zstdlib 0.1.2-x86-mingw32 → 0.2.0-x86-mingw32
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.yardopts +6 -0
- data/CHANGES.md +7 -0
- data/README.md +4 -3
- data/ext/zstdlib/extconf.rb +1 -1
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/bitstream.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/compiler.h +1 -1
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/cpu.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/debug.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/debug.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/entropy_common.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/error_private.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/error_private.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/fse.h +1 -1
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/fse_decompress.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/huf.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/mem.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/pool.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/pool.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/threading.c +2 -2
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/threading.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/xxhash.c +2 -2
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/xxhash.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/zstd_common.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/zstd_errors.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/common/zstd_internal.h +55 -2
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/fse_compress.c +2 -2
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/hist.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/hist.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/huf_compress.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_compress.c +423 -296
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_compress_internal.h +14 -11
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_double_fast.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_double_fast.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_fast.c +203 -124
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_fast.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_lazy.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_lazy.h +1 -1
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_ldm.c +1 -1
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_ldm.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_opt.c +27 -11
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstd_opt.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstdmt_compress.c +41 -49
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/compress/zstdmt_compress.h +43 -26
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/huf_decompress.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_ddict.c +4 -4
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_ddict.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_decompress.c +257 -164
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_decompress_block.c +51 -47
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_decompress_block.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/decompress/zstd_decompress_internal.h +7 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/lib/zstd.h +689 -542
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzclose.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzcompatibility.h +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzguts.h +2 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzlib.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzread.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/gzwrite.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/zstd_zlibwrapper.c +0 -0
- data/ext/zstdlib/{zstd-1.3.8 → zstd-1.4.0}/zlibWrapper/zstd_zlibwrapper.h +0 -0
- data/lib/2.2/zstdlib.so +0 -0
- data/lib/2.3/zstdlib.so +0 -0
- data/lib/2.4/zstdlib.so +0 -0
- data/lib/2.5/zstdlib.so +0 -0
- data/lib/2.6/zstdlib.so +0 -0
- metadata +58 -57
@@ -36,9 +36,9 @@ extern "C" {
|
|
36
36
|
#define ZSTD_DUBT_UNSORTED_MARK 1 /* For btlazy2 strategy, index 1 now means "unsorted".
|
37
37
|
It could be confused for a real successor at index "1", if sorted as larger than its predecessor.
|
38
38
|
It's not a big deal though : candidate will just be sorted again.
|
39
|
-
|
39
|
+
Additionally, candidate position 1 will be lost.
|
40
40
|
But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss.
|
41
|
-
The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be
|
41
|
+
The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be mishandled after table re-use with a different strategy
|
42
42
|
Constant required by ZSTD_compressBlock_btlazy2() and ZSTD_reduceTable_internal() */
|
43
43
|
|
44
44
|
|
@@ -54,6 +54,14 @@ typedef struct ZSTD_prefixDict_s {
|
|
54
54
|
ZSTD_dictContentType_e dictContentType;
|
55
55
|
} ZSTD_prefixDict;
|
56
56
|
|
57
|
+
typedef struct {
|
58
|
+
void* dictBuffer;
|
59
|
+
void const* dict;
|
60
|
+
size_t dictSize;
|
61
|
+
ZSTD_dictContentType_e dictContentType;
|
62
|
+
ZSTD_CDict* cdict;
|
63
|
+
} ZSTD_localDict;
|
64
|
+
|
57
65
|
typedef struct {
|
58
66
|
U32 CTable[HUF_CTABLE_SIZE_U32(255)];
|
59
67
|
HUF_repeat repeatMode;
|
@@ -107,6 +115,7 @@ typedef struct {
|
|
107
115
|
U32 offCodeSumBasePrice; /* to compare to log2(offreq) */
|
108
116
|
ZSTD_OptPrice_e priceType; /* prices can be determined dynamically, or follow a pre-defined cost structure */
|
109
117
|
const ZSTD_entropyCTables_t* symbolCosts; /* pre-calculated dictionary statistics */
|
118
|
+
ZSTD_literalCompressionMode_e literalCompressionMode;
|
110
119
|
} optState_t;
|
111
120
|
|
112
121
|
typedef struct {
|
@@ -188,6 +197,7 @@ struct ZSTD_CCtx_params_s {
|
|
188
197
|
* 1<<wLog, even for dictionary */
|
189
198
|
|
190
199
|
ZSTD_dictAttachPref_e attachDictPref;
|
200
|
+
ZSTD_literalCompressionMode_e literalCompressionMode;
|
191
201
|
|
192
202
|
/* Multithreading: used to pass parameters to mtctx */
|
193
203
|
int nbWorkers;
|
@@ -243,7 +253,7 @@ struct ZSTD_CCtx_s {
|
|
243
253
|
U32 frameEnded;
|
244
254
|
|
245
255
|
/* Dictionary */
|
246
|
-
|
256
|
+
ZSTD_localDict localDict;
|
247
257
|
const ZSTD_CDict* cdict;
|
248
258
|
ZSTD_prefixDict prefixDict; /* single-usage dictionary */
|
249
259
|
|
@@ -806,13 +816,6 @@ size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
|
|
806
816
|
|
807
817
|
void ZSTD_resetSeqStore(seqStore_t* ssPtr);
|
808
818
|
|
809
|
-
/*! ZSTD_compressStream_generic() :
|
810
|
-
* Private use only. To be called from zstdmt_compress.c in single-thread mode. */
|
811
|
-
size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
|
812
|
-
ZSTD_outBuffer* output,
|
813
|
-
ZSTD_inBuffer* input,
|
814
|
-
ZSTD_EndDirective const flushMode);
|
815
|
-
|
816
819
|
/*! ZSTD_getCParamsFromCDict() :
|
817
820
|
* as the name implies */
|
818
821
|
ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict);
|
@@ -839,7 +842,7 @@ size_t ZSTD_compress_advanced_internal(ZSTD_CCtx* cctx,
|
|
839
842
|
/* ZSTD_writeLastEmptyBlock() :
|
840
843
|
* output an empty Block with end-of-frame mark to complete a frame
|
841
844
|
* @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
|
842
|
-
* or an error code if `
|
845
|
+
* or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
|
843
846
|
*/
|
844
847
|
size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity);
|
845
848
|
|
File without changes
|
File without changes
|
@@ -45,7 +45,155 @@ FORCE_INLINE_TEMPLATE
|
|
45
45
|
size_t ZSTD_compressBlock_fast_generic(
|
46
46
|
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
47
47
|
void const* src, size_t srcSize,
|
48
|
-
U32 const mls
|
48
|
+
U32 const mls)
|
49
|
+
{
|
50
|
+
const ZSTD_compressionParameters* const cParams = &ms->cParams;
|
51
|
+
U32* const hashTable = ms->hashTable;
|
52
|
+
U32 const hlog = cParams->hashLog;
|
53
|
+
/* support stepSize of 0 */
|
54
|
+
size_t const stepSize = cParams->targetLength + !(cParams->targetLength) + 1;
|
55
|
+
const BYTE* const base = ms->window.base;
|
56
|
+
const BYTE* const istart = (const BYTE*)src;
|
57
|
+
/* We check ip0 (ip + 0) and ip1 (ip + 1) each loop */
|
58
|
+
const BYTE* ip0 = istart;
|
59
|
+
const BYTE* ip1;
|
60
|
+
const BYTE* anchor = istart;
|
61
|
+
const U32 prefixStartIndex = ms->window.dictLimit;
|
62
|
+
const BYTE* const prefixStart = base + prefixStartIndex;
|
63
|
+
const BYTE* const iend = istart + srcSize;
|
64
|
+
const BYTE* const ilimit = iend - HASH_READ_SIZE;
|
65
|
+
U32 offset_1=rep[0], offset_2=rep[1];
|
66
|
+
U32 offsetSaved = 0;
|
67
|
+
|
68
|
+
/* init */
|
69
|
+
ip0 += (ip0 == prefixStart);
|
70
|
+
ip1 = ip0 + 1;
|
71
|
+
{
|
72
|
+
U32 const maxRep = (U32)(ip0 - prefixStart);
|
73
|
+
if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
|
74
|
+
if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
|
75
|
+
}
|
76
|
+
|
77
|
+
/* Main Search Loop */
|
78
|
+
while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
|
79
|
+
size_t mLength;
|
80
|
+
BYTE const* ip2 = ip0 + 2;
|
81
|
+
size_t const h0 = ZSTD_hashPtr(ip0, hlog, mls);
|
82
|
+
U32 const val0 = MEM_read32(ip0);
|
83
|
+
size_t const h1 = ZSTD_hashPtr(ip1, hlog, mls);
|
84
|
+
U32 const val1 = MEM_read32(ip1);
|
85
|
+
U32 const current0 = (U32)(ip0-base);
|
86
|
+
U32 const current1 = (U32)(ip1-base);
|
87
|
+
U32 const matchIndex0 = hashTable[h0];
|
88
|
+
U32 const matchIndex1 = hashTable[h1];
|
89
|
+
BYTE const* repMatch = ip2-offset_1;
|
90
|
+
const BYTE* match0 = base + matchIndex0;
|
91
|
+
const BYTE* match1 = base + matchIndex1;
|
92
|
+
U32 offcode;
|
93
|
+
hashTable[h0] = current0; /* update hash table */
|
94
|
+
hashTable[h1] = current1; /* update hash table */
|
95
|
+
|
96
|
+
assert(ip0 + 1 == ip1);
|
97
|
+
|
98
|
+
if ((offset_1 > 0) & (MEM_read32(repMatch) == MEM_read32(ip2))) {
|
99
|
+
mLength = ip2[-1] == repMatch[-1] ? 1 : 0;
|
100
|
+
ip0 = ip2 - mLength;
|
101
|
+
match0 = repMatch - mLength;
|
102
|
+
offcode = 0;
|
103
|
+
goto _match;
|
104
|
+
}
|
105
|
+
if ((matchIndex0 > prefixStartIndex) && MEM_read32(match0) == val0) {
|
106
|
+
/* found a regular match */
|
107
|
+
goto _offset;
|
108
|
+
}
|
109
|
+
if ((matchIndex1 > prefixStartIndex) && MEM_read32(match1) == val1) {
|
110
|
+
/* found a regular match after one literal */
|
111
|
+
ip0 = ip1;
|
112
|
+
match0 = match1;
|
113
|
+
goto _offset;
|
114
|
+
}
|
115
|
+
{
|
116
|
+
size_t const step = ((ip0-anchor) >> (kSearchStrength - 1)) + stepSize;
|
117
|
+
assert(step >= 2);
|
118
|
+
ip0 += step;
|
119
|
+
ip1 += step;
|
120
|
+
continue;
|
121
|
+
}
|
122
|
+
_offset: /* Requires: ip0, match0 */
|
123
|
+
/* Compute the offset code */
|
124
|
+
offset_2 = offset_1;
|
125
|
+
offset_1 = (U32)(ip0-match0);
|
126
|
+
offcode = offset_1 + ZSTD_REP_MOVE;
|
127
|
+
mLength = 0;
|
128
|
+
/* Count the backwards match length */
|
129
|
+
while (((ip0>anchor) & (match0>prefixStart))
|
130
|
+
&& (ip0[-1] == match0[-1])) { ip0--; match0--; mLength++; } /* catch up */
|
131
|
+
|
132
|
+
_match: /* Requires: ip0, match0, offcode */
|
133
|
+
/* Count the forward length */
|
134
|
+
mLength += ZSTD_count(ip0+mLength+4, match0+mLength+4, iend) + 4;
|
135
|
+
ZSTD_storeSeq(seqStore, ip0-anchor, anchor, offcode, mLength-MINMATCH);
|
136
|
+
/* match found */
|
137
|
+
ip0 += mLength;
|
138
|
+
anchor = ip0;
|
139
|
+
ip1 = ip0 + 1;
|
140
|
+
|
141
|
+
if (ip0 <= ilimit) {
|
142
|
+
/* Fill Table */
|
143
|
+
assert(base+current0+2 > istart); /* check base overflow */
|
144
|
+
hashTable[ZSTD_hashPtr(base+current0+2, hlog, mls)] = current0+2; /* here because current+2 could be > iend-8 */
|
145
|
+
hashTable[ZSTD_hashPtr(ip0-2, hlog, mls)] = (U32)(ip0-2-base);
|
146
|
+
|
147
|
+
while ( (ip0 <= ilimit)
|
148
|
+
&& ( (offset_2>0)
|
149
|
+
& (MEM_read32(ip0) == MEM_read32(ip0 - offset_2)) )) {
|
150
|
+
/* store sequence */
|
151
|
+
size_t const rLength = ZSTD_count(ip0+4, ip0+4-offset_2, iend) + 4;
|
152
|
+
U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */
|
153
|
+
hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base);
|
154
|
+
ip0 += rLength;
|
155
|
+
ip1 = ip0 + 1;
|
156
|
+
ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
|
157
|
+
anchor = ip0;
|
158
|
+
continue; /* faster when present (confirmed on gcc-8) ... (?) */
|
159
|
+
}
|
160
|
+
}
|
161
|
+
}
|
162
|
+
|
163
|
+
/* save reps for next block */
|
164
|
+
rep[0] = offset_1 ? offset_1 : offsetSaved;
|
165
|
+
rep[1] = offset_2 ? offset_2 : offsetSaved;
|
166
|
+
|
167
|
+
/* Return the last literals size */
|
168
|
+
return iend - anchor;
|
169
|
+
}
|
170
|
+
|
171
|
+
|
172
|
+
size_t ZSTD_compressBlock_fast(
|
173
|
+
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
174
|
+
void const* src, size_t srcSize)
|
175
|
+
{
|
176
|
+
ZSTD_compressionParameters const* cParams = &ms->cParams;
|
177
|
+
U32 const mls = cParams->minMatch;
|
178
|
+
assert(ms->dictMatchState == NULL);
|
179
|
+
switch(mls)
|
180
|
+
{
|
181
|
+
default: /* includes case 3 */
|
182
|
+
case 4 :
|
183
|
+
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 4);
|
184
|
+
case 5 :
|
185
|
+
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 5);
|
186
|
+
case 6 :
|
187
|
+
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 6);
|
188
|
+
case 7 :
|
189
|
+
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 7);
|
190
|
+
}
|
191
|
+
}
|
192
|
+
|
193
|
+
FORCE_INLINE_TEMPLATE
|
194
|
+
size_t ZSTD_compressBlock_fast_dictMatchState_generic(
|
195
|
+
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
196
|
+
void const* src, size_t srcSize, U32 const mls)
|
49
197
|
{
|
50
198
|
const ZSTD_compressionParameters* const cParams = &ms->cParams;
|
51
199
|
U32* const hashTable = ms->hashTable;
|
@@ -64,46 +212,26 @@ size_t ZSTD_compressBlock_fast_generic(
|
|
64
212
|
U32 offsetSaved = 0;
|
65
213
|
|
66
214
|
const ZSTD_matchState_t* const dms = ms->dictMatchState;
|
67
|
-
const ZSTD_compressionParameters* const dictCParams =
|
68
|
-
|
69
|
-
|
70
|
-
const
|
71
|
-
|
72
|
-
const
|
73
|
-
|
74
|
-
const BYTE* const dictBase = dictMode == ZSTD_dictMatchState ?
|
75
|
-
dms->window.base : NULL;
|
76
|
-
const BYTE* const dictStart = dictMode == ZSTD_dictMatchState ?
|
77
|
-
dictBase + dictStartIndex : NULL;
|
78
|
-
const BYTE* const dictEnd = dictMode == ZSTD_dictMatchState ?
|
79
|
-
dms->window.nextSrc : NULL;
|
80
|
-
const U32 dictIndexDelta = dictMode == ZSTD_dictMatchState ?
|
81
|
-
prefixStartIndex - (U32)(dictEnd - dictBase) :
|
82
|
-
0;
|
215
|
+
const ZSTD_compressionParameters* const dictCParams = &dms->cParams ;
|
216
|
+
const U32* const dictHashTable = dms->hashTable;
|
217
|
+
const U32 dictStartIndex = dms->window.dictLimit;
|
218
|
+
const BYTE* const dictBase = dms->window.base;
|
219
|
+
const BYTE* const dictStart = dictBase + dictStartIndex;
|
220
|
+
const BYTE* const dictEnd = dms->window.nextSrc;
|
221
|
+
const U32 dictIndexDelta = prefixStartIndex - (U32)(dictEnd - dictBase);
|
83
222
|
const U32 dictAndPrefixLength = (U32)(ip - prefixStart + dictEnd - dictStart);
|
84
|
-
const U32 dictHLog =
|
85
|
-
dictCParams->hashLog : hlog;
|
86
|
-
|
87
|
-
assert(dictMode == ZSTD_noDict || dictMode == ZSTD_dictMatchState);
|
223
|
+
const U32 dictHLog = dictCParams->hashLog;
|
88
224
|
|
89
225
|
/* otherwise, we would get index underflow when translating a dict index
|
90
226
|
* into a local index */
|
91
|
-
assert(
|
92
|
-
|| prefixStartIndex >= (U32)(dictEnd - dictBase));
|
227
|
+
assert(prefixStartIndex >= (U32)(dictEnd - dictBase));
|
93
228
|
|
94
229
|
/* init */
|
95
230
|
ip += (dictAndPrefixLength == 0);
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
|
100
|
-
}
|
101
|
-
if (dictMode == ZSTD_dictMatchState) {
|
102
|
-
/* dictMatchState repCode checks don't currently handle repCode == 0
|
103
|
-
* disabling. */
|
104
|
-
assert(offset_1 <= dictAndPrefixLength);
|
105
|
-
assert(offset_2 <= dictAndPrefixLength);
|
106
|
-
}
|
231
|
+
/* dictMatchState repCode checks don't currently handle repCode == 0
|
232
|
+
* disabling. */
|
233
|
+
assert(offset_1 <= dictAndPrefixLength);
|
234
|
+
assert(offset_2 <= dictAndPrefixLength);
|
107
235
|
|
108
236
|
/* Main Search Loop */
|
109
237
|
while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
|
@@ -113,50 +241,37 @@ size_t ZSTD_compressBlock_fast_generic(
|
|
113
241
|
U32 const matchIndex = hashTable[h];
|
114
242
|
const BYTE* match = base + matchIndex;
|
115
243
|
const U32 repIndex = current + 1 - offset_1;
|
116
|
-
const BYTE* repMatch = (
|
117
|
-
&& repIndex < prefixStartIndex) ?
|
244
|
+
const BYTE* repMatch = (repIndex < prefixStartIndex) ?
|
118
245
|
dictBase + (repIndex - dictIndexDelta) :
|
119
246
|
base + repIndex;
|
120
247
|
hashTable[h] = current; /* update hash table */
|
121
248
|
|
122
|
-
if ( (
|
123
|
-
&& ((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex isn't overlapping dict + prefix */
|
249
|
+
if ( ((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex isn't overlapping dict + prefix */
|
124
250
|
&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
|
125
251
|
const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
|
126
252
|
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
|
127
253
|
ip++;
|
128
254
|
ZSTD_storeSeq(seqStore, ip-anchor, anchor, 0, mLength-MINMATCH);
|
129
|
-
} else if ( dictMode == ZSTD_noDict
|
130
|
-
&& ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
|
131
|
-
mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
|
132
|
-
ip++;
|
133
|
-
ZSTD_storeSeq(seqStore, ip-anchor, anchor, 0, mLength-MINMATCH);
|
134
255
|
} else if ( (matchIndex <= prefixStartIndex) ) {
|
135
|
-
|
136
|
-
|
137
|
-
|
138
|
-
|
139
|
-
|
140
|
-
MEM_read32(dictMatch) != MEM_read32(ip)) {
|
141
|
-
assert(stepSize >= 1);
|
142
|
-
ip += ((ip-anchor) >> kSearchStrength) + stepSize;
|
143
|
-
continue;
|
144
|
-
} else {
|
145
|
-
/* found a dict match */
|
146
|
-
U32 const offset = (U32)(current-dictMatchIndex-dictIndexDelta);
|
147
|
-
mLength = ZSTD_count_2segments(ip+4, dictMatch+4, iend, dictEnd, prefixStart) + 4;
|
148
|
-
while (((ip>anchor) & (dictMatch>dictStart))
|
149
|
-
&& (ip[-1] == dictMatch[-1])) {
|
150
|
-
ip--; dictMatch--; mLength++;
|
151
|
-
} /* catch up */
|
152
|
-
offset_2 = offset_1;
|
153
|
-
offset_1 = offset;
|
154
|
-
ZSTD_storeSeq(seqStore, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
|
155
|
-
}
|
156
|
-
} else {
|
256
|
+
size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls);
|
257
|
+
U32 const dictMatchIndex = dictHashTable[dictHash];
|
258
|
+
const BYTE* dictMatch = dictBase + dictMatchIndex;
|
259
|
+
if (dictMatchIndex <= dictStartIndex ||
|
260
|
+
MEM_read32(dictMatch) != MEM_read32(ip)) {
|
157
261
|
assert(stepSize >= 1);
|
158
262
|
ip += ((ip-anchor) >> kSearchStrength) + stepSize;
|
159
263
|
continue;
|
264
|
+
} else {
|
265
|
+
/* found a dict match */
|
266
|
+
U32 const offset = (U32)(current-dictMatchIndex-dictIndexDelta);
|
267
|
+
mLength = ZSTD_count_2segments(ip+4, dictMatch+4, iend, dictEnd, prefixStart) + 4;
|
268
|
+
while (((ip>anchor) & (dictMatch>dictStart))
|
269
|
+
&& (ip[-1] == dictMatch[-1])) {
|
270
|
+
ip--; dictMatch--; mLength++;
|
271
|
+
} /* catch up */
|
272
|
+
offset_2 = offset_1;
|
273
|
+
offset_1 = offset;
|
274
|
+
ZSTD_storeSeq(seqStore, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
|
160
275
|
}
|
161
276
|
} else if (MEM_read32(match) != MEM_read32(ip)) {
|
162
277
|
/* it's not a match, and we're not going to check the dictionary */
|
@@ -185,41 +300,27 @@ size_t ZSTD_compressBlock_fast_generic(
|
|
185
300
|
hashTable[ZSTD_hashPtr(ip-2, hlog, mls)] = (U32)(ip-2-base);
|
186
301
|
|
187
302
|
/* check immediate repcode */
|
188
|
-
|
189
|
-
|
190
|
-
|
191
|
-
|
192
|
-
|
193
|
-
|
194
|
-
|
195
|
-
|
196
|
-
|
197
|
-
|
198
|
-
|
199
|
-
|
200
|
-
|
201
|
-
|
202
|
-
|
203
|
-
|
204
|
-
continue;
|
205
|
-
}
|
206
|
-
break;
|
303
|
+
while (ip <= ilimit) {
|
304
|
+
U32 const current2 = (U32)(ip-base);
|
305
|
+
U32 const repIndex2 = current2 - offset_2;
|
306
|
+
const BYTE* repMatch2 = repIndex2 < prefixStartIndex ?
|
307
|
+
dictBase - dictIndexDelta + repIndex2 :
|
308
|
+
base + repIndex2;
|
309
|
+
if ( ((U32)((prefixStartIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
|
310
|
+
&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
|
311
|
+
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
|
312
|
+
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
|
313
|
+
U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
|
314
|
+
ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
|
315
|
+
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
|
316
|
+
ip += repLength2;
|
317
|
+
anchor = ip;
|
318
|
+
continue;
|
207
319
|
}
|
320
|
+
break;
|
208
321
|
}
|
209
|
-
|
210
|
-
|
211
|
-
while ( (ip <= ilimit)
|
212
|
-
&& ( (offset_2>0)
|
213
|
-
& (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
|
214
|
-
/* store sequence */
|
215
|
-
size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
|
216
|
-
U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */
|
217
|
-
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = (U32)(ip-base);
|
218
|
-
ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
|
219
|
-
ip += rLength;
|
220
|
-
anchor = ip;
|
221
|
-
continue; /* faster when present ... (?) */
|
222
|
-
} } } }
|
322
|
+
}
|
323
|
+
}
|
223
324
|
|
224
325
|
/* save reps for next block */
|
225
326
|
rep[0] = offset_1 ? offset_1 : offsetSaved;
|
@@ -229,28 +330,6 @@ size_t ZSTD_compressBlock_fast_generic(
|
|
229
330
|
return iend - anchor;
|
230
331
|
}
|
231
332
|
|
232
|
-
|
233
|
-
size_t ZSTD_compressBlock_fast(
|
234
|
-
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
235
|
-
void const* src, size_t srcSize)
|
236
|
-
{
|
237
|
-
ZSTD_compressionParameters const* cParams = &ms->cParams;
|
238
|
-
U32 const mls = cParams->minMatch;
|
239
|
-
assert(ms->dictMatchState == NULL);
|
240
|
-
switch(mls)
|
241
|
-
{
|
242
|
-
default: /* includes case 3 */
|
243
|
-
case 4 :
|
244
|
-
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 4, ZSTD_noDict);
|
245
|
-
case 5 :
|
246
|
-
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 5, ZSTD_noDict);
|
247
|
-
case 6 :
|
248
|
-
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 6, ZSTD_noDict);
|
249
|
-
case 7 :
|
250
|
-
return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 7, ZSTD_noDict);
|
251
|
-
}
|
252
|
-
}
|
253
|
-
|
254
333
|
size_t ZSTD_compressBlock_fast_dictMatchState(
|
255
334
|
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
256
335
|
void const* src, size_t srcSize)
|
@@ -262,13 +341,13 @@ size_t ZSTD_compressBlock_fast_dictMatchState(
|
|
262
341
|
{
|
263
342
|
default: /* includes case 3 */
|
264
343
|
case 4 :
|
265
|
-
return
|
344
|
+
return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 4);
|
266
345
|
case 5 :
|
267
|
-
return
|
346
|
+
return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 5);
|
268
347
|
case 6 :
|
269
|
-
return
|
348
|
+
return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 6);
|
270
349
|
case 7 :
|
271
|
-
return
|
350
|
+
return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 7);
|
272
351
|
}
|
273
352
|
}
|
274
353
|
|
File without changes
|
File without changes
|
@@ -19,7 +19,7 @@ extern "C" {
|
|
19
19
|
|
20
20
|
U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);
|
21
21
|
|
22
|
-
void ZSTD_preserveUnsortedMark (U32* const table, U32 const size, U32 const reducerValue); /*! used in ZSTD_reduceIndex().
|
22
|
+
void ZSTD_preserveUnsortedMark (U32* const table, U32 const size, U32 const reducerValue); /*! used in ZSTD_reduceIndex(). preemptively increase value of ZSTD_DUBT_UNSORTED_MARK */
|
23
23
|
|
24
24
|
size_t ZSTD_compressBlock_btlazy2(
|
25
25
|
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
|
@@ -429,7 +429,7 @@ size_t ZSTD_ldm_generateSequences(
|
|
429
429
|
*/
|
430
430
|
assert(ldmState->window.nextSrc >= (BYTE const*)src + srcSize);
|
431
431
|
/* The input could be very large (in zstdmt), so it must be broken up into
|
432
|
-
* chunks to enforce the
|
432
|
+
* chunks to enforce the maximum distance and handle overflow correction.
|
433
433
|
*/
|
434
434
|
assert(sequences->pos <= sequences->size);
|
435
435
|
assert(sequences->size <= sequences->capacity);
|
File without changes
|
@@ -64,9 +64,15 @@ MEM_STATIC double ZSTD_fCost(U32 price)
|
|
64
64
|
}
|
65
65
|
#endif
|
66
66
|
|
67
|
+
static int ZSTD_compressedLiterals(optState_t const* const optPtr)
|
68
|
+
{
|
69
|
+
return optPtr->literalCompressionMode != ZSTD_lcm_uncompressed;
|
70
|
+
}
|
71
|
+
|
67
72
|
static void ZSTD_setBasePrices(optState_t* optPtr, int optLevel)
|
68
73
|
{
|
69
|
-
|
74
|
+
if (ZSTD_compressedLiterals(optPtr))
|
75
|
+
optPtr->litSumBasePrice = WEIGHT(optPtr->litSum, optLevel);
|
70
76
|
optPtr->litLengthSumBasePrice = WEIGHT(optPtr->litLengthSum, optLevel);
|
71
77
|
optPtr->matchLengthSumBasePrice = WEIGHT(optPtr->matchLengthSum, optLevel);
|
72
78
|
optPtr->offCodeSumBasePrice = WEIGHT(optPtr->offCodeSum, optLevel);
|
@@ -99,6 +105,7 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
|
|
99
105
|
const BYTE* const src, size_t const srcSize,
|
100
106
|
int const optLevel)
|
101
107
|
{
|
108
|
+
int const compressedLiterals = ZSTD_compressedLiterals(optPtr);
|
102
109
|
DEBUGLOG(5, "ZSTD_rescaleFreqs (srcSize=%u)", (unsigned)srcSize);
|
103
110
|
optPtr->priceType = zop_dynamic;
|
104
111
|
|
@@ -113,9 +120,10 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
|
|
113
120
|
/* huffman table presumed generated by dictionary */
|
114
121
|
optPtr->priceType = zop_dynamic;
|
115
122
|
|
116
|
-
|
117
|
-
|
118
|
-
|
123
|
+
if (compressedLiterals) {
|
124
|
+
unsigned lit;
|
125
|
+
assert(optPtr->litFreq != NULL);
|
126
|
+
optPtr->litSum = 0;
|
119
127
|
for (lit=0; lit<=MaxLit; lit++) {
|
120
128
|
U32 const scaleLog = 11; /* scale to 2K */
|
121
129
|
U32 const bitCost = HUF_getNbBits(optPtr->symbolCosts->huf.CTable, lit);
|
@@ -163,10 +171,11 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
|
|
163
171
|
} else { /* not a dictionary */
|
164
172
|
|
165
173
|
assert(optPtr->litFreq != NULL);
|
166
|
-
|
174
|
+
if (compressedLiterals) {
|
175
|
+
unsigned lit = MaxLit;
|
167
176
|
HIST_count_simple(optPtr->litFreq, &lit, src, srcSize); /* use raw first block to init statistics */
|
177
|
+
optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
|
168
178
|
}
|
169
|
-
optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
|
170
179
|
|
171
180
|
{ unsigned ll;
|
172
181
|
for (ll=0; ll<=MaxLL; ll++)
|
@@ -190,7 +199,8 @@ ZSTD_rescaleFreqs(optState_t* const optPtr,
|
|
190
199
|
|
191
200
|
} else { /* new block : re-use previous statistics, scaled down */
|
192
201
|
|
193
|
-
|
202
|
+
if (compressedLiterals)
|
203
|
+
optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
|
194
204
|
optPtr->litLengthSum = ZSTD_downscaleStat(optPtr->litLengthFreq, MaxLL, 0);
|
195
205
|
optPtr->matchLengthSum = ZSTD_downscaleStat(optPtr->matchLengthFreq, MaxML, 0);
|
196
206
|
optPtr->offCodeSum = ZSTD_downscaleStat(optPtr->offCodeFreq, MaxOff, 0);
|
@@ -207,6 +217,10 @@ static U32 ZSTD_rawLiteralsCost(const BYTE* const literals, U32 const litLength,
|
|
207
217
|
int optLevel)
|
208
218
|
{
|
209
219
|
if (litLength == 0) return 0;
|
220
|
+
|
221
|
+
if (!ZSTD_compressedLiterals(optPtr))
|
222
|
+
return (litLength << 3) * BITCOST_MULTIPLIER; /* Uncompressed - 8 bytes per literal. */
|
223
|
+
|
210
224
|
if (optPtr->priceType == zop_predef)
|
211
225
|
return (litLength*6) * BITCOST_MULTIPLIER; /* 6 bit per literal - no statistic used */
|
212
226
|
|
@@ -310,7 +324,8 @@ static void ZSTD_updateStats(optState_t* const optPtr,
|
|
310
324
|
U32 offsetCode, U32 matchLength)
|
311
325
|
{
|
312
326
|
/* literals */
|
313
|
-
{
|
327
|
+
if (ZSTD_compressedLiterals(optPtr)) {
|
328
|
+
U32 u;
|
314
329
|
for (u=0; u < litLength; u++)
|
315
330
|
optPtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
|
316
331
|
optPtr->litSum += litLength*ZSTD_LITFREQ_ADD;
|
@@ -870,7 +885,7 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
|
|
870
885
|
/* large match -> immediate encoding */
|
871
886
|
{ U32 const maxML = matches[nbMatches-1].len;
|
872
887
|
U32 const maxOffset = matches[nbMatches-1].off;
|
873
|
-
DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffCode=%u at cPos=%u => start new
|
888
|
+
DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffCode=%u at cPos=%u => start new series",
|
874
889
|
nbMatches, maxML, maxOffset, (U32)(ip-prefixStart));
|
875
890
|
|
876
891
|
if (maxML > sufficient_len) {
|
@@ -1108,7 +1123,8 @@ static U32 ZSTD_upscaleStat(unsigned* table, U32 lastEltIndex, int bonus)
|
|
1108
1123
|
/* used in 2-pass strategy */
|
1109
1124
|
MEM_STATIC void ZSTD_upscaleStats(optState_t* optPtr)
|
1110
1125
|
{
|
1111
|
-
|
1126
|
+
if (ZSTD_compressedLiterals(optPtr))
|
1127
|
+
optPtr->litSum = ZSTD_upscaleStat(optPtr->litFreq, MaxLit, 0);
|
1112
1128
|
optPtr->litLengthSum = ZSTD_upscaleStat(optPtr->litLengthFreq, MaxLL, 0);
|
1113
1129
|
optPtr->matchLengthSum = ZSTD_upscaleStat(optPtr->matchLengthFreq, MaxML, 0);
|
1114
1130
|
optPtr->offCodeSum = ZSTD_upscaleStat(optPtr->offCodeFreq, MaxOff, 0);
|
@@ -1117,7 +1133,7 @@ MEM_STATIC void ZSTD_upscaleStats(optState_t* optPtr)
|
|
1117
1133
|
/* ZSTD_initStats_ultra():
|
1118
1134
|
* make a first compression pass, just to seed stats with more accurate starting values.
|
1119
1135
|
* only works on first block, with no dictionary and no ldm.
|
1120
|
-
* this function cannot error, hence its
|
1136
|
+
* this function cannot error, hence its contract must be respected.
|
1121
1137
|
*/
|
1122
1138
|
static void
|
1123
1139
|
ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
|
File without changes
|