Integrate dictBuilder into libzstd
This commit is contained in:
Yann Collet
+3
-3
@@ -1,6 +1,6 @@
|
||||
# ################################################################
|
||||
# ZSTD library - Makefile
|
||||
# Copyright (C) Yann Collet 2015
|
||||
# Copyright (C) Yann Collet 2015-2016
|
||||
# All rights reserved.
|
||||
#
|
||||
# BSD license
|
||||
@@ -28,7 +28,6 @@
|
||||
#
|
||||
# You can contact the author at :
|
||||
# - ZSTD homepage : http://www.zstd.net
|
||||
# - ZSTD source repository : https://github.com/Cyan4973/zstd
|
||||
# ################################################################
|
||||
|
||||
# Version numbers
|
||||
@@ -52,7 +51,7 @@ FLAGS = $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) $(MOREFLAGS)
|
||||
LIBDIR ?= $(PREFIX)/lib
|
||||
INCLUDEDIR=$(PREFIX)/include
|
||||
|
||||
ZSTD_FILES := zstd_compress.c zstd_decompress.c fse.c huff0.c
|
||||
ZSTD_FILES := zstd_compress.c zstd_decompress.c fse.c huff0.c dictBuilder.c divsufsort.c
|
||||
ZSTD_LEGACY:= legacy/zstd_v01.c legacy/zstd_v02.c legacy/zstd_v03.c legacy/zstd_v04.c
|
||||
|
||||
ifeq ($(ZSTD_LEGACY_SUPPORT), 0)
|
||||
@@ -119,6 +118,7 @@ install: libzstd libzstd.pc
|
||||
@cp -a libzstd.pc $(DESTDIR)$(LIBDIR)/pkgconfig/
|
||||
@install -m 644 libzstd.a $(DESTDIR)$(LIBDIR)/libzstd.a
|
||||
@install -m 644 zstd.h $(DESTDIR)$(INCLUDEDIR)/zstd.h
|
||||
@install -m 644 zstd.h $(DESTDIR)$(INCLUDEDIR)/zstd_buffered.h
|
||||
@echo zstd static and shared library installed
|
||||
|
||||
uninstall:
|
||||
|
||||
@@ -0,0 +1,902 @@
|
||||
/*
|
||||
dictBuilder - dictionary builder for zstd
|
||||
Copyright (C) Yann Collet 2016
|
||||
|
||||
BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
|
||||
|
||||
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.
|
||||
|
||||
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.
|
||||
|
||||
You can contact the author at :
|
||||
- Zstd source repository : https://www.zstd.net
|
||||
*/
|
||||
|
||||
/*-**************************************
|
||||
* Compiler Options
|
||||
****************************************/
|
||||
/* Disable some Visual warning messages */
|
||||
#ifdef _MSC_VER
|
||||
# define _CRT_SECURE_NO_WARNINGS /* fopen */
|
||||
# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
|
||||
#endif
|
||||
|
||||
/* Unix Large Files support (>4GB) */
|
||||
#define _FILE_OFFSET_BITS 64
|
||||
#if (defined(__sun__) && (!defined(__LP64__))) /* Sun Solaris 32-bits requires specific definitions */
|
||||
# define _LARGEFILE_SOURCE
|
||||
#elif ! defined(__LP64__) /* No point defining Large file for 64 bit */
|
||||
# define _LARGEFILE64_SOURCE
|
||||
#endif
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Dependencies
|
||||
***************************************/
|
||||
#include <stdlib.h> /* malloc, free */
|
||||
#include <string.h> /* memset */
|
||||
#include <stdio.h> /* fprintf, fopen, ftello64 */
|
||||
#include <sys/types.h> /* stat64 */
|
||||
#include <sys/stat.h> /* stat64 */
|
||||
#include <time.h> /* clock */
|
||||
|
||||
#include "mem.h" /* read */
|
||||
#include "error_private.h"
|
||||
#include "divsufsort.h"
|
||||
#include "dictBuilder_static.h"
|
||||
#include "fse.h"
|
||||
#include "huff0_static.h"
|
||||
#include "zstd_internal.h"
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Compiler specifics
|
||||
***************************************/
|
||||
#if !defined(S_ISREG)
|
||||
# define S_ISREG(x) (((x) & S_IFMT) == S_IFREG)
|
||||
#endif
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Constants
|
||||
***************************************/
|
||||
#define KB *(1 <<10)
|
||||
#define MB *(1 <<20)
|
||||
#define GB *(1U<<30)
|
||||
|
||||
#define DICTLISTSIZE 10000
|
||||
#define MEMMULT 11
|
||||
static const size_t maxMemory = (sizeof(size_t) == 4) ? (2 GB - 64 MB) : ((size_t)(512 MB) << sizeof(size_t));
|
||||
|
||||
#define NOISELENGTH 32
|
||||
#define PRIME1 2654435761U
|
||||
#define PRIME2 2246822519U
|
||||
|
||||
#define MINRATIO 4
|
||||
static const U32 g_compressionLevel_default = 5;
|
||||
static const U32 g_selectivity_default = 9;
|
||||
static const size_t g_provision_entropySize = 200;
|
||||
static const size_t g_min_fast_dictContent = 192;
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Console display
|
||||
***************************************/
|
||||
#define DISPLAY(...) fprintf(stderr, __VA_ARGS__)
|
||||
#define DISPLAYLEVEL(l, ...) if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); }
|
||||
static unsigned g_displayLevel = 0; /* 0 : no display; 1: errors; 2: default; 4: full information */
|
||||
void ZDICT_setNotificationLevel(unsigned l) { g_displayLevel=l; }
|
||||
|
||||
#define DISPLAYUPDATE(l, ...) if (g_displayLevel>=l) { \
|
||||
if (ZDICT_GetMilliSpan(g_time) > refreshRate) \
|
||||
{ g_time = clock(); DISPLAY(__VA_ARGS__); \
|
||||
if (g_displayLevel>=4) fflush(stdout); } }
|
||||
static const unsigned refreshRate = 300;
|
||||
static clock_t g_time = 0;
|
||||
|
||||
void ZDICT_printHex(U32 dlevel, const void* ptr, size_t length)
|
||||
{
|
||||
const BYTE* const b = (const BYTE*)ptr;
|
||||
size_t u;
|
||||
for (u=0; u<length; u++)
|
||||
{
|
||||
BYTE c = b[u];
|
||||
if (c<32 || c>126) c = '.'; /* non-printable char */
|
||||
DISPLAYLEVEL(dlevel, "%c", c);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Exceptions
|
||||
***************************************/
|
||||
#ifndef DEBUG
|
||||
# define DEBUG 0
|
||||
#endif
|
||||
#define DEBUGOUTPUT(...) if (DEBUG) DISPLAY(__VA_ARGS__);
|
||||
#define EXM_THROW(error, ...) \
|
||||
{ \
|
||||
DEBUGOUTPUT("Error defined at %s, line %i : \n", __FILE__, __LINE__); \
|
||||
DISPLAYLEVEL(1, "Error %i : ", error); \
|
||||
DISPLAYLEVEL(1, __VA_ARGS__); \
|
||||
DISPLAYLEVEL(1, "\n"); \
|
||||
exit(error); \
|
||||
}
|
||||
|
||||
|
||||
/*-********************************************************
|
||||
* Helper functions
|
||||
**********************************************************/
|
||||
static unsigned ZDICT_GetMilliSpan(clock_t nPrevious)
|
||||
{
|
||||
clock_t nCurrent = clock();
|
||||
unsigned nSpan = (unsigned)(((nCurrent - nPrevious) * 1000) / CLOCKS_PER_SEC);
|
||||
return nSpan;
|
||||
}
|
||||
|
||||
unsigned ZDICT_isError(size_t errorCode) { return ERR_isError(errorCode); }
|
||||
|
||||
const char* ZDICT_getErrorName(size_t errorCode) { return ERR_getErrorName(errorCode); }
|
||||
|
||||
|
||||
/*-********************************************************
|
||||
* Dictionary training functions
|
||||
**********************************************************/
|
||||
static unsigned ZDICT_NbCommonBytes (register size_t val)
|
||||
{
|
||||
if (MEM_isLittleEndian()) {
|
||||
if (MEM_64bits()) {
|
||||
# if defined(_MSC_VER) && defined(_WIN64)
|
||||
unsigned long r = 0;
|
||||
_BitScanForward64( &r, (U64)val );
|
||||
return (unsigned)(r>>3);
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3)
|
||||
return (__builtin_ctzll((U64)val) >> 3);
|
||||
# else
|
||||
static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2, 0, 3, 1, 3, 1, 4, 2, 7, 0, 2, 3, 6, 1, 5, 3, 5, 1, 3, 4, 4, 2, 5, 6, 7, 7, 0, 1, 2, 3, 3, 4, 6, 2, 6, 5, 5, 3, 4, 5, 6, 7, 1, 2, 4, 6, 4, 4, 5, 7, 2, 6, 5, 7, 6, 7, 7 };
|
||||
return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
|
||||
# endif
|
||||
} else { /* 32 bits */
|
||||
# if defined(_MSC_VER)
|
||||
unsigned long r=0;
|
||||
_BitScanForward( &r, (U32)val );
|
||||
return (unsigned)(r>>3);
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3)
|
||||
return (__builtin_ctz((U32)val) >> 3);
|
||||
# else
|
||||
static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0, 3, 2, 2, 1, 3, 2, 0, 1, 3, 3, 1, 2, 2, 2, 2, 0, 3, 1, 2, 0, 1, 0, 1, 1 };
|
||||
return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27];
|
||||
# endif
|
||||
}
|
||||
} else { /* Big Endian CPU */
|
||||
if (MEM_64bits()) {
|
||||
# if defined(_MSC_VER) && defined(_WIN64)
|
||||
unsigned long r = 0;
|
||||
_BitScanReverse64( &r, val );
|
||||
return (unsigned)(r>>3);
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3)
|
||||
return (__builtin_clzll(val) >> 3);
|
||||
# else
|
||||
unsigned r;
|
||||
const unsigned n32 = sizeof(size_t)*4; /* calculate this way due to compiler complaining in 32-bits mode */
|
||||
if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; }
|
||||
if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
|
||||
r += (!val);
|
||||
return r;
|
||||
# endif
|
||||
} else { /* 32 bits */
|
||||
# if defined(_MSC_VER)
|
||||
unsigned long r = 0;
|
||||
_BitScanReverse( &r, (unsigned long)val );
|
||||
return (unsigned)(r>>3);
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3)
|
||||
return (__builtin_clz((U32)val) >> 3);
|
||||
# else
|
||||
unsigned r;
|
||||
if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
|
||||
r += (!val);
|
||||
return r;
|
||||
# endif
|
||||
} }
|
||||
}
|
||||
|
||||
|
||||
/*! ZDICT_count() :
|
||||
Count the nb of common bytes between 2 pointers.
|
||||
Note : this function presumes end of buffer followed by noisy guard band.
|
||||
*/
|
||||
static size_t ZDICT_count(const void* pIn, const void* pMatch)
|
||||
{
|
||||
const char* const pStart = (const char*)pIn;
|
||||
for (;;) {
|
||||
size_t diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
|
||||
if (!diff) { pIn = (const char*)pIn+sizeof(size_t); pMatch = (const char*)pMatch+sizeof(size_t); continue; }
|
||||
pIn = (const char*)pIn+ZDICT_NbCommonBytes(diff);
|
||||
return (size_t)((const char*)pIn - pStart);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
typedef struct {
|
||||
U32 pos;
|
||||
U32 length;
|
||||
U32 savings;
|
||||
} dictItem;
|
||||
|
||||
static void ZDICT_initDictItem(dictItem* d)
|
||||
{
|
||||
d->pos = 1;
|
||||
d->length = 0;
|
||||
d->savings = (U32)(-1);
|
||||
}
|
||||
|
||||
|
||||
#define LLIMIT 64 /* heuristic determined experimentally */
|
||||
#define MINMATCHLENGTH 7 /* heuristic determined experimentally */
|
||||
static dictItem ZDICT_analyzePos(
|
||||
BYTE* doneMarks,
|
||||
const int* suffix, U32 start,
|
||||
const void* buffer, U32 minRatio)
|
||||
{
|
||||
U32 lengthList[LLIMIT] = {0};
|
||||
U32 cumulLength[LLIMIT] = {0};
|
||||
U32 savings[LLIMIT] = {0};
|
||||
const BYTE* b = (const BYTE*)buffer;
|
||||
size_t length;
|
||||
size_t maxLength = LLIMIT;
|
||||
size_t pos = suffix[start];
|
||||
U32 end = start;
|
||||
dictItem solution;
|
||||
|
||||
/* init */
|
||||
memset(&solution, 0, sizeof(solution));
|
||||
doneMarks[pos] = 1;
|
||||
|
||||
/* trivial repetition cases */
|
||||
if ( (MEM_read16(b+pos+0) == MEM_read16(b+pos+2))
|
||||
||(MEM_read16(b+pos+1) == MEM_read16(b+pos+3))
|
||||
||(MEM_read16(b+pos+2) == MEM_read16(b+pos+4)) ) {
|
||||
/* skip and mark segment */
|
||||
U16 u16 = MEM_read16(b+pos+4);
|
||||
U32 u, e = 6;
|
||||
while (MEM_read16(b+pos+e) == u16) e+=2 ;
|
||||
if (b[pos+e] == b[pos+e-1]) e++;
|
||||
for (u=1; u<e; u++)
|
||||
doneMarks[pos+u] = 1;
|
||||
return solution;
|
||||
}
|
||||
|
||||
/* look forward */
|
||||
do {
|
||||
end++;
|
||||
length = ZDICT_count(b + pos, b + suffix[end]);
|
||||
} while (length >=MINMATCHLENGTH);
|
||||
|
||||
/* look backward */
|
||||
do {
|
||||
length = ZDICT_count(b + pos, b + *(suffix+start-1));
|
||||
if (length >=MINMATCHLENGTH) start--;
|
||||
} while(length >= MINMATCHLENGTH);
|
||||
|
||||
/* exit if not found a minimum nb of repetitions */
|
||||
if (end-start < minRatio) {
|
||||
U32 idx;
|
||||
for(idx=start; idx<end; idx++)
|
||||
doneMarks[suffix[idx]] = 1;
|
||||
return solution;
|
||||
}
|
||||
|
||||
{
|
||||
int i;
|
||||
U32 searchLength;
|
||||
U32 refinedStart = start;
|
||||
U32 refinedEnd = end;
|
||||
|
||||
DISPLAYLEVEL(4, "\n");
|
||||
DISPLAYLEVEL(4, "found %3u matches of length >= %u at pos %7u ", (U32)(end-start), MINMATCHLENGTH, (U32)pos);
|
||||
DISPLAYLEVEL(4, "\n");
|
||||
|
||||
for (searchLength = MINMATCHLENGTH ; ; searchLength++) {
|
||||
BYTE currentChar = 0;
|
||||
U32 currentCount = 0;
|
||||
U32 currentID = refinedStart;
|
||||
U32 id;
|
||||
U32 selectedCount = 0;
|
||||
U32 selectedID = currentID;
|
||||
for (id =refinedStart; id < refinedEnd; id++) {
|
||||
if (b[ suffix[id] + searchLength] != currentChar) {
|
||||
if (currentCount > selectedCount) {
|
||||
selectedCount = currentCount;
|
||||
selectedID = currentID;
|
||||
}
|
||||
currentID = id;
|
||||
currentChar = b[ suffix[id] + searchLength];
|
||||
currentCount = 0;
|
||||
}
|
||||
currentCount ++;
|
||||
}
|
||||
if (currentCount > selectedCount) { /* for last */
|
||||
selectedCount = currentCount;
|
||||
selectedID = currentID;
|
||||
}
|
||||
|
||||
if (selectedCount < minRatio)
|
||||
break;
|
||||
refinedStart = selectedID;
|
||||
refinedEnd = refinedStart + selectedCount;
|
||||
}
|
||||
|
||||
/* evaluate gain based on new ref */
|
||||
start = refinedStart;
|
||||
pos = suffix[refinedStart];
|
||||
end = start;
|
||||
memset(lengthList, 0, sizeof(lengthList));
|
||||
|
||||
/* look forward */
|
||||
do {
|
||||
end++;
|
||||
length = ZDICT_count(b + pos, b + suffix[end]);
|
||||
if (length >= LLIMIT) length = LLIMIT-1;
|
||||
lengthList[length]++;
|
||||
} while (length >=MINMATCHLENGTH);
|
||||
|
||||
/* look backward */
|
||||
do {
|
||||
length = ZDICT_count(b + pos, b + suffix[start-1]);
|
||||
if (length >= LLIMIT) length = LLIMIT-1;
|
||||
lengthList[length]++;
|
||||
if (length >=MINMATCHLENGTH) start--;
|
||||
} while(length >= MINMATCHLENGTH);
|
||||
|
||||
/* largest useful length */
|
||||
memset(cumulLength, 0, sizeof(cumulLength));
|
||||
cumulLength[maxLength-1] = lengthList[maxLength-1];
|
||||
for (i=(int)(maxLength-2); i>=0; i--)
|
||||
cumulLength[i] = cumulLength[i+1] + lengthList[i];
|
||||
|
||||
for (i=LLIMIT-1; i>=MINMATCHLENGTH; i--) if (cumulLength[i]>=minRatio) break;
|
||||
maxLength = i;
|
||||
|
||||
/* reduce maxLength in case of final into repetitive data */
|
||||
{
|
||||
U32 l = (U32)maxLength;
|
||||
BYTE c = b[pos + maxLength-1];
|
||||
while (b[pos+l-2]==c) l--;
|
||||
maxLength = l;
|
||||
}
|
||||
if (maxLength < MINMATCHLENGTH) return solution; /* skip : no long-enough solution */
|
||||
|
||||
/* calculate savings */
|
||||
savings[5] = 0;
|
||||
for (i=MINMATCHLENGTH; i<=(int)maxLength; i++)
|
||||
savings[i] = savings[i-1] + (lengthList[i] * (i-3));
|
||||
|
||||
DISPLAYLEVEL(4, "Selected ref at position %u, of length %u : saves %u (ratio: %.2f) \n",
|
||||
(U32)pos, (U32)maxLength, savings[maxLength], (double)savings[maxLength] / maxLength);
|
||||
|
||||
solution.pos = (U32)pos;
|
||||
solution.length = (U32)maxLength;
|
||||
solution.savings = savings[maxLength];
|
||||
|
||||
/* mark positions done */
|
||||
{
|
||||
U32 id;
|
||||
U32 testedPos;
|
||||
for (id=start; id<end; id++) {
|
||||
U32 p, pEnd;
|
||||
testedPos = suffix[id];
|
||||
if (testedPos == pos)
|
||||
length = solution.length;
|
||||
else {
|
||||
length = ZDICT_count(b+pos, b+testedPos);
|
||||
if (length > solution.length) length = solution.length;
|
||||
}
|
||||
pEnd = (U32)(testedPos + length);
|
||||
for (p=testedPos; p<pEnd; p++)
|
||||
doneMarks[p] = 1;
|
||||
} } }
|
||||
|
||||
return solution;
|
||||
}
|
||||
|
||||
|
||||
/*! ZDICT_checkMerge
|
||||
check if dictItem can be merged, do it if possible
|
||||
@return : id of destination elt, 0 if not merged
|
||||
*/
|
||||
static U32 ZDICT_checkMerge(dictItem* table, dictItem elt, U32 eltNbToSkip)
|
||||
{
|
||||
const U32 tableSize = table->pos;
|
||||
const U32 max = elt.pos + (elt.length-1);
|
||||
|
||||
/* tail overlap */
|
||||
U32 u; for (u=1; u<tableSize; u++) {
|
||||
if (u==eltNbToSkip) continue;
|
||||
if ((table[u].pos > elt.pos) && (table[u].pos < max)) { /* overlap */
|
||||
/* append */
|
||||
U32 addedLength = table[u].pos - elt.pos;
|
||||
table[u].length += addedLength;
|
||||
table[u].pos = elt.pos;
|
||||
table[u].savings += elt.savings * addedLength / elt.length; /* rough approx */
|
||||
table[u].savings += elt.length / 8; /* rough approx */
|
||||
elt = table[u];
|
||||
while ((u>1) && (table[u-1].savings < elt.savings))
|
||||
table[u] = table[u-1], u--;
|
||||
table[u] = elt;
|
||||
return u;
|
||||
} }
|
||||
|
||||
/* front overlap */
|
||||
for (u=1; u<tableSize; u++) {
|
||||
if (u==eltNbToSkip) continue;
|
||||
if ((table[u].pos + table[u].length > elt.pos) && (table[u].pos < elt.pos)) { /* overlap */
|
||||
/* append */
|
||||
int addedLength = (elt.pos + elt.length) - (table[u].pos + table[u].length);
|
||||
table[u].savings += elt.length / 8; /* rough approx */
|
||||
if (addedLength > 0) { /* otherwise, already included */
|
||||
table[u].length += addedLength;
|
||||
table[u].savings += elt.savings * addedLength / elt.length; /* rough approx */
|
||||
}
|
||||
elt = table[u];
|
||||
while ((u>1) && (table[u-1].savings < elt.savings))
|
||||
table[u] = table[u-1], u--;
|
||||
table[u] = elt;
|
||||
return u;
|
||||
} }
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
|
||||
static void ZDICT_removeDictItem(dictItem* table, U32 id)
|
||||
{
|
||||
/* convention : first element is nb of elts */
|
||||
U32 max = table->pos;
|
||||
U32 u;
|
||||
if (!id) return; /* protection, should never happen */
|
||||
for (u=id; u<max-1; u++)
|
||||
table[u] = table[u+1];
|
||||
table->pos--;
|
||||
}
|
||||
|
||||
|
||||
static void ZDICT_insertDictItem(dictItem* table, U32 maxSize, dictItem elt)
|
||||
{
|
||||
/* merge if possible */
|
||||
U32 mergeId = ZDICT_checkMerge(table, elt, 0);
|
||||
if (mergeId) {
|
||||
U32 newMerge = 1;
|
||||
while (newMerge) {
|
||||
newMerge = ZDICT_checkMerge(table, table[mergeId], mergeId);
|
||||
if (newMerge) ZDICT_removeDictItem(table, mergeId);
|
||||
mergeId = newMerge;
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
||||
/* insert */
|
||||
{
|
||||
U32 current;
|
||||
U32 nextElt = table->pos;
|
||||
if (nextElt >= maxSize) nextElt = maxSize-1;
|
||||
current = nextElt-1;
|
||||
while (table[current].savings < elt.savings) {
|
||||
table[current+1] = table[current];
|
||||
current--;
|
||||
}
|
||||
table[current+1] = elt;
|
||||
table->pos = nextElt+1;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
static U32 ZDICT_dictSize(const dictItem* dictList)
|
||||
{
|
||||
U32 u, dictSize = 0;
|
||||
for (u=1; u<dictList[0].pos; u++)
|
||||
dictSize += dictList[u].length;
|
||||
return dictSize;
|
||||
}
|
||||
|
||||
|
||||
static void ZDICT_trainBuffer(dictItem* dictList, U32 dictListSize,
|
||||
const void* const buffer, const size_t bufferSize, /* buffer must end with noisy guard band */
|
||||
const size_t* fileSizes, unsigned nbFiles,
|
||||
U32 shiftRatio, unsigned maxDictSize)
|
||||
{
|
||||
int* const suffix0 = (int*)malloc((bufferSize+2)*sizeof(*suffix0));
|
||||
int* const suffix = suffix0+1;
|
||||
U32* reverseSuffix = (U32*)malloc((bufferSize)*sizeof(*reverseSuffix));
|
||||
BYTE* doneMarks = (BYTE*)malloc((bufferSize+16)*sizeof(*doneMarks)); /* +16 for overflow security */
|
||||
U32* filePos = (U32*)malloc(nbFiles * sizeof(*filePos));
|
||||
U32 minRatio = nbFiles >> shiftRatio;
|
||||
int divSuftSortResult;
|
||||
|
||||
/* init */
|
||||
DISPLAYLEVEL(2, "\r%70s\r", ""); /* clean display line */
|
||||
if (!suffix0 || !reverseSuffix || !doneMarks || !filePos)
|
||||
EXM_THROW(1, "not enough memory for ZDICT_trainBuffer");
|
||||
if (minRatio < MINRATIO) minRatio = MINRATIO;
|
||||
memset(doneMarks, 0, bufferSize+16);
|
||||
|
||||
/* sort */
|
||||
DISPLAYLEVEL(2, "sorting %u files of total size %u MB ...\n", nbFiles, (U32)(bufferSize>>20));
|
||||
divSuftSortResult = divsufsort((const unsigned char*)buffer, suffix, (int)bufferSize, 0);
|
||||
if (divSuftSortResult != 0) EXM_THROW(2, "sort failed");
|
||||
suffix[bufferSize] = (int)bufferSize; /* leads into noise */
|
||||
suffix0[0] = (int)bufferSize; /* leads into noise */
|
||||
{
|
||||
/* build reverse suffix sort */
|
||||
size_t pos;
|
||||
for (pos=0; pos < bufferSize; pos++)
|
||||
reverseSuffix[suffix[pos]] = (U32)pos;
|
||||
/* build file pos */
|
||||
filePos[0] = 0;
|
||||
for (pos=1; pos<nbFiles; pos++)
|
||||
filePos[pos] = (U32)(filePos[pos-1] + fileSizes[pos-1]);
|
||||
}
|
||||
|
||||
DISPLAYLEVEL(2, "finding patterns ... \n");
|
||||
DISPLAYLEVEL(3, "minimum ratio : %u \n", minRatio);
|
||||
|
||||
{
|
||||
U32 cursor; for (cursor=0; cursor < bufferSize; ) {
|
||||
dictItem solution;
|
||||
if (doneMarks[cursor]) { cursor++; continue; }
|
||||
solution = ZDICT_analyzePos(doneMarks, suffix, reverseSuffix[cursor], buffer, minRatio);
|
||||
if (solution.length==0) { cursor++; continue; }
|
||||
ZDICT_insertDictItem(dictList, dictListSize, solution);
|
||||
cursor += solution.length;
|
||||
DISPLAYUPDATE(2, "\r%4.2f %% \r", (double)cursor / bufferSize * 100);
|
||||
} }
|
||||
|
||||
/* limit dictionary size */
|
||||
{
|
||||
U32 max = dictList->pos; /* convention : nb of useful elts within dictList */
|
||||
U32 currentSize = 0;
|
||||
U32 n; for (n=1; n<max; n++) {
|
||||
currentSize += dictList[n].length;
|
||||
if (currentSize > maxDictSize) break;
|
||||
}
|
||||
dictList->pos = n;
|
||||
}
|
||||
|
||||
free(suffix0);
|
||||
free(reverseSuffix);
|
||||
free(doneMarks);
|
||||
free(filePos);
|
||||
}
|
||||
|
||||
|
||||
static void ZDICT_fillNoise(void* buffer, size_t length)
|
||||
{
|
||||
unsigned acc = PRIME1;
|
||||
size_t p=0;;
|
||||
|
||||
for (p=0; p<length; p++) {
|
||||
acc *= PRIME2;
|
||||
((unsigned char*)buffer)[p] = (unsigned char)(acc >> 21);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
typedef struct
|
||||
{
|
||||
ZSTD_CCtx* ref;
|
||||
ZSTD_CCtx* zc;
|
||||
void* workPlace; /* must be BLOCKSIZE allocated */
|
||||
} EStats_ress_t;
|
||||
|
||||
|
||||
static void ZDICT_countEStats(EStats_ress_t esr,
|
||||
U32* countLit, U32* offsetcodeCount, U32* matchlengthCount, U32* litlengthCount,
|
||||
const void* src, size_t srcSize)
|
||||
{
|
||||
const BYTE* bytePtr;
|
||||
const U32* u32Ptr;
|
||||
seqStore_t seqStore;
|
||||
|
||||
if (srcSize > BLOCKSIZE) srcSize = BLOCKSIZE; /* protection vs large samples */
|
||||
ZSTD_copyCCtx(esr.zc, esr.ref);
|
||||
ZSTD_compressBlock(esr.zc, esr.workPlace, BLOCKSIZE, src, srcSize);
|
||||
seqStore = ZSTD_copySeqStore(esr.zc);
|
||||
|
||||
/* count stats */
|
||||
for(bytePtr = seqStore.litStart; bytePtr < seqStore.lit; bytePtr++)
|
||||
countLit[*bytePtr]++;
|
||||
for(u32Ptr = seqStore.offsetStart; u32Ptr < seqStore.offset; u32Ptr++) {
|
||||
BYTE offcode = (BYTE)ZSTD_highbit(*u32Ptr) + 1;
|
||||
if (*u32Ptr==0) offcode=0;
|
||||
offsetcodeCount[offcode]++;
|
||||
}
|
||||
for(bytePtr = seqStore.matchLengthStart; bytePtr < seqStore.matchLength; bytePtr++)
|
||||
matchlengthCount[*bytePtr]++;
|
||||
for(bytePtr = seqStore.litLengthStart; bytePtr < seqStore.litLength; bytePtr++)
|
||||
litlengthCount[*bytePtr]++;
|
||||
}
|
||||
|
||||
|
||||
#define OFFCODE_MAX 18 /* only applicable to first block */
|
||||
static size_t ZDICT_analyzeEntropy(void* dstBuffer, size_t maxDstSize,
|
||||
unsigned compressionLevel,
|
||||
const void* srcBuffer, const size_t* fileSizes, unsigned nbFiles,
|
||||
const void* dictBuffer, size_t dictBufferSize)
|
||||
{
|
||||
U32 countLit[256];
|
||||
U32 offcodeCount[MaxOff+1];
|
||||
HUF_CREATE_STATIC_CTABLE(hufTable, 255);
|
||||
short offcodeNCount[MaxOff+1];
|
||||
U32 matchLengthCount[MaxML+1];
|
||||
short matchLengthNCount[MaxML+1];
|
||||
U32 litlengthCount[MaxLL+1];
|
||||
short litlengthNCount[MaxLL+1];
|
||||
EStats_ress_t esr;
|
||||
ZSTD_parameters params;
|
||||
U32 u, huffLog = 12, Offlog = OffFSELog, mlLog = MLFSELog, llLog = LLFSELog, total;
|
||||
size_t pos = 0, errorCode;
|
||||
size_t eSize = 0;
|
||||
|
||||
/* init */
|
||||
for (u=0; u<256; u++) countLit[u]=1; /* any character must be described */
|
||||
for (u=0; u<=OFFCODE_MAX; u++) offcodeCount[u]=1;
|
||||
for (u=0; u<=MaxML; u++) matchLengthCount[u]=1;
|
||||
for (u=0; u<=MaxLL; u++) litlengthCount[u]=1;
|
||||
esr.ref = ZSTD_createCCtx();
|
||||
esr.zc = ZSTD_createCCtx();
|
||||
esr.workPlace = malloc(BLOCKSIZE);
|
||||
if (!esr.ref || !esr.zc || !esr.workPlace) EXM_THROW(30, "Not enough memory");
|
||||
if (compressionLevel==0) compressionLevel=g_compressionLevel_default;
|
||||
params = ZSTD_getParams(compressionLevel, dictBufferSize + 15 KB);
|
||||
params.strategy = ZSTD_greedy;
|
||||
ZSTD_compressBegin_advanced(esr.ref, dictBuffer, dictBufferSize, params);
|
||||
|
||||
/* collect stats on all files */
|
||||
for (u=0; u<nbFiles; u++) {
|
||||
ZDICT_countEStats(esr,
|
||||
countLit, offcodeCount, matchLengthCount, litlengthCount,
|
||||
(const char*)srcBuffer + pos, fileSizes[u]);
|
||||
pos += fileSizes[u];
|
||||
}
|
||||
|
||||
/* analyze */
|
||||
errorCode = HUF_buildCTable (hufTable, countLit, 255, huffLog);
|
||||
if (HUF_isError(errorCode)) EXM_THROW(31, "HUF_buildCTable error");
|
||||
huffLog = (U32)errorCode;
|
||||
|
||||
total=0; for (u=0; u<=OFFCODE_MAX; u++) total+=offcodeCount[u];
|
||||
errorCode = FSE_normalizeCount(offcodeNCount, Offlog, offcodeCount, total, OFFCODE_MAX);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(32, "FSE_normalizeCount error with offcodeCount");
|
||||
Offlog = (U32)errorCode;
|
||||
|
||||
total=0; for (u=0; u<=MaxML; u++) total+=matchLengthCount[u];
|
||||
errorCode = FSE_normalizeCount(matchLengthNCount, mlLog, matchLengthCount, total, MaxML);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(33, "FSE_normalizeCount error with matchLengthCount");
|
||||
mlLog = (U32)errorCode;
|
||||
|
||||
total=0; for (u=0; u<=MaxLL; u++) total+=litlengthCount[u];
|
||||
errorCode = FSE_normalizeCount(litlengthNCount, llLog, litlengthCount, total, MaxLL);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(34, "FSE_normalizeCount error with litlengthCount");
|
||||
llLog = (U32)errorCode;
|
||||
|
||||
/* write result to buffer */
|
||||
errorCode = HUF_writeCTable(dstBuffer, maxDstSize, hufTable, 255, huffLog);
|
||||
if (HUF_isError(errorCode)) EXM_THROW(41, "HUF_writeCTable error");
|
||||
dstBuffer = (char*)dstBuffer + errorCode;
|
||||
maxDstSize -= errorCode;
|
||||
eSize += errorCode;
|
||||
|
||||
errorCode = FSE_writeNCount(dstBuffer, maxDstSize, offcodeNCount, OFFCODE_MAX, Offlog);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(42, "FSE_writeNCount error with offcodeNCount");
|
||||
dstBuffer = (char*)dstBuffer + errorCode;
|
||||
maxDstSize -= errorCode;
|
||||
eSize += errorCode;
|
||||
|
||||
errorCode = FSE_writeNCount(dstBuffer, maxDstSize, matchLengthNCount, MaxML, mlLog);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(43, "FSE_writeNCount error with matchLengthNCount");
|
||||
dstBuffer = (char*)dstBuffer + errorCode;
|
||||
maxDstSize -= errorCode;
|
||||
eSize += errorCode;
|
||||
|
||||
errorCode = FSE_writeNCount(dstBuffer, maxDstSize, litlengthNCount, MaxLL, llLog);
|
||||
if (FSE_isError(errorCode)) EXM_THROW(43, "FSE_writeNCount error with litlengthNCount");
|
||||
dstBuffer = (char*)dstBuffer + errorCode;
|
||||
maxDstSize -= errorCode;
|
||||
eSize += errorCode;
|
||||
|
||||
/* clean */
|
||||
ZSTD_freeCCtx(esr.ref);
|
||||
ZSTD_freeCCtx(esr.zc);
|
||||
free(esr.workPlace);
|
||||
|
||||
return eSize;
|
||||
}
|
||||
|
||||
|
||||
#define DIB_FASTSEGMENTSIZE 64
|
||||
/*! ZDICT_fastSampling (based on an idea by Giuseppe Ottaviano)
|
||||
Fill `dictBuffer` with stripes of size DIB_FASTSEGMENTSIZE from `samplesBuffer`
|
||||
up to `dictSize`.
|
||||
Filling starts from the end of `dictBuffer`, down to maximum possible.
|
||||
if `dictSize` is not a multiply of DIB_FASTSEGMENTSIZE, some bytes at beginning of `dictBuffer` won't be used.
|
||||
@return : amount of data written into `dictBuffer`
|
||||
or an error code
|
||||
*/
|
||||
static size_t ZDICT_fastSampling(void* dictBuffer, size_t dictSize,
|
||||
const void* samplesBuffer, size_t samplesSize)
|
||||
{
|
||||
char* dstPtr = (char*)dictBuffer + dictSize;
|
||||
const char* srcPtr = (const char*)samplesBuffer;
|
||||
size_t nbSegments = dictSize / DIB_FASTSEGMENTSIZE;
|
||||
size_t segNb, interSize;
|
||||
|
||||
if (nbSegments <= 2) return ERROR(srcSize_wrong);
|
||||
if (samplesSize < dictSize) return ERROR(srcSize_wrong);
|
||||
|
||||
/* first and last segments are part of dictionary, in case they contain interesting header/footer */
|
||||
dstPtr -= DIB_FASTSEGMENTSIZE;
|
||||
memcpy(dstPtr, srcPtr, DIB_FASTSEGMENTSIZE);
|
||||
dstPtr -= DIB_FASTSEGMENTSIZE;
|
||||
memcpy(dstPtr, srcPtr+samplesSize-DIB_FASTSEGMENTSIZE, DIB_FASTSEGMENTSIZE);
|
||||
|
||||
/* regularly copy a segment */
|
||||
interSize = (samplesSize - nbSegments*DIB_FASTSEGMENTSIZE) / (nbSegments-1);
|
||||
srcPtr += DIB_FASTSEGMENTSIZE;
|
||||
for (segNb=2; segNb < nbSegments; segNb++) {
|
||||
srcPtr += interSize;
|
||||
dstPtr -= DIB_FASTSEGMENTSIZE;
|
||||
memcpy(dstPtr, srcPtr, DIB_FASTSEGMENTSIZE);
|
||||
srcPtr += DIB_FASTSEGMENTSIZE;
|
||||
}
|
||||
|
||||
return nbSegments * DIB_FASTSEGMENTSIZE;
|
||||
}
|
||||
|
||||
|
||||
size_t ZDICT_trainFromBuffer_unsafe(
|
||||
void* dictBuffer, size_t maxDictSize,
|
||||
const void* samplesBuffer, const size_t* sampleSizes, unsigned nbSamples,
|
||||
ZDICT_params_t params)
|
||||
{
|
||||
const U32 dictListSize = MAX( MAX(DICTLISTSIZE, nbSamples), (U32)(maxDictSize/16));
|
||||
dictItem* dictList = (dictItem*)malloc(dictListSize * sizeof(*dictList));
|
||||
unsigned selectivity = params.selectivityLevel;
|
||||
unsigned compressionLevel = params.compressionLevel;
|
||||
size_t targetDictSize = maxDictSize - g_provision_entropySize;
|
||||
size_t sBuffSize;
|
||||
size_t dictSize = 0;
|
||||
|
||||
/* checks */
|
||||
if (maxDictSize <= g_provision_entropySize + g_min_fast_dictContent) return ERROR(dstSize_tooSmall);
|
||||
|
||||
/* init */
|
||||
{ unsigned u; for (u=0, sBuffSize=0; u<nbSamples; u++) sBuffSize += sampleSizes[u]; }
|
||||
if (!dictList) return ERROR(memory_allocation);
|
||||
ZDICT_initDictItem(dictList);
|
||||
if (selectivity==0) selectivity = g_selectivity_default;
|
||||
if (compressionLevel==0) compressionLevel = g_compressionLevel_default;
|
||||
|
||||
/* build dictionary */
|
||||
if (selectivity>1) { /* selectivity == 1 => fast mode */
|
||||
ZDICT_trainBuffer(dictList, dictListSize,
|
||||
samplesBuffer, sBuffSize,
|
||||
sampleSizes, nbSamples,
|
||||
selectivity, (U32)targetDictSize);
|
||||
|
||||
/* display best matches */
|
||||
if (g_displayLevel>= 3) {
|
||||
const U32 nb = 25;
|
||||
U32 u;
|
||||
U32 dictContentSize = ZDICT_dictSize(dictList);
|
||||
DISPLAYLEVEL(3, "\n %u segments found, of total size %u \n", dictList[0].pos, dictContentSize);
|
||||
DISPLAYLEVEL(3, "list %u best segments \n", nb);
|
||||
for (u=1; u<=nb; u++) {
|
||||
U32 p = dictList[u].pos;
|
||||
U32 l = dictList[u].length;
|
||||
U32 d = MIN(40, l);
|
||||
DISPLAYLEVEL(3, "%3u:%3u bytes at pos %8u, savings %7u bytes |",
|
||||
u, l, p, dictList[u].savings);
|
||||
ZDICT_printHex(3, (const char*)samplesBuffer+p, d);
|
||||
DISPLAYLEVEL(3, "| \n");
|
||||
} } }
|
||||
|
||||
/* create dictionary */
|
||||
{
|
||||
U32 dictContentSize = ZDICT_dictSize(dictList);
|
||||
size_t hSize;
|
||||
BYTE* ptr;
|
||||
U32 u;
|
||||
|
||||
/* build dict content */
|
||||
ptr = (BYTE*)dictBuffer + maxDictSize;
|
||||
for (u=1; u<dictList->pos; u++) {
|
||||
U32 l = dictList[u].length;
|
||||
ptr -= l;
|
||||
if (ptr<(BYTE*)dictBuffer) return ERROR(GENERIC); /* should not happen */
|
||||
memcpy(ptr, (const char*)samplesBuffer+dictList[u].pos, l);
|
||||
}
|
||||
|
||||
/* fast mode dict content */
|
||||
if (selectivity==1) { /* note could also be used to complete a dictionary, but not necessarily better */
|
||||
DISPLAYLEVEL(3, "\r%70s\r", ""); /* clean display line */
|
||||
DISPLAYLEVEL(3, "Adding %u KB with fast sampling \n", (U32)(targetDictSize>>10));
|
||||
dictContentSize = (U32)ZDICT_fastSampling((char*)dictBuffer + g_provision_entropySize,
|
||||
targetDictSize, samplesBuffer, sBuffSize);
|
||||
}
|
||||
|
||||
/* dictionary header */
|
||||
MEM_writeLE32(dictBuffer, ZSTD_DICT_MAGIC);
|
||||
hSize = 4;
|
||||
|
||||
/* entropic tables */
|
||||
DISPLAYLEVEL(2, "\r%70s\r", ""); /* clean display line */
|
||||
DISPLAYLEVEL(2, "statistics ... \n");
|
||||
hSize += ZDICT_analyzeEntropy((char*)dictBuffer+4, maxDictSize-4,
|
||||
compressionLevel,
|
||||
samplesBuffer, sampleSizes, nbSamples,
|
||||
(char*)dictBuffer + maxDictSize - dictContentSize, dictContentSize);
|
||||
|
||||
if (hSize + dictContentSize < maxDictSize)
|
||||
memmove((char*)dictBuffer + hSize, (char*)dictBuffer + maxDictSize - dictContentSize, dictContentSize);
|
||||
dictSize = MIN(maxDictSize, hSize+dictContentSize);
|
||||
}
|
||||
|
||||
/* clean up */
|
||||
free(dictList);
|
||||
return dictSize;
|
||||
}
|
||||
|
||||
|
||||
size_t ZDICT_trainFromBuffer_advanced(void* dictBuffer, size_t dictBufferCapacity,
|
||||
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
|
||||
ZDICT_params_t params)
|
||||
{
|
||||
size_t sBuffSize;
|
||||
void* newBuff;
|
||||
size_t result;
|
||||
|
||||
{ unsigned u; for (u=0, sBuffSize=0; u<nbSamples; u++) sBuffSize += samplesSizes[u]; }
|
||||
newBuff = malloc(sBuffSize + NOISELENGTH);
|
||||
if (!newBuff) return ERROR(memory_allocation);
|
||||
|
||||
memcpy(newBuff, samplesBuffer, sBuffSize);
|
||||
ZDICT_fillNoise((char*)newBuff + sBuffSize, NOISELENGTH); /* guard band, for end of buffer condition */
|
||||
|
||||
result = ZDICT_trainFromBuffer_unsafe(dictBuffer, dictBufferCapacity,
|
||||
newBuff, samplesSizes, nbSamples,
|
||||
params);
|
||||
free(newBuff);
|
||||
return result;
|
||||
}
|
||||
|
||||
|
||||
/* issue : samplesBuffer need to be followed by a noisy guard band.
|
||||
* work around : duplicate the buffer, and add the noise ? */
|
||||
size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
|
||||
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples)
|
||||
{
|
||||
ZDICT_params_t params;
|
||||
memset(¶ms, 0, sizeof(params));
|
||||
return ZDICT_trainFromBuffer_advanced(dictBuffer, dictBufferCapacity,
|
||||
samplesBuffer, samplesSizes, nbSamples,
|
||||
params);
|
||||
}
|
||||
|
||||
@@ -0,0 +1,67 @@
|
||||
/*
|
||||
dictBuilder header file
|
||||
Copyright (C) Yann Collet 2016
|
||||
|
||||
BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
|
||||
|
||||
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.
|
||||
|
||||
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.
|
||||
|
||||
You can contact the author at :
|
||||
- Zstd source repository : https://www.zstd.net
|
||||
*/
|
||||
|
||||
#ifndef DICTBUILDER_H_001
|
||||
#define DICTBUILDER_H_001
|
||||
|
||||
#if defined (__cplusplus)
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/*-*************************************
|
||||
* Public functions
|
||||
***************************************/
|
||||
/*! ZDICT_trainFromBuffer() :
|
||||
Train a dictionary from a memory buffer `samplesBuffer`,
|
||||
where `nbSamples` samples have been stored concatenated.
|
||||
Each sample size is provided into an orderly table `samplesSizes`.
|
||||
Resulting dictionary will be saved into `dictBuffer`.
|
||||
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
|
||||
or an error code, which can be tested by ZDICT_isError().
|
||||
*/
|
||||
size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
|
||||
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples);
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Helper functions
|
||||
***************************************/
|
||||
unsigned ZDICT_isError(size_t errorCode);
|
||||
const char* ZDICT_getErrorName(size_t errorCode);
|
||||
|
||||
|
||||
#if defined (__cplusplus)
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif
|
||||
@@ -0,0 +1,104 @@
|
||||
/*
|
||||
dictBuilder header file
|
||||
for static linking only
|
||||
Copyright (C) Yann Collet 2016
|
||||
|
||||
BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
|
||||
|
||||
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.
|
||||
|
||||
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.
|
||||
|
||||
You can contact the author at :
|
||||
- Zstd source repository : https://www.zstd.net
|
||||
*/
|
||||
|
||||
/* This library is EXPERIMENTAL, below API is not yet stable */
|
||||
|
||||
#ifndef DICTBUILDER_STATIC_H_002
|
||||
#define DICTBUILDER_STATIC_H_002
|
||||
|
||||
#if defined (__cplusplus)
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/*-*************************************
|
||||
* Dependencies
|
||||
***************************************/
|
||||
#include "dictBuilder.h"
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Public type
|
||||
***************************************/
|
||||
typedef struct {
|
||||
unsigned selectivityLevel; /* 0 means default; larger => bigger selection => larger dictionary */
|
||||
unsigned compressionLevel; /* 0 means default; target a specific zstd compression level */
|
||||
} ZDICT_params_t;
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Public functions
|
||||
***************************************/
|
||||
/*! ZDICT_trainFromBuffer_advanced() :
|
||||
Same as ZDICT_trainFromBuffer() with control over more parameters.
|
||||
`parameters` is optional and can be provided with values set to 0 to mean "default".
|
||||
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferSize`)
|
||||
or an error code, which can be tested by DiB_isError().
|
||||
note : ZDICT_trainFromBuffer_advanced() will send notifications into stderr if instructed to, using ZDICT_setNotificationLevel()
|
||||
*/
|
||||
size_t ZDICT_trainFromBuffer_advanced(void* dictBuffer, size_t dictBufferCapacity,
|
||||
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
|
||||
ZDICT_params_t parameters);
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Helper functions
|
||||
***************************************/
|
||||
/*! ZDICT_setNotificationLevel() :
|
||||
Set amount of notification to be displayed on the console.
|
||||
default : 0 = no console notification.
|
||||
1 = errors; 2 = notifications; 3 = details; 4 = debug;
|
||||
Note : not thread-safe (uses a global constant)
|
||||
*/
|
||||
void ZDICT_setNotificationLevel(unsigned l);
|
||||
|
||||
|
||||
/*-*************************************
|
||||
* Private functions
|
||||
***************************************/
|
||||
/*! ZDICT_trainFromBuffer_unsafe() :
|
||||
Same as ZDICT_trainFromBuffer_advanced(), but does not control `samplesBuffer`.
|
||||
note : `samplesBuffer` must be followed by noisy guard band to avoid out-of-buffer reads.
|
||||
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
|
||||
or an error code.
|
||||
*/
|
||||
size_t ZDICT_trainFromBuffer_unsafe(void* dictBuffer, size_t dictBufferCapacity,
|
||||
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
|
||||
ZDICT_params_t parameters);
|
||||
|
||||
|
||||
#if defined (__cplusplus)
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /* DICTBUILDER_STATIC_H_002 */
|
||||
+1905
@@ -0,0 +1,1905 @@
|
||||
/*
|
||||
* divsufsort.c for libdivsufsort-lite
|
||||
* Copyright (c) 2003-2008 Yuta Mori All Rights Reserved.
|
||||
*
|
||||
* Permission is hereby granted, free of charge, to any person
|
||||
* obtaining a copy of this software and associated documentation
|
||||
* files (the "Software"), to deal in the Software without
|
||||
* restriction, including without limitation the rights to use,
|
||||
* copy, modify, merge, publish, distribute, sublicense, and/or sell
|
||||
* copies of the Software, and to permit persons to whom the
|
||||
* Software is furnished to do so, subject to the following
|
||||
* conditions:
|
||||
*
|
||||
* The above copyright notice and this permission notice shall be
|
||||
* included in all copies or substantial portions of the Software.
|
||||
*
|
||||
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
||||
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
|
||||
* OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
||||
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
|
||||
* HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
|
||||
* WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
|
||||
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
|
||||
* OTHER DEALINGS IN THE SOFTWARE.
|
||||
*/
|
||||
|
||||
#if defined(_MSC_VER)
|
||||
#pragma warning(disable : 4244)
|
||||
#endif
|
||||
|
||||
#include <assert.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
|
||||
#include "divsufsort.h"
|
||||
|
||||
/*- Constants -*/
|
||||
#if defined(INLINE)
|
||||
# undef INLINE
|
||||
#endif
|
||||
#if !defined(INLINE)
|
||||
# define INLINE __inline
|
||||
#endif
|
||||
#if defined(ALPHABET_SIZE) && (ALPHABET_SIZE < 1)
|
||||
# undef ALPHABET_SIZE
|
||||
#endif
|
||||
#if !defined(ALPHABET_SIZE)
|
||||
# define ALPHABET_SIZE (256)
|
||||
#endif
|
||||
#define BUCKET_A_SIZE (ALPHABET_SIZE)
|
||||
#define BUCKET_B_SIZE (ALPHABET_SIZE * ALPHABET_SIZE)
|
||||
#if defined(SS_INSERTIONSORT_THRESHOLD)
|
||||
# if SS_INSERTIONSORT_THRESHOLD < 1
|
||||
# undef SS_INSERTIONSORT_THRESHOLD
|
||||
# define SS_INSERTIONSORT_THRESHOLD (1)
|
||||
# endif
|
||||
#else
|
||||
# define SS_INSERTIONSORT_THRESHOLD (8)
|
||||
#endif
|
||||
#if defined(SS_BLOCKSIZE)
|
||||
# if SS_BLOCKSIZE < 0
|
||||
# undef SS_BLOCKSIZE
|
||||
# define SS_BLOCKSIZE (0)
|
||||
# elif 32768 <= SS_BLOCKSIZE
|
||||
# undef SS_BLOCKSIZE
|
||||
# define SS_BLOCKSIZE (32767)
|
||||
# endif
|
||||
#else
|
||||
# define SS_BLOCKSIZE (1024)
|
||||
#endif
|
||||
/* minstacksize = log(SS_BLOCKSIZE) / log(3) * 2 */
|
||||
#if SS_BLOCKSIZE == 0
|
||||
# define SS_MISORT_STACKSIZE (96)
|
||||
#elif SS_BLOCKSIZE <= 4096
|
||||
# define SS_MISORT_STACKSIZE (16)
|
||||
#else
|
||||
# define SS_MISORT_STACKSIZE (24)
|
||||
#endif
|
||||
#define SS_SMERGE_STACKSIZE (32)
|
||||
#define TR_INSERTIONSORT_THRESHOLD (8)
|
||||
#define TR_STACKSIZE (64)
|
||||
|
||||
|
||||
/*- Macros -*/
|
||||
#ifndef SWAP
|
||||
# define SWAP(_a, _b) do { t = (_a); (_a) = (_b); (_b) = t; } while(0)
|
||||
#endif /* SWAP */
|
||||
#ifndef MIN
|
||||
# define MIN(_a, _b) (((_a) < (_b)) ? (_a) : (_b))
|
||||
#endif /* MIN */
|
||||
#ifndef MAX
|
||||
# define MAX(_a, _b) (((_a) > (_b)) ? (_a) : (_b))
|
||||
#endif /* MAX */
|
||||
#define STACK_PUSH(_a, _b, _c, _d)\
|
||||
do {\
|
||||
assert(ssize < STACK_SIZE);\
|
||||
stack[ssize].a = (_a), stack[ssize].b = (_b),\
|
||||
stack[ssize].c = (_c), stack[ssize++].d = (_d);\
|
||||
} while(0)
|
||||
#define STACK_PUSH5(_a, _b, _c, _d, _e)\
|
||||
do {\
|
||||
assert(ssize < STACK_SIZE);\
|
||||
stack[ssize].a = (_a), stack[ssize].b = (_b),\
|
||||
stack[ssize].c = (_c), stack[ssize].d = (_d), stack[ssize++].e = (_e);\
|
||||
} while(0)
|
||||
#define STACK_POP(_a, _b, _c, _d)\
|
||||
do {\
|
||||
assert(0 <= ssize);\
|
||||
if(ssize == 0) { return; }\
|
||||
(_a) = stack[--ssize].a, (_b) = stack[ssize].b,\
|
||||
(_c) = stack[ssize].c, (_d) = stack[ssize].d;\
|
||||
} while(0)
|
||||
#define STACK_POP5(_a, _b, _c, _d, _e)\
|
||||
do {\
|
||||
assert(0 <= ssize);\
|
||||
if(ssize == 0) { return; }\
|
||||
(_a) = stack[--ssize].a, (_b) = stack[ssize].b,\
|
||||
(_c) = stack[ssize].c, (_d) = stack[ssize].d, (_e) = stack[ssize].e;\
|
||||
} while(0)
|
||||
#define BUCKET_A(_c0) bucket_A[(_c0)]
|
||||
#if ALPHABET_SIZE == 256
|
||||
#define BUCKET_B(_c0, _c1) (bucket_B[((_c1) << 8) | (_c0)])
|
||||
#define BUCKET_BSTAR(_c0, _c1) (bucket_B[((_c0) << 8) | (_c1)])
|
||||
#else
|
||||
#define BUCKET_B(_c0, _c1) (bucket_B[(_c1) * ALPHABET_SIZE + (_c0)])
|
||||
#define BUCKET_BSTAR(_c0, _c1) (bucket_B[(_c0) * ALPHABET_SIZE + (_c1)])
|
||||
#endif
|
||||
|
||||
|
||||
/*- Private Functions -*/
|
||||
|
||||
static const int lg_table[256]= {
|
||||
-1,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,
|
||||
5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,
|
||||
6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
|
||||
6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
|
||||
7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
|
||||
7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
|
||||
7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
|
||||
7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7
|
||||
};
|
||||
|
||||
#if (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE)
|
||||
|
||||
static INLINE
|
||||
int
|
||||
ss_ilg(int n) {
|
||||
#if SS_BLOCKSIZE == 0
|
||||
return (n & 0xffff0000) ?
|
||||
((n & 0xff000000) ?
|
||||
24 + lg_table[(n >> 24) & 0xff] :
|
||||
16 + lg_table[(n >> 16) & 0xff]) :
|
||||
((n & 0x0000ff00) ?
|
||||
8 + lg_table[(n >> 8) & 0xff] :
|
||||
0 + lg_table[(n >> 0) & 0xff]);
|
||||
#elif SS_BLOCKSIZE < 256
|
||||
return lg_table[n];
|
||||
#else
|
||||
return (n & 0xff00) ?
|
||||
8 + lg_table[(n >> 8) & 0xff] :
|
||||
0 + lg_table[(n >> 0) & 0xff];
|
||||
#endif
|
||||
}
|
||||
|
||||
#endif /* (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) */
|
||||
|
||||
#if SS_BLOCKSIZE != 0
|
||||
|
||||
static const int sqq_table[256] = {
|
||||
0, 16, 22, 27, 32, 35, 39, 42, 45, 48, 50, 53, 55, 57, 59, 61,
|
||||
64, 65, 67, 69, 71, 73, 75, 76, 78, 80, 81, 83, 84, 86, 87, 89,
|
||||
90, 91, 93, 94, 96, 97, 98, 99, 101, 102, 103, 104, 106, 107, 108, 109,
|
||||
110, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126,
|
||||
128, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142,
|
||||
143, 144, 144, 145, 146, 147, 148, 149, 150, 150, 151, 152, 153, 154, 155, 155,
|
||||
156, 157, 158, 159, 160, 160, 161, 162, 163, 163, 164, 165, 166, 167, 167, 168,
|
||||
169, 170, 170, 171, 172, 173, 173, 174, 175, 176, 176, 177, 178, 178, 179, 180,
|
||||
181, 181, 182, 183, 183, 184, 185, 185, 186, 187, 187, 188, 189, 189, 190, 191,
|
||||
192, 192, 193, 193, 194, 195, 195, 196, 197, 197, 198, 199, 199, 200, 201, 201,
|
||||
202, 203, 203, 204, 204, 205, 206, 206, 207, 208, 208, 209, 209, 210, 211, 211,
|
||||
212, 212, 213, 214, 214, 215, 215, 216, 217, 217, 218, 218, 219, 219, 220, 221,
|
||||
221, 222, 222, 223, 224, 224, 225, 225, 226, 226, 227, 227, 228, 229, 229, 230,
|
||||
230, 231, 231, 232, 232, 233, 234, 234, 235, 235, 236, 236, 237, 237, 238, 238,
|
||||
239, 240, 240, 241, 241, 242, 242, 243, 243, 244, 244, 245, 245, 246, 246, 247,
|
||||
247, 248, 248, 249, 249, 250, 250, 251, 251, 252, 252, 253, 253, 254, 254, 255
|
||||
};
|
||||
|
||||
static INLINE
|
||||
int
|
||||
ss_isqrt(int x) {
|
||||
int y, e;
|
||||
|
||||
if(x >= (SS_BLOCKSIZE * SS_BLOCKSIZE)) { return SS_BLOCKSIZE; }
|
||||
e = (x & 0xffff0000) ?
|
||||
((x & 0xff000000) ?
|
||||
24 + lg_table[(x >> 24) & 0xff] :
|
||||
16 + lg_table[(x >> 16) & 0xff]) :
|
||||
((x & 0x0000ff00) ?
|
||||
8 + lg_table[(x >> 8) & 0xff] :
|
||||
0 + lg_table[(x >> 0) & 0xff]);
|
||||
|
||||
if(e >= 16) {
|
||||
y = sqq_table[x >> ((e - 6) - (e & 1))] << ((e >> 1) - 7);
|
||||
if(e >= 24) { y = (y + 1 + x / y) >> 1; }
|
||||
y = (y + 1 + x / y) >> 1;
|
||||
} else if(e >= 8) {
|
||||
y = (sqq_table[x >> ((e - 6) - (e & 1))] >> (7 - (e >> 1))) + 1;
|
||||
} else {
|
||||
return sqq_table[x] >> 4;
|
||||
}
|
||||
|
||||
return (x < (y * y)) ? y - 1 : y;
|
||||
}
|
||||
|
||||
#endif /* SS_BLOCKSIZE != 0 */
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Compares two suffixes. */
|
||||
static INLINE
|
||||
int
|
||||
ss_compare(const unsigned char *T,
|
||||
const int *p1, const int *p2,
|
||||
int depth) {
|
||||
const unsigned char *U1, *U2, *U1n, *U2n;
|
||||
|
||||
for(U1 = T + depth + *p1,
|
||||
U2 = T + depth + *p2,
|
||||
U1n = T + *(p1 + 1) + 2,
|
||||
U2n = T + *(p2 + 1) + 2;
|
||||
(U1 < U1n) && (U2 < U2n) && (*U1 == *U2);
|
||||
++U1, ++U2) {
|
||||
}
|
||||
|
||||
return U1 < U1n ?
|
||||
(U2 < U2n ? *U1 - *U2 : 1) :
|
||||
(U2 < U2n ? -1 : 0);
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
#if (SS_BLOCKSIZE != 1) && (SS_INSERTIONSORT_THRESHOLD != 1)
|
||||
|
||||
/* Insertionsort for small size groups */
|
||||
static
|
||||
void
|
||||
ss_insertionsort(const unsigned char *T, const int *PA,
|
||||
int *first, int *last, int depth) {
|
||||
int *i, *j;
|
||||
int t;
|
||||
int r;
|
||||
|
||||
for(i = last - 2; first <= i; --i) {
|
||||
for(t = *i, j = i + 1; 0 < (r = ss_compare(T, PA + t, PA + *j, depth));) {
|
||||
do { *(j - 1) = *j; } while((++j < last) && (*j < 0));
|
||||
if(last <= j) { break; }
|
||||
}
|
||||
if(r == 0) { *j = ~*j; }
|
||||
*(j - 1) = t;
|
||||
}
|
||||
}
|
||||
|
||||
#endif /* (SS_BLOCKSIZE != 1) && (SS_INSERTIONSORT_THRESHOLD != 1) */
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
#if (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE)
|
||||
|
||||
static INLINE
|
||||
void
|
||||
ss_fixdown(const unsigned char *Td, const int *PA,
|
||||
int *SA, int i, int size) {
|
||||
int j, k;
|
||||
int v;
|
||||
int c, d, e;
|
||||
|
||||
for(v = SA[i], c = Td[PA[v]]; (j = 2 * i + 1) < size; SA[i] = SA[k], i = k) {
|
||||
d = Td[PA[SA[k = j++]]];
|
||||
if(d < (e = Td[PA[SA[j]]])) { k = j; d = e; }
|
||||
if(d <= c) { break; }
|
||||
}
|
||||
SA[i] = v;
|
||||
}
|
||||
|
||||
/* Simple top-down heapsort. */
|
||||
static
|
||||
void
|
||||
ss_heapsort(const unsigned char *Td, const int *PA, int *SA, int size) {
|
||||
int i, m;
|
||||
int t;
|
||||
|
||||
m = size;
|
||||
if((size % 2) == 0) {
|
||||
m--;
|
||||
if(Td[PA[SA[m / 2]]] < Td[PA[SA[m]]]) { SWAP(SA[m], SA[m / 2]); }
|
||||
}
|
||||
|
||||
for(i = m / 2 - 1; 0 <= i; --i) { ss_fixdown(Td, PA, SA, i, m); }
|
||||
if((size % 2) == 0) { SWAP(SA[0], SA[m]); ss_fixdown(Td, PA, SA, 0, m); }
|
||||
for(i = m - 1; 0 < i; --i) {
|
||||
t = SA[0], SA[0] = SA[i];
|
||||
ss_fixdown(Td, PA, SA, 0, i);
|
||||
SA[i] = t;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Returns the median of three elements. */
|
||||
static INLINE
|
||||
int *
|
||||
ss_median3(const unsigned char *Td, const int *PA,
|
||||
int *v1, int *v2, int *v3) {
|
||||
int *t;
|
||||
if(Td[PA[*v1]] > Td[PA[*v2]]) { SWAP(v1, v2); }
|
||||
if(Td[PA[*v2]] > Td[PA[*v3]]) {
|
||||
if(Td[PA[*v1]] > Td[PA[*v3]]) { return v1; }
|
||||
else { return v3; }
|
||||
}
|
||||
return v2;
|
||||
}
|
||||
|
||||
/* Returns the median of five elements. */
|
||||
static INLINE
|
||||
int *
|
||||
ss_median5(const unsigned char *Td, const int *PA,
|
||||
int *v1, int *v2, int *v3, int *v4, int *v5) {
|
||||
int *t;
|
||||
if(Td[PA[*v2]] > Td[PA[*v3]]) { SWAP(v2, v3); }
|
||||
if(Td[PA[*v4]] > Td[PA[*v5]]) { SWAP(v4, v5); }
|
||||
if(Td[PA[*v2]] > Td[PA[*v4]]) { SWAP(v2, v4); SWAP(v3, v5); }
|
||||
if(Td[PA[*v1]] > Td[PA[*v3]]) { SWAP(v1, v3); }
|
||||
if(Td[PA[*v1]] > Td[PA[*v4]]) { SWAP(v1, v4); SWAP(v3, v5); }
|
||||
if(Td[PA[*v3]] > Td[PA[*v4]]) { return v4; }
|
||||
return v3;
|
||||
}
|
||||
|
||||
/* Returns the pivot element. */
|
||||
static INLINE
|
||||
int *
|
||||
ss_pivot(const unsigned char *Td, const int *PA, int *first, int *last) {
|
||||
int *middle;
|
||||
int t;
|
||||
|
||||
t = last - first;
|
||||
middle = first + t / 2;
|
||||
|
||||
if(t <= 512) {
|
||||
if(t <= 32) {
|
||||
return ss_median3(Td, PA, first, middle, last - 1);
|
||||
} else {
|
||||
t >>= 2;
|
||||
return ss_median5(Td, PA, first, first + t, middle, last - 1 - t, last - 1);
|
||||
}
|
||||
}
|
||||
t >>= 3;
|
||||
first = ss_median3(Td, PA, first, first + t, first + (t << 1));
|
||||
middle = ss_median3(Td, PA, middle - t, middle, middle + t);
|
||||
last = ss_median3(Td, PA, last - 1 - (t << 1), last - 1 - t, last - 1);
|
||||
return ss_median3(Td, PA, first, middle, last);
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Binary partition for substrings. */
|
||||
static INLINE
|
||||
int *
|
||||
ss_partition(const int *PA,
|
||||
int *first, int *last, int depth) {
|
||||
int *a, *b;
|
||||
int t;
|
||||
for(a = first - 1, b = last;;) {
|
||||
for(; (++a < b) && ((PA[*a] + depth) >= (PA[*a + 1] + 1));) { *a = ~*a; }
|
||||
for(; (a < --b) && ((PA[*b] + depth) < (PA[*b + 1] + 1));) { }
|
||||
if(b <= a) { break; }
|
||||
t = ~*b;
|
||||
*b = *a;
|
||||
*a = t;
|
||||
}
|
||||
if(first < a) { *first = ~*first; }
|
||||
return a;
|
||||
}
|
||||
|
||||
/* Multikey introsort for medium size groups. */
|
||||
static
|
||||
void
|
||||
ss_mintrosort(const unsigned char *T, const int *PA,
|
||||
int *first, int *last,
|
||||
int depth) {
|
||||
#define STACK_SIZE SS_MISORT_STACKSIZE
|
||||
struct { int *a, *b, c; int d; } stack[STACK_SIZE];
|
||||
const unsigned char *Td;
|
||||
int *a, *b, *c, *d, *e, *f;
|
||||
int s, t;
|
||||
int ssize;
|
||||
int limit;
|
||||
int v, x = 0;
|
||||
|
||||
for(ssize = 0, limit = ss_ilg(last - first);;) {
|
||||
|
||||
if((last - first) <= SS_INSERTIONSORT_THRESHOLD) {
|
||||
#if 1 < SS_INSERTIONSORT_THRESHOLD
|
||||
if(1 < (last - first)) { ss_insertionsort(T, PA, first, last, depth); }
|
||||
#endif
|
||||
STACK_POP(first, last, depth, limit);
|
||||
continue;
|
||||
}
|
||||
|
||||
Td = T + depth;
|
||||
if(limit-- == 0) { ss_heapsort(Td, PA, first, last - first); }
|
||||
if(limit < 0) {
|
||||
for(a = first + 1, v = Td[PA[*first]]; a < last; ++a) {
|
||||
if((x = Td[PA[*a]]) != v) {
|
||||
if(1 < (a - first)) { break; }
|
||||
v = x;
|
||||
first = a;
|
||||
}
|
||||
}
|
||||
if(Td[PA[*first] - 1] < v) {
|
||||
first = ss_partition(PA, first, a, depth);
|
||||
}
|
||||
if((a - first) <= (last - a)) {
|
||||
if(1 < (a - first)) {
|
||||
STACK_PUSH(a, last, depth, -1);
|
||||
last = a, depth += 1, limit = ss_ilg(a - first);
|
||||
} else {
|
||||
first = a, limit = -1;
|
||||
}
|
||||
} else {
|
||||
if(1 < (last - a)) {
|
||||
STACK_PUSH(first, a, depth + 1, ss_ilg(a - first));
|
||||
first = a, limit = -1;
|
||||
} else {
|
||||
last = a, depth += 1, limit = ss_ilg(a - first);
|
||||
}
|
||||
}
|
||||
continue;
|
||||
}
|
||||
|
||||
/* choose pivot */
|
||||
a = ss_pivot(Td, PA, first, last);
|
||||
v = Td[PA[*a]];
|
||||
SWAP(*first, *a);
|
||||
|
||||
/* partition */
|
||||
for(b = first; (++b < last) && ((x = Td[PA[*b]]) == v);) { }
|
||||
if(((a = b) < last) && (x < v)) {
|
||||
for(; (++b < last) && ((x = Td[PA[*b]]) <= v);) {
|
||||
if(x == v) { SWAP(*b, *a); ++a; }
|
||||
}
|
||||
}
|
||||
for(c = last; (b < --c) && ((x = Td[PA[*c]]) == v);) { }
|
||||
if((b < (d = c)) && (x > v)) {
|
||||
for(; (b < --c) && ((x = Td[PA[*c]]) >= v);) {
|
||||
if(x == v) { SWAP(*c, *d); --d; }
|
||||
}
|
||||
}
|
||||
for(; b < c;) {
|
||||
SWAP(*b, *c);
|
||||
for(; (++b < c) && ((x = Td[PA[*b]]) <= v);) {
|
||||
if(x == v) { SWAP(*b, *a); ++a; }
|
||||
}
|
||||
for(; (b < --c) && ((x = Td[PA[*c]]) >= v);) {
|
||||
if(x == v) { SWAP(*c, *d); --d; }
|
||||
}
|
||||
}
|
||||
|
||||
if(a <= d) {
|
||||
c = b - 1;
|
||||
|
||||
if((s = a - first) > (t = b - a)) { s = t; }
|
||||
for(e = first, f = b - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); }
|
||||
if((s = d - c) > (t = last - d - 1)) { s = t; }
|
||||
for(e = b, f = last - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); }
|
||||
|
||||
a = first + (b - a), c = last - (d - c);
|
||||
b = (v <= Td[PA[*a] - 1]) ? a : ss_partition(PA, a, c, depth);
|
||||
|
||||
if((a - first) <= (last - c)) {
|
||||
if((last - c) <= (c - b)) {
|
||||
STACK_PUSH(b, c, depth + 1, ss_ilg(c - b));
|
||||
STACK_PUSH(c, last, depth, limit);
|
||||
last = a;
|
||||
} else if((a - first) <= (c - b)) {
|
||||
STACK_PUSH(c, last, depth, limit);
|
||||
STACK_PUSH(b, c, depth + 1, ss_ilg(c - b));
|
||||
last = a;
|
||||
} else {
|
||||
STACK_PUSH(c, last, depth, limit);
|
||||
STACK_PUSH(first, a, depth, limit);
|
||||
first = b, last = c, depth += 1, limit = ss_ilg(c - b);
|
||||
}
|
||||
} else {
|
||||
if((a - first) <= (c - b)) {
|
||||
STACK_PUSH(b, c, depth + 1, ss_ilg(c - b));
|
||||
STACK_PUSH(first, a, depth, limit);
|
||||
first = c;
|
||||
} else if((last - c) <= (c - b)) {
|
||||
STACK_PUSH(first, a, depth, limit);
|
||||
STACK_PUSH(b, c, depth + 1, ss_ilg(c - b));
|
||||
first = c;
|
||||
} else {
|
||||
STACK_PUSH(first, a, depth, limit);
|
||||
STACK_PUSH(c, last, depth, limit);
|
||||
first = b, last = c, depth += 1, limit = ss_ilg(c - b);
|
||||
}
|
||||
}
|
||||
} else {
|
||||
limit += 1;
|
||||
if(Td[PA[*first] - 1] < v) {
|
||||
first = ss_partition(PA, first, last, depth);
|
||||
limit = ss_ilg(last - first);
|
||||
}
|
||||
depth += 1;
|
||||
}
|
||||
}
|
||||
#undef STACK_SIZE
|
||||
}
|
||||
|
||||
#endif /* (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) */
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
#if SS_BLOCKSIZE != 0
|
||||
|
||||
static INLINE
|
||||
void
|
||||
ss_blockswap(int *a, int *b, int n) {
|
||||
int t;
|
||||
for(; 0 < n; --n, ++a, ++b) {
|
||||
t = *a, *a = *b, *b = t;
|
||||
}
|
||||
}
|
||||
|
||||
static INLINE
|
||||
void
|
||||
ss_rotate(int *first, int *middle, int *last) {
|
||||
int *a, *b, t;
|
||||
int l, r;
|
||||
l = middle - first, r = last - middle;
|
||||
for(; (0 < l) && (0 < r);) {
|
||||
if(l == r) { ss_blockswap(first, middle, l); break; }
|
||||
if(l < r) {
|
||||
a = last - 1, b = middle - 1;
|
||||
t = *a;
|
||||
do {
|
||||
*a-- = *b, *b-- = *a;
|
||||
if(b < first) {
|
||||
*a = t;
|
||||
last = a;
|
||||
if((r -= l + 1) <= l) { break; }
|
||||
a -= 1, b = middle - 1;
|
||||
t = *a;
|
||||
}
|
||||
} while(1);
|
||||
} else {
|
||||
a = first, b = middle;
|
||||
t = *a;
|
||||
do {
|
||||
*a++ = *b, *b++ = *a;
|
||||
if(last <= b) {
|
||||
*a = t;
|
||||
first = a + 1;
|
||||
if((l -= r + 1) <= r) { break; }
|
||||
a += 1, b = middle;
|
||||
t = *a;
|
||||
}
|
||||
} while(1);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
static
|
||||
void
|
||||
ss_inplacemerge(const unsigned char *T, const int *PA,
|
||||
int *first, int *middle, int *last,
|
||||
int depth) {
|
||||
const int *p;
|
||||
int *a, *b;
|
||||
int len, half;
|
||||
int q, r;
|
||||
int x;
|
||||
|
||||
for(;;) {
|
||||
if(*(last - 1) < 0) { x = 1; p = PA + ~*(last - 1); }
|
||||
else { x = 0; p = PA + *(last - 1); }
|
||||
for(a = first, len = middle - first, half = len >> 1, r = -1;
|
||||
0 < len;
|
||||
len = half, half >>= 1) {
|
||||
b = a + half;
|
||||
q = ss_compare(T, PA + ((0 <= *b) ? *b : ~*b), p, depth);
|
||||
if(q < 0) {
|
||||
a = b + 1;
|
||||
half -= (len & 1) ^ 1;
|
||||
} else {
|
||||
r = q;
|
||||
}
|
||||
}
|
||||
if(a < middle) {
|
||||
if(r == 0) { *a = ~*a; }
|
||||
ss_rotate(a, middle, last);
|
||||
last -= middle - a;
|
||||
middle = a;
|
||||
if(first == middle) { break; }
|
||||
}
|
||||
--last;
|
||||
if(x != 0) { while(*--last < 0) { } }
|
||||
if(middle == last) { break; }
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Merge-forward with internal buffer. */
|
||||
static
|
||||
void
|
||||
ss_mergeforward(const unsigned char *T, const int *PA,
|
||||
int *first, int *middle, int *last,
|
||||
int *buf, int depth) {
|
||||
int *a, *b, *c, *bufend;
|
||||
int t;
|
||||
int r;
|
||||
|
||||
bufend = buf + (middle - first) - 1;
|
||||
ss_blockswap(buf, first, middle - first);
|
||||
|
||||
for(t = *(a = first), b = buf, c = middle;;) {
|
||||
r = ss_compare(T, PA + *b, PA + *c, depth);
|
||||
if(r < 0) {
|
||||
do {
|
||||
*a++ = *b;
|
||||
if(bufend <= b) { *bufend = t; return; }
|
||||
*b++ = *a;
|
||||
} while(*b < 0);
|
||||
} else if(r > 0) {
|
||||
do {
|
||||
*a++ = *c, *c++ = *a;
|
||||
if(last <= c) {
|
||||
while(b < bufend) { *a++ = *b, *b++ = *a; }
|
||||
*a = *b, *b = t;
|
||||
return;
|
||||
}
|
||||
} while(*c < 0);
|
||||
} else {
|
||||
*c = ~*c;
|
||||
do {
|
||||
*a++ = *b;
|
||||
if(bufend <= b) { *bufend = t; return; }
|
||||
*b++ = *a;
|
||||
} while(*b < 0);
|
||||
|
||||
do {
|
||||
*a++ = *c, *c++ = *a;
|
||||
if(last <= c) {
|
||||
while(b < bufend) { *a++ = *b, *b++ = *a; }
|
||||
*a = *b, *b = t;
|
||||
return;
|
||||
}
|
||||
} while(*c < 0);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Merge-backward with internal buffer. */
|
||||
static
|
||||
void
|
||||
ss_mergebackward(const unsigned char *T, const int *PA,
|
||||
int *first, int *middle, int *last,
|
||||
int *buf, int depth) {
|
||||
const int *p1, *p2;
|
||||
int *a, *b, *c, *bufend;
|
||||
int t;
|
||||
int r;
|
||||
int x;
|
||||
|
||||
bufend = buf + (last - middle) - 1;
|
||||
ss_blockswap(buf, middle, last - middle);
|
||||
|
||||
x = 0;
|
||||
if(*bufend < 0) { p1 = PA + ~*bufend; x |= 1; }
|
||||
else { p1 = PA + *bufend; }
|
||||
if(*(middle - 1) < 0) { p2 = PA + ~*(middle - 1); x |= 2; }
|
||||
else { p2 = PA + *(middle - 1); }
|
||||
for(t = *(a = last - 1), b = bufend, c = middle - 1;;) {
|
||||
r = ss_compare(T, p1, p2, depth);
|
||||
if(0 < r) {
|
||||
if(x & 1) { do { *a-- = *b, *b-- = *a; } while(*b < 0); x ^= 1; }
|
||||
*a-- = *b;
|
||||
if(b <= buf) { *buf = t; break; }
|
||||
*b-- = *a;
|
||||
if(*b < 0) { p1 = PA + ~*b; x |= 1; }
|
||||
else { p1 = PA + *b; }
|
||||
} else if(r < 0) {
|
||||
if(x & 2) { do { *a-- = *c, *c-- = *a; } while(*c < 0); x ^= 2; }
|
||||
*a-- = *c, *c-- = *a;
|
||||
if(c < first) {
|
||||
while(buf < b) { *a-- = *b, *b-- = *a; }
|
||||
*a = *b, *b = t;
|
||||
break;
|
||||
}
|
||||
if(*c < 0) { p2 = PA + ~*c; x |= 2; }
|
||||
else { p2 = PA + *c; }
|
||||
} else {
|
||||
if(x & 1) { do { *a-- = *b, *b-- = *a; } while(*b < 0); x ^= 1; }
|
||||
*a-- = ~*b;
|
||||
if(b <= buf) { *buf = t; break; }
|
||||
*b-- = *a;
|
||||
if(x & 2) { do { *a-- = *c, *c-- = *a; } while(*c < 0); x ^= 2; }
|
||||
*a-- = *c, *c-- = *a;
|
||||
if(c < first) {
|
||||
while(buf < b) { *a-- = *b, *b-- = *a; }
|
||||
*a = *b, *b = t;
|
||||
break;
|
||||
}
|
||||
if(*b < 0) { p1 = PA + ~*b; x |= 1; }
|
||||
else { p1 = PA + *b; }
|
||||
if(*c < 0) { p2 = PA + ~*c; x |= 2; }
|
||||
else { p2 = PA + *c; }
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* D&C based merge. */
|
||||
static
|
||||
void
|
||||
ss_swapmerge(const unsigned char *T, const int *PA,
|
||||
int *first, int *middle, int *last,
|
||||
int *buf, int bufsize, int depth) {
|
||||
#define STACK_SIZE SS_SMERGE_STACKSIZE
|
||||
#define GETIDX(a) ((0 <= (a)) ? (a) : (~(a)))
|
||||
#define MERGE_CHECK(a, b, c)\
|
||||
do {\
|
||||
if(((c) & 1) ||\
|
||||
(((c) & 2) && (ss_compare(T, PA + GETIDX(*((a) - 1)), PA + *(a), depth) == 0))) {\
|
||||
*(a) = ~*(a);\
|
||||
}\
|
||||
if(((c) & 4) && ((ss_compare(T, PA + GETIDX(*((b) - 1)), PA + *(b), depth) == 0))) {\
|
||||
*(b) = ~*(b);\
|
||||
}\
|
||||
} while(0)
|
||||
struct { int *a, *b, *c; int d; } stack[STACK_SIZE];
|
||||
int *l, *r, *lm, *rm;
|
||||
int m, len, half;
|
||||
int ssize;
|
||||
int check, next;
|
||||
|
||||
for(check = 0, ssize = 0;;) {
|
||||
if((last - middle) <= bufsize) {
|
||||
if((first < middle) && (middle < last)) {
|
||||
ss_mergebackward(T, PA, first, middle, last, buf, depth);
|
||||
}
|
||||
MERGE_CHECK(first, last, check);
|
||||
STACK_POP(first, middle, last, check);
|
||||
continue;
|
||||
}
|
||||
|
||||
if((middle - first) <= bufsize) {
|
||||
if(first < middle) {
|
||||
ss_mergeforward(T, PA, first, middle, last, buf, depth);
|
||||
}
|
||||
MERGE_CHECK(first, last, check);
|
||||
STACK_POP(first, middle, last, check);
|
||||
continue;
|
||||
}
|
||||
|
||||
for(m = 0, len = MIN(middle - first, last - middle), half = len >> 1;
|
||||
0 < len;
|
||||
len = half, half >>= 1) {
|
||||
if(ss_compare(T, PA + GETIDX(*(middle + m + half)),
|
||||
PA + GETIDX(*(middle - m - half - 1)), depth) < 0) {
|
||||
m += half + 1;
|
||||
half -= (len & 1) ^ 1;
|
||||
}
|
||||
}
|
||||
|
||||
if(0 < m) {
|
||||
lm = middle - m, rm = middle + m;
|
||||
ss_blockswap(lm, middle, m);
|
||||
l = r = middle, next = 0;
|
||||
if(rm < last) {
|
||||
if(*rm < 0) {
|
||||
*rm = ~*rm;
|
||||
if(first < lm) { for(; *--l < 0;) { } next |= 4; }
|
||||
next |= 1;
|
||||
} else if(first < lm) {
|
||||
for(; *r < 0; ++r) { }
|
||||
next |= 2;
|
||||
}
|
||||
}
|
||||
|
||||
if((l - first) <= (last - r)) {
|
||||
STACK_PUSH(r, rm, last, (next & 3) | (check & 4));
|
||||
middle = lm, last = l, check = (check & 3) | (next & 4);
|
||||
} else {
|
||||
if((next & 2) && (r == middle)) { next ^= 6; }
|
||||
STACK_PUSH(first, lm, l, (check & 3) | (next & 4));
|
||||
first = r, middle = rm, check = (next & 3) | (check & 4);
|
||||
}
|
||||
} else {
|
||||
if(ss_compare(T, PA + GETIDX(*(middle - 1)), PA + *middle, depth) == 0) {
|
||||
*middle = ~*middle;
|
||||
}
|
||||
MERGE_CHECK(first, last, check);
|
||||
STACK_POP(first, middle, last, check);
|
||||
}
|
||||
}
|
||||
#undef STACK_SIZE
|
||||
}
|
||||
|
||||
#endif /* SS_BLOCKSIZE != 0 */
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Substring sort */
|
||||
static
|
||||
void
|
||||
sssort(const unsigned char *T, const int *PA,
|
||||
int *first, int *last,
|
||||
int *buf, int bufsize,
|
||||
int depth, int n, int lastsuffix) {
|
||||
int *a;
|
||||
#if SS_BLOCKSIZE != 0
|
||||
int *b, *middle, *curbuf;
|
||||
int j, k, curbufsize, limit;
|
||||
#endif
|
||||
int i;
|
||||
|
||||
if(lastsuffix != 0) { ++first; }
|
||||
|
||||
#if SS_BLOCKSIZE == 0
|
||||
ss_mintrosort(T, PA, first, last, depth);
|
||||
#else
|
||||
if((bufsize < SS_BLOCKSIZE) &&
|
||||
(bufsize < (last - first)) &&
|
||||
(bufsize < (limit = ss_isqrt(last - first)))) {
|
||||
if(SS_BLOCKSIZE < limit) { limit = SS_BLOCKSIZE; }
|
||||
buf = middle = last - limit, bufsize = limit;
|
||||
} else {
|
||||
middle = last, limit = 0;
|
||||
}
|
||||
for(a = first, i = 0; SS_BLOCKSIZE < (middle - a); a += SS_BLOCKSIZE, ++i) {
|
||||
#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE
|
||||
ss_mintrosort(T, PA, a, a + SS_BLOCKSIZE, depth);
|
||||
#elif 1 < SS_BLOCKSIZE
|
||||
ss_insertionsort(T, PA, a, a + SS_BLOCKSIZE, depth);
|
||||
#endif
|
||||
curbufsize = last - (a + SS_BLOCKSIZE);
|
||||
curbuf = a + SS_BLOCKSIZE;
|
||||
if(curbufsize <= bufsize) { curbufsize = bufsize, curbuf = buf; }
|
||||
for(b = a, k = SS_BLOCKSIZE, j = i; j & 1; b -= k, k <<= 1, j >>= 1) {
|
||||
ss_swapmerge(T, PA, b - k, b, b + k, curbuf, curbufsize, depth);
|
||||
}
|
||||
}
|
||||
#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE
|
||||
ss_mintrosort(T, PA, a, middle, depth);
|
||||
#elif 1 < SS_BLOCKSIZE
|
||||
ss_insertionsort(T, PA, a, middle, depth);
|
||||
#endif
|
||||
for(k = SS_BLOCKSIZE; i != 0; k <<= 1, i >>= 1) {
|
||||
if(i & 1) {
|
||||
ss_swapmerge(T, PA, a - k, a, middle, buf, bufsize, depth);
|
||||
a -= k;
|
||||
}
|
||||
}
|
||||
if(limit != 0) {
|
||||
#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE
|
||||
ss_mintrosort(T, PA, middle, last, depth);
|
||||
#elif 1 < SS_BLOCKSIZE
|
||||
ss_insertionsort(T, PA, middle, last, depth);
|
||||
#endif
|
||||
ss_inplacemerge(T, PA, first, middle, last, depth);
|
||||
}
|
||||
#endif
|
||||
|
||||
if(lastsuffix != 0) {
|
||||
/* Insert last type B* suffix. */
|
||||
int PAi[2]; PAi[0] = PA[*(first - 1)], PAi[1] = n - 2;
|
||||
for(a = first, i = *(first - 1);
|
||||
(a < last) && ((*a < 0) || (0 < ss_compare(T, &(PAi[0]), PA + *a, depth)));
|
||||
++a) {
|
||||
*(a - 1) = *a;
|
||||
}
|
||||
*(a - 1) = i;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
static INLINE
|
||||
int
|
||||
tr_ilg(int n) {
|
||||
return (n & 0xffff0000) ?
|
||||
((n & 0xff000000) ?
|
||||
24 + lg_table[(n >> 24) & 0xff] :
|
||||
16 + lg_table[(n >> 16) & 0xff]) :
|
||||
((n & 0x0000ff00) ?
|
||||
8 + lg_table[(n >> 8) & 0xff] :
|
||||
0 + lg_table[(n >> 0) & 0xff]);
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Simple insertionsort for small size groups. */
|
||||
static
|
||||
void
|
||||
tr_insertionsort(const int *ISAd, int *first, int *last) {
|
||||
int *a, *b;
|
||||
int t, r;
|
||||
|
||||
for(a = first + 1; a < last; ++a) {
|
||||
for(t = *a, b = a - 1; 0 > (r = ISAd[t] - ISAd[*b]);) {
|
||||
do { *(b + 1) = *b; } while((first <= --b) && (*b < 0));
|
||||
if(b < first) { break; }
|
||||
}
|
||||
if(r == 0) { *b = ~*b; }
|
||||
*(b + 1) = t;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
static INLINE
|
||||
void
|
||||
tr_fixdown(const int *ISAd, int *SA, int i, int size) {
|
||||
int j, k;
|
||||
int v;
|
||||
int c, d, e;
|
||||
|
||||
for(v = SA[i], c = ISAd[v]; (j = 2 * i + 1) < size; SA[i] = SA[k], i = k) {
|
||||
d = ISAd[SA[k = j++]];
|
||||
if(d < (e = ISAd[SA[j]])) { k = j; d = e; }
|
||||
if(d <= c) { break; }
|
||||
}
|
||||
SA[i] = v;
|
||||
}
|
||||
|
||||
/* Simple top-down heapsort. */
|
||||
static
|
||||
void
|
||||
tr_heapsort(const int *ISAd, int *SA, int size) {
|
||||
int i, m;
|
||||
int t;
|
||||
|
||||
m = size;
|
||||
if((size % 2) == 0) {
|
||||
m--;
|
||||
if(ISAd[SA[m / 2]] < ISAd[SA[m]]) { SWAP(SA[m], SA[m / 2]); }
|
||||
}
|
||||
|
||||
for(i = m / 2 - 1; 0 <= i; --i) { tr_fixdown(ISAd, SA, i, m); }
|
||||
if((size % 2) == 0) { SWAP(SA[0], SA[m]); tr_fixdown(ISAd, SA, 0, m); }
|
||||
for(i = m - 1; 0 < i; --i) {
|
||||
t = SA[0], SA[0] = SA[i];
|
||||
tr_fixdown(ISAd, SA, 0, i);
|
||||
SA[i] = t;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Returns the median of three elements. */
|
||||
static INLINE
|
||||
int *
|
||||
tr_median3(const int *ISAd, int *v1, int *v2, int *v3) {
|
||||
int *t;
|
||||
if(ISAd[*v1] > ISAd[*v2]) { SWAP(v1, v2); }
|
||||
if(ISAd[*v2] > ISAd[*v3]) {
|
||||
if(ISAd[*v1] > ISAd[*v3]) { return v1; }
|
||||
else { return v3; }
|
||||
}
|
||||
return v2;
|
||||
}
|
||||
|
||||
/* Returns the median of five elements. */
|
||||
static INLINE
|
||||
int *
|
||||
tr_median5(const int *ISAd,
|
||||
int *v1, int *v2, int *v3, int *v4, int *v5) {
|
||||
int *t;
|
||||
if(ISAd[*v2] > ISAd[*v3]) { SWAP(v2, v3); }
|
||||
if(ISAd[*v4] > ISAd[*v5]) { SWAP(v4, v5); }
|
||||
if(ISAd[*v2] > ISAd[*v4]) { SWAP(v2, v4); SWAP(v3, v5); }
|
||||
if(ISAd[*v1] > ISAd[*v3]) { SWAP(v1, v3); }
|
||||
if(ISAd[*v1] > ISAd[*v4]) { SWAP(v1, v4); SWAP(v3, v5); }
|
||||
if(ISAd[*v3] > ISAd[*v4]) { return v4; }
|
||||
return v3;
|
||||
}
|
||||
|
||||
/* Returns the pivot element. */
|
||||
static INLINE
|
||||
int *
|
||||
tr_pivot(const int *ISAd, int *first, int *last) {
|
||||
int *middle;
|
||||
int t;
|
||||
|
||||
t = last - first;
|
||||
middle = first + t / 2;
|
||||
|
||||
if(t <= 512) {
|
||||
if(t <= 32) {
|
||||
return tr_median3(ISAd, first, middle, last - 1);
|
||||
} else {
|
||||
t >>= 2;
|
||||
return tr_median5(ISAd, first, first + t, middle, last - 1 - t, last - 1);
|
||||
}
|
||||
}
|
||||
t >>= 3;
|
||||
first = tr_median3(ISAd, first, first + t, first + (t << 1));
|
||||
middle = tr_median3(ISAd, middle - t, middle, middle + t);
|
||||
last = tr_median3(ISAd, last - 1 - (t << 1), last - 1 - t, last - 1);
|
||||
return tr_median3(ISAd, first, middle, last);
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
typedef struct _trbudget_t trbudget_t;
|
||||
struct _trbudget_t {
|
||||
int chance;
|
||||
int remain;
|
||||
int incval;
|
||||
int count;
|
||||
};
|
||||
|
||||
static INLINE
|
||||
void
|
||||
trbudget_init(trbudget_t *budget, int chance, int incval) {
|
||||
budget->chance = chance;
|
||||
budget->remain = budget->incval = incval;
|
||||
}
|
||||
|
||||
static INLINE
|
||||
int
|
||||
trbudget_check(trbudget_t *budget, int size) {
|
||||
if(size <= budget->remain) { budget->remain -= size; return 1; }
|
||||
if(budget->chance == 0) { budget->count += size; return 0; }
|
||||
budget->remain += budget->incval - size;
|
||||
budget->chance -= 1;
|
||||
return 1;
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
static INLINE
|
||||
void
|
||||
tr_partition(const int *ISAd,
|
||||
int *first, int *middle, int *last,
|
||||
int **pa, int **pb, int v) {
|
||||
int *a, *b, *c, *d, *e, *f;
|
||||
int t, s;
|
||||
int x = 0;
|
||||
|
||||
for(b = middle - 1; (++b < last) && ((x = ISAd[*b]) == v);) { }
|
||||
if(((a = b) < last) && (x < v)) {
|
||||
for(; (++b < last) && ((x = ISAd[*b]) <= v);) {
|
||||
if(x == v) { SWAP(*b, *a); ++a; }
|
||||
}
|
||||
}
|
||||
for(c = last; (b < --c) && ((x = ISAd[*c]) == v);) { }
|
||||
if((b < (d = c)) && (x > v)) {
|
||||
for(; (b < --c) && ((x = ISAd[*c]) >= v);) {
|
||||
if(x == v) { SWAP(*c, *d); --d; }
|
||||
}
|
||||
}
|
||||
for(; b < c;) {
|
||||
SWAP(*b, *c);
|
||||
for(; (++b < c) && ((x = ISAd[*b]) <= v);) {
|
||||
if(x == v) { SWAP(*b, *a); ++a; }
|
||||
}
|
||||
for(; (b < --c) && ((x = ISAd[*c]) >= v);) {
|
||||
if(x == v) { SWAP(*c, *d); --d; }
|
||||
}
|
||||
}
|
||||
|
||||
if(a <= d) {
|
||||
c = b - 1;
|
||||
if((s = a - first) > (t = b - a)) { s = t; }
|
||||
for(e = first, f = b - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); }
|
||||
if((s = d - c) > (t = last - d - 1)) { s = t; }
|
||||
for(e = b, f = last - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); }
|
||||
first += (b - a), last -= (d - c);
|
||||
}
|
||||
*pa = first, *pb = last;
|
||||
}
|
||||
|
||||
static
|
||||
void
|
||||
tr_copy(int *ISA, const int *SA,
|
||||
int *first, int *a, int *b, int *last,
|
||||
int depth) {
|
||||
/* sort suffixes of middle partition
|
||||
by using sorted order of suffixes of left and right partition. */
|
||||
int *c, *d, *e;
|
||||
int s, v;
|
||||
|
||||
v = b - SA - 1;
|
||||
for(c = first, d = a - 1; c <= d; ++c) {
|
||||
if((0 <= (s = *c - depth)) && (ISA[s] == v)) {
|
||||
*++d = s;
|
||||
ISA[s] = d - SA;
|
||||
}
|
||||
}
|
||||
for(c = last - 1, e = d + 1, d = b; e < d; --c) {
|
||||
if((0 <= (s = *c - depth)) && (ISA[s] == v)) {
|
||||
*--d = s;
|
||||
ISA[s] = d - SA;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
static
|
||||
void
|
||||
tr_partialcopy(int *ISA, const int *SA,
|
||||
int *first, int *a, int *b, int *last,
|
||||
int depth) {
|
||||
int *c, *d, *e;
|
||||
int s, v;
|
||||
int rank, lastrank, newrank = -1;
|
||||
|
||||
v = b - SA - 1;
|
||||
lastrank = -1;
|
||||
for(c = first, d = a - 1; c <= d; ++c) {
|
||||
if((0 <= (s = *c - depth)) && (ISA[s] == v)) {
|
||||
*++d = s;
|
||||
rank = ISA[s + depth];
|
||||
if(lastrank != rank) { lastrank = rank; newrank = d - SA; }
|
||||
ISA[s] = newrank;
|
||||
}
|
||||
}
|
||||
|
||||
lastrank = -1;
|
||||
for(e = d; first <= e; --e) {
|
||||
rank = ISA[*e];
|
||||
if(lastrank != rank) { lastrank = rank; newrank = e - SA; }
|
||||
if(newrank != rank) { ISA[*e] = newrank; }
|
||||
}
|
||||
|
||||
lastrank = -1;
|
||||
for(c = last - 1, e = d + 1, d = b; e < d; --c) {
|
||||
if((0 <= (s = *c - depth)) && (ISA[s] == v)) {
|
||||
*--d = s;
|
||||
rank = ISA[s + depth];
|
||||
if(lastrank != rank) { lastrank = rank; newrank = d - SA; }
|
||||
ISA[s] = newrank;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
static
|
||||
void
|
||||
tr_introsort(int *ISA, const int *ISAd,
|
||||
int *SA, int *first, int *last,
|
||||
trbudget_t *budget) {
|
||||
#define STACK_SIZE TR_STACKSIZE
|
||||
struct { const int *a; int *b, *c; int d, e; }stack[STACK_SIZE];
|
||||
int *a, *b, *c;
|
||||
int t;
|
||||
int v, x = 0;
|
||||
int incr = ISAd - ISA;
|
||||
int limit, next;
|
||||
int ssize, trlink = -1;
|
||||
|
||||
for(ssize = 0, limit = tr_ilg(last - first);;) {
|
||||
|
||||
if(limit < 0) {
|
||||
if(limit == -1) {
|
||||
/* tandem repeat partition */
|
||||
tr_partition(ISAd - incr, first, first, last, &a, &b, last - SA - 1);
|
||||
|
||||
/* update ranks */
|
||||
if(a < last) {
|
||||
for(c = first, v = a - SA - 1; c < a; ++c) { ISA[*c] = v; }
|
||||
}
|
||||
if(b < last) {
|
||||
for(c = a, v = b - SA - 1; c < b; ++c) { ISA[*c] = v; }
|
||||
}
|
||||
|
||||
/* push */
|
||||
if(1 < (b - a)) {
|
||||
STACK_PUSH5(NULL, a, b, 0, 0);
|
||||
STACK_PUSH5(ISAd - incr, first, last, -2, trlink);
|
||||
trlink = ssize - 2;
|
||||
}
|
||||
if((a - first) <= (last - b)) {
|
||||
if(1 < (a - first)) {
|
||||
STACK_PUSH5(ISAd, b, last, tr_ilg(last - b), trlink);
|
||||
last = a, limit = tr_ilg(a - first);
|
||||
} else if(1 < (last - b)) {
|
||||
first = b, limit = tr_ilg(last - b);
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
} else {
|
||||
if(1 < (last - b)) {
|
||||
STACK_PUSH5(ISAd, first, a, tr_ilg(a - first), trlink);
|
||||
first = b, limit = tr_ilg(last - b);
|
||||
} else if(1 < (a - first)) {
|
||||
last = a, limit = tr_ilg(a - first);
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
}
|
||||
} else if(limit == -2) {
|
||||
/* tandem repeat copy */
|
||||
a = stack[--ssize].b, b = stack[ssize].c;
|
||||
if(stack[ssize].d == 0) {
|
||||
tr_copy(ISA, SA, first, a, b, last, ISAd - ISA);
|
||||
} else {
|
||||
if(0 <= trlink) { stack[trlink].d = -1; }
|
||||
tr_partialcopy(ISA, SA, first, a, b, last, ISAd - ISA);
|
||||
}
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
} else {
|
||||
/* sorted partition */
|
||||
if(0 <= *first) {
|
||||
a = first;
|
||||
do { ISA[*a] = a - SA; } while((++a < last) && (0 <= *a));
|
||||
first = a;
|
||||
}
|
||||
if(first < last) {
|
||||
a = first; do { *a = ~*a; } while(*++a < 0);
|
||||
next = (ISA[*a] != ISAd[*a]) ? tr_ilg(a - first + 1) : -1;
|
||||
if(++a < last) { for(b = first, v = a - SA - 1; b < a; ++b) { ISA[*b] = v; } }
|
||||
|
||||
/* push */
|
||||
if(trbudget_check(budget, a - first)) {
|
||||
if((a - first) <= (last - a)) {
|
||||
STACK_PUSH5(ISAd, a, last, -3, trlink);
|
||||
ISAd += incr, last = a, limit = next;
|
||||
} else {
|
||||
if(1 < (last - a)) {
|
||||
STACK_PUSH5(ISAd + incr, first, a, next, trlink);
|
||||
first = a, limit = -3;
|
||||
} else {
|
||||
ISAd += incr, last = a, limit = next;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if(0 <= trlink) { stack[trlink].d = -1; }
|
||||
if(1 < (last - a)) {
|
||||
first = a, limit = -3;
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
}
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
}
|
||||
continue;
|
||||
}
|
||||
|
||||
if((last - first) <= TR_INSERTIONSORT_THRESHOLD) {
|
||||
tr_insertionsort(ISAd, first, last);
|
||||
limit = -3;
|
||||
continue;
|
||||
}
|
||||
|
||||
if(limit-- == 0) {
|
||||
tr_heapsort(ISAd, first, last - first);
|
||||
for(a = last - 1; first < a; a = b) {
|
||||
for(x = ISAd[*a], b = a - 1; (first <= b) && (ISAd[*b] == x); --b) { *b = ~*b; }
|
||||
}
|
||||
limit = -3;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* choose pivot */
|
||||
a = tr_pivot(ISAd, first, last);
|
||||
SWAP(*first, *a);
|
||||
v = ISAd[*first];
|
||||
|
||||
/* partition */
|
||||
tr_partition(ISAd, first, first + 1, last, &a, &b, v);
|
||||
if((last - first) != (b - a)) {
|
||||
next = (ISA[*a] != v) ? tr_ilg(b - a) : -1;
|
||||
|
||||
/* update ranks */
|
||||
for(c = first, v = a - SA - 1; c < a; ++c) { ISA[*c] = v; }
|
||||
if(b < last) { for(c = a, v = b - SA - 1; c < b; ++c) { ISA[*c] = v; } }
|
||||
|
||||
/* push */
|
||||
if((1 < (b - a)) && (trbudget_check(budget, b - a))) {
|
||||
if((a - first) <= (last - b)) {
|
||||
if((last - b) <= (b - a)) {
|
||||
if(1 < (a - first)) {
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
last = a;
|
||||
} else if(1 < (last - b)) {
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
first = b;
|
||||
} else {
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
} else if((a - first) <= (b - a)) {
|
||||
if(1 < (a - first)) {
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
last = a;
|
||||
} else {
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
} else {
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
} else {
|
||||
if((a - first) <= (b - a)) {
|
||||
if(1 < (last - b)) {
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
first = b;
|
||||
} else if(1 < (a - first)) {
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
last = a;
|
||||
} else {
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
} else if((last - b) <= (b - a)) {
|
||||
if(1 < (last - b)) {
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
STACK_PUSH5(ISAd + incr, a, b, next, trlink);
|
||||
first = b;
|
||||
} else {
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
} else {
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
ISAd += incr, first = a, last = b, limit = next;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if((1 < (b - a)) && (0 <= trlink)) { stack[trlink].d = -1; }
|
||||
if((a - first) <= (last - b)) {
|
||||
if(1 < (a - first)) {
|
||||
STACK_PUSH5(ISAd, b, last, limit, trlink);
|
||||
last = a;
|
||||
} else if(1 < (last - b)) {
|
||||
first = b;
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
} else {
|
||||
if(1 < (last - b)) {
|
||||
STACK_PUSH5(ISAd, first, a, limit, trlink);
|
||||
first = b;
|
||||
} else if(1 < (a - first)) {
|
||||
last = a;
|
||||
} else {
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if(trbudget_check(budget, last - first)) {
|
||||
limit = tr_ilg(last - first), ISAd += incr;
|
||||
} else {
|
||||
if(0 <= trlink) { stack[trlink].d = -1; }
|
||||
STACK_POP5(ISAd, first, last, limit, trlink);
|
||||
}
|
||||
}
|
||||
}
|
||||
#undef STACK_SIZE
|
||||
}
|
||||
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Tandem repeat sort */
|
||||
static
|
||||
void
|
||||
trsort(int *ISA, int *SA, int n, int depth) {
|
||||
int *ISAd;
|
||||
int *first, *last;
|
||||
trbudget_t budget;
|
||||
int t, skip, unsorted;
|
||||
|
||||
trbudget_init(&budget, tr_ilg(n) * 2 / 3, n);
|
||||
/* trbudget_init(&budget, tr_ilg(n) * 3 / 4, n); */
|
||||
for(ISAd = ISA + depth; -n < *SA; ISAd += ISAd - ISA) {
|
||||
first = SA;
|
||||
skip = 0;
|
||||
unsorted = 0;
|
||||
do {
|
||||
if((t = *first) < 0) { first -= t; skip += t; }
|
||||
else {
|
||||
if(skip != 0) { *(first + skip) = skip; skip = 0; }
|
||||
last = SA + ISA[t] + 1;
|
||||
if(1 < (last - first)) {
|
||||
budget.count = 0;
|
||||
tr_introsort(ISA, ISAd, SA, first, last, &budget);
|
||||
if(budget.count != 0) { unsorted += budget.count; }
|
||||
else { skip = first - last; }
|
||||
} else if((last - first) == 1) {
|
||||
skip = -1;
|
||||
}
|
||||
first = last;
|
||||
}
|
||||
} while(first < (SA + n));
|
||||
if(skip != 0) { *(first + skip) = skip; }
|
||||
if(unsorted == 0) { break; }
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/* Sorts suffixes of type B*. */
|
||||
static
|
||||
int
|
||||
sort_typeBstar(const unsigned char *T, int *SA,
|
||||
int *bucket_A, int *bucket_B,
|
||||
int n, int openMP) {
|
||||
int *PAb, *ISAb, *buf;
|
||||
#ifdef LIBBSC_OPENMP
|
||||
int *curbuf;
|
||||
int l;
|
||||
#endif
|
||||
int i, j, k, t, m, bufsize;
|
||||
int c0, c1;
|
||||
#ifdef LIBBSC_OPENMP
|
||||
int d0, d1;
|
||||
#endif
|
||||
(void)openMP;
|
||||
|
||||
/* Initialize bucket arrays. */
|
||||
for(i = 0; i < BUCKET_A_SIZE; ++i) { bucket_A[i] = 0; }
|
||||
for(i = 0; i < BUCKET_B_SIZE; ++i) { bucket_B[i] = 0; }
|
||||
|
||||
/* Count the number of occurrences of the first one or two characters of each
|
||||
type A, B and B* suffix. Moreover, store the beginning position of all
|
||||
type B* suffixes into the array SA. */
|
||||
for(i = n - 1, m = n, c0 = T[n - 1]; 0 <= i;) {
|
||||
/* type A suffix. */
|
||||
do { ++BUCKET_A(c1 = c0); } while((0 <= --i) && ((c0 = T[i]) >= c1));
|
||||
if(0 <= i) {
|
||||
/* type B* suffix. */
|
||||
++BUCKET_BSTAR(c0, c1);
|
||||
SA[--m] = i;
|
||||
/* type B suffix. */
|
||||
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) {
|
||||
++BUCKET_B(c0, c1);
|
||||
}
|
||||
}
|
||||
}
|
||||
m = n - m;
|
||||
/*
|
||||
note:
|
||||
A type B* suffix is lexicographically smaller than a type B suffix that
|
||||
begins with the same first two characters.
|
||||
*/
|
||||
|
||||
/* Calculate the index of start/end point of each bucket. */
|
||||
for(c0 = 0, i = 0, j = 0; c0 < ALPHABET_SIZE; ++c0) {
|
||||
t = i + BUCKET_A(c0);
|
||||
BUCKET_A(c0) = i + j; /* start point */
|
||||
i = t + BUCKET_B(c0, c0);
|
||||
for(c1 = c0 + 1; c1 < ALPHABET_SIZE; ++c1) {
|
||||
j += BUCKET_BSTAR(c0, c1);
|
||||
BUCKET_BSTAR(c0, c1) = j; /* end point */
|
||||
i += BUCKET_B(c0, c1);
|
||||
}
|
||||
}
|
||||
|
||||
if(0 < m) {
|
||||
/* Sort the type B* suffixes by their first two characters. */
|
||||
PAb = SA + n - m; ISAb = SA + m;
|
||||
for(i = m - 2; 0 <= i; --i) {
|
||||
t = PAb[i], c0 = T[t], c1 = T[t + 1];
|
||||
SA[--BUCKET_BSTAR(c0, c1)] = i;
|
||||
}
|
||||
t = PAb[m - 1], c0 = T[t], c1 = T[t + 1];
|
||||
SA[--BUCKET_BSTAR(c0, c1)] = m - 1;
|
||||
|
||||
/* Sort the type B* substrings using sssort. */
|
||||
#ifdef LIBBSC_OPENMP
|
||||
if (openMP)
|
||||
{
|
||||
buf = SA + m;
|
||||
c0 = ALPHABET_SIZE - 2, c1 = ALPHABET_SIZE - 1, j = m;
|
||||
#pragma omp parallel default(shared) private(bufsize, curbuf, k, l, d0, d1)
|
||||
{
|
||||
bufsize = (n - (2 * m)) / omp_get_num_threads();
|
||||
curbuf = buf + omp_get_thread_num() * bufsize;
|
||||
k = 0;
|
||||
for(;;) {
|
||||
#pragma omp critical(sssort_lock)
|
||||
{
|
||||
if(0 < (l = j)) {
|
||||
d0 = c0, d1 = c1;
|
||||
do {
|
||||
k = BUCKET_BSTAR(d0, d1);
|
||||
if(--d1 <= d0) {
|
||||
d1 = ALPHABET_SIZE - 1;
|
||||
if(--d0 < 0) { break; }
|
||||
}
|
||||
} while(((l - k) <= 1) && (0 < (l = k)));
|
||||
c0 = d0, c1 = d1, j = k;
|
||||
}
|
||||
}
|
||||
if(l == 0) { break; }
|
||||
sssort(T, PAb, SA + k, SA + l,
|
||||
curbuf, bufsize, 2, n, *(SA + k) == (m - 1));
|
||||
}
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
buf = SA + m, bufsize = n - (2 * m);
|
||||
for(c0 = ALPHABET_SIZE - 2, j = m; 0 < j; --c0) {
|
||||
for(c1 = ALPHABET_SIZE - 1; c0 < c1; j = i, --c1) {
|
||||
i = BUCKET_BSTAR(c0, c1);
|
||||
if(1 < (j - i)) {
|
||||
sssort(T, PAb, SA + i, SA + j,
|
||||
buf, bufsize, 2, n, *(SA + i) == (m - 1));
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
#else
|
||||
buf = SA + m, bufsize = n - (2 * m);
|
||||
for(c0 = ALPHABET_SIZE - 2, j = m; 0 < j; --c0) {
|
||||
for(c1 = ALPHABET_SIZE - 1; c0 < c1; j = i, --c1) {
|
||||
i = BUCKET_BSTAR(c0, c1);
|
||||
if(1 < (j - i)) {
|
||||
sssort(T, PAb, SA + i, SA + j,
|
||||
buf, bufsize, 2, n, *(SA + i) == (m - 1));
|
||||
}
|
||||
}
|
||||
}
|
||||
#endif
|
||||
|
||||
/* Compute ranks of type B* substrings. */
|
||||
for(i = m - 1; 0 <= i; --i) {
|
||||
if(0 <= SA[i]) {
|
||||
j = i;
|
||||
do { ISAb[SA[i]] = i; } while((0 <= --i) && (0 <= SA[i]));
|
||||
SA[i + 1] = i - j;
|
||||
if(i <= 0) { break; }
|
||||
}
|
||||
j = i;
|
||||
do { ISAb[SA[i] = ~SA[i]] = j; } while(SA[--i] < 0);
|
||||
ISAb[SA[i]] = j;
|
||||
}
|
||||
|
||||
/* Construct the inverse suffix array of type B* suffixes using trsort. */
|
||||
trsort(ISAb, SA, m, 1);
|
||||
|
||||
/* Set the sorted order of tyoe B* suffixes. */
|
||||
for(i = n - 1, j = m, c0 = T[n - 1]; 0 <= i;) {
|
||||
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) >= c1); --i, c1 = c0) { }
|
||||
if(0 <= i) {
|
||||
t = i;
|
||||
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) { }
|
||||
SA[ISAb[--j]] = ((t == 0) || (1 < (t - i))) ? t : ~t;
|
||||
}
|
||||
}
|
||||
|
||||
/* Calculate the index of start/end point of each bucket. */
|
||||
BUCKET_B(ALPHABET_SIZE - 1, ALPHABET_SIZE - 1) = n; /* end point */
|
||||
for(c0 = ALPHABET_SIZE - 2, k = m - 1; 0 <= c0; --c0) {
|
||||
i = BUCKET_A(c0 + 1) - 1;
|
||||
for(c1 = ALPHABET_SIZE - 1; c0 < c1; --c1) {
|
||||
t = i - BUCKET_B(c0, c1);
|
||||
BUCKET_B(c0, c1) = i; /* end point */
|
||||
|
||||
/* Move all type B* suffixes to the correct position. */
|
||||
for(i = t, j = BUCKET_BSTAR(c0, c1);
|
||||
j <= k;
|
||||
--i, --k) { SA[i] = SA[k]; }
|
||||
}
|
||||
BUCKET_BSTAR(c0, c0 + 1) = i - BUCKET_B(c0, c0) + 1; /* start point */
|
||||
BUCKET_B(c0, c0) = i; /* end point */
|
||||
}
|
||||
}
|
||||
|
||||
return m;
|
||||
}
|
||||
|
||||
/* Constructs the suffix array by using the sorted order of type B* suffixes. */
|
||||
static
|
||||
void
|
||||
construct_SA(const unsigned char *T, int *SA,
|
||||
int *bucket_A, int *bucket_B,
|
||||
int n, int m) {
|
||||
int *i, *j, *k;
|
||||
int s;
|
||||
int c0, c1, c2;
|
||||
|
||||
if(0 < m) {
|
||||
/* Construct the sorted order of type B suffixes by using
|
||||
the sorted order of type B* suffixes. */
|
||||
for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
|
||||
/* Scan the suffix array from right to left. */
|
||||
for(i = SA + BUCKET_BSTAR(c1, c1 + 1),
|
||||
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
|
||||
i <= j;
|
||||
--j) {
|
||||
if(0 < (s = *j)) {
|
||||
assert(T[s] == c1);
|
||||
assert(((s + 1) < n) && (T[s] <= T[s + 1]));
|
||||
assert(T[s - 1] <= T[s]);
|
||||
*j = ~s;
|
||||
c0 = T[--s];
|
||||
if((0 < s) && (T[s - 1] > c0)) { s = ~s; }
|
||||
if(c0 != c2) {
|
||||
if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; }
|
||||
k = SA + BUCKET_B(c2 = c0, c1);
|
||||
}
|
||||
assert(k < j);
|
||||
*k-- = s;
|
||||
} else {
|
||||
assert(((s == 0) && (T[s] == c1)) || (s < 0));
|
||||
*j = ~s;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Construct the suffix array by using
|
||||
the sorted order of type B suffixes. */
|
||||
k = SA + BUCKET_A(c2 = T[n - 1]);
|
||||
*k++ = (T[n - 2] < c2) ? ~(n - 1) : (n - 1);
|
||||
/* Scan the suffix array from left to right. */
|
||||
for(i = SA, j = SA + n; i < j; ++i) {
|
||||
if(0 < (s = *i)) {
|
||||
assert(T[s - 1] >= T[s]);
|
||||
c0 = T[--s];
|
||||
if((s == 0) || (T[s - 1] < c0)) { s = ~s; }
|
||||
if(c0 != c2) {
|
||||
BUCKET_A(c2) = k - SA;
|
||||
k = SA + BUCKET_A(c2 = c0);
|
||||
}
|
||||
assert(i < k);
|
||||
*k++ = s;
|
||||
} else {
|
||||
assert(s < 0);
|
||||
*i = ~s;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Constructs the burrows-wheeler transformed string directly
|
||||
by using the sorted order of type B* suffixes. */
|
||||
static
|
||||
int
|
||||
construct_BWT(const unsigned char *T, int *SA,
|
||||
int *bucket_A, int *bucket_B,
|
||||
int n, int m) {
|
||||
int *i, *j, *k, *orig;
|
||||
int s;
|
||||
int c0, c1, c2;
|
||||
|
||||
if(0 < m) {
|
||||
/* Construct the sorted order of type B suffixes by using
|
||||
the sorted order of type B* suffixes. */
|
||||
for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
|
||||
/* Scan the suffix array from right to left. */
|
||||
for(i = SA + BUCKET_BSTAR(c1, c1 + 1),
|
||||
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
|
||||
i <= j;
|
||||
--j) {
|
||||
if(0 < (s = *j)) {
|
||||
assert(T[s] == c1);
|
||||
assert(((s + 1) < n) && (T[s] <= T[s + 1]));
|
||||
assert(T[s - 1] <= T[s]);
|
||||
c0 = T[--s];
|
||||
*j = ~((int)c0);
|
||||
if((0 < s) && (T[s - 1] > c0)) { s = ~s; }
|
||||
if(c0 != c2) {
|
||||
if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; }
|
||||
k = SA + BUCKET_B(c2 = c0, c1);
|
||||
}
|
||||
assert(k < j);
|
||||
*k-- = s;
|
||||
} else if(s != 0) {
|
||||
*j = ~s;
|
||||
#ifndef NDEBUG
|
||||
} else {
|
||||
assert(T[s] == c1);
|
||||
#endif
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Construct the BWTed string by using
|
||||
the sorted order of type B suffixes. */
|
||||
k = SA + BUCKET_A(c2 = T[n - 1]);
|
||||
*k++ = (T[n - 2] < c2) ? ~((int)T[n - 2]) : (n - 1);
|
||||
/* Scan the suffix array from left to right. */
|
||||
for(i = SA, j = SA + n, orig = SA; i < j; ++i) {
|
||||
if(0 < (s = *i)) {
|
||||
assert(T[s - 1] >= T[s]);
|
||||
c0 = T[--s];
|
||||
*i = c0;
|
||||
if((0 < s) && (T[s - 1] < c0)) { s = ~((int)T[s - 1]); }
|
||||
if(c0 != c2) {
|
||||
BUCKET_A(c2) = k - SA;
|
||||
k = SA + BUCKET_A(c2 = c0);
|
||||
}
|
||||
assert(i < k);
|
||||
*k++ = s;
|
||||
} else if(s != 0) {
|
||||
*i = ~s;
|
||||
} else {
|
||||
orig = i;
|
||||
}
|
||||
}
|
||||
|
||||
return orig - SA;
|
||||
}
|
||||
|
||||
/* Constructs the burrows-wheeler transformed string directly
|
||||
by using the sorted order of type B* suffixes. */
|
||||
static
|
||||
int
|
||||
construct_BWT_indexes(const unsigned char *T, int *SA,
|
||||
int *bucket_A, int *bucket_B,
|
||||
int n, int m,
|
||||
unsigned char * num_indexes, int * indexes) {
|
||||
int *i, *j, *k, *orig;
|
||||
int s;
|
||||
int c0, c1, c2;
|
||||
|
||||
int mod = n / 8;
|
||||
{
|
||||
mod |= mod >> 1; mod |= mod >> 2;
|
||||
mod |= mod >> 4; mod |= mod >> 8;
|
||||
mod |= mod >> 16; mod >>= 1;
|
||||
|
||||
*num_indexes = (unsigned char)((n - 1) / (mod + 1));
|
||||
}
|
||||
|
||||
if(0 < m) {
|
||||
/* Construct the sorted order of type B suffixes by using
|
||||
the sorted order of type B* suffixes. */
|
||||
for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
|
||||
/* Scan the suffix array from right to left. */
|
||||
for(i = SA + BUCKET_BSTAR(c1, c1 + 1),
|
||||
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
|
||||
i <= j;
|
||||
--j) {
|
||||
if(0 < (s = *j)) {
|
||||
assert(T[s] == c1);
|
||||
assert(((s + 1) < n) && (T[s] <= T[s + 1]));
|
||||
assert(T[s - 1] <= T[s]);
|
||||
|
||||
if ((s & mod) == 0) indexes[s / (mod + 1) - 1] = j - SA;
|
||||
|
||||
c0 = T[--s];
|
||||
*j = ~((int)c0);
|
||||
if((0 < s) && (T[s - 1] > c0)) { s = ~s; }
|
||||
if(c0 != c2) {
|
||||
if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; }
|
||||
k = SA + BUCKET_B(c2 = c0, c1);
|
||||
}
|
||||
assert(k < j);
|
||||
*k-- = s;
|
||||
} else if(s != 0) {
|
||||
*j = ~s;
|
||||
#ifndef NDEBUG
|
||||
} else {
|
||||
assert(T[s] == c1);
|
||||
#endif
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Construct the BWTed string by using
|
||||
the sorted order of type B suffixes. */
|
||||
k = SA + BUCKET_A(c2 = T[n - 1]);
|
||||
if (T[n - 2] < c2) {
|
||||
if (((n - 1) & mod) == 0) indexes[(n - 1) / (mod + 1) - 1] = k - SA;
|
||||
*k++ = ~((int)T[n - 2]);
|
||||
}
|
||||
else {
|
||||
*k++ = n - 1;
|
||||
}
|
||||
|
||||
/* Scan the suffix array from left to right. */
|
||||
for(i = SA, j = SA + n, orig = SA; i < j; ++i) {
|
||||
if(0 < (s = *i)) {
|
||||
assert(T[s - 1] >= T[s]);
|
||||
|
||||
if ((s & mod) == 0) indexes[s / (mod + 1) - 1] = i - SA;
|
||||
|
||||
c0 = T[--s];
|
||||
*i = c0;
|
||||
if(c0 != c2) {
|
||||
BUCKET_A(c2) = k - SA;
|
||||
k = SA + BUCKET_A(c2 = c0);
|
||||
}
|
||||
assert(i < k);
|
||||
if((0 < s) && (T[s - 1] < c0)) {
|
||||
if ((s & mod) == 0) indexes[s / (mod + 1) - 1] = k - SA;
|
||||
*k++ = ~((int)T[s - 1]);
|
||||
} else
|
||||
*k++ = s;
|
||||
} else if(s != 0) {
|
||||
*i = ~s;
|
||||
} else {
|
||||
orig = i;
|
||||
}
|
||||
}
|
||||
|
||||
return orig - SA;
|
||||
}
|
||||
|
||||
|
||||
/*---------------------------------------------------------------------------*/
|
||||
|
||||
/*- Function -*/
|
||||
|
||||
int
|
||||
divsufsort(const unsigned char *T, int *SA, int n, int openMP) {
|
||||
int *bucket_A, *bucket_B;
|
||||
int m;
|
||||
int err = 0;
|
||||
|
||||
/* Check arguments. */
|
||||
if((T == NULL) || (SA == NULL) || (n < 0)) { return -1; }
|
||||
else if(n == 0) { return 0; }
|
||||
else if(n == 1) { SA[0] = 0; return 0; }
|
||||
else if(n == 2) { m = (T[0] < T[1]); SA[m ^ 1] = 0, SA[m] = 1; return 0; }
|
||||
|
||||
bucket_A = (int *)malloc(BUCKET_A_SIZE * sizeof(int));
|
||||
bucket_B = (int *)malloc(BUCKET_B_SIZE * sizeof(int));
|
||||
|
||||
/* Suffixsort. */
|
||||
if((bucket_A != NULL) && (bucket_B != NULL)) {
|
||||
m = sort_typeBstar(T, SA, bucket_A, bucket_B, n, openMP);
|
||||
construct_SA(T, SA, bucket_A, bucket_B, n, m);
|
||||
} else {
|
||||
err = -2;
|
||||
}
|
||||
|
||||
free(bucket_B);
|
||||
free(bucket_A);
|
||||
|
||||
return err;
|
||||
}
|
||||
|
||||
int
|
||||
divbwt(const unsigned char *T, unsigned char *U, int *A, int n, unsigned char * num_indexes, int * indexes, int openMP) {
|
||||
int *B;
|
||||
int *bucket_A, *bucket_B;
|
||||
int m, pidx, i;
|
||||
|
||||
/* Check arguments. */
|
||||
if((T == NULL) || (U == NULL) || (n < 0)) { return -1; }
|
||||
else if(n <= 1) { if(n == 1) { U[0] = T[0]; } return n; }
|
||||
|
||||
if((B = A) == NULL) { B = (int *)malloc((size_t)(n + 1) * sizeof(int)); }
|
||||
bucket_A = (int *)malloc(BUCKET_A_SIZE * sizeof(int));
|
||||
bucket_B = (int *)malloc(BUCKET_B_SIZE * sizeof(int));
|
||||
|
||||
/* Burrows-Wheeler Transform. */
|
||||
if((B != NULL) && (bucket_A != NULL) && (bucket_B != NULL)) {
|
||||
m = sort_typeBstar(T, B, bucket_A, bucket_B, n, openMP);
|
||||
|
||||
if (num_indexes == NULL || indexes == NULL) {
|
||||
pidx = construct_BWT(T, B, bucket_A, bucket_B, n, m);
|
||||
} else {
|
||||
pidx = construct_BWT_indexes(T, B, bucket_A, bucket_B, n, m, num_indexes, indexes);
|
||||
}
|
||||
|
||||
/* Copy to output string. */
|
||||
U[0] = T[n - 1];
|
||||
for(i = 0; i < pidx; ++i) { U[i + 1] = (unsigned char)B[i]; }
|
||||
for(i += 1; i < n; ++i) { U[i] = (unsigned char)B[i]; }
|
||||
pidx += 1;
|
||||
} else {
|
||||
pidx = -2;
|
||||
}
|
||||
|
||||
free(bucket_B);
|
||||
free(bucket_A);
|
||||
if(A == NULL) { free(B); }
|
||||
|
||||
return pidx;
|
||||
}
|
||||
@@ -0,0 +1,67 @@
|
||||
/*
|
||||
* divsufsort.h for libdivsufsort-lite
|
||||
* Copyright (c) 2003-2008 Yuta Mori All Rights Reserved.
|
||||
*
|
||||
* Permission is hereby granted, free of charge, to any person
|
||||
* obtaining a copy of this software and associated documentation
|
||||
* files (the "Software"), to deal in the Software without
|
||||
* restriction, including without limitation the rights to use,
|
||||
* copy, modify, merge, publish, distribute, sublicense, and/or sell
|
||||
* copies of the Software, and to permit persons to whom the
|
||||
* Software is furnished to do so, subject to the following
|
||||
* conditions:
|
||||
*
|
||||
* The above copyright notice and this permission notice shall be
|
||||
* included in all copies or substantial portions of the Software.
|
||||
*
|
||||
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
||||
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
|
||||
* OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
||||
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
|
||||
* HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
|
||||
* WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
|
||||
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
|
||||
* OTHER DEALINGS IN THE SOFTWARE.
|
||||
*/
|
||||
|
||||
#ifndef _DIVSUFSORT_H
|
||||
#define _DIVSUFSORT_H 1
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif /* __cplusplus */
|
||||
|
||||
|
||||
/*- Prototypes -*/
|
||||
|
||||
/**
|
||||
* Constructs the suffix array of a given string.
|
||||
* @param T[0..n-1] The input string.
|
||||
* @param SA[0..n-1] The output array of suffixes.
|
||||
* @param n The length of the given string.
|
||||
* @param openMP enables OpenMP optimization.
|
||||
* @return 0 if no error occurred, -1 or -2 otherwise.
|
||||
*/
|
||||
int
|
||||
divsufsort(const unsigned char *T, int *SA, int n, int openMP);
|
||||
|
||||
/**
|
||||
* Constructs the burrows-wheeler transformed string of a given string.
|
||||
* @param T[0..n-1] The input string.
|
||||
* @param U[0..n-1] The output string. (can be T)
|
||||
* @param A[0..n-1] The temporary array. (can be NULL)
|
||||
* @param n The length of the given string.
|
||||
* @param num_indexes The length of secondary indexes array. (can be NULL)
|
||||
* @param indexes The secondary indexes array. (can be NULL)
|
||||
* @param openMP enables OpenMP optimization.
|
||||
* @return The primary index if no error occurred, -1 or -2 otherwise.
|
||||
*/
|
||||
int
|
||||
divbwt(const unsigned char *T, unsigned char *U, int *A, int n, unsigned char * num_indexes, int * indexes, int openMP);
|
||||
|
||||
|
||||
#ifdef __cplusplus
|
||||
} /* extern "C" */
|
||||
#endif /* __cplusplus */
|
||||
|
||||
#endif /* _DIVSUFSORT_H */
|
||||
@@ -119,11 +119,12 @@ MEM_STATIC unsigned MEM_isLittleEndian(void)
|
||||
|
||||
#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)
|
||||
|
||||
/* violates C standard on structure alignment.
|
||||
/* violates C standard, by lying on structure alignment.
|
||||
Only use if no other choice to achieve best performance on target platform */
|
||||
MEM_STATIC U16 MEM_read16(const void* memPtr) { return *(const U16*) memPtr; }
|
||||
MEM_STATIC U32 MEM_read32(const void* memPtr) { return *(const U32*) memPtr; }
|
||||
MEM_STATIC U64 MEM_read64(const void* memPtr) { return *(const U64*) memPtr; }
|
||||
MEM_STATIC U64 MEM_readST(const void* memPtr) { return *(const size_t*) memPtr; }
|
||||
|
||||
MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; }
|
||||
MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; }
|
||||
@@ -133,11 +134,12 @@ MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; }
|
||||
|
||||
/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
|
||||
/* currently only defined for gcc and icc */
|
||||
typedef union { U16 u16; U32 u32; U64 u64; } __attribute__((packed)) unalign;
|
||||
typedef union { U16 u16; U32 u32; U64 u64; size_t st; } __attribute__((packed)) unalign;
|
||||
|
||||
MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign*)ptr)->u16; }
|
||||
MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign*)ptr)->u32; }
|
||||
MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign*)ptr)->u64; }
|
||||
MEM_STATIC U64 MEM_readST(const void* ptr) { return ((const unalign*)ptr)->st; }
|
||||
|
||||
MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign*)memPtr)->u16 = value; }
|
||||
MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign*)memPtr)->u32 = value; }
|
||||
@@ -163,6 +165,11 @@ MEM_STATIC U64 MEM_read64(const void* memPtr)
|
||||
U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
|
||||
}
|
||||
|
||||
MEM_STATIC size_t MEM_readST(const void* memPtr)
|
||||
{
|
||||
size_t val; memcpy(&val, memPtr, sizeof(val)); return val;
|
||||
}
|
||||
|
||||
MEM_STATIC void MEM_write16(void* memPtr, U16 value)
|
||||
{
|
||||
memcpy(memPtr, &value, sizeof(value));
|
||||
|
||||
+1
-1
@@ -61,7 +61,7 @@ extern "C" {
|
||||
***************************************/
|
||||
#define ZSTD_VERSION_MAJOR 0 /* for breaking interface changes */
|
||||
#define ZSTD_VERSION_MINOR 5 /* for new (non-breaking) interface capabilities */
|
||||
#define ZSTD_VERSION_RELEASE 0 /* for tweaks, bug-fixes, or development */
|
||||
#define ZSTD_VERSION_RELEASE 1 /* for tweaks, bug-fixes, or development */
|
||||
#define ZSTD_VERSION_NUMBER (ZSTD_VERSION_MAJOR *100*100 + ZSTD_VERSION_MINOR *100 + ZSTD_VERSION_RELEASE)
|
||||
ZSTDLIB_API unsigned ZSTD_versionNumber (void);
|
||||
|
||||
|
||||
+16
-62
@@ -48,7 +48,7 @@
|
||||
#endif
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Dependencies
|
||||
***************************************/
|
||||
#include <stdlib.h> /* malloc */
|
||||
@@ -59,46 +59,21 @@
|
||||
#include "zstd_internal.h"
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Constants
|
||||
***************************************/
|
||||
static const U32 g_searchStrength = 8;
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Helper functions
|
||||
***************************************/
|
||||
size_t ZSTD_compressBound(size_t srcSize) { return FSE_compressBound(srcSize) + 12; }
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Sequence storage
|
||||
***************************************/
|
||||
typedef struct {
|
||||
void* buffer;
|
||||
U32* offsetStart;
|
||||
U32* offset;
|
||||
BYTE* offCodeStart;
|
||||
BYTE* offCode;
|
||||
BYTE* litStart;
|
||||
BYTE* lit;
|
||||
BYTE* litLengthStart;
|
||||
BYTE* litLength;
|
||||
BYTE* matchLengthStart;
|
||||
BYTE* matchLength;
|
||||
BYTE* dumpsStart;
|
||||
BYTE* dumps;
|
||||
/* opt */
|
||||
U32* matchLengthFreq;
|
||||
U32* litLengthFreq;
|
||||
U32* litFreq;
|
||||
U32* offCodeFreq;
|
||||
U32 matchLengthSum;
|
||||
U32 litLengthSum;
|
||||
U32 litSum;
|
||||
U32 offCodeSum;
|
||||
} seqStore_t;
|
||||
|
||||
static void ZSTD_resetFreqs(seqStore_t* ssPtr)
|
||||
{
|
||||
unsigned u;
|
||||
@@ -129,7 +104,7 @@ static void ZSTD_resetSeqStore(seqStore_t* ssPtr)
|
||||
}
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Context memory management
|
||||
***************************************/
|
||||
struct ZSTD_CCtx_s
|
||||
@@ -159,7 +134,6 @@ struct ZSTD_CCtx_s
|
||||
FSE_CTable litlengthCTable [FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
|
||||
};
|
||||
|
||||
|
||||
ZSTD_CCtx* ZSTD_createCCtx(void)
|
||||
{
|
||||
return (ZSTD_CCtx*) calloc(1, sizeof(ZSTD_CCtx));
|
||||
@@ -172,14 +146,19 @@ size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx)
|
||||
return 0; /* reserved as a potential error code in the future */
|
||||
}
|
||||
|
||||
seqStore_t ZSTD_copySeqStore(const ZSTD_CCtx* ctx)
|
||||
{
|
||||
return ctx->seqStore;
|
||||
}
|
||||
|
||||
|
||||
static unsigned ZSTD_highbit(U32 val);
|
||||
|
||||
#define CLAMP(val,min,max) { if (val<min) val=min; else if (val>max) val=max; }
|
||||
|
||||
/** ZSTD_validateParams()
|
||||
correct params value to remain within authorized range
|
||||
optimize for srcSize if srcSize > 0 */
|
||||
/** ZSTD_validateParams() :
|
||||
correct params value to remain within authorized range,
|
||||
optimize for `srcSize` if srcSize > 0 */
|
||||
void ZSTD_validateParams(ZSTD_parameters* params)
|
||||
{
|
||||
const U32 btPlus = (params->strategy == ZSTD_btlazy2) || (params->strategy == ZSTD_opt_bt);
|
||||
@@ -800,33 +779,9 @@ MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const B
|
||||
}
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Match length counter
|
||||
***************************************/
|
||||
static size_t ZSTD_read_ARCH(const void* p) { size_t r; memcpy(&r, p, sizeof(r)); return r; }
|
||||
|
||||
static unsigned ZSTD_highbit(U32 val)
|
||||
{
|
||||
# if defined(_MSC_VER) /* Visual */
|
||||
unsigned long r=0;
|
||||
_BitScanReverse(&r, val);
|
||||
return (unsigned)r;
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3) /* GCC Intrinsic */
|
||||
return 31 - __builtin_clz(val);
|
||||
# else /* Software version */
|
||||
static const int DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
|
||||
U32 v = val;
|
||||
int r;
|
||||
v |= v >> 1;
|
||||
v |= v >> 2;
|
||||
v |= v >> 4;
|
||||
v |= v >> 8;
|
||||
v |= v >> 16;
|
||||
r = DeBruijnClz[(U32)(v * 0x07C4ACDDU) >> 27];
|
||||
return r;
|
||||
# endif
|
||||
}
|
||||
|
||||
static unsigned ZSTD_NbCommonBytes (register size_t val)
|
||||
{
|
||||
if (MEM_isLittleEndian()) {
|
||||
@@ -891,12 +846,11 @@ static size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* pInLim
|
||||
const BYTE* const pStart = pIn;
|
||||
|
||||
while ((pIn<pInLimit-(sizeof(size_t)-1))) {
|
||||
size_t diff = ZSTD_read_ARCH(pMatch) ^ ZSTD_read_ARCH(pIn);
|
||||
size_t diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
|
||||
if (!diff) { pIn+=sizeof(size_t); pMatch+=sizeof(size_t); continue; }
|
||||
pIn += ZSTD_NbCommonBytes(diff);
|
||||
return (size_t)(pIn - pStart);
|
||||
}
|
||||
|
||||
if (MEM_64bits()) if ((pIn<(pInLimit-3)) && (MEM_read32(pMatch) == MEM_read32(pIn))) { pIn+=4; pMatch+=4; }
|
||||
if ((pIn<(pInLimit-1)) && (MEM_read16(pMatch) == MEM_read16(pIn))) { pIn+=2; pMatch+=2; }
|
||||
if ((pIn<pInLimit) && (*pMatch == *pIn)) pIn++;
|
||||
@@ -904,7 +858,7 @@ static size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* pInLim
|
||||
}
|
||||
|
||||
/** ZSTD_count_2segments() :
|
||||
* can count match length with ip & match in potentially 2 different segments.
|
||||
* can count match length with `ip` & `match` in 2 different segments.
|
||||
* convention : on reaching mEnd, match count continue starting from iStart
|
||||
*/
|
||||
static size_t ZSTD_count_2segments(const BYTE* ip, const BYTE* match, const BYTE* iEnd, const BYTE* mEnd, const BYTE* iStart)
|
||||
|
||||
+55
-10
@@ -32,11 +32,7 @@
|
||||
#ifndef ZSTD_CCOMMON_H_MODULE
|
||||
#define ZSTD_CCOMMON_H_MODULE
|
||||
|
||||
#if defined (__cplusplus)
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Dependencies
|
||||
***************************************/
|
||||
#include "mem.h"
|
||||
@@ -44,14 +40,14 @@ extern "C" {
|
||||
#include "zstd_static.h"
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Common macros
|
||||
***************************************/
|
||||
#define MIN(a,b) ((a)<(b) ? (a) : (b))
|
||||
#define MAX(a,b) ((a)>(b) ? (a) : (b))
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Common constants
|
||||
***************************************/
|
||||
#define ZSTD_MAGICNUMBER 0xFD2FB525 /* v0.5 */
|
||||
@@ -130,9 +126,58 @@ MEM_STATIC void ZSTD_wildcopy(void* dst, const void* src, size_t length)
|
||||
while (op < oend);
|
||||
}
|
||||
|
||||
|
||||
#if defined (__cplusplus)
|
||||
MEM_STATIC unsigned ZSTD_highbit(U32 val)
|
||||
{
|
||||
# if defined(_MSC_VER) /* Visual */
|
||||
unsigned long r=0;
|
||||
_BitScanReverse(&r, val);
|
||||
return (unsigned)r;
|
||||
# elif defined(__GNUC__) && (__GNUC__ >= 3) /* GCC Intrinsic */
|
||||
return 31 - __builtin_clz(val);
|
||||
# else /* Software version */
|
||||
static const int DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
|
||||
U32 v = val;
|
||||
int r;
|
||||
v |= v >> 1;
|
||||
v |= v >> 2;
|
||||
v |= v >> 4;
|
||||
v |= v >> 8;
|
||||
v |= v >> 16;
|
||||
r = DeBruijnClz[(U32)(v * 0x07C4ACDDU) >> 27];
|
||||
return r;
|
||||
# endif
|
||||
}
|
||||
#endif
|
||||
|
||||
|
||||
/*-*******************************************
|
||||
* Private interfaces
|
||||
*********************************************/
|
||||
typedef struct {
|
||||
void* buffer;
|
||||
U32* offsetStart;
|
||||
U32* offset;
|
||||
BYTE* offCodeStart;
|
||||
BYTE* offCode;
|
||||
BYTE* litStart;
|
||||
BYTE* lit;
|
||||
BYTE* litLengthStart;
|
||||
BYTE* litLength;
|
||||
BYTE* matchLengthStart;
|
||||
BYTE* matchLength;
|
||||
BYTE* dumpsStart;
|
||||
BYTE* dumps;
|
||||
/* opt */
|
||||
U32* matchLengthFreq;
|
||||
U32* litLengthFreq;
|
||||
U32* litFreq;
|
||||
U32* offCodeFreq;
|
||||
U32 matchLengthSum;
|
||||
U32 litLengthSum;
|
||||
U32 litSum;
|
||||
U32 offCodeSum;
|
||||
} seqStore_t;
|
||||
|
||||
seqStore_t ZSTD_copySeqStore(const ZSTD_CCtx* ctx);
|
||||
|
||||
|
||||
#endif /* ZSTD_CCOMMON_H_MODULE */
|
||||
|
||||
+3
-4
@@ -205,7 +205,7 @@ ZSTDLIB_API size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t ds
|
||||
* Block functions
|
||||
****************************************/
|
||||
/*! Block functions produce and decode raw zstd blocks, without frame metadata.
|
||||
User will have to save and regenerate necessary information to regenerate data, such as block sizes.
|
||||
User will have to take in charge required information to regenerate data, such as block sizes.
|
||||
|
||||
A few rules to respect :
|
||||
- Uncompressed block size must be <= 128 KB
|
||||
@@ -222,18 +222,17 @@ ZSTDLIB_API size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t ds
|
||||
+ ZSTD_decompressBlock() doesn't accept uncompressed data as input !!
|
||||
*/
|
||||
|
||||
|
||||
size_t ZSTD_compressBlock (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
|
||||
size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
|
||||
|
||||
|
||||
/* *************************************
|
||||
/*-*************************************
|
||||
* Error management
|
||||
***************************************/
|
||||
#include "error_public.h"
|
||||
/*! ZSTD_getErrorCode() :
|
||||
convert a `size_t` function result into a `ZSTD_error_code` enum type,
|
||||
which can be used to compare directly with enum list within "error_public.h" */
|
||||
which can be used to compare directly with enum list published into "error_public.h" */
|
||||
ZSTD_ErrorCode ZSTD_getError(size_t code);
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user