forked from valyala/gozstd
-
Notifications
You must be signed in to change notification settings - Fork 0
/
dict.go
191 lines (162 loc) · 4.75 KB
/
dict.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
package gozstd
/*
#define ZSTD_STATIC_LINKING_ONLY
#include "zstd.h"
#define ZDICT_STATIC_LINKING_ONLY
#include "zdict.h"
#include <stdint.h> // for uintptr_t
// The following *_wrapper functions allow avoiding memory allocations
// durting calls from Go.
// See https://github.com/golang/go/issues/24450 .
static ZSTD_CDict* ZSTD_createCDict_wrapper(uintptr_t dictBuffer, size_t dictSize, int compressionLevel) {
return ZSTD_createCDict((const void *)dictBuffer, dictSize, compressionLevel);
}
static ZSTD_DDict* ZSTD_createDDict_wrapper(uintptr_t dictBuffer, size_t dictSize) {
return ZSTD_createDDict((const void *)dictBuffer, dictSize);
}
*/
import "C"
import (
"fmt"
"runtime"
"sync"
"unsafe"
)
const minDictLen = C.ZDICT_DICTSIZE_MIN
// BuildDict returns dictionary built from the given samples.
//
// The resulting dictionary size will be close to desiredDictLen.
//
// The returned dictionary may be passed to NewCDict* and NewDDict.
func BuildDict(samples [][]byte, desiredDictLen int) []byte {
if desiredDictLen < minDictLen {
desiredDictLen = minDictLen
}
dict := make([]byte, desiredDictLen)
// Calculate the total samples size.
samplesBufLen := 0
for _, sample := range samples {
if len(sample) == 0 {
// Skip empty samples.
continue
}
samplesBufLen += len(sample)
}
// Construct flat samplesBuf and samplesSizes.
samplesBuf := make([]byte, 0, samplesBufLen)
samplesSizes := make([]C.size_t, 0, len(samples))
for _, sample := range samples {
samplesBuf = append(samplesBuf, sample...)
samplesSizes = append(samplesSizes, C.size_t(len(sample)))
}
// Add fake samples if the original samples are too small.
minSamplesBufLen := int(C.ZDICT_CONTENTSIZE_MIN)
if minSamplesBufLen < minDictLen {
minSamplesBufLen = minDictLen
}
for samplesBufLen < minSamplesBufLen {
fakeSample := []byte(fmt.Sprintf("this is a fake sample %d", samplesBufLen))
samplesBuf = append(samplesBuf, fakeSample...)
samplesSizes = append(samplesSizes, C.size_t(len(fakeSample)))
samplesBufLen += len(fakeSample)
}
// Run ZDICT_trainFromBuffer under lock, since it looks like it
// is unsafe for concurrent usage (it just randomly crashes).
// TODO: remove this restriction.
buildDictLock.Lock()
result := C.ZDICT_trainFromBuffer(
unsafe.Pointer(&dict[0]),
C.size_t(len(dict)),
unsafe.Pointer(&samplesBuf[0]),
&samplesSizes[0],
C.unsigned(len(samplesSizes)))
buildDictLock.Unlock()
ensureNoError("ZDICT_trainFromBuffer", result)
dictLen := int(result)
return dict[:dictLen]
}
var buildDictLock sync.Mutex
// CDict is a dictionary used for compression.
//
// A single CDict may be re-used in concurrently running goroutines.
type CDict struct {
p *C.ZSTD_CDict
compressionLevel int
}
// NewCDict creates new CDict from the given dict.
//
// Call Release when the returned dict is no longer used.
func NewCDict(dict []byte) (*CDict, error) {
return NewCDictLevel(dict, DefaultCompressionLevel)
}
// NewCDictLevel creates new CDict from the given dict
// using the given compressionLevel.
//
// Call Release when the returned dict is no longer used.
func NewCDictLevel(dict []byte, compressionLevel int) (*CDict, error) {
if len(dict) == 0 {
return nil, fmt.Errorf("dict cannot be empty")
}
cd := &CDict{
p: C.ZSTD_createCDict_wrapper(
C.uintptr_t(uintptr(unsafe.Pointer(&dict[0]))),
C.size_t(len(dict)),
C.int(compressionLevel)),
compressionLevel: compressionLevel,
}
// Prevent from GC'ing of dict during CGO call above.
runtime.KeepAlive(dict)
runtime.SetFinalizer(cd, freeCDict)
return cd, nil
}
// Release releases resources occupied by cd.
//
// cd cannot be used after the release.
func (cd *CDict) Release() {
if cd.p == nil {
return
}
result := C.ZSTD_freeCDict(cd.p)
ensureNoError("ZSTD_freeCDict", result)
cd.p = nil
}
func freeCDict(v interface{}) {
v.(*CDict).Release()
}
// DDict is a dictionary used for decompression.
//
// A single DDict may be re-used in concurrently running goroutines.
type DDict struct {
p *C.ZSTD_DDict
}
// NewDDict creates new DDict from the given dict.
//
// Call Release when the returned dict is no longer needed.
func NewDDict(dict []byte) (*DDict, error) {
if len(dict) == 0 {
return nil, fmt.Errorf("dict cannot be empty")
}
dd := &DDict{
p: C.ZSTD_createDDict_wrapper(
C.uintptr_t(uintptr(unsafe.Pointer(&dict[0]))),
C.size_t(len(dict))),
}
// Prevent from GC'ing of dict during CGO call above.
runtime.KeepAlive(dict)
runtime.SetFinalizer(dd, freeDDict)
return dd, nil
}
// Release releases resources occupied by dd.
//
// dd cannot be used after the release.
func (dd *DDict) Release() {
if dd.p == nil {
return
}
result := C.ZSTD_freeDDict(dd.p)
ensureNoError("ZSTD_freeDDict", result)
dd.p = nil
}
func freeDDict(v interface{}) {
v.(*DDict).Release()
}