zstd-ruby 1.3.1.1 → 1.3.2.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/README.md +1 -1
- data/ext/zstdruby/libzstd/.gitignore +1 -0
- data/ext/zstdruby/libzstd/Makefile +40 -26
- data/ext/zstdruby/libzstd/README.md +68 -45
- data/ext/zstdruby/libzstd/common/bitstream.h +35 -23
- data/ext/zstdruby/libzstd/common/compiler.h +1 -0
- data/ext/zstdruby/libzstd/common/error_private.c +4 -2
- data/ext/zstdruby/libzstd/common/error_private.h +4 -4
- data/ext/zstdruby/libzstd/common/fse.h +1 -1
- data/ext/zstdruby/libzstd/common/huf.h +1 -1
- data/ext/zstdruby/libzstd/common/mem.h +1 -0
- data/ext/zstdruby/libzstd/common/pool.c +61 -46
- data/ext/zstdruby/libzstd/common/pool.h +4 -0
- data/ext/zstdruby/libzstd/common/threading.c +11 -15
- data/ext/zstdruby/libzstd/common/threading.h +52 -32
- data/ext/zstdruby/libzstd/common/zstd_common.c +2 -2
- data/ext/zstdruby/libzstd/common/zstd_errors.h +3 -1
- data/ext/zstdruby/libzstd/common/zstd_internal.h +95 -21
- data/ext/zstdruby/libzstd/compress/fse_compress.c +3 -1
- data/ext/zstdruby/libzstd/compress/huf_compress.c +4 -3
- data/ext/zstdruby/libzstd/compress/zstd_compress.c +922 -2102
- data/ext/zstdruby/libzstd/compress/zstd_compress.h +307 -0
- data/ext/zstdruby/libzstd/compress/zstd_double_fast.c +308 -0
- data/ext/zstdruby/libzstd/compress/zstd_double_fast.h +28 -0
- data/ext/zstdruby/libzstd/compress/zstd_fast.c +242 -0
- data/ext/zstdruby/libzstd/compress/zstd_fast.h +30 -0
- data/ext/zstdruby/libzstd/compress/zstd_lazy.c +749 -0
- data/ext/zstdruby/libzstd/compress/zstd_lazy.h +38 -0
- data/ext/zstdruby/libzstd/compress/zstd_ldm.c +707 -0
- data/ext/zstdruby/libzstd/compress/zstd_ldm.h +67 -0
- data/ext/zstdruby/libzstd/compress/zstd_opt.c +957 -0
- data/ext/zstdruby/libzstd/compress/zstd_opt.h +14 -922
- data/ext/zstdruby/libzstd/compress/zstdmt_compress.c +210 -133
- data/ext/zstdruby/libzstd/compress/zstdmt_compress.h +20 -3
- data/ext/zstdruby/libzstd/decompress/zstd_decompress.c +373 -196
- data/ext/zstdruby/libzstd/deprecated/zbuff.h +1 -0
- data/ext/zstdruby/libzstd/deprecated/zbuff_common.c +1 -0
- data/ext/zstdruby/libzstd/deprecated/zbuff_compress.c +1 -0
- data/ext/zstdruby/libzstd/deprecated/zbuff_decompress.c +1 -0
- data/ext/zstdruby/libzstd/dictBuilder/cover.c +33 -22
- data/ext/zstdruby/libzstd/dictBuilder/zdict.c +8 -5
- data/ext/zstdruby/libzstd/dictBuilder/zdict.h +1 -0
- data/ext/zstdruby/libzstd/dll/example/Makefile +5 -5
- data/ext/zstdruby/libzstd/legacy/zstd_legacy.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v01.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v01.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v02.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v02.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v03.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v03.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v04.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v04.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v05.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v05.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v06.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v06.h +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v07.c +1 -0
- data/ext/zstdruby/libzstd/legacy/zstd_v07.h +1 -0
- data/ext/zstdruby/libzstd/zstd.h +366 -118
- data/lib/zstd-ruby/version.rb +1 -1
- metadata +11 -1
@@ -0,0 +1,28 @@
|
|
1
|
+
/*
|
2
|
+
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
|
3
|
+
* All rights reserved.
|
4
|
+
*
|
5
|
+
* This source code is licensed under both the BSD-style license (found in the
|
6
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
7
|
+
* in the COPYING file in the root directory of this source tree).
|
8
|
+
* You may select, at your option, one of the above-listed licenses.
|
9
|
+
*/
|
10
|
+
|
11
|
+
#ifndef ZSTD_DOUBLE_FAST_H
|
12
|
+
#define ZSTD_DOUBLE_FAST_H
|
13
|
+
|
14
|
+
#include "zstd_compress.h"
|
15
|
+
|
16
|
+
#if defined (__cplusplus)
|
17
|
+
extern "C" {
|
18
|
+
#endif
|
19
|
+
|
20
|
+
void ZSTD_fillDoubleHashTable(ZSTD_CCtx* cctx, const void* end, const U32 mls);
|
21
|
+
size_t ZSTD_compressBlock_doubleFast(ZSTD_CCtx* ctx, const void* src, size_t srcSize);
|
22
|
+
size_t ZSTD_compressBlock_doubleFast_extDict(ZSTD_CCtx* ctx, const void* src, size_t srcSize);
|
23
|
+
|
24
|
+
#if defined (__cplusplus)
|
25
|
+
}
|
26
|
+
#endif
|
27
|
+
|
28
|
+
#endif /* ZSTD_DOUBLE_FAST_H */
|
@@ -0,0 +1,242 @@
|
|
1
|
+
/*
|
2
|
+
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
|
3
|
+
* All rights reserved.
|
4
|
+
*
|
5
|
+
* This source code is licensed under both the BSD-style license (found in the
|
6
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
7
|
+
* in the COPYING file in the root directory of this source tree).
|
8
|
+
* You may select, at your option, one of the above-listed licenses.
|
9
|
+
*/
|
10
|
+
|
11
|
+
#include "zstd_fast.h"
|
12
|
+
|
13
|
+
|
14
|
+
void ZSTD_fillHashTable (ZSTD_CCtx* zc, const void* end, const U32 mls)
|
15
|
+
{
|
16
|
+
U32* const hashTable = zc->hashTable;
|
17
|
+
U32 const hBits = zc->appliedParams.cParams.hashLog;
|
18
|
+
const BYTE* const base = zc->base;
|
19
|
+
const BYTE* ip = base + zc->nextToUpdate;
|
20
|
+
const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
|
21
|
+
const size_t fastHashFillStep = 3;
|
22
|
+
|
23
|
+
while(ip <= iend) {
|
24
|
+
hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip - base);
|
25
|
+
ip += fastHashFillStep;
|
26
|
+
}
|
27
|
+
}
|
28
|
+
|
29
|
+
|
30
|
+
FORCE_INLINE_TEMPLATE
|
31
|
+
size_t ZSTD_compressBlock_fast_generic(ZSTD_CCtx* cctx,
|
32
|
+
const void* src, size_t srcSize,
|
33
|
+
const U32 mls)
|
34
|
+
{
|
35
|
+
U32* const hashTable = cctx->hashTable;
|
36
|
+
U32 const hBits = cctx->appliedParams.cParams.hashLog;
|
37
|
+
seqStore_t* seqStorePtr = &(cctx->seqStore);
|
38
|
+
const BYTE* const base = cctx->base;
|
39
|
+
const BYTE* const istart = (const BYTE*)src;
|
40
|
+
const BYTE* ip = istart;
|
41
|
+
const BYTE* anchor = istart;
|
42
|
+
const U32 lowestIndex = cctx->dictLimit;
|
43
|
+
const BYTE* const lowest = base + lowestIndex;
|
44
|
+
const BYTE* const iend = istart + srcSize;
|
45
|
+
const BYTE* const ilimit = iend - HASH_READ_SIZE;
|
46
|
+
U32 offset_1=seqStorePtr->rep[0], offset_2=seqStorePtr->rep[1];
|
47
|
+
U32 offsetSaved = 0;
|
48
|
+
|
49
|
+
/* init */
|
50
|
+
ip += (ip==lowest);
|
51
|
+
{ U32 const maxRep = (U32)(ip-lowest);
|
52
|
+
if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
|
53
|
+
if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
|
54
|
+
}
|
55
|
+
|
56
|
+
/* Main Search Loop */
|
57
|
+
while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
|
58
|
+
size_t mLength;
|
59
|
+
size_t const h = ZSTD_hashPtr(ip, hBits, mls);
|
60
|
+
U32 const current = (U32)(ip-base);
|
61
|
+
U32 const matchIndex = hashTable[h];
|
62
|
+
const BYTE* match = base + matchIndex;
|
63
|
+
hashTable[h] = current; /* update hash table */
|
64
|
+
|
65
|
+
if ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1))) {
|
66
|
+
mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
|
67
|
+
ip++;
|
68
|
+
ZSTD_storeSeq(seqStorePtr, ip-anchor, anchor, 0, mLength-MINMATCH);
|
69
|
+
} else {
|
70
|
+
U32 offset;
|
71
|
+
if ( (matchIndex <= lowestIndex) || (MEM_read32(match) != MEM_read32(ip)) ) {
|
72
|
+
ip += ((ip-anchor) >> g_searchStrength) + 1;
|
73
|
+
continue;
|
74
|
+
}
|
75
|
+
mLength = ZSTD_count(ip+4, match+4, iend) + 4;
|
76
|
+
offset = (U32)(ip-match);
|
77
|
+
while (((ip>anchor) & (match>lowest)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
|
78
|
+
offset_2 = offset_1;
|
79
|
+
offset_1 = offset;
|
80
|
+
|
81
|
+
ZSTD_storeSeq(seqStorePtr, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
|
82
|
+
}
|
83
|
+
|
84
|
+
/* match found */
|
85
|
+
ip += mLength;
|
86
|
+
anchor = ip;
|
87
|
+
|
88
|
+
if (ip <= ilimit) {
|
89
|
+
/* Fill Table */
|
90
|
+
hashTable[ZSTD_hashPtr(base+current+2, hBits, mls)] = current+2; /* here because current+2 could be > iend-8 */
|
91
|
+
hashTable[ZSTD_hashPtr(ip-2, hBits, mls)] = (U32)(ip-2-base);
|
92
|
+
/* check immediate repcode */
|
93
|
+
while ( (ip <= ilimit)
|
94
|
+
&& ( (offset_2>0)
|
95
|
+
& (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
|
96
|
+
/* store sequence */
|
97
|
+
size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
|
98
|
+
{ U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; } /* swap offset_2 <=> offset_1 */
|
99
|
+
hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip-base);
|
100
|
+
ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, rLength-MINMATCH);
|
101
|
+
ip += rLength;
|
102
|
+
anchor = ip;
|
103
|
+
continue; /* faster when present ... (?) */
|
104
|
+
} } }
|
105
|
+
|
106
|
+
/* save reps for next block */
|
107
|
+
seqStorePtr->repToConfirm[0] = offset_1 ? offset_1 : offsetSaved;
|
108
|
+
seqStorePtr->repToConfirm[1] = offset_2 ? offset_2 : offsetSaved;
|
109
|
+
|
110
|
+
/* Return the last literals size */
|
111
|
+
return iend - anchor;
|
112
|
+
}
|
113
|
+
|
114
|
+
|
115
|
+
size_t ZSTD_compressBlock_fast(ZSTD_CCtx* ctx,
|
116
|
+
const void* src, size_t srcSize)
|
117
|
+
{
|
118
|
+
const U32 mls = ctx->appliedParams.cParams.searchLength;
|
119
|
+
switch(mls)
|
120
|
+
{
|
121
|
+
default: /* includes case 3 */
|
122
|
+
case 4 :
|
123
|
+
return ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 4);
|
124
|
+
case 5 :
|
125
|
+
return ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 5);
|
126
|
+
case 6 :
|
127
|
+
return ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 6);
|
128
|
+
case 7 :
|
129
|
+
return ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 7);
|
130
|
+
}
|
131
|
+
}
|
132
|
+
|
133
|
+
|
134
|
+
static size_t ZSTD_compressBlock_fast_extDict_generic(ZSTD_CCtx* ctx,
|
135
|
+
const void* src, size_t srcSize,
|
136
|
+
const U32 mls)
|
137
|
+
{
|
138
|
+
U32* hashTable = ctx->hashTable;
|
139
|
+
const U32 hBits = ctx->appliedParams.cParams.hashLog;
|
140
|
+
seqStore_t* seqStorePtr = &(ctx->seqStore);
|
141
|
+
const BYTE* const base = ctx->base;
|
142
|
+
const BYTE* const dictBase = ctx->dictBase;
|
143
|
+
const BYTE* const istart = (const BYTE*)src;
|
144
|
+
const BYTE* ip = istart;
|
145
|
+
const BYTE* anchor = istart;
|
146
|
+
const U32 lowestIndex = ctx->lowLimit;
|
147
|
+
const BYTE* const dictStart = dictBase + lowestIndex;
|
148
|
+
const U32 dictLimit = ctx->dictLimit;
|
149
|
+
const BYTE* const lowPrefixPtr = base + dictLimit;
|
150
|
+
const BYTE* const dictEnd = dictBase + dictLimit;
|
151
|
+
const BYTE* const iend = istart + srcSize;
|
152
|
+
const BYTE* const ilimit = iend - 8;
|
153
|
+
U32 offset_1=seqStorePtr->rep[0], offset_2=seqStorePtr->rep[1];
|
154
|
+
|
155
|
+
/* Search Loop */
|
156
|
+
while (ip < ilimit) { /* < instead of <=, because (ip+1) */
|
157
|
+
const size_t h = ZSTD_hashPtr(ip, hBits, mls);
|
158
|
+
const U32 matchIndex = hashTable[h];
|
159
|
+
const BYTE* matchBase = matchIndex < dictLimit ? dictBase : base;
|
160
|
+
const BYTE* match = matchBase + matchIndex;
|
161
|
+
const U32 current = (U32)(ip-base);
|
162
|
+
const U32 repIndex = current + 1 - offset_1; /* offset_1 expected <= current +1 */
|
163
|
+
const BYTE* repBase = repIndex < dictLimit ? dictBase : base;
|
164
|
+
const BYTE* repMatch = repBase + repIndex;
|
165
|
+
size_t mLength;
|
166
|
+
hashTable[h] = current; /* update hash table */
|
167
|
+
|
168
|
+
if ( (((U32)((dictLimit-1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex))
|
169
|
+
&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
|
170
|
+
const BYTE* repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
|
171
|
+
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, lowPrefixPtr) + 4;
|
172
|
+
ip++;
|
173
|
+
ZSTD_storeSeq(seqStorePtr, ip-anchor, anchor, 0, mLength-MINMATCH);
|
174
|
+
} else {
|
175
|
+
if ( (matchIndex < lowestIndex) ||
|
176
|
+
(MEM_read32(match) != MEM_read32(ip)) ) {
|
177
|
+
ip += ((ip-anchor) >> g_searchStrength) + 1;
|
178
|
+
continue;
|
179
|
+
}
|
180
|
+
{ const BYTE* matchEnd = matchIndex < dictLimit ? dictEnd : iend;
|
181
|
+
const BYTE* lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
|
182
|
+
U32 offset;
|
183
|
+
mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, lowPrefixPtr) + 4;
|
184
|
+
while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
|
185
|
+
offset = current - matchIndex;
|
186
|
+
offset_2 = offset_1;
|
187
|
+
offset_1 = offset;
|
188
|
+
ZSTD_storeSeq(seqStorePtr, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
|
189
|
+
} }
|
190
|
+
|
191
|
+
/* found a match : store it */
|
192
|
+
ip += mLength;
|
193
|
+
anchor = ip;
|
194
|
+
|
195
|
+
if (ip <= ilimit) {
|
196
|
+
/* Fill Table */
|
197
|
+
hashTable[ZSTD_hashPtr(base+current+2, hBits, mls)] = current+2;
|
198
|
+
hashTable[ZSTD_hashPtr(ip-2, hBits, mls)] = (U32)(ip-2-base);
|
199
|
+
/* check immediate repcode */
|
200
|
+
while (ip <= ilimit) {
|
201
|
+
U32 const current2 = (U32)(ip-base);
|
202
|
+
U32 const repIndex2 = current2 - offset_2;
|
203
|
+
const BYTE* repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
|
204
|
+
if ( (((U32)((dictLimit-1) - repIndex2) >= 3) & (repIndex2 > lowestIndex)) /* intentional overflow */
|
205
|
+
&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
|
206
|
+
const BYTE* const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
|
207
|
+
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, lowPrefixPtr) + 4;
|
208
|
+
U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
|
209
|
+
ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, repLength2-MINMATCH);
|
210
|
+
hashTable[ZSTD_hashPtr(ip, hBits, mls)] = current2;
|
211
|
+
ip += repLength2;
|
212
|
+
anchor = ip;
|
213
|
+
continue;
|
214
|
+
}
|
215
|
+
break;
|
216
|
+
} } }
|
217
|
+
|
218
|
+
/* save reps for next block */
|
219
|
+
seqStorePtr->repToConfirm[0] = offset_1; seqStorePtr->repToConfirm[1] = offset_2;
|
220
|
+
|
221
|
+
/* Return the last literals size */
|
222
|
+
return iend - anchor;
|
223
|
+
}
|
224
|
+
|
225
|
+
|
226
|
+
size_t ZSTD_compressBlock_fast_extDict(ZSTD_CCtx* ctx,
|
227
|
+
const void* src, size_t srcSize)
|
228
|
+
{
|
229
|
+
U32 const mls = ctx->appliedParams.cParams.searchLength;
|
230
|
+
switch(mls)
|
231
|
+
{
|
232
|
+
default: /* includes case 3 */
|
233
|
+
case 4 :
|
234
|
+
return ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 4);
|
235
|
+
case 5 :
|
236
|
+
return ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 5);
|
237
|
+
case 6 :
|
238
|
+
return ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 6);
|
239
|
+
case 7 :
|
240
|
+
return ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 7);
|
241
|
+
}
|
242
|
+
}
|
@@ -0,0 +1,30 @@
|
|
1
|
+
/*
|
2
|
+
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
|
3
|
+
* All rights reserved.
|
4
|
+
*
|
5
|
+
* This source code is licensed under both the BSD-style license (found in the
|
6
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
7
|
+
* in the COPYING file in the root directory of this source tree).
|
8
|
+
* You may select, at your option, one of the above-listed licenses.
|
9
|
+
*/
|
10
|
+
|
11
|
+
#ifndef ZSTD_FAST_H
|
12
|
+
#define ZSTD_FAST_H
|
13
|
+
|
14
|
+
#include "zstd_compress.h"
|
15
|
+
|
16
|
+
#if defined (__cplusplus)
|
17
|
+
extern "C" {
|
18
|
+
#endif
|
19
|
+
|
20
|
+
void ZSTD_fillHashTable(ZSTD_CCtx* zc, const void* end, const U32 mls);
|
21
|
+
size_t ZSTD_compressBlock_fast(ZSTD_CCtx* ctx,
|
22
|
+
const void* src, size_t srcSize);
|
23
|
+
size_t ZSTD_compressBlock_fast_extDict(ZSTD_CCtx* ctx,
|
24
|
+
const void* src, size_t srcSize);
|
25
|
+
|
26
|
+
#if defined (__cplusplus)
|
27
|
+
}
|
28
|
+
#endif
|
29
|
+
|
30
|
+
#endif /* ZSTD_FAST_H */
|
@@ -0,0 +1,749 @@
|
|
1
|
+
/*
|
2
|
+
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
|
3
|
+
* All rights reserved.
|
4
|
+
*
|
5
|
+
* This source code is licensed under both the BSD-style license (found in the
|
6
|
+
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
7
|
+
* in the COPYING file in the root directory of this source tree).
|
8
|
+
* You may select, at your option, one of the above-listed licenses.
|
9
|
+
*/
|
10
|
+
|
11
|
+
#include "zstd_lazy.h"
|
12
|
+
|
13
|
+
|
14
|
+
/*-*************************************
|
15
|
+
* Binary Tree search
|
16
|
+
***************************************/
|
17
|
+
/** ZSTD_insertBt1() : add one or multiple positions to tree.
|
18
|
+
* ip : assumed <= iend-8 .
|
19
|
+
* @return : nb of positions added */
|
20
|
+
static U32 ZSTD_insertBt1(ZSTD_CCtx* zc, const BYTE* const ip, const U32 mls, const BYTE* const iend, U32 nbCompares,
|
21
|
+
U32 extDict)
|
22
|
+
{
|
23
|
+
U32* const hashTable = zc->hashTable;
|
24
|
+
U32 const hashLog = zc->appliedParams.cParams.hashLog;
|
25
|
+
size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
|
26
|
+
U32* const bt = zc->chainTable;
|
27
|
+
U32 const btLog = zc->appliedParams.cParams.chainLog - 1;
|
28
|
+
U32 const btMask = (1 << btLog) - 1;
|
29
|
+
U32 matchIndex = hashTable[h];
|
30
|
+
size_t commonLengthSmaller=0, commonLengthLarger=0;
|
31
|
+
const BYTE* const base = zc->base;
|
32
|
+
const BYTE* const dictBase = zc->dictBase;
|
33
|
+
const U32 dictLimit = zc->dictLimit;
|
34
|
+
const BYTE* const dictEnd = dictBase + dictLimit;
|
35
|
+
const BYTE* const prefixStart = base + dictLimit;
|
36
|
+
const BYTE* match;
|
37
|
+
const U32 current = (U32)(ip-base);
|
38
|
+
const U32 btLow = btMask >= current ? 0 : current - btMask;
|
39
|
+
U32* smallerPtr = bt + 2*(current&btMask);
|
40
|
+
U32* largerPtr = smallerPtr + 1;
|
41
|
+
U32 dummy32; /* to be nullified at the end */
|
42
|
+
U32 const windowLow = zc->lowLimit;
|
43
|
+
U32 matchEndIdx = current+8;
|
44
|
+
size_t bestLength = 8;
|
45
|
+
#ifdef ZSTD_C_PREDICT
|
46
|
+
U32 predictedSmall = *(bt + 2*((current-1)&btMask) + 0);
|
47
|
+
U32 predictedLarge = *(bt + 2*((current-1)&btMask) + 1);
|
48
|
+
predictedSmall += (predictedSmall>0);
|
49
|
+
predictedLarge += (predictedLarge>0);
|
50
|
+
#endif /* ZSTD_C_PREDICT */
|
51
|
+
|
52
|
+
assert(ip <= iend-8); /* required for h calculation */
|
53
|
+
hashTable[h] = current; /* Update Hash Table */
|
54
|
+
|
55
|
+
while (nbCompares-- && (matchIndex > windowLow)) {
|
56
|
+
U32* const nextPtr = bt + 2*(matchIndex & btMask);
|
57
|
+
size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
|
58
|
+
|
59
|
+
#ifdef ZSTD_C_PREDICT /* note : can create issues when hlog small <= 11 */
|
60
|
+
const U32* predictPtr = bt + 2*((matchIndex-1) & btMask); /* written this way, as bt is a roll buffer */
|
61
|
+
if (matchIndex == predictedSmall) {
|
62
|
+
/* no need to check length, result known */
|
63
|
+
*smallerPtr = matchIndex;
|
64
|
+
if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */
|
65
|
+
smallerPtr = nextPtr+1; /* new "smaller" => larger of match */
|
66
|
+
matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
|
67
|
+
predictedSmall = predictPtr[1] + (predictPtr[1]>0);
|
68
|
+
continue;
|
69
|
+
}
|
70
|
+
if (matchIndex == predictedLarge) {
|
71
|
+
*largerPtr = matchIndex;
|
72
|
+
if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */
|
73
|
+
largerPtr = nextPtr;
|
74
|
+
matchIndex = nextPtr[0];
|
75
|
+
predictedLarge = predictPtr[0] + (predictPtr[0]>0);
|
76
|
+
continue;
|
77
|
+
}
|
78
|
+
#endif
|
79
|
+
if ((!extDict) || (matchIndex+matchLength >= dictLimit)) {
|
80
|
+
match = base + matchIndex;
|
81
|
+
if (match[matchLength] == ip[matchLength])
|
82
|
+
matchLength += ZSTD_count(ip+matchLength+1, match+matchLength+1, iend) +1;
|
83
|
+
} else {
|
84
|
+
match = dictBase + matchIndex;
|
85
|
+
matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
|
86
|
+
if (matchIndex+matchLength >= dictLimit)
|
87
|
+
match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
|
88
|
+
}
|
89
|
+
|
90
|
+
if (matchLength > bestLength) {
|
91
|
+
bestLength = matchLength;
|
92
|
+
if (matchLength > matchEndIdx - matchIndex)
|
93
|
+
matchEndIdx = matchIndex + (U32)matchLength;
|
94
|
+
}
|
95
|
+
|
96
|
+
if (ip+matchLength == iend) /* equal : no way to know if inf or sup */
|
97
|
+
break; /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */
|
98
|
+
|
99
|
+
if (match[matchLength] < ip[matchLength]) { /* necessarily within buffer */
|
100
|
+
/* match+1 is smaller than current */
|
101
|
+
*smallerPtr = matchIndex; /* update smaller idx */
|
102
|
+
commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
|
103
|
+
if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop searching */
|
104
|
+
smallerPtr = nextPtr+1; /* new "smaller" => larger of match */
|
105
|
+
matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
|
106
|
+
} else {
|
107
|
+
/* match is larger than current */
|
108
|
+
*largerPtr = matchIndex;
|
109
|
+
commonLengthLarger = matchLength;
|
110
|
+
if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop searching */
|
111
|
+
largerPtr = nextPtr;
|
112
|
+
matchIndex = nextPtr[0];
|
113
|
+
} }
|
114
|
+
|
115
|
+
*smallerPtr = *largerPtr = 0;
|
116
|
+
if (bestLength > 384) return MIN(192, (U32)(bestLength - 384)); /* speed optimization */
|
117
|
+
if (matchEndIdx > current + 8) return matchEndIdx - (current + 8);
|
118
|
+
return 1;
|
119
|
+
}
|
120
|
+
|
121
|
+
|
122
|
+
static size_t ZSTD_insertBtAndFindBestMatch (
|
123
|
+
ZSTD_CCtx* zc,
|
124
|
+
const BYTE* const ip, const BYTE* const iend,
|
125
|
+
size_t* offsetPtr,
|
126
|
+
U32 nbCompares, const U32 mls,
|
127
|
+
U32 extDict)
|
128
|
+
{
|
129
|
+
U32* const hashTable = zc->hashTable;
|
130
|
+
U32 const hashLog = zc->appliedParams.cParams.hashLog;
|
131
|
+
size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
|
132
|
+
U32* const bt = zc->chainTable;
|
133
|
+
U32 const btLog = zc->appliedParams.cParams.chainLog - 1;
|
134
|
+
U32 const btMask = (1 << btLog) - 1;
|
135
|
+
U32 matchIndex = hashTable[h];
|
136
|
+
size_t commonLengthSmaller=0, commonLengthLarger=0;
|
137
|
+
const BYTE* const base = zc->base;
|
138
|
+
const BYTE* const dictBase = zc->dictBase;
|
139
|
+
const U32 dictLimit = zc->dictLimit;
|
140
|
+
const BYTE* const dictEnd = dictBase + dictLimit;
|
141
|
+
const BYTE* const prefixStart = base + dictLimit;
|
142
|
+
const U32 current = (U32)(ip-base);
|
143
|
+
const U32 btLow = btMask >= current ? 0 : current - btMask;
|
144
|
+
const U32 windowLow = zc->lowLimit;
|
145
|
+
U32* smallerPtr = bt + 2*(current&btMask);
|
146
|
+
U32* largerPtr = bt + 2*(current&btMask) + 1;
|
147
|
+
U32 matchEndIdx = current+8;
|
148
|
+
U32 dummy32; /* to be nullified at the end */
|
149
|
+
size_t bestLength = 0;
|
150
|
+
|
151
|
+
assert(ip <= iend-8); /* required for h calculation */
|
152
|
+
hashTable[h] = current; /* Update Hash Table */
|
153
|
+
|
154
|
+
while (nbCompares-- && (matchIndex > windowLow)) {
|
155
|
+
U32* const nextPtr = bt + 2*(matchIndex & btMask);
|
156
|
+
size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
|
157
|
+
const BYTE* match;
|
158
|
+
|
159
|
+
if ((!extDict) || (matchIndex+matchLength >= dictLimit)) {
|
160
|
+
match = base + matchIndex;
|
161
|
+
if (match[matchLength] == ip[matchLength])
|
162
|
+
matchLength += ZSTD_count(ip+matchLength+1, match+matchLength+1, iend) +1;
|
163
|
+
} else {
|
164
|
+
match = dictBase + matchIndex;
|
165
|
+
matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
|
166
|
+
if (matchIndex+matchLength >= dictLimit)
|
167
|
+
match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
|
168
|
+
}
|
169
|
+
|
170
|
+
if (matchLength > bestLength) {
|
171
|
+
if (matchLength > matchEndIdx - matchIndex)
|
172
|
+
matchEndIdx = matchIndex + (U32)matchLength;
|
173
|
+
if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(current-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) )
|
174
|
+
bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + current - matchIndex;
|
175
|
+
if (ip+matchLength == iend) /* equal : no way to know if inf or sup */
|
176
|
+
break; /* drop, to guarantee consistency (miss a little bit of compression) */
|
177
|
+
}
|
178
|
+
|
179
|
+
if (match[matchLength] < ip[matchLength]) {
|
180
|
+
/* match is smaller than current */
|
181
|
+
*smallerPtr = matchIndex; /* update smaller idx */
|
182
|
+
commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
|
183
|
+
if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */
|
184
|
+
smallerPtr = nextPtr+1; /* new "smaller" => larger of match */
|
185
|
+
matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
|
186
|
+
} else {
|
187
|
+
/* match is larger than current */
|
188
|
+
*largerPtr = matchIndex;
|
189
|
+
commonLengthLarger = matchLength;
|
190
|
+
if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */
|
191
|
+
largerPtr = nextPtr;
|
192
|
+
matchIndex = nextPtr[0];
|
193
|
+
} }
|
194
|
+
|
195
|
+
*smallerPtr = *largerPtr = 0;
|
196
|
+
|
197
|
+
zc->nextToUpdate = (matchEndIdx > current + 8) ? matchEndIdx - 8 : current+1;
|
198
|
+
return bestLength;
|
199
|
+
}
|
200
|
+
|
201
|
+
|
202
|
+
void ZSTD_updateTree(ZSTD_CCtx* zc, const BYTE* const ip, const BYTE* const iend, const U32 nbCompares, const U32 mls)
|
203
|
+
{
|
204
|
+
const BYTE* const base = zc->base;
|
205
|
+
const U32 target = (U32)(ip - base);
|
206
|
+
U32 idx = zc->nextToUpdate;
|
207
|
+
|
208
|
+
while(idx < target)
|
209
|
+
idx += ZSTD_insertBt1(zc, base+idx, mls, iend, nbCompares, 0);
|
210
|
+
}
|
211
|
+
|
212
|
+
/** ZSTD_BtFindBestMatch() : Tree updater, providing best match */
|
213
|
+
static size_t ZSTD_BtFindBestMatch (
|
214
|
+
ZSTD_CCtx* zc,
|
215
|
+
const BYTE* const ip, const BYTE* const iLimit,
|
216
|
+
size_t* offsetPtr,
|
217
|
+
const U32 maxNbAttempts, const U32 mls)
|
218
|
+
{
|
219
|
+
if (ip < zc->base + zc->nextToUpdate) return 0; /* skipped area */
|
220
|
+
ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
|
221
|
+
return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 0);
|
222
|
+
}
|
223
|
+
|
224
|
+
|
225
|
+
static size_t ZSTD_BtFindBestMatch_selectMLS (
|
226
|
+
ZSTD_CCtx* zc, /* Index table will be updated */
|
227
|
+
const BYTE* ip, const BYTE* const iLimit,
|
228
|
+
size_t* offsetPtr,
|
229
|
+
const U32 maxNbAttempts, const U32 matchLengthSearch)
|
230
|
+
{
|
231
|
+
switch(matchLengthSearch)
|
232
|
+
{
|
233
|
+
default : /* includes case 3 */
|
234
|
+
case 4 : return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
|
235
|
+
case 5 : return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
|
236
|
+
case 7 :
|
237
|
+
case 6 : return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
|
238
|
+
}
|
239
|
+
}
|
240
|
+
|
241
|
+
|
242
|
+
void ZSTD_updateTree_extDict(ZSTD_CCtx* zc, const BYTE* const ip, const BYTE* const iend, const U32 nbCompares, const U32 mls)
|
243
|
+
{
|
244
|
+
const BYTE* const base = zc->base;
|
245
|
+
const U32 target = (U32)(ip - base);
|
246
|
+
U32 idx = zc->nextToUpdate;
|
247
|
+
|
248
|
+
while (idx < target) idx += ZSTD_insertBt1(zc, base+idx, mls, iend, nbCompares, 1);
|
249
|
+
}
|
250
|
+
|
251
|
+
|
252
|
+
/** Tree updater, providing best match */
|
253
|
+
static size_t ZSTD_BtFindBestMatch_extDict (
|
254
|
+
ZSTD_CCtx* zc,
|
255
|
+
const BYTE* const ip, const BYTE* const iLimit,
|
256
|
+
size_t* offsetPtr,
|
257
|
+
const U32 maxNbAttempts, const U32 mls)
|
258
|
+
{
|
259
|
+
if (ip < zc->base + zc->nextToUpdate) return 0; /* skipped area */
|
260
|
+
ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
|
261
|
+
return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 1);
|
262
|
+
}
|
263
|
+
|
264
|
+
|
265
|
+
static size_t ZSTD_BtFindBestMatch_selectMLS_extDict (
|
266
|
+
ZSTD_CCtx* zc, /* Index table will be updated */
|
267
|
+
const BYTE* ip, const BYTE* const iLimit,
|
268
|
+
size_t* offsetPtr,
|
269
|
+
const U32 maxNbAttempts, const U32 matchLengthSearch)
|
270
|
+
{
|
271
|
+
switch(matchLengthSearch)
|
272
|
+
{
|
273
|
+
default : /* includes case 3 */
|
274
|
+
case 4 : return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
|
275
|
+
case 5 : return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
|
276
|
+
case 7 :
|
277
|
+
case 6 : return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
|
278
|
+
}
|
279
|
+
}
|
280
|
+
|
281
|
+
|
282
|
+
|
283
|
+
/* *********************************
|
284
|
+
* Hash Chain
|
285
|
+
***********************************/
|
286
|
+
#define NEXT_IN_CHAIN(d, mask) chainTable[(d) & mask]
|
287
|
+
|
288
|
+
/* Update chains up to ip (excluded)
|
289
|
+
Assumption : always within prefix (i.e. not within extDict) */
|
290
|
+
U32 ZSTD_insertAndFindFirstIndex (ZSTD_CCtx* zc, const BYTE* ip, U32 mls)
|
291
|
+
{
|
292
|
+
U32* const hashTable = zc->hashTable;
|
293
|
+
const U32 hashLog = zc->appliedParams.cParams.hashLog;
|
294
|
+
U32* const chainTable = zc->chainTable;
|
295
|
+
const U32 chainMask = (1 << zc->appliedParams.cParams.chainLog) - 1;
|
296
|
+
const BYTE* const base = zc->base;
|
297
|
+
const U32 target = (U32)(ip - base);
|
298
|
+
U32 idx = zc->nextToUpdate;
|
299
|
+
|
300
|
+
while(idx < target) { /* catch up */
|
301
|
+
size_t const h = ZSTD_hashPtr(base+idx, hashLog, mls);
|
302
|
+
NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
|
303
|
+
hashTable[h] = idx;
|
304
|
+
idx++;
|
305
|
+
}
|
306
|
+
|
307
|
+
zc->nextToUpdate = target;
|
308
|
+
return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
|
309
|
+
}
|
310
|
+
|
311
|
+
|
312
|
+
/* inlining is important to hardwire a hot branch (template emulation) */
|
313
|
+
FORCE_INLINE_TEMPLATE
|
314
|
+
size_t ZSTD_HcFindBestMatch_generic (
|
315
|
+
ZSTD_CCtx* zc, /* Index table will be updated */
|
316
|
+
const BYTE* const ip, const BYTE* const iLimit,
|
317
|
+
size_t* offsetPtr,
|
318
|
+
const U32 maxNbAttempts, const U32 mls, const U32 extDict)
|
319
|
+
{
|
320
|
+
U32* const chainTable = zc->chainTable;
|
321
|
+
const U32 chainSize = (1 << zc->appliedParams.cParams.chainLog);
|
322
|
+
const U32 chainMask = chainSize-1;
|
323
|
+
const BYTE* const base = zc->base;
|
324
|
+
const BYTE* const dictBase = zc->dictBase;
|
325
|
+
const U32 dictLimit = zc->dictLimit;
|
326
|
+
const BYTE* const prefixStart = base + dictLimit;
|
327
|
+
const BYTE* const dictEnd = dictBase + dictLimit;
|
328
|
+
const U32 lowLimit = zc->lowLimit;
|
329
|
+
const U32 current = (U32)(ip-base);
|
330
|
+
const U32 minChain = current > chainSize ? current - chainSize : 0;
|
331
|
+
int nbAttempts=maxNbAttempts;
|
332
|
+
size_t ml=4-1;
|
333
|
+
|
334
|
+
/* HC4 match finder */
|
335
|
+
U32 matchIndex = ZSTD_insertAndFindFirstIndex (zc, ip, mls);
|
336
|
+
|
337
|
+
for ( ; (matchIndex>lowLimit) & (nbAttempts>0) ; nbAttempts--) {
|
338
|
+
const BYTE* match;
|
339
|
+
size_t currentMl=0;
|
340
|
+
if ((!extDict) || matchIndex >= dictLimit) {
|
341
|
+
match = base + matchIndex;
|
342
|
+
if (match[ml] == ip[ml]) /* potentially better */
|
343
|
+
currentMl = ZSTD_count(ip, match, iLimit);
|
344
|
+
} else {
|
345
|
+
match = dictBase + matchIndex;
|
346
|
+
if (MEM_read32(match) == MEM_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */
|
347
|
+
currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4;
|
348
|
+
}
|
349
|
+
|
350
|
+
/* save best solution */
|
351
|
+
if (currentMl > ml) {
|
352
|
+
ml = currentMl;
|
353
|
+
*offsetPtr = current - matchIndex + ZSTD_REP_MOVE;
|
354
|
+
if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
|
355
|
+
}
|
356
|
+
|
357
|
+
if (matchIndex <= minChain) break;
|
358
|
+
matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
|
359
|
+
}
|
360
|
+
|
361
|
+
return ml;
|
362
|
+
}
|
363
|
+
|
364
|
+
|
365
|
+
FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS (
|
366
|
+
ZSTD_CCtx* zc,
|
367
|
+
const BYTE* ip, const BYTE* const iLimit,
|
368
|
+
size_t* offsetPtr,
|
369
|
+
const U32 maxNbAttempts, const U32 matchLengthSearch)
|
370
|
+
{
|
371
|
+
switch(matchLengthSearch)
|
372
|
+
{
|
373
|
+
default : /* includes case 3 */
|
374
|
+
case 4 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 0);
|
375
|
+
case 5 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 0);
|
376
|
+
case 7 :
|
377
|
+
case 6 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 0);
|
378
|
+
}
|
379
|
+
}
|
380
|
+
|
381
|
+
|
382
|
+
FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_extDict_selectMLS (
|
383
|
+
ZSTD_CCtx* zc,
|
384
|
+
const BYTE* ip, const BYTE* const iLimit,
|
385
|
+
size_t* offsetPtr,
|
386
|
+
const U32 maxNbAttempts, const U32 matchLengthSearch)
|
387
|
+
{
|
388
|
+
switch(matchLengthSearch)
|
389
|
+
{
|
390
|
+
default : /* includes case 3 */
|
391
|
+
case 4 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 1);
|
392
|
+
case 5 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 1);
|
393
|
+
case 7 :
|
394
|
+
case 6 : return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 1);
|
395
|
+
}
|
396
|
+
}
|
397
|
+
|
398
|
+
|
399
|
+
/* *******************************
|
400
|
+
* Common parser - lazy strategy
|
401
|
+
*********************************/
|
402
|
+
FORCE_INLINE_TEMPLATE
|
403
|
+
size_t ZSTD_compressBlock_lazy_generic(ZSTD_CCtx* ctx,
|
404
|
+
const void* src, size_t srcSize,
|
405
|
+
const U32 searchMethod, const U32 depth)
|
406
|
+
{
|
407
|
+
seqStore_t* seqStorePtr = &(ctx->seqStore);
|
408
|
+
const BYTE* const istart = (const BYTE*)src;
|
409
|
+
const BYTE* ip = istart;
|
410
|
+
const BYTE* anchor = istart;
|
411
|
+
const BYTE* const iend = istart + srcSize;
|
412
|
+
const BYTE* const ilimit = iend - 8;
|
413
|
+
const BYTE* const base = ctx->base + ctx->dictLimit;
|
414
|
+
|
415
|
+
U32 const maxSearches = 1 << ctx->appliedParams.cParams.searchLog;
|
416
|
+
U32 const mls = ctx->appliedParams.cParams.searchLength;
|
417
|
+
|
418
|
+
typedef size_t (*searchMax_f)(ZSTD_CCtx* zc, const BYTE* ip, const BYTE* iLimit,
|
419
|
+
size_t* offsetPtr,
|
420
|
+
U32 maxNbAttempts, U32 matchLengthSearch);
|
421
|
+
searchMax_f const searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS : ZSTD_HcFindBestMatch_selectMLS;
|
422
|
+
U32 offset_1 = seqStorePtr->rep[0], offset_2 = seqStorePtr->rep[1], savedOffset=0;
|
423
|
+
|
424
|
+
/* init */
|
425
|
+
ip += (ip==base);
|
426
|
+
ctx->nextToUpdate3 = ctx->nextToUpdate;
|
427
|
+
{ U32 const maxRep = (U32)(ip-base);
|
428
|
+
if (offset_2 > maxRep) savedOffset = offset_2, offset_2 = 0;
|
429
|
+
if (offset_1 > maxRep) savedOffset = offset_1, offset_1 = 0;
|
430
|
+
}
|
431
|
+
|
432
|
+
/* Match Loop */
|
433
|
+
while (ip < ilimit) {
|
434
|
+
size_t matchLength=0;
|
435
|
+
size_t offset=0;
|
436
|
+
const BYTE* start=ip+1;
|
437
|
+
|
438
|
+
/* check repCode */
|
439
|
+
if ((offset_1>0) & (MEM_read32(ip+1) == MEM_read32(ip+1 - offset_1))) {
|
440
|
+
/* repcode : we take it */
|
441
|
+
matchLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
|
442
|
+
if (depth==0) goto _storeSequence;
|
443
|
+
}
|
444
|
+
|
445
|
+
/* first search (depth 0) */
|
446
|
+
{ size_t offsetFound = 99999999;
|
447
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
|
448
|
+
if (ml2 > matchLength)
|
449
|
+
matchLength = ml2, start = ip, offset=offsetFound;
|
450
|
+
}
|
451
|
+
|
452
|
+
if (matchLength < 4) {
|
453
|
+
ip += ((ip-anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
|
454
|
+
continue;
|
455
|
+
}
|
456
|
+
|
457
|
+
/* let's try to find a better solution */
|
458
|
+
if (depth>=1)
|
459
|
+
while (ip<ilimit) {
|
460
|
+
ip ++;
|
461
|
+
if ((offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
|
462
|
+
size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
|
463
|
+
int const gain2 = (int)(mlRep * 3);
|
464
|
+
int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
|
465
|
+
if ((mlRep >= 4) && (gain2 > gain1))
|
466
|
+
matchLength = mlRep, offset = 0, start = ip;
|
467
|
+
}
|
468
|
+
{ size_t offset2=99999999;
|
469
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
|
470
|
+
int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */
|
471
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
|
472
|
+
if ((ml2 >= 4) && (gain2 > gain1)) {
|
473
|
+
matchLength = ml2, offset = offset2, start = ip;
|
474
|
+
continue; /* search a better one */
|
475
|
+
} }
|
476
|
+
|
477
|
+
/* let's find an even better one */
|
478
|
+
if ((depth==2) && (ip<ilimit)) {
|
479
|
+
ip ++;
|
480
|
+
if ((offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
|
481
|
+
size_t const ml2 = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
|
482
|
+
int const gain2 = (int)(ml2 * 4);
|
483
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
|
484
|
+
if ((ml2 >= 4) && (gain2 > gain1))
|
485
|
+
matchLength = ml2, offset = 0, start = ip;
|
486
|
+
}
|
487
|
+
{ size_t offset2=99999999;
|
488
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
|
489
|
+
int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */
|
490
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
|
491
|
+
if ((ml2 >= 4) && (gain2 > gain1)) {
|
492
|
+
matchLength = ml2, offset = offset2, start = ip;
|
493
|
+
continue;
|
494
|
+
} } }
|
495
|
+
break; /* nothing found : store previous solution */
|
496
|
+
}
|
497
|
+
|
498
|
+
/* NOTE:
|
499
|
+
* start[-offset+ZSTD_REP_MOVE-1] is undefined behavior.
|
500
|
+
* (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which
|
501
|
+
* overflows the pointer, which is undefined behavior.
|
502
|
+
*/
|
503
|
+
/* catch up */
|
504
|
+
if (offset) {
|
505
|
+
while ( (start > anchor)
|
506
|
+
&& (start > base+offset-ZSTD_REP_MOVE)
|
507
|
+
&& (start[-1] == (start-offset+ZSTD_REP_MOVE)[-1]) ) /* only search for offset within prefix */
|
508
|
+
{ start--; matchLength++; }
|
509
|
+
offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
|
510
|
+
}
|
511
|
+
/* store sequence */
|
512
|
+
_storeSequence:
|
513
|
+
{ size_t const litLength = start - anchor;
|
514
|
+
ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength-MINMATCH);
|
515
|
+
anchor = ip = start + matchLength;
|
516
|
+
}
|
517
|
+
|
518
|
+
/* check immediate repcode */
|
519
|
+
while ( (ip <= ilimit)
|
520
|
+
&& ((offset_2>0)
|
521
|
+
& (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
|
522
|
+
/* store sequence */
|
523
|
+
matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
|
524
|
+
offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */
|
525
|
+
ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength-MINMATCH);
|
526
|
+
ip += matchLength;
|
527
|
+
anchor = ip;
|
528
|
+
continue; /* faster when present ... (?) */
|
529
|
+
} }
|
530
|
+
|
531
|
+
/* Save reps for next block */
|
532
|
+
seqStorePtr->repToConfirm[0] = offset_1 ? offset_1 : savedOffset;
|
533
|
+
seqStorePtr->repToConfirm[1] = offset_2 ? offset_2 : savedOffset;
|
534
|
+
|
535
|
+
/* Return the last literals size */
|
536
|
+
return iend - anchor;
|
537
|
+
}
|
538
|
+
|
539
|
+
|
540
|
+
size_t ZSTD_compressBlock_btlazy2(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
541
|
+
{
|
542
|
+
return ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 1, 2);
|
543
|
+
}
|
544
|
+
|
545
|
+
size_t ZSTD_compressBlock_lazy2(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
546
|
+
{
|
547
|
+
return ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 2);
|
548
|
+
}
|
549
|
+
|
550
|
+
size_t ZSTD_compressBlock_lazy(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
551
|
+
{
|
552
|
+
return ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 1);
|
553
|
+
}
|
554
|
+
|
555
|
+
size_t ZSTD_compressBlock_greedy(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
556
|
+
{
|
557
|
+
return ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 0);
|
558
|
+
}
|
559
|
+
|
560
|
+
|
561
|
+
FORCE_INLINE_TEMPLATE
|
562
|
+
size_t ZSTD_compressBlock_lazy_extDict_generic(ZSTD_CCtx* ctx,
|
563
|
+
const void* src, size_t srcSize,
|
564
|
+
const U32 searchMethod, const U32 depth)
|
565
|
+
{
|
566
|
+
seqStore_t* seqStorePtr = &(ctx->seqStore);
|
567
|
+
const BYTE* const istart = (const BYTE*)src;
|
568
|
+
const BYTE* ip = istart;
|
569
|
+
const BYTE* anchor = istart;
|
570
|
+
const BYTE* const iend = istart + srcSize;
|
571
|
+
const BYTE* const ilimit = iend - 8;
|
572
|
+
const BYTE* const base = ctx->base;
|
573
|
+
const U32 dictLimit = ctx->dictLimit;
|
574
|
+
const U32 lowestIndex = ctx->lowLimit;
|
575
|
+
const BYTE* const prefixStart = base + dictLimit;
|
576
|
+
const BYTE* const dictBase = ctx->dictBase;
|
577
|
+
const BYTE* const dictEnd = dictBase + dictLimit;
|
578
|
+
const BYTE* const dictStart = dictBase + ctx->lowLimit;
|
579
|
+
|
580
|
+
const U32 maxSearches = 1 << ctx->appliedParams.cParams.searchLog;
|
581
|
+
const U32 mls = ctx->appliedParams.cParams.searchLength;
|
582
|
+
|
583
|
+
typedef size_t (*searchMax_f)(ZSTD_CCtx* zc, const BYTE* ip, const BYTE* iLimit,
|
584
|
+
size_t* offsetPtr,
|
585
|
+
U32 maxNbAttempts, U32 matchLengthSearch);
|
586
|
+
searchMax_f searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS_extDict : ZSTD_HcFindBestMatch_extDict_selectMLS;
|
587
|
+
|
588
|
+
U32 offset_1 = seqStorePtr->rep[0], offset_2 = seqStorePtr->rep[1];
|
589
|
+
|
590
|
+
/* init */
|
591
|
+
ctx->nextToUpdate3 = ctx->nextToUpdate;
|
592
|
+
ip += (ip == prefixStart);
|
593
|
+
|
594
|
+
/* Match Loop */
|
595
|
+
while (ip < ilimit) {
|
596
|
+
size_t matchLength=0;
|
597
|
+
size_t offset=0;
|
598
|
+
const BYTE* start=ip+1;
|
599
|
+
U32 current = (U32)(ip-base);
|
600
|
+
|
601
|
+
/* check repCode */
|
602
|
+
{ const U32 repIndex = (U32)(current+1 - offset_1);
|
603
|
+
const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
|
604
|
+
const BYTE* const repMatch = repBase + repIndex;
|
605
|
+
if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
|
606
|
+
if (MEM_read32(ip+1) == MEM_read32(repMatch)) {
|
607
|
+
/* repcode detected we should take it */
|
608
|
+
const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
|
609
|
+
matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repEnd, prefixStart) + 4;
|
610
|
+
if (depth==0) goto _storeSequence;
|
611
|
+
} }
|
612
|
+
|
613
|
+
/* first search (depth 0) */
|
614
|
+
{ size_t offsetFound = 99999999;
|
615
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
|
616
|
+
if (ml2 > matchLength)
|
617
|
+
matchLength = ml2, start = ip, offset=offsetFound;
|
618
|
+
}
|
619
|
+
|
620
|
+
if (matchLength < 4) {
|
621
|
+
ip += ((ip-anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
|
622
|
+
continue;
|
623
|
+
}
|
624
|
+
|
625
|
+
/* let's try to find a better solution */
|
626
|
+
if (depth>=1)
|
627
|
+
while (ip<ilimit) {
|
628
|
+
ip ++;
|
629
|
+
current++;
|
630
|
+
/* check repCode */
|
631
|
+
if (offset) {
|
632
|
+
const U32 repIndex = (U32)(current - offset_1);
|
633
|
+
const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
|
634
|
+
const BYTE* const repMatch = repBase + repIndex;
|
635
|
+
if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
|
636
|
+
if (MEM_read32(ip) == MEM_read32(repMatch)) {
|
637
|
+
/* repcode detected */
|
638
|
+
const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
|
639
|
+
size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
|
640
|
+
int const gain2 = (int)(repLength * 3);
|
641
|
+
int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
|
642
|
+
if ((repLength >= 4) && (gain2 > gain1))
|
643
|
+
matchLength = repLength, offset = 0, start = ip;
|
644
|
+
} }
|
645
|
+
|
646
|
+
/* search match, depth 1 */
|
647
|
+
{ size_t offset2=99999999;
|
648
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
|
649
|
+
int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */
|
650
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
|
651
|
+
if ((ml2 >= 4) && (gain2 > gain1)) {
|
652
|
+
matchLength = ml2, offset = offset2, start = ip;
|
653
|
+
continue; /* search a better one */
|
654
|
+
} }
|
655
|
+
|
656
|
+
/* let's find an even better one */
|
657
|
+
if ((depth==2) && (ip<ilimit)) {
|
658
|
+
ip ++;
|
659
|
+
current++;
|
660
|
+
/* check repCode */
|
661
|
+
if (offset) {
|
662
|
+
const U32 repIndex = (U32)(current - offset_1);
|
663
|
+
const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
|
664
|
+
const BYTE* const repMatch = repBase + repIndex;
|
665
|
+
if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
|
666
|
+
if (MEM_read32(ip) == MEM_read32(repMatch)) {
|
667
|
+
/* repcode detected */
|
668
|
+
const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
|
669
|
+
size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
|
670
|
+
int const gain2 = (int)(repLength * 4);
|
671
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
|
672
|
+
if ((repLength >= 4) && (gain2 > gain1))
|
673
|
+
matchLength = repLength, offset = 0, start = ip;
|
674
|
+
} }
|
675
|
+
|
676
|
+
/* search match, depth 2 */
|
677
|
+
{ size_t offset2=99999999;
|
678
|
+
size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
|
679
|
+
int const gain2 = (int)(ml2*4 - ZSTD_highbit32((U32)offset2+1)); /* raw approx */
|
680
|
+
int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
|
681
|
+
if ((ml2 >= 4) && (gain2 > gain1)) {
|
682
|
+
matchLength = ml2, offset = offset2, start = ip;
|
683
|
+
continue;
|
684
|
+
} } }
|
685
|
+
break; /* nothing found : store previous solution */
|
686
|
+
}
|
687
|
+
|
688
|
+
/* catch up */
|
689
|
+
if (offset) {
|
690
|
+
U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE));
|
691
|
+
const BYTE* match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
|
692
|
+
const BYTE* const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
|
693
|
+
while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; } /* catch up */
|
694
|
+
offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
|
695
|
+
}
|
696
|
+
|
697
|
+
/* store sequence */
|
698
|
+
_storeSequence:
|
699
|
+
{ size_t const litLength = start - anchor;
|
700
|
+
ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength-MINMATCH);
|
701
|
+
anchor = ip = start + matchLength;
|
702
|
+
}
|
703
|
+
|
704
|
+
/* check immediate repcode */
|
705
|
+
while (ip <= ilimit) {
|
706
|
+
const U32 repIndex = (U32)((ip-base) - offset_2);
|
707
|
+
const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
|
708
|
+
const BYTE* const repMatch = repBase + repIndex;
|
709
|
+
if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
|
710
|
+
if (MEM_read32(ip) == MEM_read32(repMatch)) {
|
711
|
+
/* repcode detected we should take it */
|
712
|
+
const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
|
713
|
+
matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
|
714
|
+
offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset history */
|
715
|
+
ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength-MINMATCH);
|
716
|
+
ip += matchLength;
|
717
|
+
anchor = ip;
|
718
|
+
continue; /* faster when present ... (?) */
|
719
|
+
}
|
720
|
+
break;
|
721
|
+
} }
|
722
|
+
|
723
|
+
/* Save reps for next block */
|
724
|
+
seqStorePtr->repToConfirm[0] = offset_1; seqStorePtr->repToConfirm[1] = offset_2;
|
725
|
+
|
726
|
+
/* Return the last literals size */
|
727
|
+
return iend - anchor;
|
728
|
+
}
|
729
|
+
|
730
|
+
|
731
|
+
size_t ZSTD_compressBlock_greedy_extDict(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
732
|
+
{
|
733
|
+
return ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 0);
|
734
|
+
}
|
735
|
+
|
736
|
+
size_t ZSTD_compressBlock_lazy_extDict(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
737
|
+
{
|
738
|
+
return ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 1);
|
739
|
+
}
|
740
|
+
|
741
|
+
size_t ZSTD_compressBlock_lazy2_extDict(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
742
|
+
{
|
743
|
+
return ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 2);
|
744
|
+
}
|
745
|
+
|
746
|
+
size_t ZSTD_compressBlock_btlazy2_extDict(ZSTD_CCtx* ctx, const void* src, size_t srcSize)
|
747
|
+
{
|
748
|
+
return ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 1, 2);
|
749
|
+
}
|