
dst += k
}
}

54
vendor/github.com/klauspost/compress/flate/copy_test.go generated vendored Normal file
View File

@@ -0,0 +1,54 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"testing"
)
func TestForwardCopy(t *testing.T) {
testCases := []struct {
dst0, dst1 int
src0, src1 int
want string
}{
{0, 9, 0, 9, "012345678"},
{0, 5, 4, 9, "45678"},
{4, 9, 0, 5, "01230"},
{1, 6, 3, 8, "34567"},
{3, 8, 1, 6, "12121"},
{0, 9, 3, 6, "345"},
{3, 6, 0, 9, "012"},
{1, 6, 0, 9, "00000"},
{0, 4, 7, 8, "7"},
{0, 1, 6, 8, "6"},
{4, 4, 6, 9, ""},
{2, 8, 6, 6, ""},
{0, 0, 0, 0, ""},
}
for _, tc := range testCases {
b := []byte("0123456789")
n := tc.dst1 - tc.dst0
if tc.src1-tc.src0 < n {
n = tc.src1 - tc.src0
}
forwardCopy(b, tc.dst0, tc.src0, n)
got := string(b[tc.dst0 : tc.dst0+n])
if got != tc.want {
t.Errorf("dst=b[%d:%d], src=b[%d:%d]: got %q, want %q",
tc.dst0, tc.dst1, tc.src0, tc.src1, got, tc.want)
}
// Check that the bytes outside of dst[:n] were not modified.
for i, x := range b {
if i >= tc.dst0 && i < tc.dst0+n {
continue
}
if int(x) != '0'+i {
t.Errorf("dst=b[%d:%d], src=b[%d:%d]: copy overrun at b[%d]: got '%c', want '%c'",
tc.dst0, tc.dst1, tc.src0, tc.src1, i, x, '0'+i)
}
}
}
}

41
vendor/github.com/klauspost/compress/flate/crc32_amd64.go generated vendored Normal file
View File

@@ -0,0 +1,41 @@
//+build !noasm
//+build !appengine
// Copyright 2015, Klaus Post, see LICENSE for details.
package flate
import (
"github.com/klauspost/cpuid"
)
// crc32sse returns a hash for the first 4 bytes of the slice
// len(a) must be >= 4.
//go:noescape
func crc32sse(a []byte) uint32
// crc32sseAll calculates hashes for each 4-byte set in a.
// dst must be east len(a) - 4 in size.
// The size is not checked by the assembly.
//go:noescape
func crc32sseAll(a []byte, dst []uint32)
// matchLenSSE4 returns the number of matching bytes in a and b
// up to length 'max'. Both slices must be at least 'max'
// bytes in size.
//
// TODO: drop the "SSE4" name, since it doesn't use any SSE instructions.
//
//go:noescape
func matchLenSSE4(a, b []byte, max int) int
// histogram accumulates a histogram of b in h.
// h must be at least 256 entries in length,
// and must be cleared before calling this function.
//go:noescape
func histogram(b []byte, h []int32)
// Detect SSE 4.2 feature.
func init() {
useSSE42 = cpuid.CPU.SSE42()
}

213
vendor/github.com/klauspost/compress/flate/crc32_amd64.s generated vendored Normal file
View File

@@ -0,0 +1,213 @@
//+build !noasm
//+build !appengine
// Copyright 2015, Klaus Post, see LICENSE for details.
// func crc32sse(a []byte) uint32
TEXT ·crc32sse(SB), 4, $0
MOVQ a+0(FP), R10
XORQ BX, BX
// CRC32 dword (R10), EBX
BYTE $0xF2; BYTE $0x41; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0x1a
MOVL BX, ret+24(FP)
RET
// func crc32sseAll(a []byte, dst []uint32)
TEXT ·crc32sseAll(SB), 4, $0
MOVQ a+0(FP), R8 // R8: src
MOVQ a_len+8(FP), R10 // input length
MOVQ dst+24(FP), R9 // R9: dst
SUBQ $4, R10
JS end
JZ one_crc
MOVQ R10, R13
SHRQ $2, R10 // len/4
ANDQ $3, R13 // len&3
XORQ BX, BX
ADDQ $1, R13
TESTQ R10, R10
JZ rem_loop
crc_loop:
MOVQ (R8), R11
XORQ BX, BX
XORQ DX, DX
XORQ DI, DI
MOVQ R11, R12
SHRQ $8, R11
MOVQ R12, AX
MOVQ R11, CX
SHRQ $16, R12
SHRQ $16, R11
MOVQ R12, SI
// CRC32 EAX, EBX
BYTE $0xF2; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0xd8
// CRC32 ECX, EDX
BYTE $0xF2; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0xd1
// CRC32 ESI, EDI
BYTE $0xF2; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0xfe
MOVL BX, (R9)
MOVL DX, 4(R9)
MOVL DI, 8(R9)
XORQ BX, BX
MOVL R11, AX
// CRC32 EAX, EBX
BYTE $0xF2; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0xd8
MOVL BX, 12(R9)
ADDQ $16, R9
ADDQ $4, R8
XORQ BX, BX
SUBQ $1, R10
JNZ crc_loop
rem_loop:
MOVL (R8), AX
// CRC32 EAX, EBX
BYTE $0xF2; BYTE $0x0f
BYTE $0x38; BYTE $0xf1; BYTE $0xd8
MOVL BX, (R9)
ADDQ $4, R9
ADDQ $1, R8
XORQ BX, BX
SUBQ $1, R13
JNZ rem_loop
end:
RET
one_crc:
MOVQ $1, R13
XORQ BX, BX
JMP rem_loop
// func matchLenSSE4(a, b []byte, max int) int
TEXT ·matchLenSSE4(SB), 4, $0
MOVQ a_base+0(FP), SI
MOVQ b_base+24(FP), DI
MOVQ DI, DX
MOVQ max+48(FP), CX
cmp8:
// As long as we are 8 or more bytes before the end of max, we can load and
// compare 8 bytes at a time. If those 8 bytes are equal, repeat.
CMPQ CX, $8
JLT cmp1
MOVQ (SI), AX
MOVQ (DI), BX
CMPQ AX, BX
JNE bsf
ADDQ $8, SI
ADDQ $8, DI
SUBQ $8, CX
JMP cmp8
bsf:
// If those 8 bytes were not equal, XOR the two 8 byte values, and return
// the index of the first byte that differs. The BSF instruction finds the
// least significant 1 bit, the amd64 architecture is little-endian, and
// the shift by 3 converts a bit index to a byte index.
XORQ AX, BX
BSFQ BX, BX
SHRQ $3, BX
ADDQ BX, DI
// Subtract off &b[0] to convert from &b[ret] to ret, and return.
SUBQ DX, DI
MOVQ DI, ret+56(FP)
RET
cmp1:
// In the slices' tail, compare 1 byte at a time.
CMPQ CX, $0
JEQ matchLenEnd
MOVB (SI), AX
MOVB (DI), BX
CMPB AX, BX
JNE matchLenEnd
ADDQ $1, SI
ADDQ $1, DI
SUBQ $1, CX
JMP cmp1
matchLenEnd:
// Subtract off &b[0] to convert from &b[ret] to ret, and return.
SUBQ DX, DI
MOVQ DI, ret+56(FP)
RET
// func histogram(b []byte, h []int32)
TEXT ·histogram(SB), 4, $0
MOVQ b+0(FP), SI // SI: &b
MOVQ b_len+8(FP), R9 // R9: len(b)
MOVQ h+24(FP), DI // DI: Histogram
MOVQ R9, R8
SHRQ $3, R8
JZ hist1
XORQ R11, R11
loop_hist8:
MOVQ (SI), R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
MOVB R10, R11
INCL (DI)(R11*4)
SHRQ $8, R10
INCL (DI)(R10*4)
ADDQ $8, SI
DECQ R8
JNZ loop_hist8
hist1:
ANDQ $7, R9
JZ end_hist
XORQ R10, R10
loop_hist1:
MOVB (SI), R10
INCL (DI)(R10*4)
INCQ SI
DECQ R9
JNZ loop_hist1
end_hist:
RET

35
vendor/github.com/klauspost/compress/flate/crc32_noasm.go generated vendored Normal file
View File

@@ -0,0 +1,35 @@
//+build !amd64 noasm appengine
// Copyright 2015, Klaus Post, see LICENSE for details.
package flate
func init() {
useSSE42 = false
}
// crc32sse should never be called.
func crc32sse(a []byte) uint32 {
panic("no assembler")
}
// crc32sseAll should never be called.
func crc32sseAll(a []byte, dst []uint32) {
panic("no assembler")
}
// matchLenSSE4 should never be called.
func matchLenSSE4(a, b []byte, max int) int {
panic("no assembler")
return 0
}
// histogram accumulates a histogram of b in h.
//
// len(h) must be >= 256, and h's elements must be all zeroes.
func histogram(b []byte, h []int32) {
h = h[:256]
for _, t := range b {
h[t]++
}
}

1353
vendor/github.com/klauspost/compress/flate/deflate.go generated vendored Normal file
View File

File diff suppressed because it is too large Load Diff

648
vendor/github.com/klauspost/compress/flate/deflate_test.go generated vendored Normal file
View File

@@ -0,0 +1,648 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Copyright (c) 2015 Klaus Post
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"bytes"
"fmt"
"io"
"io/ioutil"
"reflect"
"strings"
"sync"
"testing"
)
type deflateTest struct {
in []byte
level int
out []byte
}
type deflateInflateTest struct {
in []byte
}
type reverseBitsTest struct {
in uint16
bitCount uint8
out uint16
}
var deflateTests = []*deflateTest{
{[]byte{}, 0, []byte{1, 0, 0, 255, 255}},
{[]byte{0x11}, BestCompression, []byte{18, 4, 4, 0, 0, 255, 255}},
{[]byte{0x11}, BestCompression, []byte{18, 4, 4, 0, 0, 255, 255}},
{[]byte{0x11}, BestCompression, []byte{18, 4, 4, 0, 0, 255, 255}},
{[]byte{0x11}, 0, []byte{0, 1, 0, 254, 255, 17, 1, 0, 0, 255, 255}},
{[]byte{0x11, 0x12}, 0, []byte{0, 2, 0, 253, 255, 17, 18, 1, 0, 0, 255, 255}},
{[]byte{0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11}, 0,
[]byte{0, 8, 0, 247, 255, 17, 17, 17, 17, 17, 17, 17, 17, 1, 0, 0, 255, 255},
},
{[]byte{}, 1, []byte{1, 0, 0, 255, 255}},
{[]byte{0x11}, BestCompression, []byte{18, 4, 4, 0, 0, 255, 255}},
{[]byte{0x11, 0x12}, BestCompression, []byte{18, 20, 2, 4, 0, 0, 255, 255}},
{[]byte{0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11}, BestCompression, []byte{18, 132, 2, 64, 0, 0, 0, 255, 255}},
{[]byte{}, 9, []byte{1, 0, 0, 255, 255}},
{[]byte{0x11}, 9, []byte{18, 4, 4, 0, 0, 255, 255}},
{[]byte{0x11, 0x12}, 9, []byte{18, 20, 2, 4, 0, 0, 255, 255}},
{[]byte{0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11}, 9, []byte{18, 132, 2, 64, 0, 0, 0, 255, 255}},
}
var deflateInflateTests = []*deflateInflateTest{
{[]byte{}},
{[]byte{0x11}},
{[]byte{0x11, 0x12}},
{[]byte{0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11, 0x11}},
{[]byte{0x11, 0x10, 0x13, 0x41, 0x21, 0x21, 0x41, 0x13, 0x87, 0x78, 0x13}},
{largeDataChunk()},
}
var reverseBitsTests = []*reverseBitsTest{
{1, 1, 1},
{1, 2, 2},
{1, 3, 4},
{1, 4, 8},
{1, 5, 16},
{17, 5, 17},
{257, 9, 257},
{29, 5, 23},
}
func largeDataChunk() []byte {
result := make([]byte, 100000)
for i := range result {
result[i] = byte(i * i & 0xFF)
}
return result
}
func TestCRCBulkOld(t *testing.T) {
for _, x := range deflateTests {
y := x.out
if len(y) >= minMatchLength {
y = append(y, y...)
for j := 4; j < len(y); j++ {
y := y[:j]
dst := make([]uint32, len(y)-minMatchLength+1)
for i := range dst {
dst[i] = uint32(i + 100)
}
bulkHash4(y, dst)
for i, val := range dst {
got := val
expect := hash4(y[i:])
if got != expect && got == uint32(i)+100 {
t.Errorf("Len:%d Index:%d, expected 0x%08x but not modified", len(y), i, expect)
} else if got != expect {
t.Errorf("Len:%d Index:%d, got 0x%08x expected:0x%08x", len(y), i, got, expect)
} else {
//t.Logf("Len:%d Index:%d OK (0x%08x)", len(y), i, got)
}
}
}
}
}
}
func TestDeflate(t *testing.T) {
for _, h := range deflateTests {
var buf bytes.Buffer
w, err := NewWriter(&buf, h.level)
if err != nil {
t.Errorf("NewWriter: %v", err)
continue
}
w.Write(h.in)
w.Close()
if !bytes.Equal(buf.Bytes(), h.out) {
t.Errorf("Deflate(%d, %x) = \n%#v, want \n%#v", h.level, h.in, buf.Bytes(), h.out)
}
}
}
// A sparseReader returns a stream consisting of 0s followed by 1<<16 1s.
// This tests missing hash references in a very large input.
type sparseReader struct {
l int64
cur int64
}
func (r *sparseReader) Read(b []byte) (n int, err error) {
if r.cur >= r.l {
return 0, io.EOF
}
n = len(b)
cur := r.cur + int64(n)
if cur > r.l {
n -= int(cur - r.l)
cur = r.l
}
for i := range b[0:n] {
if r.cur+int64(i) >= r.l-1<<16 {
b[i] = 1
} else {
b[i] = 0
}
}
r.cur = cur
return
}
func TestVeryLongSparseChunk(t *testing.T) {
if testing.Short() {
t.Skip("skipping sparse chunk during short test")
}
w, err := NewWriter(ioutil.Discard, 1)
if err != nil {
t.Errorf("NewWriter: %v", err)
return
}
if _, err = io.Copy(w, &sparseReader{l: 23E8}); err != nil {
t.Errorf("Compress failed: %v", err)
return
}
}
type syncBuffer struct {
buf bytes.Buffer
mu sync.RWMutex
closed bool
ready chan bool
}
func newSyncBuffer() *syncBuffer {
return &syncBuffer{ready: make(chan bool, 1)}
}
func (b *syncBuffer) Read(p []byte) (n int, err error) {
for {
b.mu.RLock()
n, err = b.buf.Read(p)
b.mu.RUnlock()
if n > 0 || b.closed {
return
}
<-b.ready
}
}
func (b *syncBuffer) signal() {
select {
case b.ready <- true:
default:
}
}
func (b *syncBuffer) Write(p []byte) (n int, err error) {
n, err = b.buf.Write(p)
b.signal()
return
}
func (b *syncBuffer) WriteMode() {
b.mu.Lock()
}
func (b *syncBuffer) ReadMode() {
b.mu.Unlock()
b.signal()
}
func (b *syncBuffer) Close() error {
b.closed = true
b.signal()
return nil
}
func testSync(t *testing.T, level int, input []byte, name string) {
if len(input) == 0 {
return
}
t.Logf("--testSync %d, %d, %s", level, len(input), name)
buf := newSyncBuffer()
buf1 := new(bytes.Buffer)
buf.WriteMode()
w, err := NewWriter(io.MultiWriter(buf, buf1), level)
if err != nil {
t.Errorf("NewWriter: %v", err)
return
}
r := NewReader(buf)
// Write half the input and read back.
for i := 0; i < 2; i++ {
var lo, hi int
if i == 0 {
lo, hi = 0, (len(input)+1)/2
} else {
lo, hi = (len(input)+1)/2, len(input)
}
t.Logf("#%d: write %d-%d", i, lo, hi)
if _, err := w.Write(input[lo:hi]); err != nil {
t.Errorf("testSync: write: %v", err)
return
}
if i == 0 {
if err := w.Flush(); err != nil {
t.Errorf("testSync: flush: %v", err)
return
}
} else {
if err := w.Close(); err != nil {
t.Errorf("testSync: close: %v", err)
}
}
buf.ReadMode()
out := make([]byte, hi-lo+1)
m, err := io.ReadAtLeast(r, out, hi-lo)
t.Logf("#%d: read %d", i, m)
if m != hi-lo || err != nil {
t.Errorf("testSync/%d (%d, %d, %s): read %d: %d, %v (%d left)", i, level, len(input), name, hi-lo, m, err, buf.buf.Len())
return
}
if !bytes.Equal(input[lo:hi], out[:hi-lo]) {
t.Errorf("testSync/%d: read wrong bytes: %x vs %x", i, input[lo:hi], out[:hi-lo])
return
}
// This test originally checked that after reading
// the first half of the input, there was nothing left
// in the read buffer (buf.buf.Len() != 0) but that is
// not necessarily the case: the write Flush may emit
// some extra framing bits that are not necessary
// to process to obtain the first half of the uncompressed
// data. The test ran correctly most of the time, because
// the background goroutine had usually read even
// those extra bits by now, but it's not a useful thing to
// check.
buf.WriteMode()
}
buf.ReadMode()
out := make([]byte, 10)
if n, err := r.Read(out); n > 0 || err != io.EOF {
t.Errorf("testSync (%d, %d, %s): final Read: %d, %v (hex: %x)", level, len(input), name, n, err, out[0:n])
}
if buf.buf.Len() != 0 {
t.Errorf("testSync (%d, %d, %s): extra data at end", level, len(input), name)
}
r.Close()
// stream should work for ordinary reader too
r = NewReader(buf1)
out, err = ioutil.ReadAll(r)
if err != nil {
t.Errorf("testSync: read: %s", err)
return
}
r.Close()
if !bytes.Equal(input, out) {
t.Errorf("testSync: decompress(compress(data)) != data: level=%d input=%s", level, name)
}
}
func testToFromWithLevelAndLimit(t *testing.T, level int, input []byte, name string, limit int) {
var buffer bytes.Buffer
w, err := NewWriter(&buffer, level)
if err != nil {
t.Errorf("NewWriter: %v", err)
return
}
w.Write(input)
w.Close()
if limit > 0 && buffer.Len() > limit {
t.Errorf("level: %d, len(compress(data)) = %d > limit = %d", level, buffer.Len(), limit)
return
}
if limit > 0 {
t.Logf("level: %d - Size:%.2f%%, %d b\n", level, float64(buffer.Len()*100)/float64(limit), buffer.Len())
}
r := NewReader(&buffer)
out, err := ioutil.ReadAll(r)
if err != nil {
t.Errorf("read: %s", err)
return
}
r.Close()
if !bytes.Equal(input, out) {
t.Errorf("decompress(compress(data)) != data: level=%d input=%s", level, name)
return
}
testSync(t, level, input, name)
}
func testToFromWithLimit(t *testing.T, input []byte, name string, limit [11]int) {
for i := 0; i < 10; i++ {
testToFromWithLevelAndLimit(t, i, input, name, limit[i])
}
testToFromWithLevelAndLimit(t, -2, input, name, limit[10])
}
func TestDeflateInflate(t *testing.T) {
for i, h := range deflateInflateTests {
testToFromWithLimit(t, h.in, fmt.Sprintf("#%d", i), [11]int{})
}
}
func TestReverseBits(t *testing.T) {
for _, h := range reverseBitsTests {
if v := reverseBits(h.in, h.bitCount); v != h.out {
t.Errorf("reverseBits(%v,%v) = %v, want %v",
h.in, h.bitCount, v, h.out)
}
}
}
type deflateInflateStringTest struct {
filename string
label string
limit [11]int // Number 11 is ConstantCompression
}
var deflateInflateStringTests = []deflateInflateStringTest{
{
"../testdata/e.txt",
"2.718281828...",
[...]int{100018, 67900, 50960, 51150, 50930, 50790, 50790, 50790, 50790, 50790, 43683 + 100},
},
{
"../testdata/Mark.Twain-Tom.Sawyer.txt",
"Mark.Twain-Tom.Sawyer",
[...]int{387999, 185000, 182361, 179974, 174124, 168819, 162936, 160506, 160295, 160295, 233460 + 100},
},
}
func TestDeflateInflateString(t *testing.T) {
for _, test := range deflateInflateStringTests {
gold, err := ioutil.ReadFile(test.filename)
if err != nil {
t.Error(err)
}
// Remove returns that may be present on Windows
neutral := strings.Map(func(r rune) rune {
if r != '\r' {
return r
}
return -1
}, string(gold))
testToFromWithLimit(t, []byte(neutral), test.label, test.limit)
if testing.Short() {
break
}
}
}
func TestReaderDict(t *testing.T) {
const (
dict = "hello world"
text = "hello again world"
)
var b bytes.Buffer
w, err := NewWriter(&b, 5)
if err != nil {
t.Fatalf("NewWriter: %v", err)
}
w.Write([]byte(dict))
w.Flush()
b.Reset()
w.Write([]byte(text))
w.Close()
r := NewReaderDict(&b, []byte(dict))
data, err := ioutil.ReadAll(r)
if err != nil {
t.Fatal(err)
}
if string(data) != "hello again world" {
t.Fatalf("read returned %q want %q", string(data), text)
}
}
func TestWriterDict(t *testing.T) {
const (
dict = "hello world Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua."
text = "hello world Lorem ipsum dolor sit amet"
)
// This test is sensitive to algorithm changes that skip
// data in favour of speed. Higher levels are less prone to this
// so we test level 4-9.
for l := 4; l < 9; l++ {
var b bytes.Buffer
w, err := NewWriter(&b, l)
if err != nil {
t.Fatalf("level %d, NewWriter: %v", l, err)
}
w.Write([]byte(dict))
w.Flush()
b.Reset()
w.Write([]byte(text))
w.Close()
var b1 bytes.Buffer
w, _ = NewWriterDict(&b1, l, []byte(dict))
w.Write([]byte(text))
w.Close()
if !bytes.Equal(b1.Bytes(), b.Bytes()) {
t.Errorf("level %d, writer wrote\n%v\n want\n%v", l, b1.Bytes(), b.Bytes())
}
}
}
// See http://code.google.com/p/go/issues/detail?id=2508
func TestRegression2508(t *testing.T) {
if testing.Short() {
t.Logf("test disabled with -short")
return
}
w, err := NewWriter(ioutil.Discard, 1)
if err != nil {
t.Fatalf("NewWriter: %v", err)
}
buf := make([]byte, 1024)
for i := 0; i < 131072; i++ {
if _, err := w.Write(buf); err != nil {
t.Fatalf("writer failed: %v", err)
}
}
w.Close()
}
func TestWriterReset(t *testing.T) {
for level := -2; level <= 9; level++ {
if level == -1 {
level++
}
if testing.Short() && level > 1 {
break
}
w, err := NewWriter(ioutil.Discard, level)
if err != nil {
t.Fatalf("NewWriter: %v", err)
}
buf := []byte("hello world")
for i := 0; i < 1024; i++ {
w.Write(buf)
}
w.Reset(ioutil.Discard)
wref, err := NewWriter(ioutil.Discard, level)
if err != nil {
t.Fatalf("NewWriter: %v", err)
}
// DeepEqual doesn't compare functions.
w.d.fill, wref.d.fill = nil, nil
w.d.step, wref.d.step = nil, nil
w.d.bulkHasher, wref.d.bulkHasher = nil, nil
w.d.snap, wref.d.snap = nil, nil
// hashMatch is always overwritten when used.
copy(w.d.hashMatch[:], wref.d.hashMatch[:])
if w.d.tokens.n != 0 {
t.Errorf("level %d Writer not reset after Reset. %d tokens were present", level, w.d.tokens.n)
}
// As long as the length is 0, we don't care about the content.
w.d.tokens = wref.d.tokens
// We don't care if there are values in the window, as long as it is at d.index is 0
w.d.window = wref.d.window
if !reflect.DeepEqual(w, wref) {
t.Errorf("level %d Writer not reset after Reset", level)
}
}
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriter(w, NoCompression) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriter(w, DefaultCompression) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriter(w, BestCompression) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriter(w, ConstantCompression) })
dict := []byte("we are the world")
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriterDict(w, NoCompression, dict) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriterDict(w, DefaultCompression, dict) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriterDict(w, BestCompression, dict) })
testResetOutput(t, func(w io.Writer) (*Writer, error) { return NewWriterDict(w, ConstantCompression, dict) })
}
func testResetOutput(t *testing.T, newWriter func(w io.Writer) (*Writer, error)) {
buf := new(bytes.Buffer)
w, err := newWriter(buf)
if err != nil {
t.Fatalf("NewWriter: %v", err)
}
b := []byte("hello world")
for i := 0; i < 1024; i++ {
w.Write(b)
}
w.Close()
out1 := buf.Bytes()
buf2 := new(bytes.Buffer)
w.Reset(buf2)
for i := 0; i < 1024; i++ {
w.Write(b)
}
w.Close()
out2 := buf2.Bytes()
if len(out1) != len(out2) {
t.Errorf("got %d, expected %d bytes", len(out2), len(out1))
}
if bytes.Compare(out1, out2) != 0 {
mm := 0
for i, b := range out1[:len(out2)] {
if b != out2[i] {
t.Errorf("mismatch index %d: %02x, expected %02x", i, out2[i], b)
}
mm++
if mm == 10 {
t.Fatal("Stopping")
}
}
}
t.Logf("got %d bytes", len(out1))
}
// TestBestSpeed tests that round-tripping through deflate and then inflate
// recovers the original input. The Write sizes are near the thresholds in the
// compressor.encSpeed method (0, 16, 128), as well as near maxStoreBlockSize
// (65535).
func TestBestSpeed(t *testing.T) {
abc := make([]byte, 128)
for i := range abc {
abc[i] = byte(i)
}
abcabc := bytes.Repeat(abc, 131072/len(abc))
var want []byte
testCases := [][]int{
{65536, 0},
{65536, 1},
{65536, 1, 256},
{65536, 1, 65536},
{65536, 14},
{65536, 15},
{65536, 16},
{65536, 16, 256},
{65536, 16, 65536},
{65536, 127},
{65536, 128},
{65536, 128, 256},
{65536, 128, 65536},
{65536, 129},
{65536, 65536, 256},
{65536, 65536, 65536},
}
for i, tc := range testCases {
for _, firstN := range []int{1, 65534, 65535, 65536, 65537, 131072} {
tc[0] = firstN
outer:
for _, flush := range []bool{false, true} {
buf := new(bytes.Buffer)
want = want[:0]
w, err := NewWriter(buf, BestSpeed)
if err != nil {
t.Errorf("i=%d, firstN=%d, flush=%t: NewWriter: %v", i, firstN, flush, err)
continue
}
for _, n := range tc {
want = append(want, abcabc[:n]...)
if _, err := w.Write(abcabc[:n]); err != nil {
t.Errorf("i=%d, firstN=%d, flush=%t: Write: %v", i, firstN, flush, err)
continue outer
}
if !flush {
continue
}
if err := w.Flush(); err != nil {
t.Errorf("i=%d, firstN=%d, flush=%t: Flush: %v", i, firstN, flush, err)
continue outer
}
}
if err := w.Close(); err != nil {
t.Errorf("i=%d, firstN=%d, flush=%t: Close: %v", i, firstN, flush, err)
continue
}
r := NewReader(buf)
got, err := ioutil.ReadAll(r)
if err != nil {
t.Errorf("i=%d, firstN=%d, flush=%t: ReadAll: %v", i, firstN, flush, err)
continue
}
r.Close()
if !bytes.Equal(got, want) {
t.Errorf("i=%d, firstN=%d, flush=%t: corruption during deflate-then-inflate", i, firstN, flush)
continue
}
}
}
}
}

184
vendor/github.com/klauspost/compress/flate/dict_decoder.go generated vendored Normal file
View File

@@ -0,0 +1,184 @@
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
// dictDecoder implements the LZ77 sliding dictionary as used in decompression.
// LZ77 decompresses data through sequences of two forms of commands:
//
// * Literal insertions: Runs of one or more symbols are inserted into the data
// stream as is. This is accomplished through the writeByte method for a
// single symbol, or combinations of writeSlice/writeMark for multiple symbols.
// Any valid stream must start with a literal insertion if no preset dictionary
// is used.
//
// * Backward copies: Runs of one or more symbols are copied from previously
// emitted data. Backward copies come as the tuple (dist, length) where dist
// determines how far back in the stream to copy from and length determines how
// many bytes to copy. Note that it is valid for the length to be greater than
// the distance. Since LZ77 uses forward copies, that situation is used to
// perform a form of run-length encoding on repeated runs of symbols.
// The writeCopy and tryWriteCopy are used to implement this command.
//
// For performance reasons, this implementation performs little to no sanity
// checks about the arguments. As such, the invariants documented for each
// method call must be respected.
type dictDecoder struct {
hist []byte // Sliding window history
// Invariant: 0 <= rdPos <= wrPos <= len(hist)
wrPos int // Current output position in buffer
rdPos int // Have emitted hist[:rdPos] already
full bool // Has a full window length been written yet?
}
// init initializes dictDecoder to have a sliding window dictionary of the given
// size. If a preset dict is provided, it will initialize the dictionary with
// the contents of dict.
func (dd *dictDecoder) init(size int, dict []byte) {
*dd = dictDecoder{hist: dd.hist}
if cap(dd.hist) < size {
dd.hist = make([]byte, size)
}
dd.hist = dd.hist[:size]
if len(dict) > len(dd.hist) {
dict = dict[len(dict)-len(dd.hist):]
}
dd.wrPos = copy(dd.hist, dict)
if dd.wrPos == len(dd.hist) {
dd.wrPos = 0
dd.full = true
}
dd.rdPos = dd.wrPos
}
// histSize reports the total amount of historical data in the dictionary.
func (dd *dictDecoder) histSize() int {
if dd.full {
return len(dd.hist)
}
return dd.wrPos
}
// availRead reports the number of bytes that can be flushed by readFlush.
func (dd *dictDecoder) availRead() int {
return dd.wrPos - dd.rdPos
}
// availWrite reports the available amount of output buffer space.
func (dd *dictDecoder) availWrite() int {
return len(dd.hist) - dd.wrPos
}
// writeSlice returns a slice of the available buffer to write data to.
//
// This invariant will be kept: len(s) <= availWrite()
func (dd *dictDecoder) writeSlice() []byte {
return dd.hist[dd.wrPos:]
}
// writeMark advances the writer pointer by cnt.
//
// This invariant must be kept: 0 <= cnt <= availWrite()
func (dd *dictDecoder) writeMark(cnt int) {
dd.wrPos += cnt
}
// writeByte writes a single byte to the dictionary.
//
// This invariant must be kept: 0 < availWrite()
func (dd *dictDecoder) writeByte(c byte) {
dd.hist[dd.wrPos] = c
dd.wrPos++
}
// writeCopy copies a string at a given (dist, length) to the output.
// This returns the number of bytes copied and may be less than the requested
// length if the available space in the output buffer is too small.
//
// This invariant must be kept: 0 < dist <= histSize()
func (dd *dictDecoder) writeCopy(dist, length int) int {
dstBase := dd.wrPos
dstPos := dstBase
srcPos := dstPos - dist
endPos := dstPos + length
if endPos > len(dd.hist) {
endPos = len(dd.hist)
}
// Copy non-overlapping section after destination position.
//
// This section is non-overlapping in that the copy length for this section
// is always less than or equal to the backwards distance. This can occur
// if a distance refers to data that wraps-around in the buffer.
// Thus, a backwards copy is performed here; that is, the exact bytes in
// the source prior to the copy is placed in the destination.
if srcPos < 0 {
srcPos += len(dd.hist)
dstPos += copy(dd.hist[dstPos:endPos], dd.hist[srcPos:])
srcPos = 0
}
// Copy possibly overlapping section before destination position.
//
// This section can overlap if the copy length for this section is larger
// than the backwards distance. This is allowed by LZ77 so that repeated
// strings can be succinctly represented using (dist, length) pairs.
// Thus, a forwards copy is performed here; that is, the bytes copied is
// possibly dependent on the resulting bytes in the destination as the copy
// progresses along. This is functionally equivalent to the following:
//
// for i := 0; i < endPos-dstPos; i++ {
// dd.hist[dstPos+i] = dd.hist[srcPos+i]
// }
// dstPos = endPos
//
for dstPos < endPos {
dstPos += copy(dd.hist[dstPos:endPos], dd.hist[srcPos:dstPos])
}
dd.wrPos = dstPos
return dstPos - dstBase
}
// tryWriteCopy tries to copy a string at a given (distance, length) to the
// output. This specialized version is optimized for short distances.
//
// This method is designed to be inlined for performance reasons.
//
// This invariant must be kept: 0 < dist <= histSize()
func (dd *dictDecoder) tryWriteCopy(dist, length int) int {
dstPos := dd.wrPos
endPos := dstPos + length
if dstPos < dist || endPos > len(dd.hist) {
return 0
}
dstBase := dstPos
srcPos := dstPos - dist
// Copy possibly overlapping section before destination position.
loop:
dstPos += copy(dd.hist[dstPos:endPos], dd.hist[srcPos:dstPos])
if dstPos < endPos {
goto loop // Avoid for-loop so that this function can be inlined
}
dd.wrPos = dstPos
return dstPos - dstBase
}
// readFlush returns a slice of the historical buffer that is ready to be
// emitted to the user. The data returned by readFlush must be fully consumed
// before calling any other dictDecoder methods.
func (dd *dictDecoder) readFlush() []byte {
toRead := dd.hist[dd.rdPos:dd.wrPos]
dd.rdPos = dd.wrPos
if dd.wrPos == len(dd.hist) {
dd.wrPos, dd.rdPos = 0, 0
dd.full = true
}
return toRead
}

139
vendor/github.com/klauspost/compress/flate/dict_decoder_test.go generated vendored Normal file
View File

@@ -0,0 +1,139 @@
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"bytes"
"strings"
"testing"
)
func TestDictDecoder(t *testing.T) {
const (
abc = "ABC\n"
fox = "The quick brown fox jumped over the lazy dog!\n"
poem = "The Road Not Taken\nRobert Frost\n" +
"\n" +
"Two roads diverged in a yellow wood,\n" +
"And sorry I could not travel both\n" +
"And be one traveler, long I stood\n" +
"And looked down one as far as I could\n" +
"To where it bent in the undergrowth;\n" +
"\n" +
"Then took the other, as just as fair,\n" +
"And having perhaps the better claim,\n" +
"Because it was grassy and wanted wear;\n" +
"Though as for that the passing there\n" +
"Had worn them really about the same,\n" +
"\n" +
"And both that morning equally lay\n" +
"In leaves no step had trodden black.\n" +
"Oh, I kept the first for another day!\n" +
"Yet knowing how way leads on to way,\n" +
"I doubted if I should ever come back.\n" +
"\n" +
"I shall be telling this with a sigh\n" +
"Somewhere ages and ages hence:\n" +
"Two roads diverged in a wood, and I-\n" +
"I took the one less traveled by,\n" +
"And that has made all the difference.\n"
)
var poemRefs = []struct {
dist int // Backward distance (0 if this is an insertion)
length int // Length of copy or insertion
}{
{0, 38}, {33, 3}, {0, 48}, {79, 3}, {0, 11}, {34, 5}, {0, 6}, {23, 7},
{0, 8}, {50, 3}, {0, 2}, {69, 3}, {34, 5}, {0, 4}, {97, 3}, {0, 4},
{43, 5}, {0, 6}, {7, 4}, {88, 7}, {0, 12}, {80, 3}, {0, 2}, {141, 4},
{0, 1}, {196, 3}, {0, 3}, {157, 3}, {0, 6}, {181, 3}, {0, 2}, {23, 3},
{77, 3}, {28, 5}, {128, 3}, {110, 4}, {70, 3}, {0, 4}, {85, 6}, {0, 2},
{182, 6}, {0, 4}, {133, 3}, {0, 7}, {47, 5}, {0, 20}, {112, 5}, {0, 1},
{58, 3}, {0, 8}, {59, 3}, {0, 4}, {173, 3}, {0, 5}, {114, 3}, {0, 4},
{92, 5}, {0, 2}, {71, 3}, {0, 2}, {76, 5}, {0, 1}, {46, 3}, {96, 4},
{130, 4}, {0, 3}, {360, 3}, {0, 3}, {178, 5}, {0, 7}, {75, 3}, {0, 3},
{45, 6}, {0, 6}, {299, 6}, {180, 3}, {70, 6}, {0, 1}, {48, 3}, {66, 4},
{0, 3}, {47, 5}, {0, 9}, {325, 3}, {0, 1}, {359, 3}, {318, 3}, {0, 2},
{199, 3}, {0, 1}, {344, 3}, {0, 3}, {248, 3}, {0, 10}, {310, 3}, {0, 3},
{93, 6}, {0, 3}, {252, 3}, {157, 4}, {0, 2}, {273, 5}, {0, 14}, {99, 4},
{0, 1}, {464, 4}, {0, 2}, {92, 4}, {495, 3}, {0, 1}, {322, 4}, {16, 4},
{0, 3}, {402, 3}, {0, 2}, {237, 4}, {0, 2}, {432, 4}, {0, 1}, {483, 5},
{0, 2}, {294, 4}, {0, 2}, {306, 3}, {113, 5}, {0, 1}, {26, 4}, {164, 3},
{488, 4}, {0, 1}, {542, 3}, {248, 6}, {0, 5}, {205, 3}, {0, 8}, {48, 3},
{449, 6}, {0, 2}, {192, 3}, {328, 4}, {9, 5}, {433, 3}, {0, 3}, {622, 25},
{615, 5}, {46, 5}, {0, 2}, {104, 3}, {475, 10}, {549, 3}, {0, 4}, {597, 8},
{314, 3}, {0, 1}, {473, 6}, {317, 5}, {0, 1}, {400, 3}, {0, 3}, {109, 3},
{151, 3}, {48, 4}, {0, 4}, {125, 3}, {108, 3}, {0, 2},
}
var got, want bytes.Buffer
var dd dictDecoder
dd.init(1<<11, nil)
var writeCopy = func(dist, length int) {
for length > 0 {
cnt := dd.tryWriteCopy(dist, length)
if cnt == 0 {
cnt = dd.writeCopy(dist, length)
}
length -= cnt
if dd.availWrite() == 0 {
got.Write(dd.readFlush())
}
}
}
var writeString = func(str string) {
for len(str) > 0 {
cnt := copy(dd.writeSlice(), str)
str = str[cnt:]
dd.writeMark(cnt)
if dd.availWrite() == 0 {
got.Write(dd.readFlush())
}
}
}
writeString(".")
want.WriteByte('.')
str := poem
for _, ref := range poemRefs {
if ref.dist == 0 {
writeString(str[:ref.length])
} else {
writeCopy(ref.dist, ref.length)
}
str = str[ref.length:]
}
want.WriteString(poem)
writeCopy(dd.histSize(), 33)
want.Write(want.Bytes()[:33])
writeString(abc)
writeCopy(len(abc), 59*len(abc))
want.WriteString(strings.Repeat(abc, 60))
writeString(fox)
writeCopy(len(fox), 9*len(fox))
want.WriteString(strings.Repeat(fox, 10))
writeString(".")
writeCopy(1, 9)
want.WriteString(strings.Repeat(".", 10))
writeString(strings.ToUpper(poem))
writeCopy(len(poem), 7*len(poem))
want.WriteString(strings.Repeat(strings.ToUpper(poem), 8))
writeCopy(dd.histSize(), 10)
want.Write(want.Bytes()[want.Len()-dd.histSize():][:10])
got.Write(dd.readFlush())
if got.String() != want.String() {
t.Errorf("final string mismatch:\ngot %q\nwant %q", got.String(), want.String())
}
}

260
vendor/github.com/klauspost/compress/flate/flate_test.go generated vendored Normal file
View File

@@ -0,0 +1,260 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// This test tests some internals of the flate package.
// The tests in package compress/gzip serve as the
// end-to-end test of the decompressor.
package flate
import (
"bytes"
"encoding/hex"
"io/ioutil"
"testing"
)
// The following test should not panic.
func TestIssue5915(t *testing.T) {
bits := []int{4, 0, 0, 6, 4, 3, 2, 3, 3, 4, 4, 5, 0, 0, 0, 0, 5, 5, 6,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 8, 6, 0, 11, 0, 8, 0, 6, 6, 10, 8}
var h huffmanDecoder
if h.init(bits) {
t.Fatalf("Given sequence of bits is bad, and should not succeed.")
}
}
// The following test should not panic.
func TestIssue5962(t *testing.T) {
bits := []int{4, 0, 0, 6, 4, 3, 2, 3, 3, 4, 4, 5, 0, 0, 0, 0,
5, 5, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 11}
var h huffmanDecoder
if h.init(bits) {
t.Fatalf("Given sequence of bits is bad, and should not succeed.")
}
}
// The following test should not panic.
func TestIssue6255(t *testing.T) {
bits1 := []int{1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 11}
bits2 := []int{11, 13}
var h huffmanDecoder
if !h.init(bits1) {
t.Fatalf("Given sequence of bits is good and should succeed.")
}
if h.init(bits2) {
t.Fatalf("Given sequence of bits is bad and should not succeed.")
}
}
func TestInvalidEncoding(t *testing.T) {
// Initialize Huffman decoder to recognize "0".
var h huffmanDecoder
if !h.init([]int{1}) {
t.Fatal("Failed to initialize Huffman decoder")
}
// Initialize decompressor with invalid Huffman coding.
var f decompressor
f.r = bytes.NewReader([]byte{0xff})
_, err := f.huffSym(&h)
if err == nil {
t.Fatal("Should have rejected invalid bit sequence")
}
}
func TestInvalidBits(t *testing.T) {
oversubscribed := []int{1, 2, 3, 4, 4, 5}
incomplete := []int{1, 2, 4, 4}
var h huffmanDecoder
if h.init(oversubscribed) {
t.Fatal("Should reject oversubscribed bit-length set")
}
if h.init(incomplete) {
t.Fatal("Should reject incomplete bit-length set")
}
}
func TestStreams(t *testing.T) {
// To verify any of these hexstrings as valid or invalid flate streams
// according to the C zlib library, you can use the Python wrapper library:
// >>> hex_string = "010100feff11"
// >>> import zlib
// >>> zlib.decompress(hex_string.decode("hex"), -15) # Negative means raw DEFLATE
// '\x11'
testCases := []struct {
desc string // Description of the stream
stream string // Hexstring of the input DEFLATE stream
want string // Expected result. Use "fail" to expect failure
}{{
"degenerate HCLenTree",
"05e0010000000000100000000000000000000000000000000000000000000000" +
"00000000000000000004",
"fail",
}, {
"complete HCLenTree, empty HLitTree, empty HDistTree",
"05e0010400000000000000000000000000000000000000000000000000000000" +
"00000000000000000010",
"fail",
}, {
"empty HCLenTree",
"05e0010000000000000000000000000000000000000000000000000000000000" +
"00000000000000000010",
"fail",
}, {
"complete HCLenTree, complete HLitTree, empty HDistTree, use missing HDist symbol",
"000100feff000de0010400000000100000000000000000000000000000000000" +
"0000000000000000000000000000002c",
"fail",
}, {
"complete HCLenTree, complete HLitTree, degenerate HDistTree, use missing HDist symbol",
"000100feff000de0010000000000000000000000000000000000000000000000" +
"00000000000000000610000000004070",
"fail",
}, {
"complete HCLenTree, empty HLitTree, empty HDistTree",
"05e0010400000000100400000000000000000000000000000000000000000000" +
"0000000000000000000000000008",
"fail",
}, {
"complete HCLenTree, empty HLitTree, degenerate HDistTree",
"05e0010400000000100400000000000000000000000000000000000000000000" +
"0000000000000000000800000008",
"fail",
}, {
"complete HCLenTree, degenerate HLitTree, degenerate HDistTree, use missing HLit symbol",
"05e0010400000000100000000000000000000000000000000000000000000000" +
"0000000000000000001c",
"fail",
}, {
"complete HCLenTree, complete HLitTree, too large HDistTree",
"edff870500000000200400000000000000000000000000000000000000000000" +
"000000000000000000080000000000000004",
"fail",
}, {
"complete HCLenTree, complete HLitTree, empty HDistTree, excessive repeater code",
"edfd870500000000200400000000000000000000000000000000000000000000" +
"000000000000000000e8b100",
"fail",
}, {
"complete HCLenTree, complete HLitTree, empty HDistTree of normal length 30",
"05fd01240000000000f8ffffffffffffffffffffffffffffffffffffffffffff" +
"ffffffffffffffffff07000000fe01",
"",
}, {
"complete HCLenTree, complete HLitTree, empty HDistTree of excessive length 31",
"05fe01240000000000f8ffffffffffffffffffffffffffffffffffffffffffff" +
"ffffffffffffffffff07000000fc03",
"fail",
}, {
"complete HCLenTree, over-subscribed HLitTree, empty HDistTree",
"05e001240000000000fcffffffffffffffffffffffffffffffffffffffffffff" +
"ffffffffffffffffff07f00f",
"fail",
}, {
"complete HCLenTree, under-subscribed HLitTree, empty HDistTree",
"05e001240000000000fcffffffffffffffffffffffffffffffffffffffffffff" +
"fffffffffcffffffff07f00f",
"fail",
}, {
"complete HCLenTree, complete HLitTree with single code, empty HDistTree",
"05e001240000000000f8ffffffffffffffffffffffffffffffffffffffffffff" +
"ffffffffffffffffff07f00f",
"01",
}, {
"complete HCLenTree, complete HLitTree with multiple codes, empty HDistTree",
"05e301240000000000f8ffffffffffffffffffffffffffffffffffffffffffff" +
"ffffffffffffffffff07807f",
"01",
}, {
"complete HCLenTree, complete HLitTree, degenerate HDistTree, use valid HDist symbol",
"000100feff000de0010400000000100000000000000000000000000000000000" +
"0000000000000000000000000000003c",
"00000000",
}, {
"complete HCLenTree, degenerate HLitTree, degenerate HDistTree",
"05e0010400000000100000000000000000000000000000000000000000000000" +
"0000000000000000000c",
"",
}, {
"complete HCLenTree, degenerate HLitTree, empty HDistTree",
"05e0010400000000100000000000000000000000000000000000000000000000" +
"00000000000000000004",
"",
}, {
"complete HCLenTree, complete HLitTree, empty HDistTree, spanning repeater code",
"edfd870500000000200400000000000000000000000000000000000000000000" +
"000000000000000000e8b000",
"",
}, {
"complete HCLenTree with length codes, complete HLitTree, empty HDistTree",
"ede0010400000000100000000000000000000000000000000000000000000000" +
"0000000000000000000400004000",
"",
}, {
"complete HCLenTree, complete HLitTree, degenerate HDistTree, use valid HLit symbol 284 with count 31",
"000100feff00ede0010400000000100000000000000000000000000000000000" +
"000000000000000000000000000000040000407f00",
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"0000000000000000000000000000000000000000000000000000000000000000" +
"000000",
}, {
"complete HCLenTree, complete HLitTree, degenerate HDistTree, use valid HLit and HDist symbols",
"0cc2010d00000082b0ac4aff0eb07d27060000ffff",
"616263616263",
}, {
"fixed block, use reserved symbol 287",
"33180700",
"fail",
}, {
"raw block",
"010100feff11",
"11",
}, {
"issue 10426 - over-subscribed HCLenTree causes a hang",
"344c4a4e494d4b070000ff2e2eff2e2e2e2e2eff",
"fail",
}, {
"issue 11030 - empty HDistTree unexpectedly leads to error",
"05c0070600000080400fff37a0ca",
"",
}, {
"issue 11033 - empty HDistTree unexpectedly leads to error",
"050fb109c020cca5d017dcbca044881ee1034ec149c8980bbc413c2ab35be9dc" +
"b1473449922449922411202306ee97b0383a521b4ffdcf3217f9f7d3adb701",
"3130303634342068652e706870005d05355f7ed957ff084a90925d19e3ebc6d0" +
"c6d7",
}}
for i, tc := range testCases {
data, err := hex.DecodeString(tc.stream)
if err != nil {
t.Fatal(err)
}
data, err = ioutil.ReadAll(NewReader(bytes.NewReader(data)))
if tc.want == "fail" {
if err == nil {
t.Errorf("#%d (%s): got nil error, want non-nil", i, tc.desc)
}
} else {
if err != nil {
t.Errorf("#%d (%s): %v", i, tc.desc, err)
continue
}
if got := hex.EncodeToString(data); got != tc.want {
t.Errorf("#%d (%s):\ngot %q\nwant %q", i, tc.desc, got, tc.want)
}
}
}
}

265
vendor/github.com/klauspost/compress/flate/gen.go generated vendored Normal file
View File

@@ -0,0 +1,265 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build ignore
// This program generates fixedhuff.go
// Invoke as
//
// go run gen.go -output fixedhuff.go
package main
import (
"bytes"
"flag"
"fmt"
"go/format"
"io/ioutil"
"log"
)
var filename = flag.String("output", "fixedhuff.go", "output file name")
const maxCodeLen = 16
// Note: the definition of the huffmanDecoder struct is copied from
// inflate.go, as it is private to the implementation.
// chunk & 15 is number of bits
// chunk >> 4 is value, including table link
const (
huffmanChunkBits = 9
huffmanNumChunks = 1 << huffmanChunkBits
huffmanCountMask = 15
huffmanValueShift = 4
)
type huffmanDecoder struct {
min int // the minimum code length
chunks [huffmanNumChunks]uint32 // chunks as described above
links [][]uint32 // overflow links
linkMask uint32 // mask the width of the link table
}
// Initialize Huffman decoding tables from array of code lengths.
// Following this function, h is guaranteed to be initialized into a complete
// tree (i.e., neither over-subscribed nor under-subscribed). The exception is a
// degenerate case where the tree has only a single symbol with length 1. Empty
// trees are permitted.
func (h *huffmanDecoder) init(bits []int) bool {
// Sanity enables additional runtime tests during Huffman
// table construction. It's intended to be used during
// development to supplement the currently ad-hoc unit tests.
const sanity = false
if h.min != 0 {
*h = huffmanDecoder{}
}
// Count number of codes of each length,
// compute min and max length.
var count [maxCodeLen]int
var min, max int
for _, n := range bits {
if n == 0 {
continue
}
if min == 0 || n < min {
min = n
}
if n > max {
max = n
}
count[n]++
}
// Empty tree. The decompressor.huffSym function will fail later if the tree
// is used. Technically, an empty tree is only valid for the HDIST tree and
// not the HCLEN and HLIT tree. However, a stream with an empty HCLEN tree
// is guaranteed to fail since it will attempt to use the tree to decode the
// codes for the HLIT and HDIST trees. Similarly, an empty HLIT tree is
// guaranteed to fail later since the compressed data section must be
// composed of at least one symbol (the end-of-block marker).
if max == 0 {
return true
}
code := 0
var nextcode [maxCodeLen]int
for i := min; i <= max; i++ {
code <<= 1
nextcode[i] = code
code += count[i]
}
// Check that the coding is complete (i.e., that we've
// assigned all 2-to-the-max possible bit sequences).
// Exception: To be compatible with zlib, we also need to
// accept degenerate single-code codings. See also
// TestDegenerateHuffmanCoding.
if code != 1<<uint(max) && !(code == 1 && max == 1) {
return false
}
h.min = min
if max > huffmanChunkBits {
numLinks := 1 << (uint(max) - huffmanChunkBits)
h.linkMask = uint32(numLinks - 1)
// create link tables
link := nextcode[huffmanChunkBits+1] >> 1
h.links = make([][]uint32, huffmanNumChunks-link)
for j := uint(link); j < huffmanNumChunks; j++ {
reverse := int(reverseByte[j>>8]) | int(reverseByte[j&0xff])<<8
reverse >>= uint(16 - huffmanChunkBits)
off := j - uint(link)
if sanity && h.chunks[reverse] != 0 {
panic("impossible: overwriting existing chunk")
}
h.chunks[reverse] = uint32(off<<huffmanValueShift | (huffmanChunkBits + 1))
h.links[off] = make([]uint32, numLinks)
}
}
for i, n := range bits {
if n == 0 {
continue
}
code := nextcode[n]
nextcode[n]++
chunk := uint32(i<<huffmanValueShift | n)
reverse := int(reverseByte[code>>8]) | int(reverseByte[code&0xff])<<8
reverse >>= uint(16 - n)
if n <= huffmanChunkBits {
for off := reverse; off < len(h.chunks); off += 1 << uint(n) {
// We should never need to overwrite
// an existing chunk. Also, 0 is
// never a valid chunk, because the
// lower 4 "count" bits should be
// between 1 and 15.
if sanity && h.chunks[off] != 0 {
panic("impossible: overwriting existing chunk")
}
h.chunks[off] = chunk
}
} else {
j := reverse & (huffmanNumChunks - 1)
if sanity && h.chunks[j]&huffmanCountMask != huffmanChunkBits+1 {
// Longer codes should have been
// associated with a link table above.
panic("impossible: not an indirect chunk")
}
value := h.chunks[j] >> huffmanValueShift
linktab := h.links[value]
reverse >>= huffmanChunkBits
for off := reverse; off < len(linktab); off += 1 << uint(n-huffmanChunkBits) {
if sanity && linktab[off] != 0 {
panic("impossible: overwriting existing chunk")
}
linktab[off] = chunk
}
}
}
if sanity {
// Above we've sanity checked that we never overwrote
// an existing entry. Here we additionally check that
// we filled the tables completely.
for i, chunk := range h.chunks {
if chunk == 0 {
// As an exception, in the degenerate
// single-code case, we allow odd
// chunks to be missing.
if code == 1 && i%2 == 1 {
continue
}
panic("impossible: missing chunk")
}
}
for _, linktab := range h.links {
for _, chunk := range linktab {
if chunk == 0 {
panic("impossible: missing chunk")
}
}
}
}
return true
}
func main() {
flag.Parse()
var h huffmanDecoder
var bits [288]int
initReverseByte()
for i := 0; i < 144; i++ {
bits[i] = 8
}
for i := 144; i < 256; i++ {
bits[i] = 9
}
for i := 256; i < 280; i++ {
bits[i] = 7
}
for i := 280; i < 288; i++ {
bits[i] = 8
}
h.init(bits[:])
if h.links != nil {
log.Fatal("Unexpected links table in fixed Huffman decoder")
}
var buf bytes.Buffer
fmt.Fprintf(&buf, `// Copyright 2013 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.`+"\n\n")
fmt.Fprintln(&buf, "package flate")
fmt.Fprintln(&buf)
fmt.Fprintln(&buf, "// autogenerated by go run gen.go -output fixedhuff.go, DO NOT EDIT")
fmt.Fprintln(&buf)
fmt.Fprintln(&buf, "var fixedHuffmanDecoder = huffmanDecoder{")
fmt.Fprintf(&buf, "\t%d,\n", h.min)
fmt.Fprintln(&buf, "\t[huffmanNumChunks]uint32{")
for i := 0; i < huffmanNumChunks; i++ {
if i&7 == 0 {
fmt.Fprintf(&buf, "\t\t")
} else {
fmt.Fprintf(&buf, " ")
}
fmt.Fprintf(&buf, "0x%04x,", h.chunks[i])
if i&7 == 7 {
fmt.Fprintln(&buf)
}
}
fmt.Fprintln(&buf, "\t},")
fmt.Fprintln(&buf, "\tnil, 0,")
fmt.Fprintln(&buf, "}")
data, err := format.Source(buf.Bytes())
if err != nil {
log.Fatal(err)
}
err = ioutil.WriteFile(*filename, data, 0644)
if err != nil {
log.Fatal(err)
}
}
var reverseByte [256]byte
func initReverseByte() {
for x := 0; x < 256; x++ {
var result byte
for i := uint(0); i < 8; i++ {
result |= byte(((x >> i) & 1) << (7 - i))
}
reverseByte[x] = result
}
}

701
vendor/github.com/klauspost/compress/flate/huffman_bit_writer.go generated vendored Normal file
View File

@@ -0,0 +1,701 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"io"
)
const (
// The largest offset code.
offsetCodeCount = 30
// The special code used to mark the end of a block.
endBlockMarker = 256
// The first length code.
lengthCodesStart = 257
// The number of codegen codes.
codegenCodeCount = 19
badCode = 255
// bufferFlushSize indicates the buffer size
// after which bytes are flushed to the writer.
// Should preferably be a multiple of 6, since
// we accumulate 6 bytes between writes to the buffer.
bufferFlushSize = 240
// bufferSize is the actual output byte buffer size.
// It must have additional headroom for a flush
// which can contain up to 8 bytes.
bufferSize = bufferFlushSize + 8
)
// The number of extra bits needed by length code X - LENGTH_CODES_START.
var lengthExtraBits = []int8{
/* 257 */ 0, 0, 0,
/* 260 */ 0, 0, 0, 0, 0, 1, 1, 1, 1, 2,
/* 270 */ 2, 2, 2, 3, 3, 3, 3, 4, 4, 4,
/* 280 */ 4, 5, 5, 5, 5, 0,
}
// The length indicated by length code X - LENGTH_CODES_START.
var lengthBase = []uint32{
0, 1, 2, 3, 4, 5, 6, 7, 8, 10,
12, 14, 16, 20, 24, 28, 32, 40, 48, 56,
64, 80, 96, 112, 128, 160, 192, 224, 255,
}
// offset code word extra bits.
var offsetExtraBits = []int8{
0, 0, 0, 0, 1, 1, 2, 2, 3, 3,
4, 4, 5, 5, 6, 6, 7, 7, 8, 8,
9, 9, 10, 10, 11, 11, 12, 12, 13, 13,
/* extended window */
14, 14, 15, 15, 16, 16, 17, 17, 18, 18, 19, 19, 20, 20,
}
var offsetBase = []uint32{
/* normal deflate */
0x000000, 0x000001, 0x000002, 0x000003, 0x000004,
0x000006, 0x000008, 0x00000c, 0x000010, 0x000018,
0x000020, 0x000030, 0x000040, 0x000060, 0x000080,
0x0000c0, 0x000100, 0x000180, 0x000200, 0x000300,
0x000400, 0x000600, 0x000800, 0x000c00, 0x001000,
0x001800, 0x002000, 0x003000, 0x004000, 0x006000,
/* extended window */
0x008000, 0x00c000, 0x010000, 0x018000, 0x020000,
0x030000, 0x040000, 0x060000, 0x080000, 0x0c0000,
0x100000, 0x180000, 0x200000, 0x300000,
}
// The odd order in which the codegen code sizes are written.
var codegenOrder = []uint32{16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15}
type huffmanBitWriter struct {
// writer is the underlying writer.
// Do not use it directly; use the write method, which ensures
// that Write errors are sticky.
writer io.Writer
// Data waiting to be written is bytes[0:nbytes]
// and then the low nbits of bits.
bits uint64
nbits uint
bytes [bufferSize]byte
codegenFreq [codegenCodeCount]int32
nbytes int
literalFreq []int32
offsetFreq []int32
codegen []uint8
literalEncoding *huffmanEncoder
offsetEncoding *huffmanEncoder
codegenEncoding *huffmanEncoder
err error
}
func newHuffmanBitWriter(w io.Writer) *huffmanBitWriter {
return &huffmanBitWriter{
writer: w,
literalFreq: make([]int32, maxNumLit),
offsetFreq: make([]int32, offsetCodeCount),
codegen: make([]uint8, maxNumLit+offsetCodeCount+1),
literalEncoding: newHuffmanEncoder(maxNumLit),
codegenEncoding: newHuffmanEncoder(codegenCodeCount),
offsetEncoding: newHuffmanEncoder(offsetCodeCount),
}
}
func (w *huffmanBitWriter) reset(writer io.Writer) {
w.writer = writer
w.bits, w.nbits, w.nbytes, w.err = 0, 0, 0, nil
w.bytes = [bufferSize]byte{}
}
func (w *huffmanBitWriter) flush() {
if w.err != nil {
w.nbits = 0
return
}
n := w.nbytes
for w.nbits != 0 {
w.bytes[n] = byte(w.bits)
w.bits >>= 8
if w.nbits > 8 { // Avoid underflow
w.nbits -= 8
} else {
w.nbits = 0
}
n++
}
w.bits = 0
w.write(w.bytes[:n])
w.nbytes = 0
}
func (w *huffmanBitWriter) write(b []byte) {
if w.err != nil {
return
}
_, w.err = w.writer.Write(b)
}
func (w *huffmanBitWriter) writeBits(b int32, nb uint) {
if w.err != nil {
return
}
w.bits |= uint64(b) << w.nbits
w.nbits += nb
if w.nbits >= 48 {
bits := w.bits
w.bits >>= 48
w.nbits -= 48
n := w.nbytes
bytes := w.bytes[n : n+6]
bytes[0] = byte(bits)
bytes[1] = byte(bits >> 8)
bytes[2] = byte(bits >> 16)
bytes[3] = byte(bits >> 24)
bytes[4] = byte(bits >> 32)
bytes[5] = byte(bits >> 40)
n += 6
if n >= bufferFlushSize {
w.write(w.bytes[:n])
n = 0
}
w.nbytes = n
}
}
func (w *huffmanBitWriter) writeBytes(bytes []byte) {
if w.err != nil {
return
}
n := w.nbytes
if w.nbits&7 != 0 {
w.err = InternalError("writeBytes with unfinished bits")
return
}
for w.nbits != 0 {
w.bytes[n] = byte(w.bits)
w.bits >>= 8
w.nbits -= 8
n++
}
if n != 0 {
w.write(w.bytes[:n])
}
w.nbytes = 0
w.write(bytes)
}
// RFC 1951 3.2.7 specifies a special run-length encoding for specifying
// the literal and offset lengths arrays (which are concatenated into a single
// array). This method generates that run-length encoding.
//
// The result is written into the codegen array, and the frequencies
// of each code is written into the codegenFreq array.
// Codes 0-15 are single byte codes. Codes 16-18 are followed by additional
// information. Code badCode is an end marker
//
// numLiterals The number of literals in literalEncoding
// numOffsets The number of offsets in offsetEncoding
// litenc, offenc The literal and offset encoder to use
func (w *huffmanBitWriter) generateCodegen(numLiterals int, numOffsets int, litEnc, offEnc *huffmanEncoder) {
for i := range w.codegenFreq {
w.codegenFreq[i] = 0
}
// Note that we are using codegen both as a temporary variable for holding
// a copy of the frequencies, and as the place where we put the result.
// This is fine because the output is always shorter than the input used
// so far.
codegen := w.codegen // cache
// Copy the concatenated code sizes to codegen. Put a marker at the end.
cgnl := codegen[:numLiterals]
for i := range cgnl {
cgnl[i] = uint8(litEnc.codes[i].len)
}
cgnl = codegen[numLiterals : numLiterals+numOffsets]
for i := range cgnl {
cgnl[i] = uint8(offEnc.codes[i].len)
}
codegen[numLiterals+numOffsets] = badCode
size := codegen[0]
count := 1
outIndex := 0
for inIndex := 1; size != badCode; inIndex++ {
// INVARIANT: We have seen "count" copies of size that have not yet
// had output generated for them.
nextSize := codegen[inIndex]
if nextSize == size {
count++
continue
}
// We need to generate codegen indicating "count" of size.
if size != 0 {
codegen[outIndex] = size
outIndex++
w.codegenFreq[size]++
count--
for count >= 3 {
n := 6
if n > count {
n = count
}
codegen[outIndex] = 16
outIndex++
codegen[outIndex] = uint8(n - 3)
outIndex++
w.codegenFreq[16]++
count -= n
}
} else {
for count >= 11 {
n := 138
if n > count {
n = count
}
codegen[outIndex] = 18
outIndex++
codegen[outIndex] = uint8(n - 11)
outIndex++
w.codegenFreq[18]++
count -= n
}
if count >= 3 {
// count >= 3 && count <= 10
codegen[outIndex] = 17
outIndex++
codegen[outIndex] = uint8(count - 3)
outIndex++
w.codegenFreq[17]++
count = 0
}
}
count--
for ; count >= 0; count-- {
codegen[outIndex] = size
outIndex++
w.codegenFreq[size]++
}
// Set up invariant for next time through the loop.
size = nextSize
count = 1
}
// Marker indicating the end of the codegen.
codegen[outIndex] = badCode
}
// dynamicSize returns the size of dynamically encoded data in bits.
func (w *huffmanBitWriter) dynamicSize(litEnc, offEnc *huffmanEncoder, extraBits int) (size, numCodegens int) {
numCodegens = len(w.codegenFreq)
for numCodegens > 4 && w.codegenFreq[codegenOrder[numCodegens-1]] == 0 {
numCodegens--
}
header := 3 + 5 + 5 + 4 + (3 * numCodegens) +
w.codegenEncoding.bitLength(w.codegenFreq[:]) +
int(w.codegenFreq[16])*2 +
int(w.codegenFreq[17])*3 +
int(w.codegenFreq[18])*7
size = header +
litEnc.bitLength(w.literalFreq) +
offEnc.bitLength(w.offsetFreq) +
extraBits
return size, numCodegens
}
// fixedSize returns the size of dynamically encoded data in bits.
func (w *huffmanBitWriter) fixedSize(extraBits int) int {
return 3 +
fixedLiteralEncoding.bitLength(w.literalFreq) +
fixedOffsetEncoding.bitLength(w.offsetFreq) +
extraBits
}
// storedSize calculates the stored size, including header.
// The function returns the size in bits and whether the block
// fits inside a single block.
func (w *huffmanBitWriter) storedSize(in []byte) (int, bool) {
if in == nil {
return 0, false
}
if len(in) <= maxStoreBlockSize {
return (len(in) + 5) * 8, true
}
return 0, false
}
func (w *huffmanBitWriter) writeCode(c hcode) {
if w.err != nil {
return
}
w.bits |= uint64(c.code) << w.nbits
w.nbits += uint(c.len)
if w.nbits >= 48 {
bits := w.bits
w.bits >>= 48
w.nbits -= 48
n := w.nbytes
bytes := w.bytes[n : n+6]
bytes[0] = byte(bits)
bytes[1] = byte(bits >> 8)
bytes[2] = byte(bits >> 16)
bytes[3] = byte(bits >> 24)
bytes[4] = byte(bits >> 32)
bytes[5] = byte(bits >> 40)
n += 6
if n >= bufferFlushSize {
w.write(w.bytes[:n])
n = 0
}
w.nbytes = n
}
}
// Write the header of a dynamic Huffman block to the output stream.
//
// numLiterals The number of literals specified in codegen
// numOffsets The number of offsets specified in codegen
// numCodegens The number of codegens used in codegen
func (w *huffmanBitWriter) writeDynamicHeader(numLiterals int, numOffsets int, numCodegens int, isEof bool) {
if w.err != nil {
return
}
var firstBits int32 = 4
if isEof {
firstBits = 5
}
w.writeBits(firstBits, 3)
w.writeBits(int32(numLiterals-257), 5)
w.writeBits(int32(numOffsets-1), 5)
w.writeBits(int32(numCodegens-4), 4)
for i := 0; i < numCodegens; i++ {
value := uint(w.codegenEncoding.codes[codegenOrder[i]].len)
w.writeBits(int32(value), 3)
}
i := 0
for {
var codeWord int = int(w.codegen[i])
i++
if codeWord == badCode {
break
}
w.writeCode(w.codegenEncoding.codes[uint32(codeWord)])
switch codeWord {
case 16:
w.writeBits(int32(w.codegen[i]), 2)
i++
break
case 17:
w.writeBits(int32(w.codegen[i]), 3)
i++
break
case 18:
w.writeBits(int32(w.codegen[i]), 7)
i++
break
}
}
}
func (w *huffmanBitWriter) writeStoredHeader(length int, isEof bool) {
if w.err != nil {
return
}
var flag int32
if isEof {
flag = 1
}
w.writeBits(flag, 3)
w.flush()
w.writeBits(int32(length), 16)
w.writeBits(int32(^uint16(length)), 16)
}
func (w *huffmanBitWriter) writeFixedHeader(isEof bool) {
if w.err != nil {
return
}
// Indicate that we are a fixed Huffman block
var value int32 = 2
if isEof {
value = 3
}
w.writeBits(value, 3)
}
// writeBlock will write a block of tokens with the smallest encoding.
// The original input can be supplied, and if the huffman encoded data
// is larger than the original bytes, the data will be written as a
// stored block.
// If the input is nil, the tokens will always be Huffman encoded.
func (w *huffmanBitWriter) writeBlock(tokens []token, eof bool, input []byte) {
if w.err != nil {
return
}
tokens = append(tokens, endBlockMarker)
numLiterals, numOffsets := w.indexTokens(tokens)
var extraBits int
storedSize, storable := w.storedSize(input)
if storable {
// We only bother calculating the costs of the extra bits required by
// the length of offset fields (which will be the same for both fixed
// and dynamic encoding), if we need to compare those two encodings
// against stored encoding.
for lengthCode := lengthCodesStart + 8; lengthCode < numLiterals; lengthCode++ {
// First eight length codes have extra size = 0.
extraBits += int(w.literalFreq[lengthCode]) * int(lengthExtraBits[lengthCode-lengthCodesStart])
}
for offsetCode := 4; offsetCode < numOffsets; offsetCode++ {
// First four offset codes have extra size = 0.
extraBits += int(w.offsetFreq[offsetCode]) * int(offsetExtraBits[offsetCode])
}
}
// Figure out smallest code.
// Fixed Huffman baseline.
var literalEncoding = fixedLiteralEncoding
var offsetEncoding = fixedOffsetEncoding
var size = w.fixedSize(extraBits)
// Dynamic Huffman?
var numCodegens int
// Generate codegen and codegenFrequencies, which indicates how to encode
// the literalEncoding and the offsetEncoding.
w.generateCodegen(numLiterals, numOffsets, w.literalEncoding, w.offsetEncoding)
w.codegenEncoding.generate(w.codegenFreq[:], 7)
dynamicSize, numCodegens := w.dynamicSize(w.literalEncoding, w.offsetEncoding, extraBits)
if dynamicSize < size {
size = dynamicSize
literalEncoding = w.literalEncoding
offsetEncoding = w.offsetEncoding
}
// Stored bytes?
if storable && storedSize < size {
w.writeStoredHeader(len(input), eof)
w.writeBytes(input)
return
}
// Huffman.
if literalEncoding == fixedLiteralEncoding {
w.writeFixedHeader(eof)
} else {
w.writeDynamicHeader(numLiterals, numOffsets, numCodegens, eof)
}
// Write the tokens.
w.writeTokens(tokens, literalEncoding.codes, offsetEncoding.codes)
}
// writeBlockDynamic encodes a block using a dynamic Huffman table.
// This should be used if the symbols used have a disproportionate
// histogram distribution.
// If input is supplied and the compression savings are below 1/16th of the
// input size the block is stored.
func (w *huffmanBitWriter) writeBlockDynamic(tokens []token, eof bool, input []byte) {
if w.err != nil {
return
}
tokens = append(tokens, endBlockMarker)
numLiterals, numOffsets := w.indexTokens(tokens)
// Generate codegen and codegenFrequencies, which indicates how to encode
// the literalEncoding and the offsetEncoding.
w.generateCodegen(numLiterals, numOffsets, w.literalEncoding, w.offsetEncoding)
w.codegenEncoding.generate(w.codegenFreq[:], 7)
size, numCodegens := w.dynamicSize(w.literalEncoding, w.offsetEncoding, 0)
// Store bytes, if we don't get a reasonable improvement.
if ssize, storable := w.storedSize(input); storable && ssize < (size+size>>4) {
w.writeStoredHeader(len(input), eof)
w.writeBytes(input)
return
}
// Write Huffman table.
w.writeDynamicHeader(numLiterals, numOffsets, numCodegens, eof)
// Write the tokens.
w.writeTokens(tokens, w.literalEncoding.codes, w.offsetEncoding.codes)
}
// indexTokens indexes a slice of tokens, and updates
// literalFreq and offsetFreq, and generates literalEncoding
// and offsetEncoding.
// The number of literal and offset tokens is returned.
func (w *huffmanBitWriter) indexTokens(tokens []token) (numLiterals, numOffsets int) {
for i := range w.literalFreq {
w.literalFreq[i] = 0
}
for i := range w.offsetFreq {
w.offsetFreq[i] = 0
}
for _, t := range tokens {
if t < matchType {
w.literalFreq[t.literal()]++
continue
}
length := t.length()
offset := t.offset()
w.literalFreq[lengthCodesStart+lengthCode(length)]++
w.offsetFreq[offsetCode(offset)]++
}
// get the number of literals
numLiterals = len(w.literalFreq)
for w.literalFreq[numLiterals-1] == 0 {
numLiterals--
}
// get the number of offsets
numOffsets = len(w.offsetFreq)
for numOffsets > 0 && w.offsetFreq[numOffsets-1] == 0 {
numOffsets--
}
if numOffsets == 0 {
// We haven't found a single match. If we want to go with the dynamic encoding,
// we should count at least one offset to be sure that the offset huffman tree could be encoded.
w.offsetFreq[0] = 1
numOffsets = 1
}
w.literalEncoding.generate(w.literalFreq, 15)
w.offsetEncoding.generate(w.offsetFreq, 15)
return
}
// writeTokens writes a slice of tokens to the output.
// codes for literal and offset encoding must be supplied.
func (w *huffmanBitWriter) writeTokens(tokens []token, leCodes, oeCodes []hcode) {
if w.err != nil {
return
}
for _, t := range tokens {
if t < matchType {
w.writeCode(leCodes[t.literal()])
continue
}
// Write the length
length := t.length()
lengthCode := lengthCode(length)
w.writeCode(leCodes[lengthCode+lengthCodesStart])
extraLengthBits := uint(lengthExtraBits[lengthCode])
if extraLengthBits > 0 {
extraLength := int32(length - lengthBase[lengthCode])
w.writeBits(extraLength, extraLengthBits)
}
// Write the offset
offset := t.offset()
offsetCode := offsetCode(offset)
w.writeCode(oeCodes[offsetCode])
extraOffsetBits := uint(offsetExtraBits[offsetCode])
if extraOffsetBits > 0 {
extraOffset := int32(offset - offsetBase[offsetCode])
w.writeBits(extraOffset, extraOffsetBits)
}
}
}
// huffOffset is a static offset encoder used for huffman only encoding.
// It can be reused since we will not be encoding offset values.
var huffOffset *huffmanEncoder
func init() {
w := newHuffmanBitWriter(nil)
w.offsetFreq[0] = 1
huffOffset = newHuffmanEncoder(offsetCodeCount)
huffOffset.generate(w.offsetFreq, 15)
}
// writeBlockHuff encodes a block of bytes as either
// Huffman encoded literals or uncompressed bytes if the
// results only gains very little from compression.
func (w *huffmanBitWriter) writeBlockHuff(eof bool, input []byte) {
if w.err != nil {
return
}
// Clear histogram
for i := range w.literalFreq {
w.literalFreq[i] = 0
}
// Add everything as literals
histogram(input, w.literalFreq)
w.literalFreq[endBlockMarker] = 1
const numLiterals = endBlockMarker + 1
const numOffsets = 1
w.literalEncoding.generate(w.literalFreq, 15)
// Figure out smallest code.
// Always use dynamic Huffman or Store
var numCodegens int
// Generate codegen and codegenFrequencies, which indicates how to encode
// the literalEncoding and the offsetEncoding.
w.generateCodegen(numLiterals, numOffsets, w.literalEncoding, huffOffset)
w.codegenEncoding.generate(w.codegenFreq[:], 7)
size, numCodegens := w.dynamicSize(w.literalEncoding, huffOffset, 0)
// Store bytes, if we don't get a reasonable improvement.
if ssize, storable := w.storedSize(input); storable && ssize < (size+size>>4) {
w.writeStoredHeader(len(input), eof)
w.writeBytes(input)
return
}
// Huffman.
w.writeDynamicHeader(numLiterals, numOffsets, numCodegens, eof)
encoding := w.literalEncoding.codes[:257]
n := w.nbytes
for _, t := range input {
// Bitwriting inlined, ~30% speedup
c := encoding[t]
w.bits |= uint64(c.code) << w.nbits
w.nbits += uint(c.len)
if w.nbits < 48 {
continue
}
// Store 6 bytes
bits := w.bits
w.bits >>= 48
w.nbits -= 48
bytes := w.bytes[n : n+6]
bytes[0] = byte(bits)
bytes[1] = byte(bits >> 8)
bytes[2] = byte(bits >> 16)
bytes[3] = byte(bits >> 24)
bytes[4] = byte(bits >> 32)
bytes[5] = byte(bits >> 40)
n += 6
if n < bufferFlushSize {
continue
}
w.write(w.bytes[:n])
if w.err != nil {
return // Return early in the event of write failures
}
n = 0
}
w.nbytes = n
w.writeCode(encoding[endBlockMarker])
}

366
vendor/github.com/klauspost/compress/flate/huffman_bit_writer_test.go generated vendored Normal file
View File

File diff suppressed because one or more lines are too long

344
vendor/github.com/klauspost/compress/flate/huffman_code.go generated vendored Normal file
View File

@@ -0,0 +1,344 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"math"
"sort"
)
// hcode is a huffman code with a bit code and bit length.
type hcode struct {
code, len uint16
}
type huffmanEncoder struct {
codes []hcode
freqcache []literalNode
bitCount [17]int32
lns byLiteral // stored to avoid repeated allocation in generate
lfs byFreq // stored to avoid repeated allocation in generate
}
type literalNode struct {
literal uint16
freq int32
}
// A levelInfo describes the state of the constructed tree for a given depth.
type levelInfo struct {
// Our level. for better printing
level int32
// The frequency of the last node at this level
lastFreq int32
// The frequency of the next character to add to this level
nextCharFreq int32
// The frequency of the next pair (from level below) to add to this level.
// Only valid if the "needed" value of the next lower level is 0.
nextPairFreq int32
// The number of chains remaining to generate for this level before moving
// up to the next level
needed int32
}
// set sets the code and length of an hcode.
func (h *hcode) set(code uint16, length uint16) {
h.len = length
h.code = code
}
func maxNode() literalNode { return literalNode{math.MaxUint16, math.MaxInt32} }
func newHuffmanEncoder(size int) *huffmanEncoder {
return &huffmanEncoder{codes: make([]hcode, size)}
}
// Generates a HuffmanCode corresponding to the fixed literal table
func generateFixedLiteralEncoding() *huffmanEncoder {
h := newHuffmanEncoder(maxNumLit)
codes := h.codes
var ch uint16
for ch = 0; ch < maxNumLit; ch++ {
var bits uint16
var size uint16
switch {
case ch < 144:
// size 8, 000110000 .. 10111111
bits = ch + 48
size = 8
break
case ch < 256:
// size 9, 110010000 .. 111111111
bits = ch + 400 - 144
size = 9
break
case ch < 280:
// size 7, 0000000 .. 0010111
bits = ch - 256
size = 7
break
default:
// size 8, 11000000 .. 11000111
bits = ch + 192 - 280
size = 8
}
codes[ch] = hcode{code: reverseBits(bits, byte(size)), len: size}
}
return h
}
func generateFixedOffsetEncoding() *huffmanEncoder {
h := newHuffmanEncoder(30)
codes := h.codes
for ch := range codes {
codes[ch] = hcode{code: reverseBits(uint16(ch), 5), len: 5}
}
return h
}
var fixedLiteralEncoding *huffmanEncoder = generateFixedLiteralEncoding()
var fixedOffsetEncoding *huffmanEncoder = generateFixedOffsetEncoding()
func (h *huffmanEncoder) bitLength(freq []int32) int {
var total int
for i, f := range freq {
if f != 0 {
total += int(f) * int(h.codes[i].len)
}
}
return total
}
const maxBitsLimit = 16
// Return the number of literals assigned to each bit size in the Huffman encoding
//
// This method is only called when list.length >= 3
// The cases of 0, 1, and 2 literals are handled by special case code.
//
// list An array of the literals with non-zero frequencies
// and their associated frequencies. The array is in order of increasing
// frequency, and has as its last element a special element with frequency
// MaxInt32
// maxBits The maximum number of bits that should be used to encode any literal.
// Must be less than 16.
// return An integer array in which array[i] indicates the number of literals
// that should be encoded in i bits.
func (h *huffmanEncoder) bitCounts(list []literalNode, maxBits int32) []int32 {
if maxBits >= maxBitsLimit {
panic("flate: maxBits too large")
}
n := int32(len(list))
list = list[0 : n+1]
list[n] = maxNode()
// The tree can't have greater depth than n - 1, no matter what. This
// saves a little bit of work in some small cases
if maxBits > n-1 {
maxBits = n - 1
}
// Create information about each of the levels.
// A bogus "Level 0" whose sole purpose is so that
// level1.prev.needed==0. This makes level1.nextPairFreq
// be a legitimate value that never gets chosen.
var levels [maxBitsLimit]levelInfo
// leafCounts[i] counts the number of literals at the left
// of ancestors of the rightmost node at level i.
// leafCounts[i][j] is the number of literals at the left
// of the level j ancestor.
var leafCounts [maxBitsLimit][maxBitsLimit]int32
for level := int32(1); level <= maxBits; level++ {
// For every level, the first two items are the first two characters.
// We initialize the levels as if we had already figured this out.
levels[level] = levelInfo{
level: level,
lastFreq: list[1].freq,
nextCharFreq: list[2].freq,
nextPairFreq: list[0].freq + list[1].freq,
}
leafCounts[level][level] = 2
if level == 1 {
levels[level].nextPairFreq = math.MaxInt32
}
}
// We need a total of 2*n - 2 items at top level and have already generated 2.
levels[maxBits].needed = 2*n - 4
level := maxBits
for {
l := &levels[level]
if l.nextPairFreq == math.MaxInt32 && l.nextCharFreq == math.MaxInt32 {
// We've run out of both leafs and pairs.
// End all calculations for this level.
// To make sure we never come back to this level or any lower level,
// set nextPairFreq impossibly large.
l.needed = 0
levels[level+1].nextPairFreq = math.MaxInt32
level++
continue
}
prevFreq := l.lastFreq
if l.nextCharFreq < l.nextPairFreq {
// The next item on this row is a leaf node.
n := leafCounts[level][level] + 1
l.lastFreq = l.nextCharFreq
// Lower leafCounts are the same of the previous node.
leafCounts[level][level] = n
l.nextCharFreq = list[n].freq
} else {
// The next item on this row is a pair from the previous row.
// nextPairFreq isn't valid until we generate two
// more values in the level below
l.lastFreq = l.nextPairFreq
// Take leaf counts from the lower level, except counts[level] remains the same.
copy(leafCounts[level][:level], leafCounts[level-1][:level])
levels[l.level-1].needed = 2
}
if l.needed--; l.needed == 0 {
// We've done everything we need to do for this level.
// Continue calculating one level up. Fill in nextPairFreq
// of that level with the sum of the two nodes we've just calculated on
// this level.
if l.level == maxBits {
// All done!
break
}
levels[l.level+1].nextPairFreq = prevFreq + l.lastFreq
level++
} else {
// If we stole from below, move down temporarily to replenish it.
for levels[level-1].needed > 0 {
level--
}
}
}
// Somethings is wrong if at the end, the top level is null or hasn't used
// all of the leaves.
if leafCounts[maxBits][maxBits] != n {
panic("leafCounts[maxBits][maxBits] != n")
}
bitCount := h.bitCount[:maxBits+1]
bits := 1
counts := &leafCounts[maxBits]
for level := maxBits; level > 0; level-- {
// chain.leafCount gives the number of literals requiring at least "bits"
// bits to encode.
bitCount[bits] = counts[level] - counts[level-1]
bits++
}
return bitCount
}
// Look at the leaves and assign them a bit count and an encoding as specified
// in RFC 1951 3.2.2
func (h *huffmanEncoder) assignEncodingAndSize(bitCount []int32, list []literalNode) {
code := uint16(0)
for n, bits := range bitCount {
code <<= 1
if n == 0 || bits == 0 {
continue
}
// The literals list[len(list)-bits] .. list[len(list)-bits]
// are encoded using "bits" bits, and get the values
// code, code + 1, .... The code values are
// assigned in literal order (not frequency order).
chunk := list[len(list)-int(bits):]
h.lns.sort(chunk)
for _, node := range chunk {
h.codes[node.literal] = hcode{code: reverseBits(code, uint8(n)), len: uint16(n)}
code++
}
list = list[0 : len(list)-int(bits)]
}
}
// Update this Huffman Code object to be the minimum code for the specified frequency count.
//
// freq An array of frequencies, in which frequency[i] gives the frequency of literal i.
// maxBits The maximum number of bits to use for any literal.
func (h *huffmanEncoder) generate(freq []int32, maxBits int32) {
if h.freqcache == nil {
// Allocate a reusable buffer with the longest possible frequency table.
// Possible lengths are codegenCodeCount, offsetCodeCount and maxNumLit.
// The largest of these is maxNumLit, so we allocate for that case.
h.freqcache = make([]literalNode, maxNumLit+1)
}
list := h.freqcache[:len(freq)+1]
// Number of non-zero literals
count := 0
// Set list to be the set of all non-zero literals and their frequencies
for i, f := range freq {
if f != 0 {
list[count] = literalNode{uint16(i), f}
count++
} else {
list[count] = literalNode{}
h.codes[i].len = 0
}
}
list[len(freq)] = literalNode{}
list = list[:count]
if count <= 2 {
// Handle the small cases here, because they are awkward for the general case code. With
// two or fewer literals, everything has bit length 1.
for i, node := range list {
// "list" is in order of increasing literal value.
h.codes[node.literal].set(uint16(i), 1)
}
return
}
h.lfs.sort(list)
// Get the number of literals for each bit count
bitCount := h.bitCounts(list, maxBits)
// And do the assignment
h.assignEncodingAndSize(bitCount, list)
}
type byLiteral []literalNode
func (s *byLiteral) sort(a []literalNode) {
*s = byLiteral(a)
sort.Sort(s)
}
func (s byLiteral) Len() int { return len(s) }
func (s byLiteral) Less(i, j int) bool {
return s[i].literal < s[j].literal
}
func (s byLiteral) Swap(i, j int) { s[i], s[j] = s[j], s[i] }
type byFreq []literalNode
func (s *byFreq) sort(a []literalNode) {
*s = byFreq(a)
sort.Sort(s)
}
func (s byFreq) Len() int { return len(s) }
func (s byFreq) Less(i, j int) bool {
if s[i].freq == s[j].freq {
return s[i].literal < s[j].literal
}
return s[i].freq < s[j].freq
}
func (s byFreq) Swap(i, j int) { s[i], s[j] = s[j], s[i] }

846
vendor/github.com/klauspost/compress/flate/inflate.go generated vendored Normal file
View File

@@ -0,0 +1,846 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// Package flate implements the DEFLATE compressed data format, described in
// RFC 1951. The gzip and zlib packages implement access to DEFLATE-based file
// formats.
package flate
import (
"bufio"
"io"
"strconv"
"sync"
)
const (
maxCodeLen = 16 // max length of Huffman code
// The next three numbers come from the RFC section 3.2.7, with the
// additional proviso in section 3.2.5 which implies that distance codes
// 30 and 31 should never occur in compressed data.
maxNumLit = 286
maxNumDist = 30
numCodes = 19 // number of codes in Huffman meta-code
)
// Initialize the fixedHuffmanDecoder only once upon first use.
var fixedOnce sync.Once
var fixedHuffmanDecoder huffmanDecoder
// A CorruptInputError reports the presence of corrupt input at a given offset.
type CorruptInputError int64
func (e CorruptInputError) Error() string {
return "flate: corrupt input before offset " + strconv.FormatInt(int64(e), 10)
}
// An InternalError reports an error in the flate code itself.
type InternalError string
func (e InternalError) Error() string { return "flate: internal error: " + string(e) }
// A ReadError reports an error encountered while reading input.
//
// Deprecated: No longer returned.
type ReadError struct {
Offset int64 // byte offset where error occurred
Err error // error returned by underlying Read
}
func (e *ReadError) Error() string {
return "flate: read error at offset " + strconv.FormatInt(e.Offset, 10) + ": " + e.Err.Error()
}
// A WriteError reports an error encountered while writing output.
//
// Deprecated: No longer returned.
type WriteError struct {
Offset int64 // byte offset where error occurred
Err error // error returned by underlying Write
}
func (e *WriteError) Error() string {
return "flate: write error at offset " + strconv.FormatInt(e.Offset, 10) + ": " + e.Err.Error()
}
// Resetter resets a ReadCloser returned by NewReader or NewReaderDict to
// to switch to a new underlying Reader. This permits reusing a ReadCloser
// instead of allocating a new one.
type Resetter interface {
// Reset discards any buffered data and resets the Resetter as if it was
// newly initialized with the given reader.
Reset(r io.Reader, dict []byte) error
}
// The data structure for decoding Huffman tables is based on that of
// zlib. There is a lookup table of a fixed bit width (huffmanChunkBits),
// For codes smaller than the table width, there are multiple entries
// (each combination of trailing bits has the same value). For codes
// larger than the table width, the table contains a link to an overflow
// table. The width of each entry in the link table is the maximum code
// size minus the chunk width.
//
// Note that you can do a lookup in the table even without all bits
// filled. Since the extra bits are zero, and the DEFLATE Huffman codes
// have the property that shorter codes come before longer ones, the
// bit length estimate in the result is a lower bound on the actual
// number of bits.
//
// See the following:
// http://www.gzip.org/algorithm.txt
// chunk & 15 is number of bits
// chunk >> 4 is value, including table link
const (
huffmanChunkBits = 9
huffmanNumChunks = 1 << huffmanChunkBits
huffmanCountMask = 15
huffmanValueShift = 4
)
type huffmanDecoder struct {
min int // the minimum code length
chunks [huffmanNumChunks]uint32 // chunks as described above
links [][]uint32 // overflow links
linkMask uint32 // mask the width of the link table
}
// Initialize Huffman decoding tables from array of code lengths.
// Following this function, h is guaranteed to be initialized into a complete
// tree (i.e., neither over-subscribed nor under-subscribed). The exception is a
// degenerate case where the tree has only a single symbol with length 1. Empty
// trees are permitted.
func (h *huffmanDecoder) init(bits []int) bool {
// Sanity enables additional runtime tests during Huffman
// table construction. It's intended to be used during
// development to supplement the currently ad-hoc unit tests.
const sanity = false
if h.min != 0 {
*h = huffmanDecoder{}
}
// Count number of codes of each length,
// compute min and max length.
var count [maxCodeLen]int
var min, max int
for _, n := range bits {
if n == 0 {
continue
}
if min == 0 || n < min {
min = n
}
if n > max {
max = n
}
count[n]++
}
// Empty tree. The decompressor.huffSym function will fail later if the tree
// is used. Technically, an empty tree is only valid for the HDIST tree and
// not the HCLEN and HLIT tree. However, a stream with an empty HCLEN tree
// is guaranteed to fail since it will attempt to use the tree to decode the
// codes for the HLIT and HDIST trees. Similarly, an empty HLIT tree is
// guaranteed to fail later since the compressed data section must be
// composed of at least one symbol (the end-of-block marker).
if max == 0 {
return true
}
code := 0
var nextcode [maxCodeLen]int
for i := min; i <= max; i++ {
code <<= 1
nextcode[i] = code
code += count[i]
}
// Check that the coding is complete (i.e., that we've
// assigned all 2-to-the-max possible bit sequences).
// Exception: To be compatible with zlib, we also need to
// accept degenerate single-code codings. See also
// TestDegenerateHuffmanCoding.
if code != 1<<uint(max) && !(code == 1 && max == 1) {
return false
}
h.min = min
if max > huffmanChunkBits {
numLinks := 1 << (uint(max) - huffmanChunkBits)
h.linkMask = uint32(numLinks - 1)
// create link tables
link := nextcode[huffmanChunkBits+1] >> 1
h.links = make([][]uint32, huffmanNumChunks-link)
for j := uint(link); j < huffmanNumChunks; j++ {
reverse := int(reverseByte[j>>8]) | int(reverseByte[j&0xff])<<8
reverse >>= uint(16 - huffmanChunkBits)
off := j - uint(link)
if sanity && h.chunks[reverse] != 0 {
panic("impossible: overwriting existing chunk")
}
h.chunks[reverse] = uint32(off<<huffmanValueShift | (huffmanChunkBits + 1))
h.links[off] = make([]uint32, numLinks)
}
}
for i, n := range bits {
if n == 0 {
continue
}
code := nextcode[n]
nextcode[n]++
chunk := uint32(i<<huffmanValueShift | n)
reverse := int(reverseByte[code>>8]) | int(reverseByte[code&0xff])<<8
reverse >>= uint(16 - n)
if n <= huffmanChunkBits {
for off := reverse; off < len(h.chunks); off += 1 << uint(n) {
// We should never need to overwrite
// an existing chunk. Also, 0 is
// never a valid chunk, because the
// lower 4 "count" bits should be
// between 1 and 15.
if sanity && h.chunks[off] != 0 {
panic("impossible: overwriting existing chunk")
}
h.chunks[off] = chunk
}
} else {
j := reverse & (huffmanNumChunks - 1)
if sanity && h.chunks[j]&huffmanCountMask != huffmanChunkBits+1 {
// Longer codes should have been
// associated with a link table above.
panic("impossible: not an indirect chunk")
}
value := h.chunks[j] >> huffmanValueShift
linktab := h.links[value]
reverse >>= huffmanChunkBits
for off := reverse; off < len(linktab); off += 1 << uint(n-huffmanChunkBits) {
if sanity && linktab[off] != 0 {
panic("impossible: overwriting existing chunk")
}
linktab[off] = chunk
}
}
}
if sanity {
// Above we've sanity checked that we never overwrote
// an existing entry. Here we additionally check that
// we filled the tables completely.
for i, chunk := range h.chunks {
if chunk == 0 {
// As an exception, in the degenerate
// single-code case, we allow odd
// chunks to be missing.
if code == 1 && i%2 == 1 {
continue
}
panic("impossible: missing chunk")
}
}
for _, linktab := range h.links {
for _, chunk := range linktab {
if chunk == 0 {
panic("impossible: missing chunk")
}
}
}
}
return true
}
// The actual read interface needed by NewReader.
// If the passed in io.Reader does not also have ReadByte,
// the NewReader will introduce its own buffering.
type Reader interface {
io.Reader
io.ByteReader
}
// Decompress state.
type decompressor struct {
// Input source.
r Reader
roffset int64
// Input bits, in top of b.
b uint32
nb uint
// Huffman decoders for literal/length, distance.
h1, h2 huffmanDecoder
// Length arrays used to define Huffman codes.
bits *[maxNumLit + maxNumDist]int
codebits *[numCodes]int
// Output history, buffer.
dict dictDecoder
// Temporary buffer (avoids repeated allocation).
buf [4]byte
// Next step in the decompression,
// and decompression state.
step func(*decompressor)
stepState int
final bool
err error
toRead []byte
hl, hd *huffmanDecoder
copyLen int
copyDist int
}
func (f *decompressor) nextBlock() {
for f.nb < 1+2 {
if f.err = f.moreBits(); f.err != nil {
return
}
}
f.final = f.b&1 == 1
f.b >>= 1
typ := f.b & 3
f.b >>= 2
f.nb -= 1 + 2
switch typ {
case 0:
f.dataBlock()
case 1:
// compressed, fixed Huffman tables
f.hl = &fixedHuffmanDecoder
f.hd = nil
f.huffmanBlock()
case 2:
// compressed, dynamic Huffman tables
if f.err = f.readHuffman(); f.err != nil {
break
}
f.hl = &f.h1
f.hd = &f.h2
f.huffmanBlock()
default:
// 3 is reserved.
f.err = CorruptInputError(f.roffset)
}
}
func (f *decompressor) Read(b []byte) (int, error) {
for {
if len(f.toRead) > 0 {
n := copy(b, f.toRead)
f.toRead = f.toRead[n:]
if len(f.toRead) == 0 {
return n, f.err
}
return n, nil
}
if f.err != nil {
return 0, f.err
}
f.step(f)
if f.err != nil && len(f.toRead) == 0 {
f.toRead = f.dict.readFlush() // Flush what's left in case of error
}
}
}
// Support the io.WriteTo interface for io.Copy and friends.
func (f *decompressor) WriteTo(w io.Writer) (int64, error) {
total := int64(0)
flushed := false
for {
if len(f.toRead) > 0 {
n, err := w.Write(f.toRead)
total += int64(n)
if err != nil {
f.err = err
return total, err
}
if n != len(f.toRead) {
return total, io.ErrShortWrite
}
f.toRead = f.toRead[:0]
}
if f.err != nil && flushed {
if f.err == io.EOF {
return total, nil
}
return total, f.err
}
if f.err == nil {
f.step(f)
}
if len(f.toRead) == 0 && f.err != nil && !flushed {
f.toRead = f.dict.readFlush() // Flush what's left in case of error
flushed = true
}
}
}
func (f *decompressor) Close() error {
if f.err == io.EOF {
return nil
}
return f.err
}
// RFC 1951 section 3.2.7.
// Compression with dynamic Huffman codes
var codeOrder = [...]int{16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15}
func (f *decompressor) readHuffman() error {
// HLIT[5], HDIST[5], HCLEN[4].
for f.nb < 5+5+4 {
if err := f.moreBits(); err != nil {
return err
}
}
nlit := int(f.b&0x1F) + 257
if nlit > maxNumLit {
return CorruptInputError(f.roffset)
}
f.b >>= 5
ndist := int(f.b&0x1F) + 1
if ndist > maxNumDist {
return CorruptInputError(f.roffset)
}
f.b >>= 5
nclen := int(f.b&0xF) + 4
// numCodes is 19, so nclen is always valid.
f.b >>= 4
f.nb -= 5 + 5 + 4
// (HCLEN+4)*3 bits: code lengths in the magic codeOrder order.
for i := 0; i < nclen; i++ {
for f.nb < 3 {
if err := f.moreBits(); err != nil {
return err
}
}
f.codebits[codeOrder[i]] = int(f.b & 0x7)
f.b >>= 3
f.nb -= 3
}
for i := nclen; i < len(codeOrder); i++ {
f.codebits[codeOrder[i]] = 0
}
if !f.h1.init(f.codebits[0:]) {
return CorruptInputError(f.roffset)
}
// HLIT + 257 code lengths, HDIST + 1 code lengths,
// using the code length Huffman code.
for i, n := 0, nlit+ndist; i < n; {
x, err := f.huffSym(&f.h1)
if err != nil {
return err
}
if x < 16 {
// Actual length.
f.bits[i] = x
i++
continue
}
// Repeat previous length or zero.
var rep int
var nb uint
var b int
switch x {
default:
return InternalError("unexpected length code")
case 16:
rep = 3
nb = 2
if i == 0 {
return CorruptInputError(f.roffset)
}
b = f.bits[i-1]
case 17:
rep = 3
nb = 3
b = 0
case 18:
rep = 11
nb = 7
b = 0
}
for f.nb < nb {
if err := f.moreBits(); err != nil {
return err
}
}
rep += int(f.b & uint32(1<<nb-1))
f.b >>= nb
f.nb -= nb
if i+rep > n {
return CorruptInputError(f.roffset)
}
for j := 0; j < rep; j++ {
f.bits[i] = b
i++
}
}
if !f.h1.init(f.bits[0:nlit]) || !f.h2.init(f.bits[nlit:nlit+ndist]) {
return CorruptInputError(f.roffset)
}
// As an optimization, we can initialize the min bits to read at a time
// for the HLIT tree to the length of the EOB marker since we know that
// every block must terminate with one. This preserves the property that
// we never read any extra bytes after the end of the DEFLATE stream.
if f.h1.min < f.bits[endBlockMarker] {
f.h1.min = f.bits[endBlockMarker]
}
return nil
}
// Decode a single Huffman block from f.
// hl and hd are the Huffman states for the lit/length values
// and the distance values, respectively. If hd == nil, using the
// fixed distance encoding associated with fixed Huffman blocks.
func (f *decompressor) huffmanBlock() {
const (
stateInit = iota // Zero value must be stateInit
stateDict
)
switch f.stepState {
case stateInit:
goto readLiteral
case stateDict:
goto copyHistory
}
readLiteral:
// Read literal and/or (length, distance) according to RFC section 3.2.3.
{
v, err := f.huffSym(f.hl)
if err != nil {
f.err = err
return
}
var n uint // number of bits extra
var length int
switch {
case v < 256:
f.dict.writeByte(byte(v))
if f.dict.availWrite() == 0 {
f.toRead = f.dict.readFlush()
f.step = (*decompressor).huffmanBlock
f.stepState = stateInit
return
}
goto readLiteral
case v == 256:
f.finishBlock()
return
// otherwise, reference to older data
case v < 265:
length = v - (257 - 3)
n = 0
case v < 269:
length = v*2 - (265*2 - 11)
n = 1
case v < 273:
length = v*4 - (269*4 - 19)
n = 2
case v < 277:
length = v*8 - (273*8 - 35)
n = 3
case v < 281:
length = v*16 - (277*16 - 67)
n = 4
case v < 285:
length = v*32 - (281*32 - 131)
n = 5
case v < maxNumLit:
length = 258
n = 0
default:
f.err = CorruptInputError(f.roffset)
return
}
if n > 0 {
for f.nb < n {
if err = f.moreBits(); err != nil {
f.err = err
return
}
}
length += int(f.b & uint32(1<<n-1))
f.b >>= n
f.nb -= n
}
var dist int
if f.hd == nil {
for f.nb < 5 {
if err = f.moreBits(); err != nil {
f.err = err
return
}
}
dist = int(reverseByte[(f.b&0x1F)<<3])
f.b >>= 5
f.nb -= 5
} else {
if dist, err = f.huffSym(f.hd); err != nil {
f.err = err
return
}
}
switch {
case dist < 4:
dist++
case dist < maxNumDist:
nb := uint(dist-2) >> 1
// have 1 bit in bottom of dist, need nb more.
extra := (dist & 1) << nb
for f.nb < nb {
if err = f.moreBits(); err != nil {
f.err = err
return
}
}
extra |= int(f.b & uint32(1<<nb-1))
f.b >>= nb
f.nb -= nb
dist = 1<<(nb+1) + 1 + extra
default:
f.err = CorruptInputError(f.roffset)
return
}
// No check on length; encoding can be prescient.
if dist > f.dict.histSize() {
f.err = CorruptInputError(f.roffset)
return
}
f.copyLen, f.copyDist = length, dist
goto copyHistory
}
copyHistory:
// Perform a backwards copy according to RFC section 3.2.3.
{
cnt := f.dict.tryWriteCopy(f.copyDist, f.copyLen)
if cnt == 0 {
cnt = f.dict.writeCopy(f.copyDist, f.copyLen)
}
f.copyLen -= cnt
if f.dict.availWrite() == 0 || f.copyLen > 0 {
f.toRead = f.dict.readFlush()
f.step = (*decompressor).huffmanBlock // We need to continue this work
f.stepState = stateDict
return
}
goto readLiteral
}
}
// Copy a single uncompressed data block from input to output.
func (f *decompressor) dataBlock() {
// Uncompressed.
// Discard current half-byte.
f.nb = 0
f.b = 0
// Length then ones-complement of length.
nr, err := io.ReadFull(f.r, f.buf[0:4])
f.roffset += int64(nr)
if err != nil {
if err == io.EOF {
err = io.ErrUnexpectedEOF
}
f.err = err
return
}
n := int(f.buf[0]) | int(f.buf[1])<<8
nn := int(f.buf[2]) | int(f.buf[3])<<8
if uint16(nn) != uint16(^n) {
f.err = CorruptInputError(f.roffset)
return
}
if n == 0 {
f.toRead = f.dict.readFlush()
f.finishBlock()
return
}
f.copyLen = n
f.copyData()
}
// copyData copies f.copyLen bytes from the underlying reader into f.hist.
// It pauses for reads when f.hist is full.
func (f *decompressor) copyData() {
buf := f.dict.writeSlice()
if len(buf) > f.copyLen {
buf = buf[:f.copyLen]
}
cnt, err := io.ReadFull(f.r, buf)
f.roffset += int64(cnt)
f.copyLen -= cnt
f.dict.writeMark(cnt)
if err != nil {
if err == io.EOF {
err = io.ErrUnexpectedEOF
}
f.err = err
return
}
if f.dict.availWrite() == 0 || f.copyLen > 0 {
f.toRead = f.dict.readFlush()
f.step = (*decompressor).copyData
return
}
f.finishBlock()
}
func (f *decompressor) finishBlock() {
if f.final {
if f.dict.availRead() > 0 {
f.toRead = f.dict.readFlush()
}
f.err = io.EOF
}
f.step = (*decompressor).nextBlock
}
func (f *decompressor) moreBits() error {
c, err := f.r.ReadByte()
if err != nil {
if err == io.EOF {
err = io.ErrUnexpectedEOF
}
return err
}
f.roffset++
f.b |= uint32(c) << f.nb
f.nb += 8
return nil
}
// Read the next Huffman-encoded symbol from f according to h.
func (f *decompressor) huffSym(h *huffmanDecoder) (int, error) {
// Since a huffmanDecoder can be empty or be composed of a degenerate tree
// with single element, huffSym must error on these two edge cases. In both
// cases, the chunks slice will be 0 for the invalid sequence, leading it
// satisfy the n == 0 check below.
n := uint(h.min)
for {
for f.nb < n {
if err := f.moreBits(); err != nil {
return 0, err
}
}
chunk := h.chunks[f.b&(huffmanNumChunks-1)]
n = uint(chunk & huffmanCountMask)
if n > huffmanChunkBits {
chunk = h.links[chunk>>huffmanValueShift][(f.b>>huffmanChunkBits)&h.linkMask]
n = uint(chunk & huffmanCountMask)
}
if n <= f.nb {
if n == 0 {
f.err = CorruptInputError(f.roffset)
return 0, f.err
}
f.b >>= n
f.nb -= n
return int(chunk >> huffmanValueShift), nil
}
}
}
func makeReader(r io.Reader) Reader {
if rr, ok := r.(Reader); ok {
return rr
}
return bufio.NewReader(r)
}
func fixedHuffmanDecoderInit() {
fixedOnce.Do(func() {
// These come from the RFC section 3.2.6.
var bits [288]int
for i := 0; i < 144; i++ {
bits[i] = 8
}
for i := 144; i < 256; i++ {
bits[i] = 9
}
for i := 256; i < 280; i++ {
bits[i] = 7
}
for i := 280; i < 288; i++ {
bits[i] = 8
}
fixedHuffmanDecoder.init(bits[:])
})
}
func (f *decompressor) Reset(r io.Reader, dict []byte) error {
*f = decompressor{
r: makeReader(r),
bits: f.bits,
codebits: f.codebits,
dict: f.dict,
step: (*decompressor).nextBlock,
}
f.dict.init(maxMatchOffset, dict)
return nil
}
// NewReader returns a new ReadCloser that can be used
// to read the uncompressed version of r.
// If r does not also implement io.ByteReader,
// the decompressor may read more data than necessary from r.
// It is the caller's responsibility to call Close on the ReadCloser
// when finished reading.
//
// The ReadCloser returned by NewReader also implements Resetter.
func NewReader(r io.Reader) io.ReadCloser {
fixedHuffmanDecoderInit()
var f decompressor
f.r = makeReader(r)
f.bits = new([maxNumLit + maxNumDist]int)
f.codebits = new([numCodes]int)
f.step = (*decompressor).nextBlock
f.dict.init(maxMatchOffset, nil)
return &f
}
// NewReaderDict is like NewReader but initializes the reader
// with a preset dictionary. The returned Reader behaves as if
// the uncompressed data stream started with the given dictionary,
// which has already been read. NewReaderDict is typically used
// to read data compressed by NewWriterDict.
//
// The ReadCloser returned by NewReader also implements Resetter.
func NewReaderDict(r io.Reader, dict []byte) io.ReadCloser {
fixedHuffmanDecoderInit()
var f decompressor
f.r = makeReader(r)
f.bits = new([maxNumLit + maxNumDist]int)
f.codebits = new([numCodes]int)
f.step = (*decompressor).nextBlock
f.dict.init(maxMatchOffset, dict)
return &f
}

282
vendor/github.com/klauspost/compress/flate/inflate_test.go generated vendored Normal file
View File

@@ -0,0 +1,282 @@
// Copyright 2014 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"bytes"
"crypto/rand"
"io"
"io/ioutil"
"strconv"
"strings"
"testing"
)
func TestReset(t *testing.T) {
ss := []string{
"lorem ipsum izzle fo rizzle",
"the quick brown fox jumped over",
}
deflated := make([]bytes.Buffer, 2)
for i, s := range ss {
w, _ := NewWriter(&deflated[i], 1)
w.Write([]byte(s))
w.Close()
}
inflated := make([]bytes.Buffer, 2)
f := NewReader(&deflated[0])
io.Copy(&inflated[0], f)
f.(Resetter).Reset(&deflated[1], nil)
io.Copy(&inflated[1], f)
f.Close()
for i, s := range ss {
if s != inflated[i].String() {
t.Errorf("inflated[%d]:\ngot %q\nwant %q", i, inflated[i], s)
}
}
}
func TestReaderTruncated(t *testing.T) {
vectors := []struct{ input, output string }{
{"\x00", ""},
{"\x00\f", ""},
{"\x00\f\x00", ""},
{"\x00\f\x00\xf3\xff", ""},
{"\x00\f\x00\xf3\xffhello", "hello"},
{"\x00\f\x00\xf3\xffhello, world", "hello, world"},
{"\x02", ""},
{"\xf2H\xcd", "He"},
{"\xf2H͙0a\u0084\t", "Hel\x90\x90\x90\x90\x90"},
{"\xf2H͙0a\u0084\t\x00", "Hel\x90\x90\x90\x90\x90"},
}
for i, v := range vectors {
r := strings.NewReader(v.input)
zr := NewReader(r)
b, err := ioutil.ReadAll(zr)
if err != io.ErrUnexpectedEOF {
t.Errorf("test %d, error mismatch: got %v, want io.ErrUnexpectedEOF", i, err)
}
if string(b) != v.output {
t.Errorf("test %d, output mismatch: got %q, want %q", i, b, v.output)
}
}
}
func TestResetDict(t *testing.T) {
dict := []byte("the lorem fox")
ss := []string{
"lorem ipsum izzle fo rizzle",
"the quick brown fox jumped over",
}
deflated := make([]bytes.Buffer, len(ss))
for i, s := range ss {
w, _ := NewWriterDict(&deflated[i], DefaultCompression, dict)
w.Write([]byte(s))
w.Close()
}
inflated := make([]bytes.Buffer, len(ss))
f := NewReader(nil)
for i := range inflated {
f.(Resetter).Reset(&deflated[i], dict)
io.Copy(&inflated[i], f)
}
f.Close()
for i, s := range ss {
if s != inflated[i].String() {
t.Errorf("inflated[%d]:\ngot %q\nwant %q", i, inflated[i], s)
}
}
}
// Tests ported from zlib/test/infcover.c
type infTest struct {
hex string
id string
n int
}
var infTests = []infTest{
{"0 0 0 0 0", "invalid stored block lengths", 1},
{"3 0", "fixed", 0},
{"6", "invalid block type", 1},
{"1 1 0 fe ff 0", "stored", 0},
{"fc 0 0", "too many length or distance symbols", 1},
{"4 0 fe ff", "invalid code lengths set", 1},
{"4 0 24 49 0", "invalid bit length repeat", 1},
{"4 0 24 e9 ff ff", "invalid bit length repeat", 1},
{"4 0 24 e9 ff 6d", "invalid code -- missing end-of-block", 1},
{"4 80 49 92 24 49 92 24 71 ff ff 93 11 0", "invalid literal/lengths set", 1},
{"4 80 49 92 24 49 92 24 f b4 ff ff c3 84", "invalid distances set", 1},
{"4 c0 81 8 0 0 0 0 20 7f eb b 0 0", "invalid literal/length code", 1},
{"2 7e ff ff", "invalid distance code", 1},
{"c c0 81 0 0 0 0 0 90 ff 6b 4 0", "invalid distance too far back", 1},
// also trailer mismatch just in inflate()
{"1f 8b 8 0 0 0 0 0 0 0 3 0 0 0 0 1", "incorrect data check", -1},
{"1f 8b 8 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 1", "incorrect length check", -1},
{"5 c0 21 d 0 0 0 80 b0 fe 6d 2f 91 6c", "pull 17", 0},
{"5 e0 81 91 24 cb b2 2c 49 e2 f 2e 8b 9a 47 56 9f fb fe ec d2 ff 1f", "long code", 0},
{"ed c0 1 1 0 0 0 40 20 ff 57 1b 42 2c 4f", "length extra", 0},
{"ed cf c1 b1 2c 47 10 c4 30 fa 6f 35 1d 1 82 59 3d fb be 2e 2a fc f c", "long distance and extra", 0},
{"ed c0 81 0 0 0 0 80 a0 fd a9 17 a9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6", "window end", 0},
}
func TestInflate(t *testing.T) {
for _, test := range infTests {
hex := strings.Split(test.hex, " ")
data := make([]byte, len(hex))
for i, h := range hex {
b, _ := strconv.ParseInt(h, 16, 32)
data[i] = byte(b)
}
buf := bytes.NewReader(data)
r := NewReader(buf)
_, err := io.Copy(ioutil.Discard, r)
if (test.n == 0 && err == nil) || (test.n != 0 && err != nil) {
t.Logf("%q: OK:", test.id)
t.Logf(" - got %v", err)
continue
}
if test.n == 0 && err != nil {
t.Errorf("%q: Expected no error, but got %v", test.id, err)
continue
}
if test.n != 0 && err == nil {
t.Errorf("%q:Expected an error, but got none", test.id)
continue
}
t.Fatal(test.n, err)
}
for _, test := range infOutTests {
hex := strings.Split(test.hex, " ")
data := make([]byte, len(hex))
for i, h := range hex {
b, _ := strconv.ParseInt(h, 16, 32)
data[i] = byte(b)
}
buf := bytes.NewReader(data)
r := NewReader(buf)
_, err := io.Copy(ioutil.Discard, r)
if test.err == (err != nil) {
t.Logf("%q: OK:", test.id)
t.Logf(" - got %v", err)
continue
}
if test.err == false && err != nil {
t.Errorf("%q: Expected no error, but got %v", test.id, err)
continue
}
if test.err && err == nil {
t.Errorf("%q: Expected an error, but got none", test.id)
continue
}
t.Fatal(test.err, err)
}
}
// Tests ported from zlib/test/infcover.c
// Since zlib inflate is push (writer) instead of pull (reader)
// some of the window size tests have been removed, since they
// are irrelevant.
type infOutTest struct {
hex string
id string
step int
win int
length int
err bool
}
var infOutTests = []infOutTest{
{"2 8 20 80 0 3 0", "inflate_fast TYPE return", 0, -15, 258, false},
{"63 18 5 40 c 0", "window wrap", 3, -8, 300, false},
{"e5 e0 81 ad 6d cb b2 2c c9 01 1e 59 63 ae 7d ee fb 4d fd b5 35 41 68 ff 7f 0f 0 0 0", "fast length extra bits", 0, -8, 258, true},
{"25 fd 81 b5 6d 59 b6 6a 49 ea af 35 6 34 eb 8c b9 f6 b9 1e ef 67 49 50 fe ff ff 3f 0 0", "fast distance extra bits", 0, -8, 258, true},
{"3 7e 0 0 0 0 0", "fast invalid distance code", 0, -8, 258, true},
{"1b 7 0 0 0 0 0", "fast invalid literal/length code", 0, -8, 258, true},
{"d c7 1 ae eb 38 c 4 41 a0 87 72 de df fb 1f b8 36 b1 38 5d ff ff 0", "fast 2nd level codes and too far back", 0, -8, 258, true},
{"63 18 5 8c 10 8 0 0 0 0", "very common case", 0, -8, 259, false},
{"63 60 60 18 c9 0 8 18 18 18 26 c0 28 0 29 0 0 0", "contiguous and wrap around window", 6, -8, 259, false},
{"63 0 3 0 0 0 0 0", "copy direct from output", 0, -8, 259, false},
{"1f 8b 0 0", "bad gzip method", 0, 31, 0, true},
{"1f 8b 8 80", "bad gzip flags", 0, 31, 0, true},
{"77 85", "bad zlib method", 0, 15, 0, true},
{"78 9c", "bad zlib window size", 0, 8, 0, true},
{"1f 8b 8 1e 0 0 0 0 0 0 1 0 0 0 0 0 0", "bad header crc", 0, 47, 1, true},
{"1f 8b 8 2 0 0 0 0 0 0 1d 26 3 0 0 0 0 0 0 0 0 0", "check gzip length", 0, 47, 0, true},
{"78 90", "bad zlib header check", 0, 47, 0, true},
{"8 b8 0 0 0 1", "need dictionary", 0, 8, 0, true},
{"63 18 68 30 d0 0 0", "force split window update", 4, -8, 259, false},
{"3 0", "use fixed blocks", 0, -15, 1, false},
{"", "bad window size", 0, 1, 0, true},
}
func TestWriteTo(t *testing.T) {
input := make([]byte, 100000)
n, err := rand.Read(input)
if err != nil {
t.Fatal(err)
}
if n != len(input) {
t.Fatal("did not fill buffer")
}
compressed := &bytes.Buffer{}
w, err := NewWriter(compressed, -2)
if err != nil {
t.Fatal(err)
}
n, err = w.Write(input)
if err != nil {
t.Fatal(err)
}
if n != len(input) {
t.Fatal("did not fill buffer")
}
w.Close()
buf := compressed.Bytes()
dec := NewReader(bytes.NewBuffer(buf))
// ReadAll does not use WriteTo, but we wrap it in a NopCloser to be sure.
readall, err := ioutil.ReadAll(ioutil.NopCloser(dec))
if err != nil {
t.Fatal(err)
}
if len(readall) != len(input) {
t.Fatal("did not decompress everything")
}
dec = NewReader(bytes.NewBuffer(buf))
wtbuf := &bytes.Buffer{}
written, err := dec.(io.WriterTo).WriteTo(wtbuf)
if err != nil {
t.Fatal(err)
}
if written != int64(len(input)) {
t.Error("Returned length did not match, expected", len(input), "got", written)
}
if wtbuf.Len() != len(input) {
t.Error("Actual Length did not match, expected", len(input), "got", wtbuf.Len())
}
if bytes.Compare(wtbuf.Bytes(), input) != 0 {
t.Fatal("output did not match input")
}
}

97
vendor/github.com/klauspost/compress/flate/reader_test.go generated vendored Normal file
View File

@@ -0,0 +1,97 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"bytes"
"io"
"io/ioutil"
"runtime"
"strings"
"testing"
)
func TestNlitOutOfRange(t *testing.T) {
// Trying to decode this bogus flate data, which has a Huffman table
// with nlit=288, should not panic.
io.Copy(ioutil.Discard, NewReader(strings.NewReader(
"\xfc\xfe\x36\xe7\x5e\x1c\xef\xb3\x55\x58\x77\xb6\x56\xb5\x43\xf4"+
"\x6f\xf2\xd2\xe6\x3d\x99\xa0\x85\x8c\x48\xeb\xf8\xda\x83\x04\x2a"+
"\x75\xc4\xf8\x0f\x12\x11\xb9\xb4\x4b\x09\xa0\xbe\x8b\x91\x4c")))
}
const (
digits = iota
twain
)
var testfiles = []string{
// Digits is the digits of the irrational number e. Its decimal representation
// does not repeat, but there are only 10 possible digits, so it should be
// reasonably compressible.
digits: "../testdata/e.txt",
// Twain is Project Gutenberg's edition of Mark Twain's classic English novel.
twain: "../testdata/Mark.Twain-Tom.Sawyer.txt",
}
func benchmarkDecode(b *testing.B, testfile, level, n int) {
b.ReportAllocs()
b.StopTimer()
b.SetBytes(int64(n))
buf0, err := ioutil.ReadFile(testfiles[testfile])
if err != nil {
b.Fatal(err)
}
if len(buf0) == 0 {
b.Fatalf("test file %q has no data", testfiles[testfile])
}
compressed := new(bytes.Buffer)
w, err := NewWriter(compressed, level)
if err != nil {
b.Fatal(err)
}
for i := 0; i < n; i += len(buf0) {
if len(buf0) > n-i {
buf0 = buf0[:n-i]
}
io.Copy(w, bytes.NewReader(buf0))
}
w.Close()
buf1 := compressed.Bytes()
buf0, compressed, w = nil, nil, nil
runtime.GC()
b.StartTimer()
for i := 0; i < b.N; i++ {
io.Copy(ioutil.Discard, NewReader(bytes.NewReader(buf1)))
}
}
// These short names are so that gofmt doesn't break the BenchmarkXxx function
// bodies below over multiple lines.
const (
constant = ConstantCompression
speed = BestSpeed
default_ = DefaultCompression
compress = BestCompression
)
func BenchmarkDecodeDigitsSpeed1e4(b *testing.B) { benchmarkDecode(b, digits, speed, 1e4) }
func BenchmarkDecodeDigitsSpeed1e5(b *testing.B) { benchmarkDecode(b, digits, speed, 1e5) }
func BenchmarkDecodeDigitsSpeed1e6(b *testing.B) { benchmarkDecode(b, digits, speed, 1e6) }
func BenchmarkDecodeDigitsDefault1e4(b *testing.B) { benchmarkDecode(b, digits, default_, 1e4) }
func BenchmarkDecodeDigitsDefault1e5(b *testing.B) { benchmarkDecode(b, digits, default_, 1e5) }
func BenchmarkDecodeDigitsDefault1e6(b *testing.B) { benchmarkDecode(b, digits, default_, 1e6) }
func BenchmarkDecodeDigitsCompress1e4(b *testing.B) { benchmarkDecode(b, digits, compress, 1e4) }
func BenchmarkDecodeDigitsCompress1e5(b *testing.B) { benchmarkDecode(b, digits, compress, 1e5) }
func BenchmarkDecodeDigitsCompress1e6(b *testing.B) { benchmarkDecode(b, digits, compress, 1e6) }
func BenchmarkDecodeTwainSpeed1e4(b *testing.B) { benchmarkDecode(b, twain, speed, 1e4) }
func BenchmarkDecodeTwainSpeed1e5(b *testing.B) { benchmarkDecode(b, twain, speed, 1e5) }
func BenchmarkDecodeTwainSpeed1e6(b *testing.B) { benchmarkDecode(b, twain, speed, 1e6) }
func BenchmarkDecodeTwainDefault1e4(b *testing.B) { benchmarkDecode(b, twain, default_, 1e4) }
func BenchmarkDecodeTwainDefault1e5(b *testing.B) { benchmarkDecode(b, twain, default_, 1e5) }
func BenchmarkDecodeTwainDefault1e6(b *testing.B) { benchmarkDecode(b, twain, default_, 1e6) }
func BenchmarkDecodeTwainCompress1e4(b *testing.B) { benchmarkDecode(b, twain, compress, 1e4) }
func BenchmarkDecodeTwainCompress1e5(b *testing.B) { benchmarkDecode(b, twain, compress, 1e5) }
func BenchmarkDecodeTwainCompress1e6(b *testing.B) { benchmarkDecode(b, twain, compress, 1e6) }

48
vendor/github.com/klauspost/compress/flate/reverse_bits.go generated vendored Normal file
View File

@@ -0,0 +1,48 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
var reverseByte = [256]byte{
0x00, 0x80, 0x40, 0xc0, 0x20, 0xa0, 0x60, 0xe0,
0x10, 0x90, 0x50, 0xd0, 0x30, 0xb0, 0x70, 0xf0,
0x08, 0x88, 0x48, 0xc8, 0x28, 0xa8, 0x68, 0xe8,
0x18, 0x98, 0x58, 0xd8, 0x38, 0xb8, 0x78, 0xf8,
0x04, 0x84, 0x44, 0xc4, 0x24, 0xa4, 0x64, 0xe4,
0x14, 0x94, 0x54, 0xd4, 0x34, 0xb4, 0x74, 0xf4,
0x0c, 0x8c, 0x4c, 0xcc, 0x2c, 0xac, 0x6c, 0xec,
0x1c, 0x9c, 0x5c, 0xdc, 0x3c, 0xbc, 0x7c, 0xfc,
0x02, 0x82, 0x42, 0xc2, 0x22, 0xa2, 0x62, 0xe2,
0x12, 0x92, 0x52, 0xd2, 0x32, 0xb2, 0x72, 0xf2,
0x0a, 0x8a, 0x4a, 0xca, 0x2a, 0xaa, 0x6a, 0xea,
0x1a, 0x9a, 0x5a, 0xda, 0x3a, 0xba, 0x7a, 0xfa,
0x06, 0x86, 0x46, 0xc6, 0x26, 0xa6, 0x66, 0xe6,
0x16, 0x96, 0x56, 0xd6, 0x36, 0xb6, 0x76, 0xf6,
0x0e, 0x8e, 0x4e, 0xce, 0x2e, 0xae, 0x6e, 0xee,
0x1e, 0x9e, 0x5e, 0xde, 0x3e, 0xbe, 0x7e, 0xfe,
0x01, 0x81, 0x41, 0xc1, 0x21, 0xa1, 0x61, 0xe1,
0x11, 0x91, 0x51, 0xd1, 0x31, 0xb1, 0x71, 0xf1,
0x09, 0x89, 0x49, 0xc9, 0x29, 0xa9, 0x69, 0xe9,
0x19, 0x99, 0x59, 0xd9, 0x39, 0xb9, 0x79, 0xf9,
0x05, 0x85, 0x45, 0xc5, 0x25, 0xa5, 0x65, 0xe5,
0x15, 0x95, 0x55, 0xd5, 0x35, 0xb5, 0x75, 0xf5,
0x0d, 0x8d, 0x4d, 0xcd, 0x2d, 0xad, 0x6d, 0xed,
0x1d, 0x9d, 0x5d, 0xdd, 0x3d, 0xbd, 0x7d, 0xfd,
0x03, 0x83, 0x43, 0xc3, 0x23, 0xa3, 0x63, 0xe3,
0x13, 0x93, 0x53, 0xd3, 0x33, 0xb3, 0x73, 0xf3,
0x0b, 0x8b, 0x4b, 0xcb, 0x2b, 0xab, 0x6b, 0xeb,
0x1b, 0x9b, 0x5b, 0xdb, 0x3b, 0xbb, 0x7b, 0xfb,
0x07, 0x87, 0x47, 0xc7, 0x27, 0xa7, 0x67, 0xe7,
0x17, 0x97, 0x57, 0xd7, 0x37, 0xb7, 0x77, 0xf7,
0x0f, 0x8f, 0x4f, 0xcf, 0x2f, 0xaf, 0x6f, 0xef,
0x1f, 0x9f, 0x5f, 0xdf, 0x3f, 0xbf, 0x7f, 0xff,
}
func reverseUint16(v uint16) uint16 {
return uint16(reverseByte[v>>8]) | uint16(reverseByte[v&0xFF])<<8
}
func reverseBits(number uint16, bitLength byte) uint16 {
return reverseUint16(number << uint8(16-bitLength))
}

900
vendor/github.com/klauspost/compress/flate/snappy.go generated vendored Normal file
View File

@@ -0,0 +1,900 @@
// Copyright 2011 The Snappy-Go Authors. All rights reserved.
// Modified for deflate by Klaus Post (c) 2015.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
// emitLiteral writes a literal chunk and returns the number of bytes written.
func emitLiteral(dst *tokens, lit []byte) {
ol := int(dst.n)
for i, v := range lit {
dst.tokens[(i+ol)&maxStoreBlockSize] = token(v)
}
dst.n += uint16(len(lit))
}
// emitCopy writes a copy chunk and returns the number of bytes written.
func emitCopy(dst *tokens, offset, length int) {
dst.tokens[dst.n] = matchToken(uint32(length-3), uint32(offset-minOffsetSize))
dst.n++
}
type snappyEnc interface {
Encode(dst *tokens, src []byte)
Reset()
}
func newSnappy(level int) snappyEnc {
switch level {
case 1:
return &snappyL1{}
case 2:
return &snappyL2{snappyGen: snappyGen{cur: maxStoreBlockSize, prev: make([]byte, 0, maxStoreBlockSize)}}
case 3:
return &snappyL3{snappyGen: snappyGen{cur: maxStoreBlockSize, prev: make([]byte, 0, maxStoreBlockSize)}}
case 4:
return &snappyL4{snappyL3{snappyGen: snappyGen{cur: maxStoreBlockSize, prev: make([]byte, 0, maxStoreBlockSize)}}}
default:
panic("invalid level specified")
}
}
const (
tableBits = 14 // Bits used in the table
tableSize = 1 << tableBits // Size of the table
tableMask = tableSize - 1 // Mask for table indices. Redundant, but can eliminate bounds checks.
tableShift = 32 - tableBits // Right-shift to get the tableBits most significant bits of a uint32.
baseMatchOffset = 1 // The smallest match offset
baseMatchLength = 3 // The smallest match length per the RFC section 3.2.5
maxMatchOffset = 1 << 15 // The largest match offset
)
func load32(b []byte, i int) uint32 {
b = b[i : i+4 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24
}
func load64(b []byte, i int) uint64 {
b = b[i : i+8 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 |
uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56
}
func hash(u uint32) uint32 {
return (u * 0x1e35a7bd) >> tableShift
}
// snappyL1 encapsulates level 1 compression
type snappyL1 struct{}
func (e *snappyL1) Reset() {}
func (e *snappyL1) Encode(dst *tokens, src []byte) {
const (
inputMargin = 16 - 1
minNonLiteralBlockSize = 1 + 1 + inputMargin
)
// This check isn't in the Snappy implementation, but there, the caller
// instead of the callee handles this case.
if len(src) < minNonLiteralBlockSize {
// We do not fill the token table.
// This will be picked up by caller.
dst.n = uint16(len(src))
return
}
// Initialize the hash table.
//
// The table element type is uint16, as s < sLimit and sLimit < len(src)
// and len(src) <= maxStoreBlockSize and maxStoreBlockSize == 65535.
var table [tableSize]uint16
// sLimit is when to stop looking for offset/length copies. The inputMargin
// lets us use a fast path for emitLiteral in the main loop, while we are
// looking for copies.
sLimit := len(src) - inputMargin
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := 0
// The encoded form must start with a literal, as there are no previous
// bytes to copy, so we start looking for hash matches at s == 1.
s := 1
nextHash := hash(load32(src, s))
for {
// Copied from the C++ snappy implementation:
//
// Heuristic match skipping: If 32 bytes are scanned with no matches
// found, start looking only at every other byte. If 32 more bytes are
// scanned (or skipped), look at every third byte, etc.. When a match
// is found, immediately go back to looking at every byte. This is a
// small loss (~5% performance, ~0.1% density) for compressible data
// due to more bookkeeping, but for non-compressible data (such as
// JPEG) it's a huge win since the compressor quickly "realizes" the
// data is incompressible and doesn't bother looking for matches
// everywhere.
//
// The "skip" variable keeps track of how many bytes there are since
// the last match; dividing it by 32 (ie. right-shifting by five) gives
// the number of bytes to move ahead for each iteration.
skip := 32
nextS := s
candidate := 0
for {
s = nextS
bytesBetweenHashLookups := skip >> 5
nextS = s + bytesBetweenHashLookups
skip += bytesBetweenHashLookups
if nextS > sLimit {
goto emitRemainder
}
candidate = int(table[nextHash&tableMask])
table[nextHash&tableMask] = uint16(s)
nextHash = hash(load32(src, nextS))
if s-candidate <= maxMatchOffset && load32(src, s) == load32(src, candidate) {
break
}
}
// A 4-byte match has been found. We'll later see if more than 4 bytes
// match. But, prior to the match, src[nextEmit:s] are unmatched. Emit
// them as literal bytes.
emitLiteral(dst, src[nextEmit:s])
// Call emitCopy, and then see if another emitCopy could be our next
// move. Repeat until we find no match for the input immediately after
// what was consumed by the last emitCopy call.
//
// If we exit this loop normally then we need to call emitLiteral next,
// though we don't yet know how big the literal will be. We handle that
// by proceeding to the next iteration of the main loop. We also can
// exit this loop via goto if we get close to exhausting the input.
for {
// Invariant: we have a 4-byte match at s, and no need to emit any
// literal bytes prior to s.
base := s
// Extend the 4-byte match as long as possible.
//
// This is an inlined version of Snappy's:
// s = extendMatch(src, candidate+4, s+4)
s += 4
s1 := base + maxMatchLength
if s1 > len(src) {
s1 = len(src)
}
a := src[s:s1]
b := src[candidate+4:]
b = b[:len(a)]
l := len(a)
for i := range a {
if a[i] != b[i] {
l = i
break
}
}
s += l
// matchToken is flate's equivalent of Snappy's emitCopy.
dst.tokens[dst.n] = matchToken(uint32(s-base-baseMatchLength), uint32(base-candidate-baseMatchOffset))
dst.n++
nextEmit = s
if s >= sLimit {
goto emitRemainder
}
// We could immediately start working at s now, but to improve
// compression we first update the hash table at s-1 and at s. If
// another emitCopy is not our next move, also calculate nextHash
// at s+1. At least on GOARCH=amd64, these three hash calculations
// are faster as one load64 call (with some shifts) instead of
// three load32 calls.
x := load64(src, s-1)
prevHash := hash(uint32(x >> 0))
table[prevHash&tableMask] = uint16(s - 1)
currHash := hash(uint32(x >> 8))
candidate = int(table[currHash&tableMask])
table[currHash&tableMask] = uint16(s)
if s-candidate > maxMatchOffset || uint32(x>>8) != load32(src, candidate) {
nextHash = hash(uint32(x >> 16))
s++
break
}
}
}
emitRemainder:
if nextEmit < len(src) {
emitLiteral(dst, src[nextEmit:])
}
}
type tableEntry struct {
val uint32
offset int32
}
func load3232(b []byte, i int32) uint32 {
b = b[i : i+4 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24
}
func load6432(b []byte, i int32) uint64 {
b = b[i : i+8 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 |
uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56
}
// snappyGen maintains the table for matches,
// and the previous byte block for level 2.
// This is the generic implementation.
type snappyGen struct {
prev []byte
cur int32
}
// snappyGen maintains the table for matches,
// and the previous byte block for level 2.
// This is the generic implementation.
type snappyL2 struct {
snappyGen
table [tableSize]tableEntry
}
// EncodeL2 uses a similar algorithm to level 1, but is capable
// of matching across blocks giving better compression at a small slowdown.
func (e *snappyL2) Encode(dst *tokens, src []byte) {
const (
inputMargin = 8 - 1
minNonLiteralBlockSize = 1 + 1 + inputMargin
)
// Protect against e.cur wraparound.
if e.cur > 1<<30 {
for i := range e.table[:] {
e.table[i] = tableEntry{}
}
e.cur = maxStoreBlockSize
}
// This check isn't in the Snappy implementation, but there, the caller
// instead of the callee handles this case.
if len(src) < minNonLiteralBlockSize {
// We do not fill the token table.
// This will be picked up by caller.
dst.n = uint16(len(src))
e.cur += maxStoreBlockSize
e.prev = e.prev[:0]
return
}
// sLimit is when to stop looking for offset/length copies. The inputMargin
// lets us use a fast path for emitLiteral in the main loop, while we are
// looking for copies.
sLimit := int32(len(src) - inputMargin)
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := int32(0)
s := int32(0)
cv := load3232(src, s)
nextHash := hash(cv)
for {
// Copied from the C++ snappy implementation:
//
// Heuristic match skipping: If 32 bytes are scanned with no matches
// found, start looking only at every other byte. If 32 more bytes are
// scanned (or skipped), look at every third byte, etc.. When a match
// is found, immediately go back to looking at every byte. This is a
// small loss (~5% performance, ~0.1% density) for compressible data
// due to more bookkeeping, but for non-compressible data (such as
// JPEG) it's a huge win since the compressor quickly "realizes" the
// data is incompressible and doesn't bother looking for matches
// everywhere.
//
// The "skip" variable keeps track of how many bytes there are since
// the last match; dividing it by 32 (ie. right-shifting by five) gives
// the number of bytes to move ahead for each iteration.
skip := int32(32)
nextS := s
var candidate tableEntry
for {
s = nextS
bytesBetweenHashLookups := skip >> 5
nextS = s + bytesBetweenHashLookups
skip += bytesBetweenHashLookups
if nextS > sLimit {
goto emitRemainder
}
candidate = e.table[nextHash&tableMask]
now := load3232(src, nextS)
e.table[nextHash&tableMask] = tableEntry{offset: s + e.cur, val: cv}
nextHash = hash(now)
offset := s - (candidate.offset - e.cur)
if offset > maxMatchOffset || cv != candidate.val {
// Out of range or not matched.
cv = now
continue
}
break
}
// A 4-byte match has been found. We'll later see if more than 4 bytes
// match. But, prior to the match, src[nextEmit:s] are unmatched. Emit
// them as literal bytes.
emitLiteral(dst, src[nextEmit:s])
// Call emitCopy, and then see if another emitCopy could be our next
// move. Repeat until we find no match for the input immediately after
// what was consumed by the last emitCopy call.
//
// If we exit this loop normally then we need to call emitLiteral next,
// though we don't yet know how big the literal will be. We handle that
// by proceeding to the next iteration of the main loop. We also can
// exit this loop via goto if we get close to exhausting the input.
for {
// Invariant: we have a 4-byte match at s, and no need to emit any
// literal bytes prior to s.
// Extend the 4-byte match as long as possible.
//
s += 4
t := candidate.offset - e.cur + 4
l := e.matchlen(s, t, src)
// matchToken is flate's equivalent of Snappy's emitCopy. (length,offset)
dst.tokens[dst.n] = matchToken(uint32(l+4-baseMatchLength), uint32(s-t-baseMatchOffset))
dst.n++
s += l
nextEmit = s
if s >= sLimit {
t += l
// Index first pair after match end.
if int(t+4) < len(src) && t > 0 {
cv := load3232(src, t)
e.table[hash(cv)&tableMask] = tableEntry{offset: t + e.cur, val: cv}
}
goto emitRemainder
}
// We could immediately start working at s now, but to improve
// compression we first update the hash table at s-1 and at s. If
// another emitCopy is not our next move, also calculate nextHash
// at s+1. At least on GOARCH=amd64, these three hash calculations
// are faster as one load64 call (with some shifts) instead of
// three load32 calls.
x := load6432(src, s-1)
prevHash := hash(uint32(x))
e.table[prevHash&tableMask] = tableEntry{offset: e.cur + s - 1, val: uint32(x)}
x >>= 8
currHash := hash(uint32(x))
candidate = e.table[currHash&tableMask]
e.table[currHash&tableMask] = tableEntry{offset: e.cur + s, val: uint32(x)}
offset := s - (candidate.offset - e.cur)
if offset > maxMatchOffset || uint32(x) != candidate.val {
cv = uint32(x >> 8)
nextHash = hash(cv)
s++
break
}
}
}
emitRemainder:
if int(nextEmit) < len(src) {
emitLiteral(dst, src[nextEmit:])
}
e.cur += int32(len(src))
e.prev = e.prev[:len(src)]
copy(e.prev, src)
}
type tableEntryPrev struct {
Cur tableEntry
Prev tableEntry
}
// snappyL3
type snappyL3 struct {
snappyGen
table [tableSize]tableEntryPrev
}
// Encode uses a similar algorithm to level 2, will check up to two candidates.
func (e *snappyL3) Encode(dst *tokens, src []byte) {
const (
inputMargin = 8 - 1
minNonLiteralBlockSize = 1 + 1 + inputMargin
)
// Protect against e.cur wraparound.
if e.cur > 1<<30 {
for i := range e.table[:] {
e.table[i] = tableEntryPrev{}
}
e.snappyGen = snappyGen{cur: maxStoreBlockSize, prev: e.prev[:0]}
}
// This check isn't in the Snappy implementation, but there, the caller
// instead of the callee handles this case.
if len(src) < minNonLiteralBlockSize {
// We do not fill the token table.
// This will be picked up by caller.
dst.n = uint16(len(src))
e.cur += maxStoreBlockSize
e.prev = e.prev[:0]
return
}
// sLimit is when to stop looking for offset/length copies. The inputMargin
// lets us use a fast path for emitLiteral in the main loop, while we are
// looking for copies.
sLimit := int32(len(src) - inputMargin)
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := int32(0)
s := int32(0)
cv := load3232(src, s)
nextHash := hash(cv)
for {
// Copied from the C++ snappy implementation:
//
// Heuristic match skipping: If 32 bytes are scanned with no matches
// found, start looking only at every other byte. If 32 more bytes are
// scanned (or skipped), look at every third byte, etc.. When a match
// is found, immediately go back to looking at every byte. This is a
// small loss (~5% performance, ~0.1% density) for compressible data
// due to more bookkeeping, but for non-compressible data (such as
// JPEG) it's a huge win since the compressor quickly "realizes" the
// data is incompressible and doesn't bother looking for matches
// everywhere.
//
// The "skip" variable keeps track of how many bytes there are since
// the last match; dividing it by 32 (ie. right-shifting by five) gives
// the number of bytes to move ahead for each iteration.
skip := int32(32)
nextS := s
var candidate tableEntry
for {
s = nextS
bytesBetweenHashLookups := skip >> 5
nextS = s + bytesBetweenHashLookups
skip += bytesBetweenHashLookups
if nextS > sLimit {
goto emitRemainder
}
candidates := e.table[nextHash&tableMask]
now := load3232(src, nextS)
e.table[nextHash&tableMask] = tableEntryPrev{Prev: candidates.Cur, Cur: tableEntry{offset: s + e.cur, val: cv}}
nextHash = hash(now)
// Check both candidates
candidate = candidates.Cur
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
break
}
} else {
// We only check if value mismatches.
// Offset will always be invalid in other cases.
candidate = candidates.Prev
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
break
}
}
}
cv = now
}
// A 4-byte match has been found. We'll later see if more than 4 bytes
// match. But, prior to the match, src[nextEmit:s] are unmatched. Emit
// them as literal bytes.
emitLiteral(dst, src[nextEmit:s])
// Call emitCopy, and then see if another emitCopy could be our next
// move. Repeat until we find no match for the input immediately after
// what was consumed by the last emitCopy call.
//
// If we exit this loop normally then we need to call emitLiteral next,
// though we don't yet know how big the literal will be. We handle that
// by proceeding to the next iteration of the main loop. We also can
// exit this loop via goto if we get close to exhausting the input.
for {
// Invariant: we have a 4-byte match at s, and no need to emit any
// literal bytes prior to s.
// Extend the 4-byte match as long as possible.
//
s += 4
t := candidate.offset - e.cur + 4
l := e.matchlen(s, t, src)
// matchToken is flate's equivalent of Snappy's emitCopy. (length,offset)
dst.tokens[dst.n] = matchToken(uint32(l+4-baseMatchLength), uint32(s-t-baseMatchOffset))
dst.n++
s += l
nextEmit = s
if s >= sLimit {
t += l
// Index first pair after match end.
if int(t+4) < len(src) && t > 0 {
cv := load3232(src, t)
nextHash = hash(cv)
e.table[nextHash&tableMask] = tableEntryPrev{
Prev: e.table[nextHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + t, val: cv},
}
}
goto emitRemainder
}
// We could immediately start working at s now, but to improve
// compression we first update the hash table at s-3 to s. If
// another emitCopy is not our next move, also calculate nextHash
// at s+1. At least on GOARCH=amd64, these three hash calculations
// are faster as one load64 call (with some shifts) instead of
// three load32 calls.
x := load6432(src, s-3)
prevHash := hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 3, val: uint32(x)},
}
x >>= 8
prevHash = hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 2, val: uint32(x)},
}
x >>= 8
prevHash = hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 1, val: uint32(x)},
}
x >>= 8
currHash := hash(uint32(x))
candidates := e.table[currHash&tableMask]
cv = uint32(x)
e.table[currHash&tableMask] = tableEntryPrev{
Prev: candidates.Cur,
Cur: tableEntry{offset: s + e.cur, val: cv},
}
// Check both candidates
candidate = candidates.Cur
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
continue
}
} else {
// We only check if value mismatches.
// Offset will always be invalid in other cases.
candidate = candidates.Prev
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
continue
}
}
}
cv = uint32(x >> 8)
nextHash = hash(cv)
s++
break
}
}
emitRemainder:
if int(nextEmit) < len(src) {
emitLiteral(dst, src[nextEmit:])
}
e.cur += int32(len(src))
e.prev = e.prev[:len(src)]
copy(e.prev, src)
}
// snappyL4
type snappyL4 struct {
snappyL3
}
// Encode uses a similar algorithm to level 3,
// but will check up to two candidates if first isn't long enough.
func (e *snappyL4) Encode(dst *tokens, src []byte) {
const (
inputMargin = 8 - 3
minNonLiteralBlockSize = 1 + 1 + inputMargin
matchLenGood = 12
)
// Protect against e.cur wraparound.
if e.cur > 1<<30 {
for i := range e.table[:] {
e.table[i] = tableEntryPrev{}
}
e.snappyGen = snappyGen{cur: maxStoreBlockSize, prev: e.prev[:0]}
}
// This check isn't in the Snappy implementation, but there, the caller
// instead of the callee handles this case.
if len(src) < minNonLiteralBlockSize {
// We do not fill the token table.
// This will be picked up by caller.
dst.n = uint16(len(src))
e.cur += maxStoreBlockSize
e.prev = e.prev[:0]
return
}
// sLimit is when to stop looking for offset/length copies. The inputMargin
// lets us use a fast path for emitLiteral in the main loop, while we are
// looking for copies.
sLimit := int32(len(src) - inputMargin)
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := int32(0)
s := int32(0)
cv := load3232(src, s)
nextHash := hash(cv)
for {
// Copied from the C++ snappy implementation:
//
// Heuristic match skipping: If 32 bytes are scanned with no matches
// found, start looking only at every other byte. If 32 more bytes are
// scanned (or skipped), look at every third byte, etc.. When a match
// is found, immediately go back to looking at every byte. This is a
// small loss (~5% performance, ~0.1% density) for compressible data
// due to more bookkeeping, but for non-compressible data (such as
// JPEG) it's a huge win since the compressor quickly "realizes" the
// data is incompressible and doesn't bother looking for matches
// everywhere.
//
// The "skip" variable keeps track of how many bytes there are since
// the last match; dividing it by 32 (ie. right-shifting by five) gives
// the number of bytes to move ahead for each iteration.
skip := int32(32)
nextS := s
var candidate tableEntry
var candidateAlt tableEntry
for {
s = nextS
bytesBetweenHashLookups := skip >> 5
nextS = s + bytesBetweenHashLookups
skip += bytesBetweenHashLookups
if nextS > sLimit {
goto emitRemainder
}
candidates := e.table[nextHash&tableMask]
now := load3232(src, nextS)
e.table[nextHash&tableMask] = tableEntryPrev{Prev: candidates.Cur, Cur: tableEntry{offset: s + e.cur, val: cv}}
nextHash = hash(now)
// Check both candidates
candidate = candidates.Cur
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset < maxMatchOffset {
offset = s - (candidates.Prev.offset - e.cur)
if cv == candidates.Prev.val && offset < maxMatchOffset {
candidateAlt = candidates.Prev
}
break
}
} else {
// We only check if value mismatches.
// Offset will always be invalid in other cases.
candidate = candidates.Prev
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset < maxMatchOffset {
break
}
}
}
cv = now
}
// A 4-byte match has been found. We'll later see if more than 4 bytes
// match. But, prior to the match, src[nextEmit:s] are unmatched. Emit
// them as literal bytes.
emitLiteral(dst, src[nextEmit:s])
// Call emitCopy, and then see if another emitCopy could be our next
// move. Repeat until we find no match for the input immediately after
// what was consumed by the last emitCopy call.
//
// If we exit this loop normally then we need to call emitLiteral next,
// though we don't yet know how big the literal will be. We handle that
// by proceeding to the next iteration of the main loop. We also can
// exit this loop via goto if we get close to exhausting the input.
for {
// Invariant: we have a 4-byte match at s, and no need to emit any
// literal bytes prior to s.
// Extend the 4-byte match as long as possible.
//
s += 4
t := candidate.offset - e.cur + 4
l := e.matchlen(s, t, src)
// Try alternative candidate if match length < matchLenGood.
if l < matchLenGood-4 && candidateAlt.offset != 0 {
t2 := candidateAlt.offset - e.cur + 4
l2 := e.matchlen(s, t2, src)
if l2 > l {
l = l2
t = t2
}
}
// matchToken is flate's equivalent of Snappy's emitCopy. (length,offset)
dst.tokens[dst.n] = matchToken(uint32(l+4-baseMatchLength), uint32(s-t-baseMatchOffset))
dst.n++
s += l
nextEmit = s
if s >= sLimit {
t += l
// Index first pair after match end.
if int(t+4) < len(src) && t > 0 {
cv := load3232(src, t)
nextHash = hash(cv)
e.table[nextHash&tableMask] = tableEntryPrev{
Prev: e.table[nextHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + t, val: cv},
}
}
goto emitRemainder
}
// We could immediately start working at s now, but to improve
// compression we first update the hash table at s-3 to s. If
// another emitCopy is not our next move, also calculate nextHash
// at s+1. At least on GOARCH=amd64, these three hash calculations
// are faster as one load64 call (with some shifts) instead of
// three load32 calls.
x := load6432(src, s-3)
prevHash := hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 3, val: uint32(x)},
}
x >>= 8
prevHash = hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 2, val: uint32(x)},
}
x >>= 8
prevHash = hash(uint32(x))
e.table[prevHash&tableMask] = tableEntryPrev{
Prev: e.table[prevHash&tableMask].Cur,
Cur: tableEntry{offset: e.cur + s - 1, val: uint32(x)},
}
x >>= 8
currHash := hash(uint32(x))
candidates := e.table[currHash&tableMask]
cv = uint32(x)
e.table[currHash&tableMask] = tableEntryPrev{
Prev: candidates.Cur,
Cur: tableEntry{offset: s + e.cur, val: cv},
}
// Check both candidates
candidate = candidates.Cur
candidateAlt = tableEntry{}
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
offset = s - (candidates.Prev.offset - e.cur)
if cv == candidates.Prev.val && offset <= maxMatchOffset {
candidateAlt = candidates.Prev
}
continue
}
} else {
// We only check if value mismatches.
// Offset will always be invalid in other cases.
candidate = candidates.Prev
if cv == candidate.val {
offset := s - (candidate.offset - e.cur)
if offset <= maxMatchOffset {
continue
}
}
}
cv = uint32(x >> 8)
nextHash = hash(cv)
s++
break
}
}
emitRemainder:
if int(nextEmit) < len(src) {
emitLiteral(dst, src[nextEmit:])
}
e.cur += int32(len(src))
e.prev = e.prev[:len(src)]
copy(e.prev, src)
}
func (e *snappyGen) matchlen(s, t int32, src []byte) int32 {
s1 := int(s) + maxMatchLength - 4
if s1 > len(src) {
s1 = len(src)
}
// If we are inside the current block
if t >= 0 {
b := src[t:]
a := src[s:s1]
b = b[:len(a)]
// Extend the match to be as long as possible.
for i := range a {
if a[i] != b[i] {
return int32(i)
}
}
return int32(len(a))
}
// We found a match in the previous block.
tp := int32(len(e.prev)) + t
if tp < 0 {
return 0
}
// Extend the match to be as long as possible.
a := src[s:s1]
b := e.prev[tp:]
if len(b) > len(a) {
b = b[:len(a)]
}
a = a[:len(b)]
for i := range b {
if a[i] != b[i] {
return int32(i)
}
}
// If we reached our limit, we matched everything we are
// allowed to in the previous block and we return.
n := int32(len(b))
if int(s+n) == s1 {
return n
}
// Continue looking for more matches in the current block.
a = src[s+n : s1]
b = src[:len(a)]
for i := range a {
if a[i] != b[i] {
return int32(i) + n
}
}
return int32(len(a)) + n
}
// Reset the encoding table.
func (e *snappyGen) Reset() {
e.prev = e.prev[:0]
e.cur += maxMatchOffset
}

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.golden generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.in generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-null-max.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.golden generated vendored Normal file
View File

Binary file not shown.

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.in generated vendored Normal file
View File

@@ -0,0 +1 @@
3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825342117067982148086513282306647093844609550582231725359408128481117450284102701938521105559644622948954930381964428810975665933446128475648233786783165271201909145648566923460348610454326648213393607260249141273724587006606315588174881520920962829254091715364367892590360011330530548820466521384146951941511609433057270365759591953092186117381932611793105118548074462379962749567351885752724891227938183011949129833673362440656643086021394946395224737190702179860943702770539217176293176752384674818467669405132000568127145263560827785771342757789609173637178721468440901224953430146549585371050792279689258923542019956112129021960864034418159813629774771309960518707211349999998372978049951059731732816096318595024459455346908302642522308253344685035261931188171010003137838752886587533208381420617177669147303598253490428755468731159562863882353787593751957781857780532171226806613001927876611195909216420198938095257201065485863278865936153381827968230301952035301852968995773622599413891249721775283479131515574857242454150695950829533116861727855889075098381754637464939319255060400927701671139009848824012858361603563707660104710181942955596198946767837449448255379774726847104047534646208046684259069491293313677028989152104752162056966024058038150193511253382430035587640247496473263914199272604269922796782354781636009341721641219924586315030286182974555706749838505494588586926995690927210797509302955321165344987202755960236480665499119881834797753566369807426542527862551818417574672890977772793800081647060016145249192173217214772350141441973568548161361157352552133475741849468438523323907394143334547762416862518983569485562099219222184272550254256887671790494601653466804988627232791786085784383827967976681454100953883786360950680064225125205117392984896084128488626945604241965285022210661186306744278622039194945047123713786960956364371917287467764657573962413890865832645995813390478027590099465764078951269468398352595709825822620522489407726719478268482601476990902640136394437455305068203496252451749399651431429809190659250937221696461515709858387410597885959772975498930161753928468138268683868942774155991855925245953959431049972524680845987273644695848653836736222626099124608051243884390451244136549762780797715691435997700129616089441694868555848406353422072225828488648158456028506016842739452267467678895252138522549954666727823986456596116354886230577456498035593634568174324112515076069479451096596094025228879710893145669136867228748940560101503308617928680920874760917824938589009714909675985261365549781893129784821682998948722658804857564014270477555132379641451523746234364542858444795265867821051141354735739523113427166102135969536231442952484937187110145765403590279934403742007310578539062198387447808478489683321445713868751943506430218453191048481005370614680674919278191197939952061419663428754440643745123718192179998391015919561814675142691239748940907186494231961567945208095146550225231603881930142093762137855956638937787083039069792077346722182562599661501421503068038447734549202605414665925201497442850732518666002132434088190710486331734649651453905796268561005508106658796998163574736384052571459102897064140110971206280439039759515677157700420337869936007230558763176359421873125147120532928191826186125867321579198414848829164470609575270695722091756711672291098169091528017350671274858322287183520935396572512108357915136988209144421006751033467110314126711136990865851639831501970165151168517143765761835155650884909989859982387345528331635507647918535893226185489632132933089857064204675259070915481416549859461637180

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-pi.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.golden generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.in generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-1k.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.golden generated vendored Normal file
View File

Binary file not shown.

4
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.in generated vendored Normal file
View File

@@ -0,0 +1,4 @@
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
<EFBFBD><EFBFBD><EFBFBD>vH
<EFBFBD><EFBFBD>%<25><><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><17>ɷ<EFBFBD><C9B7><06>}<18><>><07><><EFBFBD>ls<6C><73><EFBFBD>m<EFBFBD>IGH<15><><EFBFBD><EFBFBD>1Y<31>4<EFBFBD>[<5B><> 0ˆ[|]o#<23>
<EFBFBD>-#<23><><EFBFBD>ul<><6C><EFBFBD>pf<70><66>ٱ<EFBFBD>n<EFBFBD>Y<EFBFBD>ԀY<D480>w<EFBFBD>C8ɯ02<30> F=gn<67>r<EFBFBD>N!O<><4F><EFBFBD>{<04><><03><05>k<EFBFBD>*<2A>w(<28><>b<EFBFBD> <20><1F>kQC9/<2F><>lu><3E>5<EFBFBD>C.<2E><>u<EFBFBD><75><EFBFBD>

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-limit.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-max.golden generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-rand-max.in generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.golden generated vendored Normal file
View File

Binary file not shown.

2
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.in generated vendored Normal file
View File

@@ -0,0 +1,2 @@
101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010101010
232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323232323

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-shifts.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.golden generated vendored Normal file
View File

Binary file not shown.

14
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.in generated vendored Normal file
View File

@@ -0,0 +1,14 @@
//Copyright2009ThGoAuthor.Allrightrrvd.
//UofthiourccodigovrndbyBSD-tyl
//licnthtcnbfoundinthLICENSEfil.
pckgmin
import"o"
funcmin(){
vrb=mk([]byt,65535)
f,_:=o.Crt("huffmn-null-mx.in")
f.Writ(b)
}
ABCDEFGHIJKLMNOPQRSTUVXxyz!"#¤%&/?"

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-text-shift.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.dyn.expect generated vendored Normal file
View File

@@ -0,0 +1 @@
<1C>_K<5F>0<1C><><EFBFBD><05><0E><>`K<><4B>0Aasě)^<5E>H<EFBFBD><48><EFBFBD><EFBFBD><EFBFBD>Iɟb߻<><DFBB>_><3E>4

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.dyn.expect-noinput generated vendored Normal file
View File

@@ -0,0 +1 @@
<1C>_K<5F>0<1C><><EFBFBD><05><0E><>`K<><4B>0Aasě)^<5E>H<EFBFBD><48><EFBFBD><EFBFBD><EFBFBD>Iɟb߻<><DFBB>_><3E>4

3
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.golden generated vendored Normal file
View File

@@ -0,0 +1,3 @@
<04>AK<41>0<07><>x<>ß<EFBFBD>Z<EFBFBD><5A><EFBFBD><EFBFBD>LP<4C>a<EFBFBD>!<21>x<EFBFBD><78>AD<41><44>I<13>&#I<>E<EFBFBD><45><EFBFBD><EFBFBD><06>p]<5D>Lƿ<4C><C6BF><16>F<EFBFBD>p<><70> 1<>88<38>h<07><13>$<24><><EFBFBD>5S<35><53>- <09>F66!<21>)v<>.<2E><02>0<EFBFBD>Y<EFBFBD><59><1E><02><><EFBFBD><EFBFBD>&<26><> S<><53><EFBFBD>N|d<>2:<3A><>
t<EFBFBD>|둍<><EB918D><EFBFBD>xz9<7A><39><EFBFBD><EFBFBD><EFBFBD><EFBFBD>骺<EFBFBD><04><><EFBFBD><EFBFBD>Ɏ<EFBFBD>3<><33>
&&=<3D><0E><><EFBFBD><EFBFBD><EFBFBD>ô<EFBFBD>UD<55>=Fu<46><75><EFBFBD><EFBFBD>]<5D><>q<EFBFBD><71><EFBFBD><EFBFBD>UL+<2B><><17><><08>>FQY<51><59>LZ<4C><5A>o<EFBFBD><6F><EFBFBD>fTߵ<54>EŴ<45><C5B4>{<7B>Yʶb<CAB6>e<EFBFBD>

13
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.in generated vendored Normal file
View File

@@ -0,0 +1,13 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package main
import "os"
func main() {
var b = make([]byte, 65535)
f, _ := os.Create("huffman-null-max.in")
f.Write(b)
}

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.wb.expect generated vendored Normal file
View File

@@ -0,0 +1 @@
<1C>_K<5F>0<1C><><EFBFBD><05><0E><>`K<><4B>0Aasě)^<5E>H<EFBFBD><48><EFBFBD><EFBFBD><EFBFBD>Iɟb߻<><DFBB>_><3E>4

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-text.wb.expect-noinput generated vendored Normal file
View File

@@ -0,0 +1 @@
<1C>_K<5F>0<1C><><EFBFBD><05><0E><>`K<><4B>0Aasě)^<5E>H<EFBFBD><48><EFBFBD><EFBFBD><EFBFBD>Iɟb߻<><DFBB>_><3E>4

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.dyn.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.golden generated vendored Normal file
View File

Binary file not shown.

1
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.in generated vendored Normal file
View File

@@ -0,0 +1 @@
00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.wb.expect generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/huffman-zero.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/null-long-match.dyn.expect-noinput generated vendored Normal file
View File

Binary file not shown.

BIN
vendor/github.com/klauspost/compress/flate/testdata/null-long-match.wb.expect-noinput generated vendored Normal file
View File

Binary file not shown.

115
vendor/github.com/klauspost/compress/flate/token.go generated vendored Normal file
View File

@@ -0,0 +1,115 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import "fmt"
const (
// 2 bits: type 0 = literal 1=EOF 2=Match 3=Unused
// 8 bits: xlength = length - MIN_MATCH_LENGTH
// 22 bits xoffset = offset - MIN_OFFSET_SIZE, or literal
lengthShift = 22
offsetMask = 1<<lengthShift - 1
typeMask = 3 << 30
literalType = 0 << 30
matchType = 1 << 30
)
// The length code for length X (MIN_MATCH_LENGTH <= X <= MAX_MATCH_LENGTH)
// is lengthCodes[length - MIN_MATCH_LENGTH]
var lengthCodes = [...]uint32{
0, 1, 2, 3, 4, 5, 6, 7, 8, 8,
9, 9, 10, 10, 11, 11, 12, 12, 12, 12,
13, 13, 13, 13, 14, 14, 14, 14, 15, 15,
15, 15, 16, 16, 16, 16, 16, 16, 16, 16,
17, 17, 17, 17, 17, 17, 17, 17, 18, 18,
18, 18, 18, 18, 18, 18, 19, 19, 19, 19,
19, 19, 19, 19, 20, 20, 20, 20, 20, 20,
20, 20, 20, 20, 20, 20, 20, 20, 20, 20,
21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
21, 21, 21, 21, 21, 21, 22, 22, 22, 22,
22, 22, 22, 22, 22, 22, 22, 22, 22, 22,
22, 22, 23, 23, 23, 23, 23, 23, 23, 23,
23, 23, 23, 23, 23, 23, 23, 23, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
25, 25, 25, 25, 25, 25, 25, 25, 25, 25,
25, 25, 25, 25, 25, 25, 25, 25, 25, 25,
25, 25, 25, 25, 25, 25, 25, 25, 25, 25,
25, 25, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 26, 26, 26, 26, 26, 26,
26, 26, 26, 26, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
27, 27, 27, 27, 27, 28,
}
var offsetCodes = [...]uint32{
0, 1, 2, 3, 4, 4, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7,
8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9,
10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
}
type token uint32
type tokens struct {
tokens [maxStoreBlockSize + 1]token
n uint16 // Must be able to contain maxStoreBlockSize
}
// Convert a literal into a literal token.
func literalToken(literal uint32) token { return token(literalType + literal) }
// Convert a < xlength, xoffset > pair into a match token.
func matchToken(xlength uint32, xoffset uint32) token {
return token(matchType + xlength<<lengthShift + xoffset)
}
func matchTokend(xlength uint32, xoffset uint32) token {
if xlength > maxMatchLength || xoffset > maxMatchOffset {
panic(fmt.Sprintf("Invalid match: len: %d, offset: %d\n", xlength, xoffset))
return token(matchType)
}
return token(matchType + xlength<<lengthShift + xoffset)
}
// Returns the type of a token
func (t token) typ() uint32 { return uint32(t) & typeMask }
// Returns the literal of a literal token
func (t token) literal() uint32 { return uint32(t - literalType) }
// Returns the extra offset of a match token
func (t token) offset() uint32 { return uint32(t) & offsetMask }
func (t token) length() uint32 { return uint32((t - matchType) >> lengthShift) }
func lengthCode(len uint32) uint32 { return lengthCodes[len] }
// Returns the offset code corresponding to a specific offset
func offsetCode(off uint32) uint32 {
if off < uint32(len(offsetCodes)) {
return offsetCodes[off]
} else if off>>7 < uint32(len(offsetCodes)) {
return offsetCodes[off>>7] + 14
} else {
return offsetCodes[off>>14] + 28
}
}

258
vendor/github.com/klauspost/compress/flate/writer_test.go generated vendored Normal file
View File

@@ -0,0 +1,258 @@
// Copyright 2012 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package flate
import (
"bytes"
"fmt"
"io"
"io/ioutil"
"math/rand"
"runtime"
"testing"
)
func benchmarkEncoder(b *testing.B, testfile, level, n int) {
b.StopTimer()
b.SetBytes(int64(n))
buf0, err := ioutil.ReadFile(testfiles[testfile])
if err != nil {
b.Fatal(err)
}
if len(buf0) == 0 {
b.Fatalf("test file %q has no data", testfiles[testfile])
}
buf1 := make([]byte, n)
for i := 0; i < n; i += len(buf0) {
if len(buf0) > n-i {
buf0 = buf0[:n-i]
}
copy(buf1[i:], buf0)
}
buf0 = nil
runtime.GC()
w, err := NewWriter(ioutil.Discard, level)
b.StartTimer()
for i := 0; i < b.N; i++ {
w.Reset(ioutil.Discard)
_, err = w.Write(buf1)
if err != nil {
b.Fatal(err)
}
err = w.Close()
if err != nil {
b.Fatal(err)
}
}
}
func BenchmarkEncodeDigitsConstant1e4(b *testing.B) { benchmarkEncoder(b, digits, constant, 1e4) }
func BenchmarkEncodeDigitsConstant1e5(b *testing.B) { benchmarkEncoder(b, digits, constant, 1e5) }
func BenchmarkEncodeDigitsConstant1e6(b *testing.B) { benchmarkEncoder(b, digits, constant, 1e6) }
func BenchmarkEncodeDigitsSpeed1e4(b *testing.B) { benchmarkEncoder(b, digits, speed, 1e4) }
func BenchmarkEncodeDigitsSpeed1e5(b *testing.B) { benchmarkEncoder(b, digits, speed, 1e5) }
func BenchmarkEncodeDigitsSpeed1e6(b *testing.B) { benchmarkEncoder(b, digits, speed, 1e6) }
func BenchmarkEncodeDigitsDefault1e4(b *testing.B) { benchmarkEncoder(b, digits, default_, 1e4) }
func BenchmarkEncodeDigitsDefault1e5(b *testing.B) { benchmarkEncoder(b, digits, default_, 1e5) }
func BenchmarkEncodeDigitsDefault1e6(b *testing.B) { benchmarkEncoder(b, digits, default_, 1e6) }
func BenchmarkEncodeDigitsCompress1e4(b *testing.B) { benchmarkEncoder(b, digits, compress, 1e4) }
func BenchmarkEncodeDigitsCompress1e5(b *testing.B) { benchmarkEncoder(b, digits, compress, 1e5) }
func BenchmarkEncodeDigitsCompress1e6(b *testing.B) { benchmarkEncoder(b, digits, compress, 1e6) }
func BenchmarkEncodeTwainConstant1e4(b *testing.B) { benchmarkEncoder(b, twain, constant, 1e4) }
func BenchmarkEncodeTwainConstant1e5(b *testing.B) { benchmarkEncoder(b, twain, constant, 1e5) }
func BenchmarkEncodeTwainConstant1e6(b *testing.B) { benchmarkEncoder(b, twain, constant, 1e6) }
func BenchmarkEncodeTwainSpeed1e4(b *testing.B) { benchmarkEncoder(b, twain, speed, 1e4) }
func BenchmarkEncodeTwainSpeed1e5(b *testing.B) { benchmarkEncoder(b, twain, speed, 1e5) }
func BenchmarkEncodeTwainSpeed1e6(b *testing.B) { benchmarkEncoder(b, twain, speed, 1e6) }
func BenchmarkEncodeTwainDefault1e4(b *testing.B) { benchmarkEncoder(b, twain, default_, 1e4) }
func BenchmarkEncodeTwainDefault1e5(b *testing.B) { benchmarkEncoder(b, twain, default_, 1e5) }
func BenchmarkEncodeTwainDefault1e6(b *testing.B) { benchmarkEncoder(b, twain, default_, 1e6) }
func BenchmarkEncodeTwainCompress1e4(b *testing.B) { benchmarkEncoder(b, twain, compress, 1e4) }
func BenchmarkEncodeTwainCompress1e5(b *testing.B) { benchmarkEncoder(b, twain, compress, 1e5) }
func BenchmarkEncodeTwainCompress1e6(b *testing.B) { benchmarkEncoder(b, twain, compress, 1e6) }
// A writer that fails after N writes.
type errorWriter struct {
N int
}
func (e *errorWriter) Write(b []byte) (int, error) {
if e.N <= 0 {
return 0, io.ErrClosedPipe
}
e.N--
return len(b), nil
}
// Test if errors from the underlying writer is passed upwards.
func TestWriteError(t *testing.T) {
buf := new(bytes.Buffer)
n := 65536
if !testing.Short() {
n *= 4
}
for i := 0; i < n; i++ {
fmt.Fprintf(buf, "asdasfasf%d%dfghfgujyut%dyutyu\n", i, i, i)
}
in := buf.Bytes()
// We create our own buffer to control number of writes.
copyBuf := make([]byte, 128)
for l := 0; l < 10; l++ {
for fail := 1; fail <= 256; fail *= 2 {
// Fail after 'fail' writes
ew := &errorWriter{N: fail}
w, err := NewWriter(ew, l)
if err != nil {
t.Fatalf("NewWriter: level %d: %v", l, err)
}
n, err := copyBuffer(w, bytes.NewBuffer(in), copyBuf)
if err == nil {
t.Fatalf("Level %d: Expected an error, writer was %#v", l, ew)
}
n2, err := w.Write([]byte{1, 2, 2, 3, 4, 5})
if n2 != 0 {
t.Fatal("Level", l, "Expected 0 length write, got", n)
}
if err == nil {
t.Fatal("Level", l, "Expected an error")
}
err = w.Flush()
if err == nil {
t.Fatal("Level", l, "Expected an error on flush")
}
err = w.Close()
if err == nil {
t.Fatal("Level", l, "Expected an error on close")
}
w.Reset(ioutil.Discard)
n2, err = w.Write([]byte{1, 2, 3, 4, 5, 6})
if err != nil {
t.Fatal("Level", l, "Got unexpected error after reset:", err)
}
if n2 == 0 {
t.Fatal("Level", l, "Got 0 length write, expected > 0")
}
if testing.Short() {
return
}
}
}
}
func TestDeterministicL1(t *testing.T) { testDeterministic(1, t) }
func TestDeterministicL2(t *testing.T) { testDeterministic(2, t) }
func TestDeterministicL3(t *testing.T) { testDeterministic(3, t) }
func TestDeterministicL4(t *testing.T) { testDeterministic(4, t) }
func TestDeterministicL5(t *testing.T) { testDeterministic(5, t) }
func TestDeterministicL6(t *testing.T) { testDeterministic(6, t) }
func TestDeterministicL7(t *testing.T) { testDeterministic(7, t) }
func TestDeterministicL8(t *testing.T) { testDeterministic(8, t) }
func TestDeterministicL9(t *testing.T) { testDeterministic(9, t) }
func TestDeterministicL0(t *testing.T) { testDeterministic(0, t) }
func TestDeterministicLM2(t *testing.T) { testDeterministic(-2, t) }
func testDeterministic(i int, t *testing.T) {
// Test so much we cross a good number of block boundaries.
var length = maxStoreBlockSize*30 + 500
if testing.Short() {
length /= 10
}
// Create a random, but compressible stream.
rng := rand.New(rand.NewSource(1))
t1 := make([]byte, length)
for i := range t1 {
t1[i] = byte(rng.Int63() & 7)
}
// Do our first encode.
var b1 bytes.Buffer
br := bytes.NewBuffer(t1)
w, err := NewWriter(&b1, i)
if err != nil {
t.Fatal(err)
}
// Use a very small prime sized buffer.
cbuf := make([]byte, 787)
_, err = copyBuffer(w, br, cbuf)
if err != nil {
t.Fatal(err)
}
w.Close()
// We choose a different buffer size,
// bigger than a maximum block, and also a prime.
var b2 bytes.Buffer
cbuf = make([]byte, 81761)
br2 := bytes.NewBuffer(t1)
w2, err := NewWriter(&b2, i)
if err != nil {
t.Fatal(err)
}
_, err = copyBuffer(w2, br2, cbuf)
if err != nil {
t.Fatal(err)
}
w2.Close()
b1b := b1.Bytes()
b2b := b2.Bytes()
if !bytes.Equal(b1b, b2b) {
t.Errorf("level %d did not produce deterministic result, result mismatch, len(a) = %d, len(b) = %d", i, len(b1b), len(b2b))
}
// Test using io.WriterTo interface.
var b3 bytes.Buffer
br = bytes.NewBuffer(t1)
w, err = NewWriter(&b3, i)
if err != nil {
t.Fatal(err)
}
_, err = br.WriteTo(w)
if err != nil {
t.Fatal(err)
}
w.Close()
b3b := b3.Bytes()
if !bytes.Equal(b1b, b3b) {
t.Errorf("level %d (io.WriterTo) did not produce deterministic result, result mismatch, len(a) = %d, len(b) = %d", i, len(b1b), len(b3b))
}
}
// copyBuffer is a copy of io.CopyBuffer, since we want to support older go versions.
// This is modified to never use io.WriterTo or io.ReaderFrom interfaces.
func copyBuffer(dst io.Writer, src io.Reader, buf []byte) (written int64, err error) {
if buf == nil {
buf = make([]byte, 32*1024)
}
for {
nr, er := src.Read(buf)
if nr > 0 {
nw, ew := dst.Write(buf[0:nr])
if nw > 0 {
written += int64(nw)
}
if ew != nil {
err = ew
break
}
if nr != nw {
err = io.ErrShortWrite
break
}
}
if er == io.EOF {
break
}
if er != nil {
err = er
break
}
}
return written, err
}

128
vendor/github.com/klauspost/compress/gzip/example_test.go generated vendored Normal file
View File

@@ -0,0 +1,128 @@
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package gzip_test
import (
"bytes"
"compress/gzip"
"fmt"
"io"
"log"
"os"
"time"
)
func Example_writerReader() {
var buf bytes.Buffer
zw := gzip.NewWriter(&buf)
// Setting the Header fields is optional.
zw.Name = "a-new-hope.txt"
zw.Comment = "an epic space opera by George Lucas"
zw.ModTime = time.Date(1977, time.May, 25, 0, 0, 0, 0, time.UTC)
_, err := zw.Write([]byte("A long time ago in a galaxy far, far away..."))
if err != nil {
log.Fatal(err)
}
if err := zw.Close(); err != nil {
log.Fatal(err)
}
zr, err := gzip.NewReader(&buf)
if err != nil {
log.Fatal(err)
}
fmt.Printf("Name: %s\nComment: %s\nModTime: %s\n\n", zr.Name, zr.Comment, zr.ModTime.UTC())
if _, err := io.Copy(os.Stdout, zr); err != nil {
log.Fatal(err)
}
if err := zr.Close(); err != nil {
log.Fatal(err)
}
// Output:
// Name: a-new-hope.txt
// Comment: an epic space opera by George Lucas
// ModTime: 1977-05-25 00:00:00 +0000 UTC
//
// A long time ago in a galaxy far, far away...
}
func ExampleReader_Multistream() {
var buf bytes.Buffer
zw := gzip.NewWriter(&buf)
var files = []struct {
name string
comment string
modTime time.Time
data string
}{
{"file-1.txt", "file-header-1", time.Date(2006, time.February, 1, 3, 4, 5, 0, time.UTC), "Hello Gophers - 1"},
{"file-2.txt", "file-header-2", time.Date(2007, time.March, 2, 4, 5, 6, 1, time.UTC), "Hello Gophers - 2"},
}
for _, file := range files {
zw.Name = file.name
zw.Comment = file.comment
zw.ModTime = file.modTime
if _, err := zw.Write([]byte(file.data)); err != nil {
log.Fatal(err)
}
if err := zw.Close(); err != nil {
log.Fatal(err)
}
zw.Reset(&buf)
}
zr, err := gzip.NewReader(&buf)
if err != nil {
log.Fatal(err)
}
for {
zr.Multistream(false)
fmt.Printf("Name: %s\nComment: %s\nModTime: %s\n\n", zr.Name, zr.Comment, zr.ModTime.UTC())
if _, err := io.Copy(os.Stdout, zr); err != nil {
log.Fatal(err)
}
fmt.Println("\n")
err = zr.Reset(&buf)
if err == io.EOF {
break
}
if err != nil {
log.Fatal(err)
}
}
if err := zr.Close(); err != nil {
log.Fatal(err)
}
// Output:
// Name: file-1.txt
// Comment: file-header-1
// ModTime: 2006-02-01 03:04:05 +0000 UTC
//
// Hello Gophers - 1
//
// Name: file-2.txt
// Comment: file-header-2
// ModTime: 2007-03-02 04:05:06 +0000 UTC
//
// Hello Gophers - 2
}

344
vendor/github.com/klauspost/compress/gzip/gunzip.go generated vendored Normal file
View File

@@ -0,0 +1,344 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// Package gzip implements reading and writing of gzip format compressed files,
// as specified in RFC 1952.
package gzip
import (
"bufio"
"encoding/binary"
"errors"
"io"
"time"
"github.com/klauspost/compress/flate"
"github.com/klauspost/crc32"
)
const (
gzipID1 = 0x1f
gzipID2 = 0x8b
gzipDeflate = 8
flagText = 1 << 0
flagHdrCrc = 1 << 1
flagExtra = 1 << 2
flagName = 1 << 3
flagComment = 1 << 4
)
var (
// ErrChecksum is returned when reading GZIP data that has an invalid checksum.
ErrChecksum = errors.New("gzip: invalid checksum")
// ErrHeader is returned when reading GZIP data that has an invalid header.
ErrHeader = errors.New("gzip: invalid header")
)
var le = binary.LittleEndian
// noEOF converts io.EOF to io.ErrUnexpectedEOF.
func noEOF(err error) error {
if err == io.EOF {
return io.ErrUnexpectedEOF
}
return err
}
// The gzip file stores a header giving metadata about the compressed file.
// That header is exposed as the fields of the Writer and Reader structs.
//
// Strings must be UTF-8 encoded and may only contain Unicode code points
// U+0001 through U+00FF, due to limitations of the GZIP file format.
type Header struct {
Comment string // comment
Extra []byte // "extra data"
ModTime time.Time // modification time
Name string // file name
OS byte // operating system type
}
// A Reader is an io.Reader that can be read to retrieve
// uncompressed data from a gzip-format compressed file.
//
// In general, a gzip file can be a concatenation of gzip files,
// each with its own header. Reads from the Reader
// return the concatenation of the uncompressed data of each.
// Only the first header is recorded in the Reader fields.
//
// Gzip files store a length and checksum of the uncompressed data.
// The Reader will return a ErrChecksum when Read
// reaches the end of the uncompressed data if it does not
// have the expected length or checksum. Clients should treat data
// returned by Read as tentative until they receive the io.EOF
// marking the end of the data.
type Reader struct {
Header // valid after NewReader or Reader.Reset
r flate.Reader
decompressor io.ReadCloser
digest uint32 // CRC-32, IEEE polynomial (section 8)
size uint32 // Uncompressed size (section 2.3.1)
buf [512]byte
err error
multistream bool
}
// NewReader creates a new Reader reading the given reader.
// If r does not also implement io.ByteReader,
// the decompressor may read more data than necessary from r.
//
// It is the caller's responsibility to call Close on the Reader when done.
//
// The Reader.Header fields will be valid in the Reader returned.
func NewReader(r io.Reader) (*Reader, error) {
z := new(Reader)
if err := z.Reset(r); err != nil {
return nil, err
}
return z, nil
}
// Reset discards the Reader z's state and makes it equivalent to the
// result of its original state from NewReader, but reading from r instead.
// This permits reusing a Reader rather than allocating a new one.
func (z *Reader) Reset(r io.Reader) error {
*z = Reader{
decompressor: z.decompressor,
multistream: true,
}
if rr, ok := r.(flate.Reader); ok {
z.r = rr
} else {
z.r = bufio.NewReader(r)
}
z.Header, z.err = z.readHeader()
return z.err
}
// Multistream controls whether the reader supports multistream files.
//
// If enabled (the default), the Reader expects the input to be a sequence
// of individually gzipped data streams, each with its own header and
// trailer, ending at EOF. The effect is that the concatenation of a sequence
// of gzipped files is treated as equivalent to the gzip of the concatenation
// of the sequence. This is standard behavior for gzip readers.
//
// Calling Multistream(false) disables this behavior; disabling the behavior
// can be useful when reading file formats that distinguish individual gzip
// data streams or mix gzip data streams with other data streams.
// In this mode, when the Reader reaches the end of the data stream,
// Read returns io.EOF. If the underlying reader implements io.ByteReader,
// it will be left positioned just after the gzip stream.
// To start the next stream, call z.Reset(r) followed by z.Multistream(false).
// If there is no next stream, z.Reset(r) will return io.EOF.
func (z *Reader) Multistream(ok bool) {
z.multistream = ok
}
// readString reads a NUL-terminated string from z.r.
// It treats the bytes read as being encoded as ISO 8859-1 (Latin-1) and
// will output a string encoded using UTF-8.
// This method always updates z.digest with the data read.
func (z *Reader) readString() (string, error) {
var err error
needConv := false
for i := 0; ; i++ {
if i >= len(z.buf) {
return "", ErrHeader
}
z.buf[i], err = z.r.ReadByte()
if err != nil {
return "", err
}
if z.buf[i] > 0x7f {
needConv = true
}
if z.buf[i] == 0 {
// Digest covers the NUL terminator.
z.digest = crc32.Update(z.digest, crc32.IEEETable, z.buf[:i+1])
// Strings are ISO 8859-1, Latin-1 (RFC 1952, section 2.3.1).
if needConv {
s := make([]rune, 0, i)
for _, v := range z.buf[:i] {
s = append(s, rune(v))
}
return string(s), nil
}
return string(z.buf[:i]), nil
}
}
}
// readHeader reads the GZIP header according to section 2.3.1.
// This method does not set z.err.
func (z *Reader) readHeader() (hdr Header, err error) {
if _, err = io.ReadFull(z.r, z.buf[:10]); err != nil {
// RFC 1952, section 2.2, says the following:
// A gzip file consists of a series of "members" (compressed data sets).
//
// Other than this, the specification does not clarify whether a
// "series" is defined as "one or more" or "zero or more". To err on the
// side of caution, Go interprets this to mean "zero or more".
// Thus, it is okay to return io.EOF here.
return hdr, err
}
if z.buf[0] != gzipID1 || z.buf[1] != gzipID2 || z.buf[2] != gzipDeflate {
return hdr, ErrHeader
}
flg := z.buf[3]
hdr.ModTime = time.Unix(int64(le.Uint32(z.buf[4:8])), 0)
// z.buf[8] is XFL and is currently ignored.
hdr.OS = z.buf[9]
z.digest = crc32.ChecksumIEEE(z.buf[:10])
if flg&flagExtra != 0 {
if _, err = io.ReadFull(z.r, z.buf[:2]); err != nil {
return hdr, noEOF(err)
}
z.digest = crc32.Update(z.digest, crc32.IEEETable, z.buf[:2])
data := make([]byte, le.Uint16(z.buf[:2]))
if _, err = io.ReadFull(z.r, data); err != nil {
return hdr, noEOF(err)
}
z.digest = crc32.Update(z.digest, crc32.IEEETable, data)
hdr.Extra = data
}
var s string
if flg&flagName != 0 {
if s, err = z.readString(); err != nil {
return hdr, err
}
hdr.Name = s
}
if flg&flagComment != 0 {
if s, err = z.readString(); err != nil {
return hdr, err
}
hdr.Comment = s
}
if flg&flagHdrCrc != 0 {
if _, err = io.ReadFull(z.r, z.buf[:2]); err != nil {
return hdr, noEOF(err)
}
digest := le.Uint16(z.buf[:2])
if digest != uint16(z.digest) {
return hdr, ErrHeader
}
}
z.digest = 0
if z.decompressor == nil {
z.decompressor = flate.NewReader(z.r)
} else {
z.decompressor.(flate.Resetter).Reset(z.r, nil)
}
return hdr, nil
}
// Read implements io.Reader, reading uncompressed bytes from its underlying Reader.
func (z *Reader) Read(p []byte) (n int, err error) {
if z.err != nil {
return 0, z.err
}
n, z.err = z.decompressor.Read(p)
z.digest = crc32.Update(z.digest, crc32.IEEETable, p[:n])
z.size += uint32(n)
if z.err != io.EOF {
// In the normal case we return here.
return n, z.err
}
// Finished file; check checksum and size.
if _, err := io.ReadFull(z.r, z.buf[:8]); err != nil {
z.err = noEOF(err)
return n, z.err
}
digest := le.Uint32(z.buf[:4])
size := le.Uint32(z.buf[4:8])
if digest != z.digest || size != z.size {
z.err = ErrChecksum
return n, z.err
}
z.digest, z.size = 0, 0
// File is ok; check if there is another.
if !z.multistream {
return n, io.EOF
}
z.err = nil // Remove io.EOF
if _, z.err = z.readHeader(); z.err != nil {
return n, z.err
}
// Read from next file, if necessary.
if n > 0 {
return n, nil
}
return z.Read(p)
}
// Support the io.WriteTo interface for io.Copy and friends.
func (z *Reader) WriteTo(w io.Writer) (int64, error) {
total := int64(0)
crcWriter := crc32.NewIEEE()
for {
if z.err != nil {
if z.err == io.EOF {
return total, nil
}
return total, z.err
}
// We write both to output and digest.
mw := io.MultiWriter(w, crcWriter)
n, err := z.decompressor.(io.WriterTo).WriteTo(mw)
total += n
z.size += uint32(n)
if err != nil {
z.err = err
return total, z.err
}
// Finished file; check checksum + size.
if _, err := io.ReadFull(z.r, z.buf[0:8]); err != nil {
if err == io.EOF {
err = io.ErrUnexpectedEOF
}
z.err = err
return total, err
}
z.digest = crcWriter.Sum32()
digest := le.Uint32(z.buf[:4])
size := le.Uint32(z.buf[4:8])
if digest != z.digest || size != z.size {
z.err = ErrChecksum
return total, z.err
}
z.digest, z.size = 0, 0
// File is ok; check if there is another.
if !z.multistream {
return total, nil
}
crcWriter.Reset()
z.err = nil // Remove io.EOF
if _, z.err = z.readHeader(); z.err != nil {
if z.err == io.EOF {
return total, nil
}
return total, z.err
}
}
}
// Close closes the Reader. It does not close the underlying io.Reader.
// In order for the GZIP checksum to be verified, the reader must be
// fully consumed until the io.EOF.
func (z *Reader) Close() error { return z.decompressor.Close() }

682
vendor/github.com/klauspost/compress/gzip/gunzip_test.go generated vendored Normal file
View File

@@ -0,0 +1,682 @@
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package gzip
import (
"bytes"
oldgz "compress/gzip"
"crypto/rand"
"io"
"io/ioutil"
"os"
"strings"
"testing"
"time"
"github.com/klauspost/compress/flate"
)
type gunzipTest struct {
name string
desc string
raw string
gzip []byte
err error
}
var gunzipTests = []gunzipTest{
{ // has 1 empty fixed-huffman block
"empty.txt",
"empty.txt",
"",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xf7, 0x5e, 0x14, 0x4a,
0x00, 0x03, 0x65, 0x6d, 0x70, 0x74, 0x79, 0x2e,
0x74, 0x78, 0x74, 0x00, 0x03, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
},
nil,
},
{
"",
"empty - with no file name",
"",
[]byte{
0x1f, 0x8b, 0x08, 0x00, 0x00, 0x09, 0x6e, 0x88,
0x00, 0xff, 0x01, 0x00, 0x00, 0xff, 0xff, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
},
nil,
},
{ // has 1 non-empty fixed huffman block
"hello.txt",
"hello.txt",
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0x0c, 0x00,
0x00, 0x00,
},
nil,
},
{ // concatenation
"hello.txt",
"hello.txt x2",
"hello world\n" +
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0x0c, 0x00,
0x00, 0x00,
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0x0c, 0x00,
0x00, 0x00,
},
nil,
},
{ // has a fixed huffman block with some length-distance pairs
"shesells.txt",
"shesells.txt",
"she sells seashells by the seashore\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0x72, 0x66, 0x8b, 0x4a,
0x00, 0x03, 0x73, 0x68, 0x65, 0x73, 0x65, 0x6c,
0x6c, 0x73, 0x2e, 0x74, 0x78, 0x74, 0x00, 0x2b,
0xce, 0x48, 0x55, 0x28, 0x4e, 0xcd, 0xc9, 0x29,
0x06, 0x92, 0x89, 0xc5, 0x19, 0x60, 0x56, 0x52,
0xa5, 0x42, 0x09, 0x58, 0x18, 0x28, 0x90, 0x5f,
0x94, 0xca, 0x05, 0x00, 0x76, 0xb0, 0x3b, 0xeb,
0x24, 0x00, 0x00, 0x00,
},
nil,
},
{ // has dynamic huffman blocks
"gettysburg",
"gettysburg",
" Four score and seven years ago our fathers brought forth on\n" +
"this continent, a new nation, conceived in Liberty, and dedicated\n" +
"to the proposition that all men are created equal.\n" +
" Now we are engaged in a great Civil War, testing whether that\n" +
"nation, or any nation so conceived and so dedicated, can long\n" +
"endure.\n" +
" We are met on a great battle-field of that war.\n" +
" We have come to dedicate a portion of that field, as a final\n" +
"resting place for those who here gave their lives that that\n" +
"nation might live. It is altogether fitting and proper that\n" +
"we should do this.\n" +
" But, in a larger sense, we can not dedicate — we can not\n" +
"consecrate — we can not hallow — this ground.\n" +
" The brave men, living and dead, who struggled here, have\n" +
"consecrated it, far above our poor power to add or detract.\n" +
"The world will little note, nor long remember what we say here,\n" +
"but it can never forget what they did here.\n" +
" It is for us the living, rather, to be dedicated here to the\n" +
"unfinished work which they who fought here have thus far so\n" +
"nobly advanced. It is rather for us to be here dedicated to\n" +
"the great task remaining before us — that from these honored\n" +
"dead we take increased devotion to that cause for which they\n" +
"gave the last full measure of devotion —\n" +
" that we here highly resolve that these dead shall not have\n" +
"died in vain — that this nation, under God, shall have a new\n" +
"birth of freedom — and that government of the people, by the\n" +
"people, for the people, shall not perish from this earth.\n" +
"\n" +
"Abraham Lincoln, November 19, 1863, Gettysburg, Pennsylvania\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xd1, 0x12, 0x2b, 0x4a,
0x00, 0x03, 0x67, 0x65, 0x74, 0x74, 0x79, 0x73,
0x62, 0x75, 0x72, 0x67, 0x00, 0x65, 0x54, 0xcd,
0x6e, 0xd4, 0x30, 0x10, 0xbe, 0xfb, 0x29, 0xe6,
0x01, 0x42, 0xa5, 0x0a, 0x09, 0xc1, 0x11, 0x90,
0x40, 0x48, 0xa8, 0xe2, 0x80, 0xd4, 0xf3, 0x24,
0x9e, 0x24, 0x56, 0xbd, 0x9e, 0xc5, 0x76, 0x76,
0x95, 0x1b, 0x0f, 0xc1, 0x13, 0xf2, 0x24, 0x7c,
0x63, 0x77, 0x9b, 0x4a, 0x5c, 0xaa, 0x6e, 0x6c,
0xcf, 0x7c, 0x7f, 0x33, 0x44, 0x5f, 0x74, 0xcb,
0x54, 0x26, 0xcd, 0x42, 0x9c, 0x3c, 0x15, 0xb9,
0x48, 0xa2, 0x5d, 0x38, 0x17, 0xe2, 0x45, 0xc9,
0x4e, 0x67, 0xae, 0xab, 0xe0, 0xf7, 0x98, 0x75,
0x5b, 0xd6, 0x4a, 0xb3, 0xe6, 0xba, 0x92, 0x26,
0x57, 0xd7, 0x50, 0x68, 0xd2, 0x54, 0x43, 0x92,
0x54, 0x07, 0x62, 0x4a, 0x72, 0xa5, 0xc4, 0x35,
0x68, 0x1a, 0xec, 0x60, 0x92, 0x70, 0x11, 0x4f,
0x21, 0xd1, 0xf7, 0x30, 0x4a, 0xae, 0xfb, 0xd0,
0x9a, 0x78, 0xf1, 0x61, 0xe2, 0x2a, 0xde, 0x55,
0x25, 0xd4, 0xa6, 0x73, 0xd6, 0xb3, 0x96, 0x60,
0xef, 0xf0, 0x9b, 0x2b, 0x71, 0x8c, 0x74, 0x02,
0x10, 0x06, 0xac, 0x29, 0x8b, 0xdd, 0x25, 0xf9,
0xb5, 0x71, 0xbc, 0x73, 0x44, 0x0f, 0x7a, 0xa5,
0xab, 0xb4, 0x33, 0x49, 0x0b, 0x2f, 0xbd, 0x03,
0xd3, 0x62, 0x17, 0xe9, 0x73, 0xb8, 0x84, 0x48,
0x8f, 0x9c, 0x07, 0xaa, 0x52, 0x00, 0x6d, 0xa1,
0xeb, 0x2a, 0xc6, 0xa0, 0x95, 0x76, 0x37, 0x78,
0x9a, 0x81, 0x65, 0x7f, 0x46, 0x4b, 0x45, 0x5f,
0xe1, 0x6d, 0x42, 0xe8, 0x01, 0x13, 0x5c, 0x38,
0x51, 0xd4, 0xb4, 0x38, 0x49, 0x7e, 0xcb, 0x62,
0x28, 0x1e, 0x3b, 0x82, 0x93, 0x54, 0x48, 0xf1,
0xd2, 0x7d, 0xe4, 0x5a, 0xa3, 0xbc, 0x99, 0x83,
0x44, 0x4f, 0x3a, 0x77, 0x36, 0x57, 0xce, 0xcf,
0x2f, 0x56, 0xbe, 0x80, 0x90, 0x9e, 0x84, 0xea,
0x51, 0x1f, 0x8f, 0xcf, 0x90, 0xd4, 0x60, 0xdc,
0x5e, 0xb4, 0xf7, 0x10, 0x0b, 0x26, 0xe0, 0xff,
0xc4, 0xd1, 0xe5, 0x67, 0x2e, 0xe7, 0xc8, 0x93,
0x98, 0x05, 0xb8, 0xa8, 0x45, 0xc0, 0x4d, 0x09,
0xdc, 0x84, 0x16, 0x2b, 0x0d, 0x9a, 0x21, 0x53,
0x04, 0x8b, 0xd2, 0x0b, 0xbd, 0xa2, 0x4c, 0xa7,
0x60, 0xee, 0xd9, 0xe1, 0x1d, 0xd1, 0xb7, 0x4a,
0x30, 0x8f, 0x63, 0xd5, 0xa5, 0x8b, 0x33, 0x87,
0xda, 0x1a, 0x18, 0x79, 0xf3, 0xe3, 0xa6, 0x17,
0x94, 0x2e, 0xab, 0x6e, 0xa0, 0xe3, 0xcd, 0xac,
0x50, 0x8c, 0xca, 0xa7, 0x0d, 0x76, 0x37, 0xd1,
0x23, 0xe7, 0x05, 0x57, 0x8b, 0xa4, 0x22, 0x83,
0xd9, 0x62, 0x52, 0x25, 0xad, 0x07, 0xbb, 0xbf,
0xbf, 0xff, 0xbc, 0xfa, 0xee, 0x20, 0x73, 0x91,
0x29, 0xff, 0x7f, 0x02, 0x71, 0x62, 0x84, 0xb5,
0xf6, 0xb5, 0x25, 0x6b, 0x41, 0xde, 0x92, 0xb7,
0x76, 0x3f, 0x91, 0x91, 0x31, 0x1b, 0x41, 0x84,
0x62, 0x30, 0x0a, 0x37, 0xa4, 0x5e, 0x18, 0x3a,
0x99, 0x08, 0xa5, 0xe6, 0x6d, 0x59, 0x22, 0xec,
0x33, 0x39, 0x86, 0x26, 0xf5, 0xab, 0x66, 0xc8,
0x08, 0x20, 0xcf, 0x0c, 0xd7, 0x47, 0x45, 0x21,
0x0b, 0xf6, 0x59, 0xd5, 0xfe, 0x5c, 0x8d, 0xaa,
0x12, 0x7b, 0x6f, 0xa1, 0xf0, 0x52, 0x33, 0x4f,
0xf5, 0xce, 0x59, 0xd3, 0xab, 0x66, 0x10, 0xbf,
0x06, 0xc4, 0x31, 0x06, 0x73, 0xd6, 0x80, 0xa2,
0x78, 0xc2, 0x45, 0xcb, 0x03, 0x65, 0x39, 0xc9,
0x09, 0xd1, 0x06, 0x04, 0x33, 0x1a, 0x5a, 0xf1,
0xde, 0x01, 0xb8, 0x71, 0x83, 0xc4, 0xb5, 0xb3,
0xc3, 0x54, 0x65, 0x33, 0x0d, 0x5a, 0xf7, 0x9b,
0x90, 0x7c, 0x27, 0x1f, 0x3a, 0x58, 0xa3, 0xd8,
0xfd, 0x30, 0x5f, 0xb7, 0xd2, 0x66, 0xa2, 0x93,
0x1c, 0x28, 0xb7, 0xe9, 0x1b, 0x0c, 0xe1, 0x28,
0x47, 0x26, 0xbb, 0xe9, 0x7d, 0x7e, 0xdc, 0x96,
0x10, 0x92, 0x50, 0x56, 0x7c, 0x06, 0xe2, 0x27,
0xb4, 0x08, 0xd3, 0xda, 0x7b, 0x98, 0x34, 0x73,
0x9f, 0xdb, 0xf6, 0x62, 0xed, 0x31, 0x41, 0x13,
0xd3, 0xa2, 0xa8, 0x4b, 0x3a, 0xc6, 0x1d, 0xe4,
0x2f, 0x8c, 0xf8, 0xfb, 0x97, 0x64, 0xf4, 0xb6,
0x2f, 0x80, 0x5a, 0xf3, 0x56, 0xe0, 0x40, 0x50,
0xd5, 0x19, 0xd0, 0x1e, 0xfc, 0xca, 0xe5, 0xc9,
0xd4, 0x60, 0x00, 0x81, 0x2e, 0xa3, 0xcc, 0xb6,
0x52, 0xf0, 0xb4, 0xdb, 0x69, 0x99, 0xce, 0x7a,
0x32, 0x4c, 0x08, 0xed, 0xaa, 0x10, 0x10, 0xe3,
0x6f, 0xee, 0x99, 0x68, 0x95, 0x9f, 0x04, 0x71,
0xb2, 0x49, 0x2f, 0x62, 0xa6, 0x5e, 0xb4, 0xef,
0x02, 0xed, 0x4f, 0x27, 0xde, 0x4a, 0x0f, 0xfd,
0xc1, 0xcc, 0xdd, 0x02, 0x8f, 0x08, 0x16, 0x54,
0xdf, 0xda, 0xca, 0xe0, 0x82, 0xf1, 0xb4, 0x31,
0x7a, 0xa9, 0x81, 0xfe, 0x90, 0xb7, 0x3e, 0xdb,
0xd3, 0x35, 0xc0, 0x20, 0x80, 0x33, 0x46, 0x4a,
0x63, 0xab, 0xd1, 0x0d, 0x29, 0xd2, 0xe2, 0x84,
0xb8, 0xdb, 0xfa, 0xe9, 0x89, 0x44, 0x86, 0x7c,
0xe8, 0x0b, 0xe6, 0x02, 0x6a, 0x07, 0x9b, 0x96,
0xd0, 0xdb, 0x2e, 0x41, 0x4c, 0xa1, 0xd5, 0x57,
0x45, 0x14, 0xfb, 0xe3, 0xa6, 0x72, 0x5b, 0x87,
0x6e, 0x0c, 0x6d, 0x5b, 0xce, 0xe0, 0x2f, 0xe2,
0x21, 0x81, 0x95, 0xb0, 0xe8, 0xb6, 0x32, 0x0b,
0xb2, 0x98, 0x13, 0x52, 0x5d, 0xfb, 0xec, 0x63,
0x17, 0x8a, 0x9e, 0x23, 0x22, 0x36, 0xee, 0xcd,
0xda, 0xdb, 0xcf, 0x3e, 0xf1, 0xc7, 0xf1, 0x01,
0x12, 0x93, 0x0a, 0xeb, 0x6f, 0xf2, 0x02, 0x15,
0x96, 0x77, 0x5d, 0xef, 0x9c, 0xfb, 0x88, 0x91,
0x59, 0xf9, 0x84, 0xdd, 0x9b, 0x26, 0x8d, 0x80,
0xf9, 0x80, 0x66, 0x2d, 0xac, 0xf7, 0x1f, 0x06,
0xba, 0x7f, 0xff, 0xee, 0xed, 0x40, 0x5f, 0xa5,
0xd6, 0xbd, 0x8c, 0x5b, 0x46, 0xd2, 0x7e, 0x48,
0x4a, 0x65, 0x8f, 0x08, 0x42, 0x60, 0xf7, 0x0f,
0xb9, 0x16, 0x0b, 0x0c, 0x1a, 0x06, 0x00, 0x00,
},
nil,
},
{ // has 1 non-empty fixed huffman block then garbage
"hello.txt",
"hello.txt + garbage",
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0x0c, 0x00,
0x00, 0x00, 'g', 'a', 'r', 'b', 'a', 'g', 'e', '!', '!', '!',
},
ErrHeader,
},
{ // has 1 non-empty fixed huffman block not enough header
"hello.txt",
"hello.txt + garbage",
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0x0c, 0x00,
0x00, 0x00, gzipID1,
},
io.ErrUnexpectedEOF,
},
{ // has 1 non-empty fixed huffman block but corrupt checksum
"hello.txt",
"hello.txt + corrupt checksum",
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0xff, 0xff, 0xff, 0xff, 0x0c, 0x00,
0x00, 0x00,
},
ErrChecksum,
},
{ // has 1 non-empty fixed huffman block but corrupt size
"hello.txt",
"hello.txt + corrupt size",
"hello world\n",
[]byte{
0x1f, 0x8b, 0x08, 0x08, 0xc8, 0x58, 0x13, 0x4a,
0x00, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x2e,
0x74, 0x78, 0x74, 0x00, 0xcb, 0x48, 0xcd, 0xc9,
0xc9, 0x57, 0x28, 0xcf, 0x2f, 0xca, 0x49, 0xe1,
0x02, 0x00, 0x2d, 0x3b, 0x08, 0xaf, 0xff, 0x00,
0x00, 0x00,
},
ErrChecksum,
},
{
"f1l3n4m3.tXt",
"header with all fields used",
"",
[]byte{
0x1f, 0x8b, 0x08, 0x1e, 0x70, 0xf0, 0xf9, 0x4a,
0x00, 0xaa, 0x09, 0x00, 0x7a, 0x7a, 0x05, 0x00,
0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x31, 0x6c,
0x33, 0x6e, 0x34, 0x6d, 0x33, 0x2e, 0x74, 0x58,
0x74, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,
0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,
0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,
0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e,
0x1f, 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26,
0x27, 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e,
0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,
0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e,
0x3f, 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46,
0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e,
0x4f, 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56,
0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e,
0x5f, 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66,
0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e,
0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76,
0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e,
0x7f, 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86,
0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e,
0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96,
0x97, 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e,
0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6,
0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae,
0xaf, 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6,
0xb7, 0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe,
0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6,
0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce,
0xcf, 0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6,
0xd7, 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde,
0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6,
0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee,
0xef, 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6,
0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe,
0xff, 0x00, 0x92, 0xfd, 0x01, 0x00, 0x00, 0xff,
0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00,
},
nil,
},
{
"",
"truncated gzip file amid raw-block",
"hello",
[]byte{
0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff,
0x00, 0x0c, 0x00, 0xf3, 0xff, 0x68, 0x65, 0x6c, 0x6c, 0x6f,
},
io.ErrUnexpectedEOF,
},
{
"",
"truncated gzip file amid fixed-block",
"He",
[]byte{
0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff,
0xf2, 0x48, 0xcd,
},
io.ErrUnexpectedEOF,
},
}
func TestDecompressor(t *testing.T) {
b := new(bytes.Buffer)
for _, tt := range gunzipTests {
in := bytes.NewReader(tt.gzip)
gzip, err := NewReader(in)
if err != nil {
t.Errorf("%s: NewReader: %s", tt.name, err)
continue
}
defer gzip.Close()
if tt.name != gzip.Name {
t.Errorf("%s: got name %s", tt.name, gzip.Name)
}
b.Reset()
n, err := io.Copy(b, gzip)
if err != tt.err {
t.Errorf("%s: io.Copy: %v want %v", tt.name, err, tt.err)
}
s := b.String()
if s != tt.raw {
t.Errorf("%s: got %d-byte %q want %d-byte %q", tt.name, n, s, len(tt.raw), tt.raw)
}
// Test Reader Reset.
in = bytes.NewReader(tt.gzip)
err = gzip.Reset(in)
if err != nil {
t.Errorf("%s: Reset: %s", tt.name, err)
continue
}
if tt.name != gzip.Name {
t.Errorf("%s: got name %s", tt.name, gzip.Name)
}
b.Reset()
n, err = io.Copy(b, gzip)
if err != tt.err {
t.Errorf("%s: io.Copy: %v want %v", tt.name, err, tt.err)
}
s = b.String()
if s != tt.raw {
t.Errorf("%s: got %d-byte %q want %d-byte %q", tt.name, n, s, len(tt.raw), tt.raw)
}
}
}
func TestIssue6550(t *testing.T) {
f, err := os.Open("testdata/issue6550.gz")
if err != nil {
t.Fatal(err)
}
gzip, err := NewReader(f)
if err != nil {
t.Fatalf("NewReader(testdata/issue6550.gz): %v", err)
}
defer gzip.Close()
done := make(chan bool, 1)
go func() {
_, err := io.Copy(ioutil.Discard, gzip)
if err == nil {
t.Errorf("Copy succeeded")
} else {
t.Logf("Copy failed (correctly): %v", err)
}
done <- true
}()
select {
case <-time.After(1 * time.Second):
t.Errorf("Copy hung")
case <-done:
// ok
}
}
func TestInitialReset(t *testing.T) {
var r Reader
if err := r.Reset(bytes.NewReader(gunzipTests[1].gzip)); err != nil {
t.Error(err)
}
var buf bytes.Buffer
if _, err := io.Copy(&buf, &r); err != nil {
t.Error(err)
}
if s := buf.String(); s != gunzipTests[1].raw {
t.Errorf("got %q want %q", s, gunzipTests[1].raw)
}
}
func TestMultistreamFalse(t *testing.T) {
// Find concatenation test.
var tt gunzipTest
for _, tt = range gunzipTests {
if strings.HasSuffix(tt.desc, " x2") {
goto Found
}
}
t.Fatal("cannot find hello.txt x2 in gunzip tests")
Found:
br := bytes.NewReader(tt.gzip)
var r Reader
if err := r.Reset(br); err != nil {
t.Fatalf("first reset: %v", err)
}
// Expect two streams with "hello world\n", then real EOF.
const hello = "hello world\n"
r.Multistream(false)
data, err := ioutil.ReadAll(&r)
if string(data) != hello || err != nil {
t.Fatalf("first stream = %q, %v, want %q, %v", string(data), err, hello, nil)
}
if err := r.Reset(br); err != nil {
t.Fatalf("second reset: %v", err)
}
r.Multistream(false)
data, err = ioutil.ReadAll(&r)
if string(data) != hello || err != nil {
t.Fatalf("second stream = %q, %v, want %q, %v", string(data), err, hello, nil)
}
if err := r.Reset(br); err != io.EOF {
t.Fatalf("third reset: err=%v, want io.EOF", err)
}
}
func TestWriteTo(t *testing.T) {
input := make([]byte, 100000)
n, err := rand.Read(input)
if err != nil {
t.Fatal(err)
}
if n != len(input) {
t.Fatal("did not fill buffer")
}
compressed := &bytes.Buffer{}
// Do it twice to test MultiStream functionality
for i := 0; i < 2; i++ {
w, err := NewWriterLevel(compressed, -2)
if err != nil {
t.Fatal(err)
}
n, err = w.Write(input)
if err != nil {
t.Fatal(err)
}
if n != len(input) {
t.Fatal("did not fill buffer")
}
w.Close()
}
input = append(input, input...)
buf := compressed.Bytes()
dec, err := NewReader(bytes.NewBuffer(buf))
if err != nil {
t.Fatal(err)
}
// ReadAll does not use WriteTo, but we wrap it in a NopCloser to be sure.
readall, err := ioutil.ReadAll(ioutil.NopCloser(dec))
if err != nil {
t.Fatal(err)
}
if len(readall) != len(input) {
t.Errorf("did not decompress everything, want %d, got %d", len(input), len(readall))
}
if bytes.Compare(readall, input) != 0 {
t.Error("output did not match input")
}
dec, err = NewReader(bytes.NewBuffer(buf))
if err != nil {
t.Fatal(err)
}
wtbuf := &bytes.Buffer{}
written, err := dec.WriteTo(wtbuf)
if err != nil {
t.Fatal(err)
}
if written != int64(len(input)) {
t.Error("Returned length did not match, expected", len(input), "got", written)
}
if wtbuf.Len() != len(input) {
t.Error("Actual Length did not match, expected", len(input), "got", wtbuf.Len())
}
if bytes.Compare(wtbuf.Bytes(), input) != 0 {
t.Fatal("output did not match input")
}
}
func TestNilStream(t *testing.T) {
// Go liberally interprets RFC 1952 section 2.2 to mean that a gzip file
// consist of zero or more members. Thus, we test that a nil stream is okay.
_, err := NewReader(bytes.NewReader(nil))
if err != io.EOF {
t.Fatalf("NewReader(nil) on empty stream: got %v, want io.EOF", err)
}
}
func TestTruncatedStreams(t *testing.T) {
const data = "\x1f\x8b\b\x04\x00\tn\x88\x00\xff\a\x00foo bar\xcbH\xcd\xc9\xc9\xd7Q(\xcf/\xcaI\x01\x04:r\xab\xff\f\x00\x00\x00"
// Intentionally iterate starting with at least one byte in the stream.
for i := 1; i < len(data)-1; i++ {
r, err := NewReader(strings.NewReader(data[:i]))
if err != nil {
if err != io.ErrUnexpectedEOF {
t.Errorf("NewReader(%d) on truncated stream: got %v, want %v", i, err, io.ErrUnexpectedEOF)
}
continue
}
_, err = io.Copy(ioutil.Discard, r)
if ferr, ok := err.(*flate.ReadError); ok {
err = ferr.Err
}
if err != io.ErrUnexpectedEOF {
t.Errorf("io.Copy(%d) on truncated stream: got %v, want %v", i, err, io.ErrUnexpectedEOF)
}
}
}
func BenchmarkGunzipCopy(b *testing.B) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dst := &bytes.Buffer{}
w, _ := NewWriterLevel(dst, 1)
_, err := w.Write(dat)
if err != nil {
b.Fatal(err)
}
w.Close()
input := dst.Bytes()
b.SetBytes(int64(len(dat)))
b.ResetTimer()
for n := 0; n < b.N; n++ {
r, err := NewReader(bytes.NewBuffer(input))
if err != nil {
b.Fatal(err)
}
_, err = io.Copy(ioutil.Discard, r)
if err != nil {
b.Fatal(err)
}
}
}
func BenchmarkGunzipNoWriteTo(b *testing.B) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dst := &bytes.Buffer{}
w, _ := NewWriterLevel(dst, 1)
_, err := w.Write(dat)
if err != nil {
b.Fatal(err)
}
w.Close()
input := dst.Bytes()
r, err := NewReader(bytes.NewBuffer(input))
if err != nil {
b.Fatal(err)
}
b.SetBytes(int64(len(dat)))
b.ResetTimer()
for n := 0; n < b.N; n++ {
err := r.Reset(bytes.NewBuffer(input))
if err != nil {
b.Fatal(err)
}
_, err = io.Copy(ioutil.Discard, ioutil.NopCloser(r))
if err != nil {
b.Fatal(err)
}
}
}
func BenchmarkGunzipStdlib(b *testing.B) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dst := &bytes.Buffer{}
w, _ := NewWriterLevel(dst, 1)
_, err := w.Write(dat)
if err != nil {
b.Fatal(err)
}
w.Close()
input := dst.Bytes()
r, err := oldgz.NewReader(bytes.NewBuffer(input))
if err != nil {
b.Fatal(err)
}
b.SetBytes(int64(len(dat)))
b.ResetTimer()
for n := 0; n < b.N; n++ {
err := r.Reset(bytes.NewBuffer(input))
if err != nil {
b.Fatal(err)
}
_, err = io.Copy(ioutil.Discard, r)
if err != nil {
b.Fatal(err)
}
}
}

251
vendor/github.com/klauspost/compress/gzip/gzip.go generated vendored Normal file
View File

@@ -0,0 +1,251 @@
// Copyright 2010 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package gzip
import (
"errors"
"fmt"
"io"
"github.com/klauspost/compress/flate"
"github.com/klauspost/crc32"
)
// These constants are copied from the flate package, so that code that imports
// "compress/gzip" does not also have to import "compress/flate".
const (
NoCompression = flate.NoCompression
BestSpeed = flate.BestSpeed
BestCompression = flate.BestCompression
DefaultCompression = flate.DefaultCompression
ConstantCompression = flate.ConstantCompression
HuffmanOnly = flate.HuffmanOnly
)
// A Writer is an io.WriteCloser.
// Writes to a Writer are compressed and written to w.
type Writer struct {
Header // written at first call to Write, Flush, or Close
w io.Writer
level int
wroteHeader bool
compressor *flate.Writer
digest uint32 // CRC-32, IEEE polynomial (section 8)
size uint32 // Uncompressed size (section 2.3.1)
closed bool
buf [10]byte
err error
}
// NewWriter returns a new Writer.
// Writes to the returned writer are compressed and written to w.
//
// It is the caller's responsibility to call Close on the WriteCloser when done.
// Writes may be buffered and not flushed until Close.
//
// Callers that wish to set the fields in Writer.Header must do so before
// the first call to Write, Flush, or Close.
func NewWriter(w io.Writer) *Writer {
z, _ := NewWriterLevel(w, DefaultCompression)
return z
}
// NewWriterLevel is like NewWriter but specifies the compression level instead
// of assuming DefaultCompression.
//
// The compression level can be DefaultCompression, NoCompression, or any
// integer value between BestSpeed and BestCompression inclusive. The error
// returned will be nil if the level is valid.
func NewWriterLevel(w io.Writer, level int) (*Writer, error) {
if level < HuffmanOnly || level > BestCompression {
return nil, fmt.Errorf("gzip: invalid compression level: %d", level)
}
z := new(Writer)
z.init(w, level)
return z, nil
}
func (z *Writer) init(w io.Writer, level int) {
compressor := z.compressor
if compressor != nil {
compressor.Reset(w)
}
*z = Writer{
Header: Header{
OS: 255, // unknown
},
w: w,
level: level,
compressor: compressor,
}
}
// Reset discards the Writer z's state and makes it equivalent to the
// result of its original state from NewWriter or NewWriterLevel, but
// writing to w instead. This permits reusing a Writer rather than
// allocating a new one.
func (z *Writer) Reset(w io.Writer) {
z.init(w, z.level)
}
// writeBytes writes a length-prefixed byte slice to z.w.
func (z *Writer) writeBytes(b []byte) error {
if len(b) > 0xffff {
return errors.New("gzip.Write: Extra data is too large")
}
le.PutUint16(z.buf[:2], uint16(len(b)))
_, err := z.w.Write(z.buf[:2])
if err != nil {
return err
}
_, err = z.w.Write(b)
return err
}
// writeString writes a UTF-8 string s in GZIP's format to z.w.
// GZIP (RFC 1952) specifies that strings are NUL-terminated ISO 8859-1 (Latin-1).
func (z *Writer) writeString(s string) (err error) {
// GZIP stores Latin-1 strings; error if non-Latin-1; convert if non-ASCII.
needconv := false
for _, v := range s {
if v == 0 || v > 0xff {
return errors.New("gzip.Write: non-Latin-1 header string")
}
if v > 0x7f {
needconv = true
}
}
if needconv {
b := make([]byte, 0, len(s))
for _, v := range s {
b = append(b, byte(v))
}
_, err = z.w.Write(b)
} else {
_, err = io.WriteString(z.w, s)
}
if err != nil {
return err
}
// GZIP strings are NUL-terminated.
z.buf[0] = 0
_, err = z.w.Write(z.buf[:1])
return err
}
// Write writes a compressed form of p to the underlying io.Writer. The
// compressed bytes are not necessarily flushed until the Writer is closed.
func (z *Writer) Write(p []byte) (int, error) {
if z.err != nil {
return 0, z.err
}
var n int
// Write the GZIP header lazily.
if !z.wroteHeader {
z.wroteHeader = true
z.buf[0] = gzipID1
z.buf[1] = gzipID2
z.buf[2] = gzipDeflate
z.buf[3] = 0
if z.Extra != nil {
z.buf[3] |= 0x04
}
if z.Name != "" {
z.buf[3] |= 0x08
}
if z.Comment != "" {
z.buf[3] |= 0x10
}
le.PutUint32(z.buf[4:8], uint32(z.ModTime.Unix()))
if z.level == BestCompression {
z.buf[8] = 2
} else if z.level == BestSpeed {
z.buf[8] = 4
} else {
z.buf[8] = 0
}
z.buf[9] = z.OS
n, z.err = z.w.Write(z.buf[:10])
if z.err != nil {
return n, z.err
}
if z.Extra != nil {
z.err = z.writeBytes(z.Extra)
if z.err != nil {
return n, z.err
}
}
if z.Name != "" {
z.err = z.writeString(z.Name)
if z.err != nil {
return n, z.err
}
}
if z.Comment != "" {
z.err = z.writeString(z.Comment)
if z.err != nil {
return n, z.err
}
}
if z.compressor == nil {
z.compressor, _ = flate.NewWriter(z.w, z.level)
}
}
z.size += uint32(len(p))
z.digest = crc32.Update(z.digest, crc32.IEEETable, p)
n, z.err = z.compressor.Write(p)
return n, z.err
}
// Flush flushes any pending compressed data to the underlying writer.
//
// It is useful mainly in compressed network protocols, to ensure that
// a remote reader has enough data to reconstruct a packet. Flush does
// not return until the data has been written. If the underlying
// writer returns an error, Flush returns that error.
//
// In the terminology of the zlib library, Flush is equivalent to Z_SYNC_FLUSH.
func (z *Writer) Flush() error {
if z.err != nil {
return z.err
}
if z.closed {
return nil
}
if !z.wroteHeader {
z.Write(nil)
if z.err != nil {
return z.err
}
}
z.err = z.compressor.Flush()
return z.err
}
// Close closes the Writer, flushing any unwritten data to the underlying
// io.Writer, but does not close the underlying io.Writer.
func (z *Writer) Close() error {
if z.err != nil {
return z.err
}
if z.closed {
return nil
}
z.closed = true
if !z.wroteHeader {
z.Write(nil)
if z.err != nil {
return z.err
}
}
z.err = z.compressor.Close()
if z.err != nil {
return z.err
}
le.PutUint32(z.buf[:4], z.digest)
le.PutUint32(z.buf[4:8], z.size)
_, z.err = z.w.Write(z.buf[:8])
return z.err
}

519
vendor/github.com/klauspost/compress/gzip/gzip_test.go generated vendored Normal file
View File

@@ -0,0 +1,519 @@
// Copyright 2010 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package gzip
import (
"bufio"
"bytes"
oldgz "compress/gzip"
"io"
"io/ioutil"
"math/rand"
"testing"
"time"
)
// TestEmpty tests that an empty payload still forms a valid GZIP stream.
func TestEmpty(t *testing.T) {
buf := new(bytes.Buffer)
if err := NewWriter(buf).Close(); err != nil {
t.Fatalf("Writer.Close: %v", err)
}
r, err := NewReader(buf)
if err != nil {
t.Fatalf("NewReader: %v", err)
}
b, err := ioutil.ReadAll(r)
if err != nil {
t.Fatalf("ReadAll: %v", err)
}
if len(b) != 0 {
t.Fatalf("got %d bytes, want 0", len(b))
}
if err := r.Close(); err != nil {
t.Fatalf("Reader.Close: %v", err)
}
}
// TestRoundTrip tests that gzipping and then gunzipping is the identity
// function.
func TestRoundTrip(t *testing.T) {
buf := new(bytes.Buffer)
w := NewWriter(buf)
w.Comment = "comment"
w.Extra = []byte("extra")
w.ModTime = time.Unix(1e8, 0)
w.Name = "name"
if _, err := w.Write([]byte("payload")); err != nil {
t.Fatalf("Write: %v", err)
}
if err := w.Close(); err != nil {
t.Fatalf("Writer.Close: %v", err)
}
r, err := NewReader(buf)
if err != nil {
t.Fatalf("NewReader: %v", err)
}
b, err := ioutil.ReadAll(r)
if err != nil {
t.Fatalf("ReadAll: %v", err)
}
if string(b) != "payload" {
t.Fatalf("payload is %q, want %q", string(b), "payload")
}
if r.Comment != "comment" {
t.Fatalf("comment is %q, want %q", r.Comment, "comment")
}
if string(r.Extra) != "extra" {
t.Fatalf("extra is %q, want %q", r.Extra, "extra")
}
if r.ModTime.Unix() != 1e8 {
t.Fatalf("mtime is %d, want %d", r.ModTime.Unix(), uint32(1e8))
}
if r.Name != "name" {
t.Fatalf("name is %q, want %q", r.Name, "name")
}
if err := r.Close(); err != nil {
t.Fatalf("Reader.Close: %v", err)
}
}
// TestLatin1 tests the internal functions for converting to and from Latin-1.
func TestLatin1(t *testing.T) {
latin1 := []byte{0xc4, 'u', 0xdf, 'e', 'r', 'u', 'n', 'g', 0}
utf8 := "Äußerung"
z := Reader{r: bufio.NewReader(bytes.NewReader(latin1))}
s, err := z.readString()
if err != nil {
t.Fatalf("readString: %v", err)
}
if s != utf8 {
t.Fatalf("read latin-1: got %q, want %q", s, utf8)
}
buf := bytes.NewBuffer(make([]byte, 0, len(latin1)))
c := Writer{w: buf}
if err = c.writeString(utf8); err != nil {
t.Fatalf("writeString: %v", err)
}
s = buf.String()
if s != string(latin1) {
t.Fatalf("write utf-8: got %q, want %q", s, string(latin1))
}
}
// TestLatin1RoundTrip tests that metadata that is representable in Latin-1
// survives a round trip.
func TestLatin1RoundTrip(t *testing.T) {
testCases := []struct {
name string
ok bool
}{
{"", true},
{"ASCII is OK", true},
{"unless it contains a NUL\x00", false},
{"no matter where \x00 occurs", false},
{"\x00\x00\x00", false},
{"Látin-1 also passes (U+00E1)", true},
{"but LĀtin Extended-A (U+0100) does not", false},
{"neither does 日本語", false},
{"invalid UTF-8 also \xffails", false},
{"\x00 as does Látin-1 with NUL", false},
}
for _, tc := range testCases {
buf := new(bytes.Buffer)
w := NewWriter(buf)
w.Name = tc.name
err := w.Close()
if (err == nil) != tc.ok {
t.Errorf("Writer.Close: name = %q, err = %v", tc.name, err)
continue
}
if !tc.ok {
continue
}
r, err := NewReader(buf)
if err != nil {
t.Errorf("NewReader: %v", err)
continue
}
_, err = ioutil.ReadAll(r)
if err != nil {
t.Errorf("ReadAll: %v", err)
continue
}
if r.Name != tc.name {
t.Errorf("name is %q, want %q", r.Name, tc.name)
continue
}
if err := r.Close(); err != nil {
t.Errorf("Reader.Close: %v", err)
continue
}
}
}
func TestWriterFlush(t *testing.T) {
buf := new(bytes.Buffer)
w := NewWriter(buf)
w.Comment = "comment"
w.Extra = []byte("extra")
w.ModTime = time.Unix(1e8, 0)
w.Name = "name"
n0 := buf.Len()
if n0 != 0 {
t.Fatalf("buffer size = %d before writes; want 0", n0)
}
if err := w.Flush(); err != nil {
t.Fatal(err)
}
n1 := buf.Len()
if n1 == 0 {
t.Fatal("no data after first flush")
}
w.Write([]byte("x"))
n2 := buf.Len()
if n1 != n2 {
t.Fatalf("after writing a single byte, size changed from %d to %d; want no change", n1, n2)
}
if err := w.Flush(); err != nil {
t.Fatal(err)
}
n3 := buf.Len()
if n2 == n3 {
t.Fatal("Flush didn't flush any data")
}
}
// Multiple gzip files concatenated form a valid gzip file.
func TestConcat(t *testing.T) {
var buf bytes.Buffer
w := NewWriter(&buf)
w.Write([]byte("hello "))
w.Close()
w = NewWriter(&buf)
w.Write([]byte("world\n"))
w.Close()
r, err := NewReader(&buf)
data, err := ioutil.ReadAll(r)
if string(data) != "hello world\n" || err != nil {
t.Fatalf("ReadAll = %q, %v, want %q, nil", data, err, "hello world")
}
}
func TestWriterReset(t *testing.T) {
buf := new(bytes.Buffer)
buf2 := new(bytes.Buffer)
z := NewWriter(buf)
msg := []byte("hello world")
z.Write(msg)
z.Close()
z.Reset(buf2)
z.Write(msg)
z.Close()
if buf.String() != buf2.String() {
t.Errorf("buf2 %q != original buf of %q", buf2.String(), buf.String())
}
}
var testbuf []byte
func testFile(i, level int, t *testing.T) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dl := len(dat)
if len(testbuf) != i*dl {
// Make results predictable
testbuf = make([]byte, i*dl)
for j := 0; j < i; j++ {
copy(testbuf[j*dl:j*dl+dl], dat)
}
}
br := bytes.NewBuffer(testbuf)
var buf bytes.Buffer
w, err := NewWriterLevel(&buf, DefaultCompression)
if err != nil {
t.Fatal(err)
}
n, err := io.Copy(w, br)
if err != nil {
t.Fatal(err)
}
if int(n) != len(testbuf) {
t.Fatal("Short write:", n, "!=", testbuf)
}
err = w.Close()
if err != nil {
t.Fatal(err)
}
r, err := NewReader(&buf)
if err != nil {
t.Fatal(err.Error())
}
decoded, err := ioutil.ReadAll(r)
if err != nil {
t.Fatal(err.Error())
}
if !bytes.Equal(testbuf, decoded) {
t.Errorf("decoded content does not match.")
}
}
func TestFile1xM2(t *testing.T) { testFile(1, -2, t) }
func TestFile1xM1(t *testing.T) { testFile(1, -1, t) }
func TestFile1x0(t *testing.T) { testFile(1, 0, t) }
func TestFile1x1(t *testing.T) { testFile(1, 1, t) }
func TestFile1x2(t *testing.T) { testFile(1, 2, t) }
func TestFile1x3(t *testing.T) { testFile(1, 3, t) }
func TestFile1x4(t *testing.T) { testFile(1, 4, t) }
func TestFile1x5(t *testing.T) { testFile(1, 5, t) }
func TestFile1x6(t *testing.T) { testFile(1, 6, t) }
func TestFile1x7(t *testing.T) { testFile(1, 7, t) }
func TestFile1x8(t *testing.T) { testFile(1, 8, t) }
func TestFile1x9(t *testing.T) { testFile(1, 9, t) }
func TestFile10(t *testing.T) { testFile(10, DefaultCompression, t) }
func TestFile50(t *testing.T) {
if testing.Short() {
t.Skip("skipping during short test")
}
testFile(50, DefaultCompression, t)
}
func TestFile200(t *testing.T) {
if testing.Short() {
t.Skip("skipping during short test")
}
testFile(200, BestSpeed, t)
}
func testBigGzip(i int, t *testing.T) {
if len(testbuf) != i {
// Make results predictable
rand.Seed(1337)
testbuf = make([]byte, i)
for idx := range testbuf {
testbuf[idx] = byte(65 + rand.Intn(20))
}
}
c := BestCompression
if testing.Short() {
c = BestSpeed
}
br := bytes.NewBuffer(testbuf)
var buf bytes.Buffer
w, err := NewWriterLevel(&buf, c)
if err != nil {
t.Fatal(err)
}
n, err := io.Copy(w, br)
if err != nil {
t.Fatal(err)
}
if int(n) != len(testbuf) {
t.Fatal("Short write:", n, "!=", len(testbuf))
}
err = w.Close()
if err != nil {
t.Fatal(err.Error())
}
r, err := NewReader(&buf)
if err != nil {
t.Fatal(err.Error())
}
decoded, err := ioutil.ReadAll(r)
if err != nil {
t.Fatal(err.Error())
}
if !bytes.Equal(testbuf, decoded) {
t.Errorf("decoded content does not match.")
}
}
func TestGzip1K(t *testing.T) { testBigGzip(1000, t) }
func TestGzip100K(t *testing.T) { testBigGzip(100000, t) }
func TestGzip1M(t *testing.T) {
if testing.Short() {
t.Skip("skipping during short test")
}
testBigGzip(1000000, t)
}
func TestGzip10M(t *testing.T) {
if testing.Short() {
t.Skip("skipping during short test")
}
testBigGzip(10000000, t)
}
// Test if two runs produce identical results.
func TestDeterministicLM2(t *testing.T) { testDeterm(-2, t) }
// Level 0 is not deterministic since it depends on the size of each write.
// func TestDeterministicL0(t *testing.T) { testDeterm(0, t) }
func TestDeterministicL1(t *testing.T) { testDeterm(1, t) }
func TestDeterministicL2(t *testing.T) { testDeterm(2, t) }
func TestDeterministicL3(t *testing.T) { testDeterm(3, t) }
func TestDeterministicL4(t *testing.T) { testDeterm(4, t) }
func TestDeterministicL5(t *testing.T) { testDeterm(5, t) }
func TestDeterministicL6(t *testing.T) { testDeterm(6, t) }
func TestDeterministicL7(t *testing.T) { testDeterm(7, t) }
func TestDeterministicL8(t *testing.T) { testDeterm(8, t) }
func TestDeterministicL9(t *testing.T) { testDeterm(9, t) }
func testDeterm(i int, t *testing.T) {
var length = 500000
if testing.Short() {
length = 100000
}
rand.Seed(1337)
t1 := make([]byte, length)
for idx := range t1 {
t1[idx] = byte(65 + rand.Intn(8))
}
br := bytes.NewBuffer(t1)
var b1 bytes.Buffer
w, err := NewWriterLevel(&b1, i)
if err != nil {
t.Fatal(err)
}
_, err = io.Copy(w, br)
if err != nil {
t.Fatal(err)
}
w.Flush()
w.Close()
// We recreate the buffer, so we have a goos chance of getting a
// different memory address.
rand.Seed(1337)
t2 := make([]byte, length)
for idx := range t2 {
t2[idx] = byte(65 + rand.Intn(8))
}
br2 := bytes.NewBuffer(t2)
var b2 bytes.Buffer
w2, err := NewWriterLevel(&b2, i)
if err != nil {
t.Fatal(err)
}
// We write the same data, but with a different size than
// the default copy.
for {
_, err = io.CopyN(w2, br2, 1234)
if err == io.EOF {
err = nil
break
} else if err != nil {
break
}
}
if err != nil {
t.Fatal(err)
}
w2.Flush()
w2.Close()
b1b := b1.Bytes()
b2b := b2.Bytes()
if bytes.Compare(b1b, b2b) != 0 {
t.Fatalf("Level %d did not produce deterministric result, len(a) = %d, len(b) = %d", i, len(b1b), len(b2b))
}
}
func BenchmarkGzipLM2(b *testing.B) { benchmarkGzipN(b, -2) }
func BenchmarkGzipL1(b *testing.B) { benchmarkGzipN(b, 1) }
func BenchmarkGzipL2(b *testing.B) { benchmarkGzipN(b, 2) }
func BenchmarkGzipL3(b *testing.B) { benchmarkGzipN(b, 3) }
func BenchmarkGzipL4(b *testing.B) { benchmarkGzipN(b, 4) }
func BenchmarkGzipL5(b *testing.B) { benchmarkGzipN(b, 5) }
func BenchmarkGzipL6(b *testing.B) { benchmarkGzipN(b, 6) }
func BenchmarkGzipL7(b *testing.B) { benchmarkGzipN(b, 7) }
func BenchmarkGzipL8(b *testing.B) { benchmarkGzipN(b, 8) }
func BenchmarkGzipL9(b *testing.B) { benchmarkGzipN(b, 9) }
func benchmarkGzipN(b *testing.B, level int) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
b.SetBytes(int64(len(dat)))
w, _ := NewWriterLevel(ioutil.Discard, level)
b.ResetTimer()
for n := 0; n < b.N; n++ {
w.Reset(ioutil.Discard)
n, err := w.Write(dat)
if n != len(dat) {
panic("short write")
}
if err != nil {
panic(err)
}
err = w.Close()
if err != nil {
panic(err)
}
}
}
func BenchmarkOldGzipL1(b *testing.B) { benchmarkOldGzipN(b, 1) }
func BenchmarkOldGzipL2(b *testing.B) { benchmarkOldGzipN(b, 2) }
func BenchmarkOldGzipL3(b *testing.B) { benchmarkOldGzipN(b, 3) }
func BenchmarkOldGzipL4(b *testing.B) { benchmarkOldGzipN(b, 4) }
func BenchmarkOldGzipL5(b *testing.B) { benchmarkOldGzipN(b, 5) }
func BenchmarkOldGzipL6(b *testing.B) { benchmarkOldGzipN(b, 6) }
func BenchmarkOldGzipL7(b *testing.B) { benchmarkOldGzipN(b, 7) }
func BenchmarkOldGzipL8(b *testing.B) { benchmarkOldGzipN(b, 8) }
func BenchmarkOldGzipL9(b *testing.B) { benchmarkOldGzipN(b, 9) }
func benchmarkOldGzipN(b *testing.B, level int) {
dat, _ := ioutil.ReadFile("testdata/test.json")
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
dat = append(dat, dat...)
b.SetBytes(int64(len(dat)))
w, _ := oldgz.NewWriterLevel(ioutil.Discard, level)
b.ResetTimer()
for n := 0; n < b.N; n++ {
w.Reset(ioutil.Discard)
n, err := w.Write(dat)
if n != len(dat) {
panic("short write")
}
if err != nil {
panic(err)
}
err = w.Close()
if err != nil {
panic(err)
}
}
}

BIN
vendor/github.com/klauspost/compress/gzip/testdata/issue6550.gz generated vendored Normal file
View File

Binary file not shown.

5902
vendor/github.com/klauspost/compress/gzip/testdata/test.json generated vendored Normal file
View File

File diff suppressed because it is too large Load Diff

16
vendor/github.com/klauspost/compress/snappy/.gitignore generated vendored Normal file
View File

@@ -0,0 +1,16 @@
cmd/snappytool/snappytool
testdata/bench
# These explicitly listed benchmark data files are for an obsolete version of
# snappy_test.go.
testdata/alice29.txt
testdata/asyoulik.txt
testdata/fireworks.jpeg
testdata/geo.protodata
testdata/html
testdata/html_x_4
testdata/kppkn.gtb
testdata/lcet10.txt
testdata/paper-100k.pdf
testdata/plrabn12.txt
testdata/urls.10K

15
vendor/github.com/klauspost/compress/snappy/AUTHORS generated vendored Normal file
View File

@@ -0,0 +1,15 @@
# This is the official list of Snappy-Go authors for copyright purposes.
# This file is distinct from the CONTRIBUTORS files.
# See the latter for an explanation.
# Names should be added to this file as
# Name or Organization <email address>
# The email address is not required for organizations.
# Please keep the list sorted.
Damian Gryski <dgryski@gmail.com>
Google Inc.
Jan Mercl <0xjnml@gmail.com>
Rodolfo Carvalho <rhcarvalho@gmail.com>
Sebastien Binet <seb.binet@gmail.com>

37
vendor/github.com/klauspost/compress/snappy/CONTRIBUTORS generated vendored Normal file
View File

@@ -0,0 +1,37 @@
# This is the official list of people who can contribute
# (and typically have contributed) code to the Snappy-Go repository.
# The AUTHORS file lists the copyright holders; this file
# lists people. For example, Google employees are listed here
# but not in AUTHORS, because Google holds the copyright.
#
# The submission process automatically checks to make sure
# that people submitting code are listed in this file (by email address).
#
# Names should be added to this file only after verifying that
# the individual or the individual's organization has agreed to
# the appropriate Contributor License Agreement, found here:
#
# http://code.google.com/legal/individual-cla-v1.0.html
# http://code.google.com/legal/corporate-cla-v1.0.html
#
# The agreement for individuals can be filled out on the web.
#
# When adding J Random Contributor's name to this file,
# either J's name or J's organization's name should be
# added to the AUTHORS file, depending on whether the
# individual or corporate CLA was used.
# Names should be added to this file like so:
# Name <email address>
# Please keep the list sorted.
Damian Gryski <dgryski@gmail.com>
Jan Mercl <0xjnml@gmail.com>
Kai Backman <kaib@golang.org>
Marc-Antoine Ruel <maruel@chromium.org>
Nigel Tao <nigeltao@golang.org>
Rob Pike <r@golang.org>
Rodolfo Carvalho <rhcarvalho@gmail.com>
Russ Cox <rsc@golang.org>
Sebastien Binet <seb.binet@gmail.com>

27
vendor/github.com/klauspost/compress/snappy/LICENSE generated vendored Normal file
View File

@@ -0,0 +1,27 @@
Copyright (c) 2011 The Snappy-Go Authors. All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above
copyright notice, this list of conditions and the following disclaimer
in the documentation and/or other materials provided with the
distribution.
* Neither the name of Google Inc. nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

107
vendor/github.com/klauspost/compress/snappy/README generated vendored Normal file
View File

@@ -0,0 +1,107 @@
The Snappy compression format in the Go programming language.
To download and install from source:
$ go get github.com/golang/snappy
Unless otherwise noted, the Snappy-Go source files are distributed
under the BSD-style license found in the LICENSE file.
Benchmarks.
The golang/snappy benchmarks include compressing (Z) and decompressing (U) ten
or so files, the same set used by the C++ Snappy code (github.com/google/snappy
and note the "google", not "golang"). On an "Intel(R) Core(TM) i7-3770 CPU @
3.40GHz", Go's GOARCH=amd64 numbers as of 2016-05-29:
"go test -test.bench=."
_UFlat0-8 2.19GB/s ± 0% html
_UFlat1-8 1.41GB/s ± 0% urls
_UFlat2-8 23.5GB/s ± 2% jpg
_UFlat3-8 1.91GB/s ± 0% jpg_200
_UFlat4-8 14.0GB/s ± 1% pdf
_UFlat5-8 1.97GB/s ± 0% html4
_UFlat6-8 814MB/s ± 0% txt1
_UFlat7-8 785MB/s ± 0% txt2
_UFlat8-8 857MB/s ± 0% txt3
_UFlat9-8 719MB/s ± 1% txt4
_UFlat10-8 2.84GB/s ± 0% pb
_UFlat11-8 1.05GB/s ± 0% gaviota
_ZFlat0-8 1.04GB/s ± 0% html
_ZFlat1-8 534MB/s ± 0% urls
_ZFlat2-8 15.7GB/s ± 1% jpg
_ZFlat3-8 740MB/s ± 3% jpg_200
_ZFlat4-8 9.20GB/s ± 1% pdf
_ZFlat5-8 991MB/s ± 0% html4
_ZFlat6-8 379MB/s ± 0% txt1
_ZFlat7-8 352MB/s ± 0% txt2
_ZFlat8-8 396MB/s ± 1% txt3
_ZFlat9-8 327MB/s ± 1% txt4
_ZFlat10-8 1.33GB/s ± 1% pb
_ZFlat11-8 605MB/s ± 1% gaviota
"go test -test.bench=. -tags=noasm"
_UFlat0-8 621MB/s ± 2% html
_UFlat1-8 494MB/s ± 1% urls
_UFlat2-8 23.2GB/s ± 1% jpg
_UFlat3-8 1.12GB/s ± 1% jpg_200
_UFlat4-8 4.35GB/s ± 1% pdf
_UFlat5-8 609MB/s ± 0% html4
_UFlat6-8 296MB/s ± 0% txt1
_UFlat7-8 288MB/s ± 0% txt2
_UFlat8-8 309MB/s ± 1% txt3
_UFlat9-8 280MB/s ± 1% txt4
_UFlat10-8 753MB/s ± 0% pb
_UFlat11-8 400MB/s ± 0% gaviota
_ZFlat0-8 409MB/s ± 1% html
_ZFlat1-8 250MB/s ± 1% urls
_ZFlat2-8 12.3GB/s ± 1% jpg
_ZFlat3-8 132MB/s ± 0% jpg_200
_ZFlat4-8 2.92GB/s ± 0% pdf
_ZFlat5-8 405MB/s ± 1% html4
_ZFlat6-8 179MB/s ± 1% txt1
_ZFlat7-8 170MB/s ± 1% txt2
_ZFlat8-8 189MB/s ± 1% txt3
_ZFlat9-8 164MB/s ± 1% txt4
_ZFlat10-8 479MB/s ± 1% pb
_ZFlat11-8 270MB/s ± 1% gaviota
For comparison (Go's encoded output is byte-for-byte identical to C++'s), here
are the numbers from C++ Snappy's
make CXXFLAGS="-O2 -DNDEBUG -g" clean snappy_unittest.log && cat snappy_unittest.log
BM_UFlat/0 2.4GB/s html
BM_UFlat/1 1.4GB/s urls
BM_UFlat/2 21.8GB/s jpg
BM_UFlat/3 1.5GB/s jpg_200
BM_UFlat/4 13.3GB/s pdf
BM_UFlat/5 2.1GB/s html4
BM_UFlat/6 1.0GB/s txt1
BM_UFlat/7 959.4MB/s txt2
BM_UFlat/8 1.0GB/s txt3
BM_UFlat/9 864.5MB/s txt4
BM_UFlat/10 2.9GB/s pb
BM_UFlat/11 1.2GB/s gaviota
BM_ZFlat/0 944.3MB/s html (22.31 %)
BM_ZFlat/1 501.6MB/s urls (47.78 %)
BM_ZFlat/2 14.3GB/s jpg (99.95 %)
BM_ZFlat/3 538.3MB/s jpg_200 (73.00 %)
BM_ZFlat/4 8.3GB/s pdf (83.30 %)
BM_ZFlat/5 903.5MB/s html4 (22.52 %)
BM_ZFlat/6 336.0MB/s txt1 (57.88 %)
BM_ZFlat/7 312.3MB/s txt2 (61.91 %)
BM_ZFlat/8 353.1MB/s txt3 (54.99 %)
BM_ZFlat/9 289.9MB/s txt4 (66.26 %)
BM_ZFlat/10 1.2GB/s pb (19.68 %)
BM_ZFlat/11 527.4MB/s gaviota (37.72 %)

77
vendor/github.com/klauspost/compress/snappy/cmd/snappytool/main.cpp generated vendored Normal file
View File

@@ -0,0 +1,77 @@
/*
To build the snappytool binary:
g++ main.cpp /usr/lib/libsnappy.a -o snappytool
or, if you have built the C++ snappy library from source:
g++ main.cpp /path/to/your/snappy/.libs/libsnappy.a -o snappytool
after running "make" from your snappy checkout directory.
*/
#include <errno.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include "snappy.h"
#define N 1000000
char dst[N];
char src[N];
int main(int argc, char** argv) {
// Parse args.
if (argc != 2) {
fprintf(stderr, "exactly one of -d or -e must be given\n");
return 1;
}
bool decode = strcmp(argv[1], "-d") == 0;
bool encode = strcmp(argv[1], "-e") == 0;
if (decode == encode) {
fprintf(stderr, "exactly one of -d or -e must be given\n");
return 1;
}
// Read all of stdin into src[:s].
size_t s = 0;
while (1) {
if (s == N) {
fprintf(stderr, "input too large\n");
return 1;
}
ssize_t n = read(0, src+s, N-s);
if (n == 0) {
break;
}
if (n < 0) {
fprintf(stderr, "read error: %s\n", strerror(errno));
// TODO: handle EAGAIN, EINTR?
return 1;
}
s += n;
}
// Encode or decode src[:s] to dst[:d], and write to stdout.
size_t d = 0;
if (encode) {
if (N < snappy::MaxCompressedLength(s)) {
fprintf(stderr, "input too large after encoding\n");
return 1;
}
snappy::RawCompress(src, s, dst, &d);
} else {
if (!snappy::GetUncompressedLength(src, s, &d)) {
fprintf(stderr, "could not get uncompressed length\n");
return 1;
}
if (N < d) {
fprintf(stderr, "input too large after decoding\n");
return 1;
}
if (!snappy::RawUncompress(src, s, dst)) {
fprintf(stderr, "input was not valid Snappy-compressed data\n");
return 1;
}
}
write(1, dst, d);
return 0;
}

237
vendor/github.com/klauspost/compress/snappy/decode.go generated vendored Normal file
View File

@@ -0,0 +1,237 @@
// Copyright 2011 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package snappy
import (
"encoding/binary"
"errors"
"io"
)
var (
// ErrCorrupt reports that the input is invalid.
ErrCorrupt = errors.New("snappy: corrupt input")
// ErrTooLarge reports that the uncompressed length is too large.
ErrTooLarge = errors.New("snappy: decoded block is too large")
// ErrUnsupported reports that the input isn't supported.
ErrUnsupported = errors.New("snappy: unsupported input")
errUnsupportedLiteralLength = errors.New("snappy: unsupported literal length")
)
// DecodedLen returns the length of the decoded block.
func DecodedLen(src []byte) (int, error) {
v, _, err := decodedLen(src)
return v, err
}
// decodedLen returns the length of the decoded block and the number of bytes
// that the length header occupied.
func decodedLen(src []byte) (blockLen, headerLen int, err error) {
v, n := binary.Uvarint(src)
if n <= 0 || v > 0xffffffff {
return 0, 0, ErrCorrupt
}
const wordSize = 32 << (^uint(0) >> 32 & 1)
if wordSize == 32 && v > 0x7fffffff {
return 0, 0, ErrTooLarge
}
return int(v), n, nil
}
const (
decodeErrCodeCorrupt = 1
decodeErrCodeUnsupportedLiteralLength = 2
)
// Decode returns the decoded form of src. The returned slice may be a sub-
// slice of dst if dst was large enough to hold the entire decoded block.
// Otherwise, a newly allocated slice will be returned.
//
// The dst and src must not overlap. It is valid to pass a nil dst.
func Decode(dst, src []byte) ([]byte, error) {
dLen, s, err := decodedLen(src)
if err != nil {
return nil, err
}
if dLen <= len(dst) {
dst = dst[:dLen]
} else {
dst = make([]byte, dLen)
}
switch decode(dst, src[s:]) {
case 0:
return dst, nil
case decodeErrCodeUnsupportedLiteralLength:
return nil, errUnsupportedLiteralLength
}
return nil, ErrCorrupt
}
// NewReader returns a new Reader that decompresses from r, using the framing
// format described at
// https://github.com/google/snappy/blob/master/framing_format.txt
func NewReader(r io.Reader) *Reader {
return &Reader{
r: r,
decoded: make([]byte, maxBlockSize),
buf: make([]byte, maxEncodedLenOfMaxBlockSize+checksumSize),
}
}
// Reader is an io.Reader that can read Snappy-compressed bytes.
type Reader struct {
r io.Reader
err error
decoded []byte
buf []byte
// decoded[i:j] contains decoded bytes that have not yet been passed on.
i, j int
readHeader bool
}
// Reset discards any buffered data, resets all state, and switches the Snappy
// reader to read from r. This permits reusing a Reader rather than allocating
// a new one.
func (r *Reader) Reset(reader io.Reader) {
r.r = reader
r.err = nil
r.i = 0
r.j = 0
r.readHeader = false
}
func (r *Reader) readFull(p []byte, allowEOF bool) (ok bool) {
if _, r.err = io.ReadFull(r.r, p); r.err != nil {
if r.err == io.ErrUnexpectedEOF || (r.err == io.EOF && !allowEOF) {
r.err = ErrCorrupt
}
return false
}
return true
}
// Read satisfies the io.Reader interface.
func (r *Reader) Read(p []byte) (int, error) {
if r.err != nil {
return 0, r.err
}
for {
if r.i < r.j {
n := copy(p, r.decoded[r.i:r.j])
r.i += n
return n, nil
}
if !r.readFull(r.buf[:4], true) {
return 0, r.err
}
chunkType := r.buf[0]
if !r.readHeader {
if chunkType != chunkTypeStreamIdentifier {
r.err = ErrCorrupt
return 0, r.err
}
r.readHeader = true
}
chunkLen := int(r.buf[1]) | int(r.buf[2])<<8 | int(r.buf[3])<<16
if chunkLen > len(r.buf) {
r.err = ErrUnsupported
return 0, r.err
}
// The chunk types are specified at
// https://github.com/google/snappy/blob/master/framing_format.txt
switch chunkType {
case chunkTypeCompressedData:
// Section 4.2. Compressed data (chunk type 0x00).
if chunkLen < checksumSize {
r.err = ErrCorrupt
return 0, r.err
}
buf := r.buf[:chunkLen]
if !r.readFull(buf, false) {
return 0, r.err
}
checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
buf = buf[checksumSize:]
n, err := DecodedLen(buf)
if err != nil {
r.err = err
return 0, r.err
}
if n > len(r.decoded) {
r.err = ErrCorrupt
return 0, r.err
}
if _, err := Decode(r.decoded, buf); err != nil {
r.err = err
return 0, r.err
}
if crc(r.decoded[:n]) != checksum {
r.err = ErrCorrupt
return 0, r.err
}
r.i, r.j = 0, n
continue
case chunkTypeUncompressedData:
// Section 4.3. Uncompressed data (chunk type 0x01).
if chunkLen < checksumSize {
r.err = ErrCorrupt
return 0, r.err
}
buf := r.buf[:checksumSize]
if !r.readFull(buf, false) {
return 0, r.err
}
checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
// Read directly into r.decoded instead of via r.buf.
n := chunkLen - checksumSize
if n > len(r.decoded) {
r.err = ErrCorrupt
return 0, r.err
}
if !r.readFull(r.decoded[:n], false) {
return 0, r.err
}
if crc(r.decoded[:n]) != checksum {
r.err = ErrCorrupt
return 0, r.err
}
r.i, r.j = 0, n
continue
case chunkTypeStreamIdentifier:
// Section 4.1. Stream identifier (chunk type 0xff).
if chunkLen != len(magicBody) {
r.err = ErrCorrupt
return 0, r.err
}
if !r.readFull(r.buf[:len(magicBody)], false) {
return 0, r.err
}
for i := 0; i < len(magicBody); i++ {
if r.buf[i] != magicBody[i] {
r.err = ErrCorrupt
return 0, r.err
}
}
continue
}
if chunkType <= 0x7f {
// Section 4.5. Reserved unskippable chunks (chunk types 0x02-0x7f).
r.err = ErrUnsupported
return 0, r.err
}
// Section 4.4 Padding (chunk type 0xfe).
// Section 4.6. Reserved skippable chunks (chunk types 0x80-0xfd).
if !r.readFull(r.buf[:chunkLen], false) {
return 0, r.err
}
}
}

14
vendor/github.com/klauspost/compress/snappy/decode_amd64.go generated vendored Normal file
View File

@@ -0,0 +1,14 @@
// Copyright 2016 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !appengine
// +build gc
// +build !noasm
package snappy
// decode has the same semantics as in decode_other.go.
//
//go:noescape
func decode(dst, src []byte) int

490
vendor/github.com/klauspost/compress/snappy/decode_amd64.s generated vendored Normal file
View File

@@ -0,0 +1,490 @@
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !appengine
// +build gc
// +build !noasm
#include "textflag.h"
// The asm code generally follows the pure Go code in decode_other.go, except
// where marked with a "!!!".
// func decode(dst, src []byte) int
//
// All local variables fit into registers. The non-zero stack size is only to
// spill registers and push args when issuing a CALL. The register allocation:
// - AX scratch
// - BX scratch
// - CX length or x
// - DX offset
// - SI &src[s]
// - DI &dst[d]
// + R8 dst_base
// + R9 dst_len
// + R10 dst_base + dst_len
// + R11 src_base
// + R12 src_len
// + R13 src_base + src_len
// - R14 used by doCopy
// - R15 used by doCopy
//
// The registers R8-R13 (marked with a "+") are set at the start of the
// function, and after a CALL returns, and are not otherwise modified.
//
// The d variable is implicitly DI - R8, and len(dst)-d is R10 - DI.
// The s variable is implicitly SI - R11, and len(src)-s is R13 - SI.
TEXT ·decode(SB), NOSPLIT, $48-56
// Initialize SI, DI and R8-R13.
MOVQ dst_base+0(FP), R8
MOVQ dst_len+8(FP), R9
MOVQ R8, DI
MOVQ R8, R10
ADDQ R9, R10
MOVQ src_base+24(FP), R11
MOVQ src_len+32(FP), R12
MOVQ R11, SI
MOVQ R11, R13
ADDQ R12, R13
loop:
// for s < len(src)
CMPQ SI, R13
JEQ end
// CX = uint32(src[s])
//
// switch src[s] & 0x03
MOVBLZX (SI), CX
MOVL CX, BX
ANDL $3, BX
CMPL BX, $1
JAE tagCopy
// ----------------------------------------
// The code below handles literal tags.
// case tagLiteral:
// x := uint32(src[s] >> 2)
// switch
SHRL $2, CX
CMPL CX, $60
JAE tagLit60Plus
// case x < 60:
// s++
INCQ SI
doLit:
// This is the end of the inner "switch", when we have a literal tag.
//
// We assume that CX == x and x fits in a uint32, where x is the variable
// used in the pure Go decode_other.go code.
// length = int(x) + 1
//
// Unlike the pure Go code, we don't need to check if length <= 0 because
// CX can hold 64 bits, so the increment cannot overflow.
INCQ CX
// Prepare to check if copying length bytes will run past the end of dst or
// src.
//
// AX = len(dst) - d
// BX = len(src) - s
MOVQ R10, AX
SUBQ DI, AX
MOVQ R13, BX
SUBQ SI, BX
// !!! Try a faster technique for short (16 or fewer bytes) copies.
//
// if length > 16 || len(dst)-d < 16 || len(src)-s < 16 {
// goto callMemmove // Fall back on calling runtime·memmove.
// }
//
// The C++ snappy code calls this TryFastAppend. It also checks len(src)-s
// against 21 instead of 16, because it cannot assume that all of its input
// is contiguous in memory and so it needs to leave enough source bytes to
// read the next tag without refilling buffers, but Go's Decode assumes
// contiguousness (the src argument is a []byte).
CMPQ CX, $16
JGT callMemmove
CMPQ AX, $16
JLT callMemmove
CMPQ BX, $16
JLT callMemmove
// !!! Implement the copy from src to dst as a 16-byte load and store.
// (Decode's documentation says that dst and src must not overlap.)
//
// This always copies 16 bytes, instead of only length bytes, but that's
// OK. If the input is a valid Snappy encoding then subsequent iterations
// will fix up the overrun. Otherwise, Decode returns a nil []byte (and a
// non-nil error), so the overrun will be ignored.
//
// Note that on amd64, it is legal and cheap to issue unaligned 8-byte or
// 16-byte loads and stores. This technique probably wouldn't be as
// effective on architectures that are fussier about alignment.
MOVOU 0(SI), X0
MOVOU X0, 0(DI)
// d += length
// s += length
ADDQ CX, DI
ADDQ CX, SI
JMP loop
callMemmove:
// if length > len(dst)-d || length > len(src)-s { etc }
CMPQ CX, AX
JGT errCorrupt
CMPQ CX, BX
JGT errCorrupt
// copy(dst[d:], src[s:s+length])
//
// This means calling runtime·memmove(&dst[d], &src[s], length), so we push
// DI, SI and CX as arguments. Coincidentally, we also need to spill those
// three registers to the stack, to save local variables across the CALL.
MOVQ DI, 0(SP)
MOVQ SI, 8(SP)
MOVQ CX, 16(SP)
MOVQ DI, 24(SP)
MOVQ SI, 32(SP)
MOVQ CX, 40(SP)
CALL runtime·memmove(SB)
// Restore local variables: unspill registers from the stack and
// re-calculate R8-R13.
MOVQ 24(SP), DI
MOVQ 32(SP), SI
MOVQ 40(SP), CX
MOVQ dst_base+0(FP), R8
MOVQ dst_len+8(FP), R9
MOVQ R8, R10
ADDQ R9, R10
MOVQ src_base+24(FP), R11
MOVQ src_len+32(FP), R12
MOVQ R11, R13
ADDQ R12, R13
// d += length
// s += length
ADDQ CX, DI
ADDQ CX, SI
JMP loop
tagLit60Plus:
// !!! This fragment does the
//
// s += x - 58; if uint(s) > uint(len(src)) { etc }
//
// checks. In the asm version, we code it once instead of once per switch case.
ADDQ CX, SI
SUBQ $58, SI
MOVQ SI, BX
SUBQ R11, BX
CMPQ BX, R12
JA errCorrupt
// case x == 60:
CMPL CX, $61
JEQ tagLit61
JA tagLit62Plus
// x = uint32(src[s-1])
MOVBLZX -1(SI), CX
JMP doLit
tagLit61:
// case x == 61:
// x = uint32(src[s-2]) | uint32(src[s-1])<<8
MOVWLZX -2(SI), CX
JMP doLit
tagLit62Plus:
CMPL CX, $62
JA tagLit63
// case x == 62:
// x = uint32(src[s-3]) | uint32(src[s-2])<<8 | uint32(src[s-1])<<16
MOVWLZX -3(SI), CX
MOVBLZX -1(SI), BX
SHLL $16, BX
ORL BX, CX
JMP doLit
tagLit63:
// case x == 63:
// x = uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24
MOVL -4(SI), CX
JMP doLit
// The code above handles literal tags.
// ----------------------------------------
// The code below handles copy tags.
tagCopy4:
// case tagCopy4:
// s += 5
ADDQ $5, SI
// if uint(s) > uint(len(src)) { etc }
MOVQ SI, BX
SUBQ R11, BX
CMPQ BX, R12
JA errCorrupt
// length = 1 + int(src[s-5])>>2
SHRQ $2, CX
INCQ CX
// offset = int(uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24)
MOVLQZX -4(SI), DX
JMP doCopy
tagCopy2:
// case tagCopy2:
// s += 3
ADDQ $3, SI
// if uint(s) > uint(len(src)) { etc }
MOVQ SI, BX
SUBQ R11, BX
CMPQ BX, R12
JA errCorrupt
// length = 1 + int(src[s-3])>>2
SHRQ $2, CX
INCQ CX
// offset = int(uint32(src[s-2]) | uint32(src[s-1])<<8)
MOVWQZX -2(SI), DX
JMP doCopy
tagCopy:
// We have a copy tag. We assume that:
// - BX == src[s] & 0x03
// - CX == src[s]
CMPQ BX, $2
JEQ tagCopy2
JA tagCopy4
// case tagCopy1:
// s += 2
ADDQ $2, SI
// if uint(s) > uint(len(src)) { etc }
MOVQ SI, BX
SUBQ R11, BX
CMPQ BX, R12
JA errCorrupt
// offset = int(uint32(src[s-2])&0xe0<<3 | uint32(src[s-1]))
MOVQ CX, DX
ANDQ $0xe0, DX
SHLQ $3, DX
MOVBQZX -1(SI), BX
ORQ BX, DX
// length = 4 + int(src[s-2])>>2&0x7
SHRQ $2, CX
ANDQ $7, CX
ADDQ $4, CX
doCopy:
// This is the end of the outer "switch", when we have a copy tag.
//
// We assume that:
// - CX == length && CX > 0
// - DX == offset
// if offset <= 0 { etc }
CMPQ DX, $0
JLE errCorrupt
// if d < offset { etc }
MOVQ DI, BX
SUBQ R8, BX
CMPQ BX, DX
JLT errCorrupt
// if length > len(dst)-d { etc }
MOVQ R10, BX
SUBQ DI, BX
CMPQ CX, BX
JGT errCorrupt
// forwardCopy(dst[d:d+length], dst[d-offset:]); d += length
//
// Set:
// - R14 = len(dst)-d
// - R15 = &dst[d-offset]
MOVQ R10, R14
SUBQ DI, R14
MOVQ DI, R15
SUBQ DX, R15
// !!! Try a faster technique for short (16 or fewer bytes) forward copies.
//
// First, try using two 8-byte load/stores, similar to the doLit technique
// above. Even if dst[d:d+length] and dst[d-offset:] can overlap, this is
// still OK if offset >= 8. Note that this has to be two 8-byte load/stores
// and not one 16-byte load/store, and the first store has to be before the
// second load, due to the overlap if offset is in the range [8, 16).
//
// if length > 16 || offset < 8 || len(dst)-d < 16 {
// goto slowForwardCopy
// }
// copy 16 bytes
// d += length
CMPQ CX, $16
JGT slowForwardCopy
CMPQ DX, $8
JLT slowForwardCopy
CMPQ R14, $16
JLT slowForwardCopy
MOVQ 0(R15), AX
MOVQ AX, 0(DI)
MOVQ 8(R15), BX
MOVQ BX, 8(DI)
ADDQ CX, DI
JMP loop
slowForwardCopy:
// !!! If the forward copy is longer than 16 bytes, or if offset < 8, we
// can still try 8-byte load stores, provided we can overrun up to 10 extra
// bytes. As above, the overrun will be fixed up by subsequent iterations
// of the outermost loop.
//
// The C++ snappy code calls this technique IncrementalCopyFastPath. Its
// commentary says:
//
// ----
//
// The main part of this loop is a simple copy of eight bytes at a time
// until we've copied (at least) the requested amount of bytes. However,
// if d and d-offset are less than eight bytes apart (indicating a
// repeating pattern of length < 8), we first need to expand the pattern in
// order to get the correct results. For instance, if the buffer looks like
// this, with the eight-byte <d-offset> and <d> patterns marked as
// intervals:
//
// abxxxxxxxxxxxx
// [------] d-offset
// [------] d
//
// a single eight-byte copy from <d-offset> to <d> will repeat the pattern
// once, after which we can move <d> two bytes without moving <d-offset>:
//
// ababxxxxxxxxxx
// [------] d-offset
// [------] d
//
// and repeat the exercise until the two no longer overlap.
//
// This allows us to do very well in the special case of one single byte
// repeated many times, without taking a big hit for more general cases.
//
// The worst case of extra writing past the end of the match occurs when
// offset == 1 and length == 1; the last copy will read from byte positions
// [0..7] and write to [4..11], whereas it was only supposed to write to
// position 1. Thus, ten excess bytes.
//
// ----
//
// That "10 byte overrun" worst case is confirmed by Go's
// TestSlowForwardCopyOverrun, which also tests the fixUpSlowForwardCopy
// and finishSlowForwardCopy algorithm.
//
// if length > len(dst)-d-10 {
// goto verySlowForwardCopy
// }
SUBQ $10, R14
CMPQ CX, R14
JGT verySlowForwardCopy
makeOffsetAtLeast8:
// !!! As above, expand the pattern so that offset >= 8 and we can use
// 8-byte load/stores.
//
// for offset < 8 {
// copy 8 bytes from dst[d-offset:] to dst[d:]
// length -= offset
// d += offset
// offset += offset
// // The two previous lines together means that d-offset, and therefore
// // R15, is unchanged.
// }
CMPQ DX, $8
JGE fixUpSlowForwardCopy
MOVQ (R15), BX
MOVQ BX, (DI)
SUBQ DX, CX
ADDQ DX, DI
ADDQ DX, DX
JMP makeOffsetAtLeast8
fixUpSlowForwardCopy:
// !!! Add length (which might be negative now) to d (implied by DI being
// &dst[d]) so that d ends up at the right place when we jump back to the
// top of the loop. Before we do that, though, we save DI to AX so that, if
// length is positive, copying the remaining length bytes will write to the
// right place.
MOVQ DI, AX
ADDQ CX, DI
finishSlowForwardCopy:
// !!! Repeat 8-byte load/stores until length <= 0. Ending with a negative
// length means that we overrun, but as above, that will be fixed up by
// subsequent iterations of the outermost loop.
CMPQ CX, $0
JLE loop
MOVQ (R15), BX
MOVQ BX, (AX)
ADDQ $8, R15
ADDQ $8, AX
SUBQ $8, CX
JMP finishSlowForwardCopy
verySlowForwardCopy:
// verySlowForwardCopy is a simple implementation of forward copy. In C
// parlance, this is a do/while loop instead of a while loop, since we know
// that length > 0. In Go syntax:
//
// for {
// dst[d] = dst[d - offset]
// d++
// length--
// if length == 0 {
// break
// }
// }
MOVB (R15), BX
MOVB BX, (DI)
INCQ R15
INCQ DI
DECQ CX
JNZ verySlowForwardCopy
JMP loop
// The code above handles copy tags.
// ----------------------------------------
end:
// This is the end of the "for s < len(src)".
//
// if d != len(dst) { etc }
CMPQ DI, R10
JNE errCorrupt
// return 0
MOVQ $0, ret+48(FP)
RET
errCorrupt:
// return decodeErrCodeCorrupt
MOVQ $1, ret+48(FP)
RET

101
vendor/github.com/klauspost/compress/snappy/decode_other.go generated vendored Normal file
View File

@@ -0,0 +1,101 @@
// Copyright 2016 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !amd64 appengine !gc noasm
package snappy
// decode writes the decoding of src to dst. It assumes that the varint-encoded
// length of the decompressed bytes has already been read, and that len(dst)
// equals that length.
//
// It returns 0 on success or a decodeErrCodeXxx error code on failure.
func decode(dst, src []byte) int {
var d, s, offset, length int
for s < len(src) {
switch src[s] & 0x03 {
case tagLiteral:
x := uint32(src[s] >> 2)
switch {
case x < 60:
s++
case x == 60:
s += 2
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
x = uint32(src[s-1])
case x == 61:
s += 3
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
x = uint32(src[s-2]) | uint32(src[s-1])<<8
case x == 62:
s += 4
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
x = uint32(src[s-3]) | uint32(src[s-2])<<8 | uint32(src[s-1])<<16
case x == 63:
s += 5
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
x = uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24
}
length = int(x) + 1
if length <= 0 {
return decodeErrCodeUnsupportedLiteralLength
}
if length > len(dst)-d || length > len(src)-s {
return decodeErrCodeCorrupt
}
copy(dst[d:], src[s:s+length])
d += length
s += length
continue
case tagCopy1:
s += 2
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
length = 4 + int(src[s-2])>>2&0x7
offset = int(uint32(src[s-2])&0xe0<<3 | uint32(src[s-1]))
case tagCopy2:
s += 3
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
length = 1 + int(src[s-3])>>2
offset = int(uint32(src[s-2]) | uint32(src[s-1])<<8)
case tagCopy4:
s += 5
if uint(s) > uint(len(src)) { // The uint conversions catch overflow from the previous line.
return decodeErrCodeCorrupt
}
length = 1 + int(src[s-5])>>2
offset = int(uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24)
}
if offset <= 0 || d < offset || length > len(dst)-d {
return decodeErrCodeCorrupt
}
// Copy from an earlier sub-slice of dst to a later sub-slice. Unlike
// the built-in copy function, this byte-by-byte copy always runs
// forwards, even if the slices overlap. Conceptually, this is:
//
// d += forwardCopy(dst[d:d+length], dst[d-offset:])
for end := d + length; d != end; d++ {
dst[d] = dst[d-offset]
}
}
if d != len(dst) {
return decodeErrCodeCorrupt
}
return 0
}

285
vendor/github.com/klauspost/compress/snappy/encode.go generated vendored Normal file
View File

@@ -0,0 +1,285 @@
// Copyright 2011 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package snappy
import (
"encoding/binary"
"errors"
"io"
)
// Encode returns the encoded form of src. The returned slice may be a sub-
// slice of dst if dst was large enough to hold the entire encoded block.
// Otherwise, a newly allocated slice will be returned.
//
// The dst and src must not overlap. It is valid to pass a nil dst.
func Encode(dst, src []byte) []byte {
if n := MaxEncodedLen(len(src)); n < 0 {
panic(ErrTooLarge)
} else if len(dst) < n {
dst = make([]byte, n)
}
// The block starts with the varint-encoded length of the decompressed bytes.
d := binary.PutUvarint(dst, uint64(len(src)))
for len(src) > 0 {
p := src
src = nil
if len(p) > maxBlockSize {
p, src = p[:maxBlockSize], p[maxBlockSize:]
}
if len(p) < minNonLiteralBlockSize {
d += emitLiteral(dst[d:], p)
} else {
d += encodeBlock(dst[d:], p)
}
}
return dst[:d]
}
// inputMargin is the minimum number of extra input bytes to keep, inside
// encodeBlock's inner loop. On some architectures, this margin lets us
// implement a fast path for emitLiteral, where the copy of short (<= 16 byte)
// literals can be implemented as a single load to and store from a 16-byte
// register. That literal's actual length can be as short as 1 byte, so this
// can copy up to 15 bytes too much, but that's OK as subsequent iterations of
// the encoding loop will fix up the copy overrun, and this inputMargin ensures
// that we don't overrun the dst and src buffers.
const inputMargin = 16 - 1
// minNonLiteralBlockSize is the minimum size of the input to encodeBlock that
// could be encoded with a copy tag. This is the minimum with respect to the
// algorithm used by encodeBlock, not a minimum enforced by the file format.
//
// The encoded output must start with at least a 1 byte literal, as there are
// no previous bytes to copy. A minimal (1 byte) copy after that, generated
// from an emitCopy call in encodeBlock's main loop, would require at least
// another inputMargin bytes, for the reason above: we want any emitLiteral
// calls inside encodeBlock's main loop to use the fast path if possible, which
// requires being able to overrun by inputMargin bytes. Thus,
// minNonLiteralBlockSize equals 1 + 1 + inputMargin.
//
// The C++ code doesn't use this exact threshold, but it could, as discussed at
// https://groups.google.com/d/topic/snappy-compression/oGbhsdIJSJ8/discussion
// The difference between Go (2+inputMargin) and C++ (inputMargin) is purely an
// optimization. It should not affect the encoded form. This is tested by
// TestSameEncodingAsCppShortCopies.
const minNonLiteralBlockSize = 1 + 1 + inputMargin
// MaxEncodedLen returns the maximum length of a snappy block, given its
// uncompressed length.
//
// It will return a negative value if srcLen is too large to encode.
func MaxEncodedLen(srcLen int) int {
n := uint64(srcLen)
if n > 0xffffffff {
return -1
}
// Compressed data can be defined as:
// compressed := item* literal*
// item := literal* copy
//
// The trailing literal sequence has a space blowup of at most 62/60
// since a literal of length 60 needs one tag byte + one extra byte
// for length information.
//
// Item blowup is trickier to measure. Suppose the "copy" op copies
// 4 bytes of data. Because of a special check in the encoding code,
// we produce a 4-byte copy only if the offset is < 65536. Therefore
// the copy op takes 3 bytes to encode, and this type of item leads
// to at most the 62/60 blowup for representing literals.
//
// Suppose the "copy" op copies 5 bytes of data. If the offset is big
// enough, it will take 5 bytes to encode the copy op. Therefore the
// worst case here is a one-byte literal followed by a five-byte copy.
// That is, 6 bytes of input turn into 7 bytes of "compressed" data.
//
// This last factor dominates the blowup, so the final estimate is:
n = 32 + n + n/6
if n > 0xffffffff {
return -1
}
return int(n)
}
var errClosed = errors.New("snappy: Writer is closed")
// NewWriter returns a new Writer that compresses to w.
//
// The Writer returned does not buffer writes. There is no need to Flush or
// Close such a Writer.
//
// Deprecated: the Writer returned is not suitable for many small writes, only
// for few large writes. Use NewBufferedWriter instead, which is efficient
// regardless of the frequency and shape of the writes, and remember to Close
// that Writer when done.
func NewWriter(w io.Writer) *Writer {
return &Writer{
w: w,
obuf: make([]byte, obufLen),
}
}
// NewBufferedWriter returns a new Writer that compresses to w, using the
// framing format described at
// https://github.com/google/snappy/blob/master/framing_format.txt
//
// The Writer returned buffers writes. Users must call Close to guarantee all
// data has been forwarded to the underlying io.Writer. They may also call
// Flush zero or more times before calling Close.
func NewBufferedWriter(w io.Writer) *Writer {
return &Writer{
w: w,
ibuf: make([]byte, 0, maxBlockSize),
obuf: make([]byte, obufLen),
}
}
// Writer is an io.Writer than can write Snappy-compressed bytes.
type Writer struct {
w io.Writer
err error
// ibuf is a buffer for the incoming (uncompressed) bytes.
//
// Its use is optional. For backwards compatibility, Writers created by the
// NewWriter function have ibuf == nil, do not buffer incoming bytes, and
// therefore do not need to be Flush'ed or Close'd.
ibuf []byte
// obuf is a buffer for the outgoing (compressed) bytes.
obuf []byte
// wroteStreamHeader is whether we have written the stream header.
wroteStreamHeader bool
}
// Reset discards the writer's state and switches the Snappy writer to write to
// w. This permits reusing a Writer rather than allocating a new one.
func (w *Writer) Reset(writer io.Writer) {
w.w = writer
w.err = nil
if w.ibuf != nil {
w.ibuf = w.ibuf[:0]
}
w.wroteStreamHeader = false
}
// Write satisfies the io.Writer interface.
func (w *Writer) Write(p []byte) (nRet int, errRet error) {
if w.ibuf == nil {
// Do not buffer incoming bytes. This does not perform or compress well
// if the caller of Writer.Write writes many small slices. This
// behavior is therefore deprecated, but still supported for backwards
// compatibility with code that doesn't explicitly Flush or Close.
return w.write(p)
}
// The remainder of this method is based on bufio.Writer.Write from the
// standard library.
for len(p) > (cap(w.ibuf)-len(w.ibuf)) && w.err == nil {
var n int
if len(w.ibuf) == 0 {
// Large write, empty buffer.
// Write directly from p to avoid copy.
n, _ = w.write(p)
} else {
n = copy(w.ibuf[len(w.ibuf):cap(w.ibuf)], p)
w.ibuf = w.ibuf[:len(w.ibuf)+n]
w.Flush()
}
nRet += n
p = p[n:]
}
if w.err != nil {
return nRet, w.err
}
n := copy(w.ibuf[len(w.ibuf):cap(w.ibuf)], p)
w.ibuf = w.ibuf[:len(w.ibuf)+n]
nRet += n
return nRet, nil
}
func (w *Writer) write(p []byte) (nRet int, errRet error) {
if w.err != nil {
return 0, w.err
}
for len(p) > 0 {
obufStart := len(magicChunk)
if !w.wroteStreamHeader {
w.wroteStreamHeader = true
copy(w.obuf, magicChunk)
obufStart = 0
}
var uncompressed []byte
if len(p) > maxBlockSize {
uncompressed, p = p[:maxBlockSize], p[maxBlockSize:]
} else {
uncompressed, p = p, nil
}
checksum := crc(uncompressed)
// Compress the buffer, discarding the result if the improvement
// isn't at least 12.5%.
compressed := Encode(w.obuf[obufHeaderLen:], uncompressed)
chunkType := uint8(chunkTypeCompressedData)
chunkLen := 4 + len(compressed)
obufEnd := obufHeaderLen + len(compressed)
if len(compressed) >= len(uncompressed)-len(uncompressed)/8 {
chunkType = chunkTypeUncompressedData
chunkLen = 4 + len(uncompressed)
obufEnd = obufHeaderLen
}
// Fill in the per-chunk header that comes before the body.
w.obuf[len(magicChunk)+0] = chunkType
w.obuf[len(magicChunk)+1] = uint8(chunkLen >> 0)
w.obuf[len(magicChunk)+2] = uint8(chunkLen >> 8)
w.obuf[len(magicChunk)+3] = uint8(chunkLen >> 16)
w.obuf[len(magicChunk)+4] = uint8(checksum >> 0)
w.obuf[len(magicChunk)+5] = uint8(checksum >> 8)
w.obuf[len(magicChunk)+6] = uint8(checksum >> 16)
w.obuf[len(magicChunk)+7] = uint8(checksum >> 24)
if _, err := w.w.Write(w.obuf[obufStart:obufEnd]); err != nil {
w.err = err
return nRet, err
}
if chunkType == chunkTypeUncompressedData {
if _, err := w.w.Write(uncompressed); err != nil {
w.err = err
return nRet, err
}
}
nRet += len(uncompressed)
}
return nRet, nil
}
// Flush flushes the Writer to its underlying io.Writer.
func (w *Writer) Flush() error {
if w.err != nil {
return w.err
}
if len(w.ibuf) == 0 {
return nil
}
w.write(w.ibuf)
w.ibuf = w.ibuf[:0]
return w.err
}
// Close calls Flush and then closes the Writer.
func (w *Writer) Close() error {
w.Flush()
ret := w.err
if w.err == nil {
w.err = errClosed
}
return ret
}

29
vendor/github.com/klauspost/compress/snappy/encode_amd64.go generated vendored Normal file
View File

@@ -0,0 +1,29 @@
// Copyright 2016 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !appengine
// +build gc
// +build !noasm
package snappy
// emitLiteral has the same semantics as in encode_other.go.
//
//go:noescape
func emitLiteral(dst, lit []byte) int
// emitCopy has the same semantics as in encode_other.go.
//
//go:noescape
func emitCopy(dst []byte, offset, length int) int
// extendMatch has the same semantics as in encode_other.go.
//
//go:noescape
func extendMatch(src []byte, i, j int) int
// encodeBlock has the same semantics as in encode_other.go.
//
//go:noescape
func encodeBlock(dst, src []byte) (d int)

730
vendor/github.com/klauspost/compress/snappy/encode_amd64.s generated vendored Normal file
View File

@@ -0,0 +1,730 @@
// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !appengine
// +build gc
// +build !noasm
#include "textflag.h"
// The XXX lines assemble on Go 1.4, 1.5 and 1.7, but not 1.6, due to a
// Go toolchain regression. See https://github.com/golang/go/issues/15426 and
// https://github.com/golang/snappy/issues/29
//
// As a workaround, the package was built with a known good assembler, and
// those instructions were disassembled by "objdump -d" to yield the
// 4e 0f b7 7c 5c 78 movzwq 0x78(%rsp,%r11,2),%r15
// style comments, in AT&T asm syntax. Note that rsp here is a physical
// register, not Go/asm's SP pseudo-register (see https://golang.org/doc/asm).
// The instructions were then encoded as "BYTE $0x.." sequences, which assemble
// fine on Go 1.6.
// The asm code generally follows the pure Go code in encode_other.go, except
// where marked with a "!!!".
// ----------------------------------------------------------------------------
// func emitLiteral(dst, lit []byte) int
//
// All local variables fit into registers. The register allocation:
// - AX len(lit)
// - BX n
// - DX return value
// - DI &dst[i]
// - R10 &lit[0]
//
// The 24 bytes of stack space is to call runtime·memmove.
//
// The unusual register allocation of local variables, such as R10 for the
// source pointer, matches the allocation used at the call site in encodeBlock,
// which makes it easier to manually inline this function.
TEXT ·emitLiteral(SB), NOSPLIT, $24-56
MOVQ dst_base+0(FP), DI
MOVQ lit_base+24(FP), R10
MOVQ lit_len+32(FP), AX
MOVQ AX, DX
MOVL AX, BX
SUBL $1, BX
CMPL BX, $60
JLT oneByte
CMPL BX, $256
JLT twoBytes
threeBytes:
MOVB $0xf4, 0(DI)
MOVW BX, 1(DI)
ADDQ $3, DI
ADDQ $3, DX
JMP memmove
twoBytes:
MOVB $0xf0, 0(DI)
MOVB BX, 1(DI)
ADDQ $2, DI
ADDQ $2, DX
JMP memmove
oneByte:
SHLB $2, BX
MOVB BX, 0(DI)
ADDQ $1, DI
ADDQ $1, DX
memmove:
MOVQ DX, ret+48(FP)
// copy(dst[i:], lit)
//
// This means calling runtime·memmove(&dst[i], &lit[0], len(lit)), so we push
// DI, R10 and AX as arguments.
MOVQ DI, 0(SP)
MOVQ R10, 8(SP)
MOVQ AX, 16(SP)
CALL runtime·memmove(SB)
RET
// ----------------------------------------------------------------------------
// func emitCopy(dst []byte, offset, length int) int
//
// All local variables fit into registers. The register allocation:
// - AX length
// - SI &dst[0]
// - DI &dst[i]
// - R11 offset
//
// The unusual register allocation of local variables, such as R11 for the
// offset, matches the allocation used at the call site in encodeBlock, which
// makes it easier to manually inline this function.
TEXT ·emitCopy(SB), NOSPLIT, $0-48
MOVQ dst_base+0(FP), DI
MOVQ DI, SI
MOVQ offset+24(FP), R11
MOVQ length+32(FP), AX
loop0:
// for length >= 68 { etc }
CMPL AX, $68
JLT step1
// Emit a length 64 copy, encoded as 3 bytes.
MOVB $0xfe, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
SUBL $64, AX
JMP loop0
step1:
// if length > 64 { etc }
CMPL AX, $64
JLE step2
// Emit a length 60 copy, encoded as 3 bytes.
MOVB $0xee, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
SUBL $60, AX
step2:
// if length >= 12 || offset >= 2048 { goto step3 }
CMPL AX, $12
JGE step3
CMPL R11, $2048
JGE step3
// Emit the remaining copy, encoded as 2 bytes.
MOVB R11, 1(DI)
SHRL $8, R11
SHLB $5, R11
SUBB $4, AX
SHLB $2, AX
ORB AX, R11
ORB $1, R11
MOVB R11, 0(DI)
ADDQ $2, DI
// Return the number of bytes written.
SUBQ SI, DI
MOVQ DI, ret+40(FP)
RET
step3:
// Emit the remaining copy, encoded as 3 bytes.
SUBL $1, AX
SHLB $2, AX
ORB $2, AX
MOVB AX, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
// Return the number of bytes written.
SUBQ SI, DI
MOVQ DI, ret+40(FP)
RET
// ----------------------------------------------------------------------------
// func extendMatch(src []byte, i, j int) int
//
// All local variables fit into registers. The register allocation:
// - DX &src[0]
// - SI &src[j]
// - R13 &src[len(src) - 8]
// - R14 &src[len(src)]
// - R15 &src[i]
//
// The unusual register allocation of local variables, such as R15 for a source
// pointer, matches the allocation used at the call site in encodeBlock, which
// makes it easier to manually inline this function.
TEXT ·extendMatch(SB), NOSPLIT, $0-48
MOVQ src_base+0(FP), DX
MOVQ src_len+8(FP), R14
MOVQ i+24(FP), R15
MOVQ j+32(FP), SI
ADDQ DX, R14
ADDQ DX, R15
ADDQ DX, SI
MOVQ R14, R13
SUBQ $8, R13
cmp8:
// As long as we are 8 or more bytes before the end of src, we can load and
// compare 8 bytes at a time. If those 8 bytes are equal, repeat.
CMPQ SI, R13
JA cmp1
MOVQ (R15), AX
MOVQ (SI), BX
CMPQ AX, BX
JNE bsf
ADDQ $8, R15
ADDQ $8, SI
JMP cmp8
bsf:
// If those 8 bytes were not equal, XOR the two 8 byte values, and return
// the index of the first byte that differs. The BSF instruction finds the
// least significant 1 bit, the amd64 architecture is little-endian, and
// the shift by 3 converts a bit index to a byte index.
XORQ AX, BX
BSFQ BX, BX
SHRQ $3, BX
ADDQ BX, SI
// Convert from &src[ret] to ret.
SUBQ DX, SI
MOVQ SI, ret+40(FP)
RET
cmp1:
// In src's tail, compare 1 byte at a time.
CMPQ SI, R14
JAE extendMatchEnd
MOVB (R15), AX
MOVB (SI), BX
CMPB AX, BX
JNE extendMatchEnd
ADDQ $1, R15
ADDQ $1, SI
JMP cmp1
extendMatchEnd:
// Convert from &src[ret] to ret.
SUBQ DX, SI
MOVQ SI, ret+40(FP)
RET
// ----------------------------------------------------------------------------
// func encodeBlock(dst, src []byte) (d int)
//
// All local variables fit into registers, other than "var table". The register
// allocation:
// - AX . .
// - BX . .
// - CX 56 shift (note that amd64 shifts by non-immediates must use CX).
// - DX 64 &src[0], tableSize
// - SI 72 &src[s]
// - DI 80 &dst[d]
// - R9 88 sLimit
// - R10 . &src[nextEmit]
// - R11 96 prevHash, currHash, nextHash, offset
// - R12 104 &src[base], skip
// - R13 . &src[nextS], &src[len(src) - 8]
// - R14 . len(src), bytesBetweenHashLookups, &src[len(src)], x
// - R15 112 candidate
//
// The second column (56, 64, etc) is the stack offset to spill the registers
// when calling other functions. We could pack this slightly tighter, but it's
// simpler to have a dedicated spill map independent of the function called.
//
// "var table [maxTableSize]uint16" takes up 32768 bytes of stack space. An
// extra 56 bytes, to call other functions, and an extra 64 bytes, to spill
// local variables (registers) during calls gives 32768 + 56 + 64 = 32888.
TEXT ·encodeBlock(SB), 0, $32888-56
MOVQ dst_base+0(FP), DI
MOVQ src_base+24(FP), SI
MOVQ src_len+32(FP), R14
// shift, tableSize := uint32(32-8), 1<<8
MOVQ $24, CX
MOVQ $256, DX
calcShift:
// for ; tableSize < maxTableSize && tableSize < len(src); tableSize *= 2 {
// shift--
// }
CMPQ DX, $16384
JGE varTable
CMPQ DX, R14
JGE varTable
SUBQ $1, CX
SHLQ $1, DX
JMP calcShift
varTable:
// var table [maxTableSize]uint16
//
// In the asm code, unlike the Go code, we can zero-initialize only the
// first tableSize elements. Each uint16 element is 2 bytes and each MOVOU
// writes 16 bytes, so we can do only tableSize/8 writes instead of the
// 2048 writes that would zero-initialize all of table's 32768 bytes.
SHRQ $3, DX
LEAQ table-32768(SP), BX
PXOR X0, X0
memclr:
MOVOU X0, 0(BX)
ADDQ $16, BX
SUBQ $1, DX
JNZ memclr
// !!! DX = &src[0]
MOVQ SI, DX
// sLimit := len(src) - inputMargin
MOVQ R14, R9
SUBQ $15, R9
// !!! Pre-emptively spill CX, DX and R9 to the stack. Their values don't
// change for the rest of the function.
MOVQ CX, 56(SP)
MOVQ DX, 64(SP)
MOVQ R9, 88(SP)
// nextEmit := 0
MOVQ DX, R10
// s := 1
ADDQ $1, SI
// nextHash := hash(load32(src, s), shift)
MOVL 0(SI), R11
IMULL $0x1e35a7bd, R11
SHRL CX, R11
outer:
// for { etc }
// skip := 32
MOVQ $32, R12
// nextS := s
MOVQ SI, R13
// candidate := 0
MOVQ $0, R15
inner0:
// for { etc }
// s := nextS
MOVQ R13, SI
// bytesBetweenHashLookups := skip >> 5
MOVQ R12, R14
SHRQ $5, R14
// nextS = s + bytesBetweenHashLookups
ADDQ R14, R13
// skip += bytesBetweenHashLookups
ADDQ R14, R12
// if nextS > sLimit { goto emitRemainder }
MOVQ R13, AX
SUBQ DX, AX
CMPQ AX, R9
JA emitRemainder
// candidate = int(table[nextHash])
// XXX: MOVWQZX table-32768(SP)(R11*2), R15
// XXX: 4e 0f b7 7c 5c 78 movzwq 0x78(%rsp,%r11,2),%r15
BYTE $0x4e
BYTE $0x0f
BYTE $0xb7
BYTE $0x7c
BYTE $0x5c
BYTE $0x78
// table[nextHash] = uint16(s)
MOVQ SI, AX
SUBQ DX, AX
// XXX: MOVW AX, table-32768(SP)(R11*2)
// XXX: 66 42 89 44 5c 78 mov %ax,0x78(%rsp,%r11,2)
BYTE $0x66
BYTE $0x42
BYTE $0x89
BYTE $0x44
BYTE $0x5c
BYTE $0x78
// nextHash = hash(load32(src, nextS), shift)
MOVL 0(R13), R11
IMULL $0x1e35a7bd, R11
SHRL CX, R11
// if load32(src, s) != load32(src, candidate) { continue } break
MOVL 0(SI), AX
MOVL (DX)(R15*1), BX
CMPL AX, BX
JNE inner0
fourByteMatch:
// As per the encode_other.go code:
//
// A 4-byte match has been found. We'll later see etc.
// !!! Jump to a fast path for short (<= 16 byte) literals. See the comment
// on inputMargin in encode.go.
MOVQ SI, AX
SUBQ R10, AX
CMPQ AX, $16
JLE emitLiteralFastPath
// ----------------------------------------
// Begin inline of the emitLiteral call.
//
// d += emitLiteral(dst[d:], src[nextEmit:s])
MOVL AX, BX
SUBL $1, BX
CMPL BX, $60
JLT inlineEmitLiteralOneByte
CMPL BX, $256
JLT inlineEmitLiteralTwoBytes
inlineEmitLiteralThreeBytes:
MOVB $0xf4, 0(DI)
MOVW BX, 1(DI)
ADDQ $3, DI
JMP inlineEmitLiteralMemmove
inlineEmitLiteralTwoBytes:
MOVB $0xf0, 0(DI)
MOVB BX, 1(DI)
ADDQ $2, DI
JMP inlineEmitLiteralMemmove
inlineEmitLiteralOneByte:
SHLB $2, BX
MOVB BX, 0(DI)
ADDQ $1, DI
inlineEmitLiteralMemmove:
// Spill local variables (registers) onto the stack; call; unspill.
//
// copy(dst[i:], lit)
//
// This means calling runtime·memmove(&dst[i], &lit[0], len(lit)), so we push
// DI, R10 and AX as arguments.
MOVQ DI, 0(SP)
MOVQ R10, 8(SP)
MOVQ AX, 16(SP)
ADDQ AX, DI // Finish the "d +=" part of "d += emitLiteral(etc)".
MOVQ SI, 72(SP)
MOVQ DI, 80(SP)
MOVQ R15, 112(SP)
CALL runtime·memmove(SB)
MOVQ 56(SP), CX
MOVQ 64(SP), DX
MOVQ 72(SP), SI
MOVQ 80(SP), DI
MOVQ 88(SP), R9
MOVQ 112(SP), R15
JMP inner1
inlineEmitLiteralEnd:
// End inline of the emitLiteral call.
// ----------------------------------------
emitLiteralFastPath:
// !!! Emit the 1-byte encoding "uint8(len(lit)-1)<<2".
MOVB AX, BX
SUBB $1, BX
SHLB $2, BX
MOVB BX, (DI)
ADDQ $1, DI
// !!! Implement the copy from lit to dst as a 16-byte load and store.
// (Encode's documentation says that dst and src must not overlap.)
//
// This always copies 16 bytes, instead of only len(lit) bytes, but that's
// OK. Subsequent iterations will fix up the overrun.
//
// Note that on amd64, it is legal and cheap to issue unaligned 8-byte or
// 16-byte loads and stores. This technique probably wouldn't be as
// effective on architectures that are fussier about alignment.
MOVOU 0(R10), X0
MOVOU X0, 0(DI)
ADDQ AX, DI
inner1:
// for { etc }
// base := s
MOVQ SI, R12
// !!! offset := base - candidate
MOVQ R12, R11
SUBQ R15, R11
SUBQ DX, R11
// ----------------------------------------
// Begin inline of the extendMatch call.
//
// s = extendMatch(src, candidate+4, s+4)
// !!! R14 = &src[len(src)]
MOVQ src_len+32(FP), R14
ADDQ DX, R14
// !!! R13 = &src[len(src) - 8]
MOVQ R14, R13
SUBQ $8, R13
// !!! R15 = &src[candidate + 4]
ADDQ $4, R15
ADDQ DX, R15
// !!! s += 4
ADDQ $4, SI
inlineExtendMatchCmp8:
// As long as we are 8 or more bytes before the end of src, we can load and
// compare 8 bytes at a time. If those 8 bytes are equal, repeat.
CMPQ SI, R13
JA inlineExtendMatchCmp1
MOVQ (R15), AX
MOVQ (SI), BX
CMPQ AX, BX
JNE inlineExtendMatchBSF
ADDQ $8, R15
ADDQ $8, SI
JMP inlineExtendMatchCmp8
inlineExtendMatchBSF:
// If those 8 bytes were not equal, XOR the two 8 byte values, and return
// the index of the first byte that differs. The BSF instruction finds the
// least significant 1 bit, the amd64 architecture is little-endian, and
// the shift by 3 converts a bit index to a byte index.
XORQ AX, BX
BSFQ BX, BX
SHRQ $3, BX
ADDQ BX, SI
JMP inlineExtendMatchEnd
inlineExtendMatchCmp1:
// In src's tail, compare 1 byte at a time.
CMPQ SI, R14
JAE inlineExtendMatchEnd
MOVB (R15), AX
MOVB (SI), BX
CMPB AX, BX
JNE inlineExtendMatchEnd
ADDQ $1, R15
ADDQ $1, SI
JMP inlineExtendMatchCmp1
inlineExtendMatchEnd:
// End inline of the extendMatch call.
// ----------------------------------------
// ----------------------------------------
// Begin inline of the emitCopy call.
//
// d += emitCopy(dst[d:], base-candidate, s-base)
// !!! length := s - base
MOVQ SI, AX
SUBQ R12, AX
inlineEmitCopyLoop0:
// for length >= 68 { etc }
CMPL AX, $68
JLT inlineEmitCopyStep1
// Emit a length 64 copy, encoded as 3 bytes.
MOVB $0xfe, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
SUBL $64, AX
JMP inlineEmitCopyLoop0
inlineEmitCopyStep1:
// if length > 64 { etc }
CMPL AX, $64
JLE inlineEmitCopyStep2
// Emit a length 60 copy, encoded as 3 bytes.
MOVB $0xee, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
SUBL $60, AX
inlineEmitCopyStep2:
// if length >= 12 || offset >= 2048 { goto inlineEmitCopyStep3 }
CMPL AX, $12
JGE inlineEmitCopyStep3
CMPL R11, $2048
JGE inlineEmitCopyStep3
// Emit the remaining copy, encoded as 2 bytes.
MOVB R11, 1(DI)
SHRL $8, R11
SHLB $5, R11
SUBB $4, AX
SHLB $2, AX
ORB AX, R11
ORB $1, R11
MOVB R11, 0(DI)
ADDQ $2, DI
JMP inlineEmitCopyEnd
inlineEmitCopyStep3:
// Emit the remaining copy, encoded as 3 bytes.
SUBL $1, AX
SHLB $2, AX
ORB $2, AX
MOVB AX, 0(DI)
MOVW R11, 1(DI)
ADDQ $3, DI
inlineEmitCopyEnd:
// End inline of the emitCopy call.
// ----------------------------------------
// nextEmit = s
MOVQ SI, R10
// if s >= sLimit { goto emitRemainder }
MOVQ SI, AX
SUBQ DX, AX
CMPQ AX, R9
JAE emitRemainder
// As per the encode_other.go code:
//
// We could immediately etc.
// x := load64(src, s-1)
MOVQ -1(SI), R14
// prevHash := hash(uint32(x>>0), shift)
MOVL R14, R11
IMULL $0x1e35a7bd, R11
SHRL CX, R11
// table[prevHash] = uint16(s-1)
MOVQ SI, AX
SUBQ DX, AX
SUBQ $1, AX
// XXX: MOVW AX, table-32768(SP)(R11*2)
// XXX: 66 42 89 44 5c 78 mov %ax,0x78(%rsp,%r11,2)
BYTE $0x66
BYTE $0x42
BYTE $0x89
BYTE $0x44
BYTE $0x5c
BYTE $0x78
// currHash := hash(uint32(x>>8), shift)
SHRQ $8, R14
MOVL R14, R11
IMULL $0x1e35a7bd, R11
SHRL CX, R11
// candidate = int(table[currHash])
// XXX: MOVWQZX table-32768(SP)(R11*2), R15
// XXX: 4e 0f b7 7c 5c 78 movzwq 0x78(%rsp,%r11,2),%r15
BYTE $0x4e
BYTE $0x0f
BYTE $0xb7
BYTE $0x7c
BYTE $0x5c
BYTE $0x78
// table[currHash] = uint16(s)
ADDQ $1, AX
// XXX: MOVW AX, table-32768(SP)(R11*2)
// XXX: 66 42 89 44 5c 78 mov %ax,0x78(%rsp,%r11,2)
BYTE $0x66
BYTE $0x42
BYTE $0x89
BYTE $0x44
BYTE $0x5c
BYTE $0x78
// if uint32(x>>8) == load32(src, candidate) { continue }
MOVL (DX)(R15*1), BX
CMPL R14, BX
JEQ inner1
// nextHash = hash(uint32(x>>16), shift)
SHRQ $8, R14
MOVL R14, R11
IMULL $0x1e35a7bd, R11
SHRL CX, R11
// s++
ADDQ $1, SI
// break out of the inner1 for loop, i.e. continue the outer loop.
JMP outer
emitRemainder:
// if nextEmit < len(src) { etc }
MOVQ src_len+32(FP), AX
ADDQ DX, AX
CMPQ R10, AX
JEQ encodeBlockEnd
// d += emitLiteral(dst[d:], src[nextEmit:])
//
// Push args.
MOVQ DI, 0(SP)
MOVQ $0, 8(SP) // Unnecessary, as the callee ignores it, but conservative.
MOVQ $0, 16(SP) // Unnecessary, as the callee ignores it, but conservative.
MOVQ R10, 24(SP)
SUBQ R10, AX
MOVQ AX, 32(SP)
MOVQ AX, 40(SP) // Unnecessary, as the callee ignores it, but conservative.
// Spill local variables (registers) onto the stack; call; unspill.
MOVQ DI, 80(SP)
CALL ·emitLiteral(SB)
MOVQ 80(SP), DI
// Finish the "d +=" part of "d += emitLiteral(etc)".
ADDQ 48(SP), DI
encodeBlockEnd:
MOVQ dst_base+0(FP), AX
SUBQ AX, DI
MOVQ DI, d+48(FP)
RET

238
vendor/github.com/klauspost/compress/snappy/encode_other.go generated vendored Normal file
View File

@@ -0,0 +1,238 @@
// Copyright 2016 The Snappy-Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build !amd64 appengine !gc noasm
package snappy
func load32(b []byte, i int) uint32 {
b = b[i : i+4 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint32(b[0]) | uint32(b[1])<<8 | uint32(b[2])<<16 | uint32(b[3])<<24
}
func load64(b []byte, i int) uint64 {
b = b[i : i+8 : len(b)] // Help the compiler eliminate bounds checks on the next line.
return uint64(b[0]) | uint64(b[1])<<8 | uint64(b[2])<<16 | uint64(b[3])<<24 |
uint64(b[4])<<32 | uint64(b[5])<<40 | uint64(b[6])<<48 | uint64(b[7])<<56
}
// emitLiteral writes a literal chunk and returns the number of bytes written.
//
// It assumes that:
// dst is long enough to hold the encoded bytes
// 1 <= len(lit) && len(lit) <= 65536
func emitLiteral(dst, lit []byte) int {
i, n := 0, uint(len(lit)-1)
switch {
case n < 60:
dst[0] = uint8(n)<<2 | tagLiteral
i = 1
case n < 1<<8:
dst[0] = 60<<2 | tagLiteral
dst[1] = uint8(n)
i = 2
default:
dst[0] = 61<<2 | tagLiteral
dst[1] = uint8(n)
dst[2] = uint8(n >> 8)
i = 3
}
return i + copy(dst[i:], lit)
}
// emitCopy writes a copy chunk and returns the number of bytes written.
//
// It assumes that:
// dst is long enough to hold the encoded bytes
// 1 <= offset && offset <= 65535
// 4 <= length && length <= 65535
func emitCopy(dst []byte, offset, length int) int {
i := 0
// The maximum length for a single tagCopy1 or tagCopy2 op is 64 bytes. The
// threshold for this loop is a little higher (at 68 = 64 + 4), and the
// length emitted down below is is a little lower (at 60 = 64 - 4), because
// it's shorter to encode a length 67 copy as a length 60 tagCopy2 followed
// by a length 7 tagCopy1 (which encodes as 3+2 bytes) than to encode it as
// a length 64 tagCopy2 followed by a length 3 tagCopy2 (which encodes as
// 3+3 bytes). The magic 4 in the 64±4 is because the minimum length for a
// tagCopy1 op is 4 bytes, which is why a length 3 copy has to be an
// encodes-as-3-bytes tagCopy2 instead of an encodes-as-2-bytes tagCopy1.
for length >= 68 {
// Emit a length 64 copy, encoded as 3 bytes.
dst[i+0] = 63<<2 | tagCopy2
dst[i+1] = uint8(offset)
dst[i+2] = uint8(offset >> 8)
i += 3
length -= 64
}
if length > 64 {
// Emit a length 60 copy, encoded as 3 bytes.
dst[i+0] = 59<<2 | tagCopy2
dst[i+1] = uint8(offset)
dst[i+2] = uint8(offset >> 8)
i += 3
length -= 60
}
if length >= 12 || offset >= 2048 {
// Emit the remaining copy, encoded as 3 bytes.
dst[i+0] = uint8(length-1)<<2 | tagCopy2
dst[i+1] = uint8(offset)
dst[i+2] = uint8(offset >> 8)
return i + 3
}
// Emit the remaining copy, encoded as 2 bytes.
dst[i+0] = uint8(offset>>8)<<5 | uint8(length-4)<<2 | tagCopy1
dst[i+1] = uint8(offset)
return i + 2
}
// extendMatch returns the largest k such that k <= len(src) and that
// src[i:i+k-j] and src[j:k] have the same contents.
//
// It assumes that:
// 0 <= i && i < j && j <= len(src)
func extendMatch(src []byte, i, j int) int {
for ; j < len(src) && src[i] == src[j]; i, j = i+1, j+1 {
}
return j
}
func hash(u, shift uint32) uint32 {
return (u * 0x1e35a7bd) >> shift
}
// encodeBlock encodes a non-empty src to a guaranteed-large-enough dst. It
// assumes that the varint-encoded length of the decompressed bytes has already
// been written.
//
// It also assumes that:
// len(dst) >= MaxEncodedLen(len(src)) &&
// minNonLiteralBlockSize <= len(src) && len(src) <= maxBlockSize
func encodeBlock(dst, src []byte) (d int) {
// Initialize the hash table. Its size ranges from 1<<8 to 1<<14 inclusive.
// The table element type is uint16, as s < sLimit and sLimit < len(src)
// and len(src) <= maxBlockSize and maxBlockSize == 65536.
const (
maxTableSize = 1 << 14
// tableMask is redundant, but helps the compiler eliminate bounds
// checks.
tableMask = maxTableSize - 1
)
shift := uint32(32 - 8)
for tableSize := 1 << 8; tableSize < maxTableSize && tableSize < len(src); tableSize *= 2 {
shift--
}
// In Go, all array elements are zero-initialized, so there is no advantage
// to a smaller tableSize per se. However, it matches the C++ algorithm,
// and in the asm versions of this code, we can get away with zeroing only
// the first tableSize elements.
var table [maxTableSize]uint16
// sLimit is when to stop looking for offset/length copies. The inputMargin
// lets us use a fast path for emitLiteral in the main loop, while we are
// looking for copies.
sLimit := len(src) - inputMargin
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := 0
// The encoded form must start with a literal, as there are no previous
// bytes to copy, so we start looking for hash matches at s == 1.
s := 1
nextHash := hash(load32(src, s), shift)
for {
// Copied from the C++ snappy implementation:
//
// Heuristic match skipping: If 32 bytes are scanned with no matches
// found, start looking only at every other byte. If 32 more bytes are
// scanned (or skipped), look at every third byte, etc.. When a match
// is found, immediately go back to looking at every byte. This is a
// small loss (~5% performance, ~0.1% density) for compressible data
// due to more bookkeeping, but for non-compressible data (such as
// JPEG) it's a huge win since the compressor quickly "realizes" the
// data is incompressible and doesn't bother looking for matches
// everywhere.
//
// The "skip" variable keeps track of how many bytes there are since
// the last match; dividing it by 32 (ie. right-shifting by five) gives
// the number of bytes to move ahead for each iteration.
skip := 32
nextS := s
candidate := 0
for {
s = nextS
bytesBetweenHashLookups := skip >> 5
nextS = s + bytesBetweenHashLookups
skip += bytesBetweenHashLookups
if nextS > sLimit {
goto emitRemainder
}
candidate = int(table[nextHash&tableMask])
table[nextHash&tableMask] = uint16(s)
nextHash = hash(load32(src, nextS), shift)
if load32(src, s) == load32(src, candidate) {
break
}
}
// A 4-byte match has been found. We'll later see if more than 4 bytes
// match. But, prior to the match, src[nextEmit:s] are unmatched. Emit
// them as literal bytes.
d += emitLiteral(dst[d:], src[nextEmit:s])
// Call emitCopy, and then see if another emitCopy could be our next
// move. Repeat until we find no match for the input immediately after
// what was consumed by the last emitCopy call.
//
// If we exit this loop normally then we need to call emitLiteral next,
// though we don't yet know how big the literal will be. We handle that
// by proceeding to the next iteration of the main loop. We also can
// exit this loop via goto if we get close to exhausting the input.
for {
// Invariant: we have a 4-byte match at s, and no need to emit any
// literal bytes prior to s.
base := s
// Extend the 4-byte match as long as possible.
//
// This is an inlined version of:
// s = extendMatch(src, candidate+4, s+4)
s += 4
for i := candidate + 4; s < len(src) && src[i] == src[s]; i, s = i+1, s+1 {
}
d += emitCopy(dst[d:], base-candidate, s-base)
nextEmit = s
if s >= sLimit {
goto emitRemainder
}
// We could immediately start working at s now, but to improve
// compression we first update the hash table at s-1 and at s. If
// another emitCopy is not our next move, also calculate nextHash
// at s+1. At least on GOARCH=amd64, these three hash calculations
// are faster as one load64 call (with some shifts) instead of
// three load32 calls.
x := load64(src, s-1)
prevHash := hash(uint32(x>>0), shift)
table[prevHash&tableMask] = uint16(s - 1)
currHash := hash(uint32(x>>8), shift)
candidate = int(table[currHash&tableMask])
table[currHash&tableMask] = uint16(s)
if uint32(x>>8) != load32(src, candidate) {
nextHash = hash(uint32(x>>16), shift)
s++
break
}
}
}
emitRemainder:
if nextEmit < len(src) {
d += emitLiteral(dst[d:], src[nextEmit:])
}
return d
}

1965
vendor/github.com/klauspost/compress/snappy/golden_test.go generated vendored Normal file
View File

File diff suppressed because it is too large Load Diff

Some files were not shown because too many files have changed in this diff Show More