facebook
/
zstd
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Refactor RANDOM

dev
Jennifer Liu 2018-07-16 16:03:04 -07:00
parent 0e5fbc10fa
commit b5806d33db
4 changed files with 314 additions and 361 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
contrib/randomDictBuilder/Makefile
View File

@ -14,14 +14,14 @@ rand:
./main $(ARG) ./main $(ARG)
main: random.o main.o libzstd.a main: main.o random.o libzstd.a
gcc random.o main.o libzstd.a -o main gcc main.o random.o libzstd.a -o main
main.o: main.c main.o: main.c $(PROGRAM_FILES)
gcc -c main.c -I ../../programs -I ../../lib/common -I ../../lib -I ../../lib/dictBuilder -I random.h gcc -c main.c $(PROGRAM_FILES) -I random.h -I ../../programs -I ../../lib/common -I ../../lib -I ../../lib/dictBuilder
random.o: $(PROGRAM_FILES) random.c random.o: random.c
gcc -c $(PROGRAM_FILES) -I ../../programs -I ../../lib/common -I random.h random.c gcc -c random.c -I random.h -I ../../lib/common -I ../../lib/dictBuilder
libzstd.a: libzstd.a:
$(MAKE) -C ../../lib libzstd.a $(MAKE) -C ../../lib libzstd.a

297
contrib/randomDictBuilder/main.c
View File

@ -3,13 +3,45 @@
#include <string.h> /* strcmp, strlen */ #include <string.h> /* strcmp, strlen */
#include <errno.h> /* errno */ #include <errno.h> /* errno */
#include <ctype.h> #include <ctype.h>
#include "fileio.h" /* stdinmark, stdoutmark, ZSTD_EXTENSION */
#include "random.h" #include "random.h"
#include "fileio.h" /* stdinmark, stdoutmark, ZSTD_EXTENSION */
#include "platform.h" /* Large Files support */
#include "util.h" #include "util.h"
#include "zdict.h"
/*-*************************************
* Console display
***************************************/
#define DISPLAY(...) fprintf(stderr, __VA_ARGS__) #define DISPLAY(...) fprintf(stderr, __VA_ARGS__)
#define DISPLAYLEVEL(l, ...) if (displayLevel>=l) { DISPLAY(__VA_ARGS__); } #define DISPLAYLEVEL(l, ...) if (displayLevel>=l) { DISPLAY(__VA_ARGS__); }
static const U64 g_refreshRate = SEC_TO_MICRO / 6;
static UTIL_time_t g_displayClock = UTIL_TIME_INITIALIZER;
#define DISPLAYUPDATE(l, ...) { if (displayLevel>=l) { \
if ((UTIL_clockSpanMicro(g_displayClock) > g_refreshRate) || (displayLevel>=4)) \
{ g_displayClock = UTIL_getTime(); DISPLAY(__VA_ARGS__); \
if (displayLevel>=4) fflush(stderr); } } }
/*-*************************************
* 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__); \
DISPLAY("Error %i : ", error); \
DISPLAY(__VA_ARGS__); \
DISPLAY("\n"); \
exit(error); \
}
/*-*************************************
* Constants
***************************************/
static const unsigned g_defaultMaxDictSize = 110 KB; static const unsigned g_defaultMaxDictSize = 110 KB;
#define DEFAULT_CLEVEL 3 #define DEFAULT_CLEVEL 3
#define DEFAULT_INPUTFILE "" #define DEFAULT_INPUTFILE ""
@ -17,7 +49,33 @@ static const unsigned g_defaultMaxDictSize = 110 KB;
#define DEFAULT_OUTPUTFILE "defaultDict" #define DEFAULT_OUTPUTFILE "defaultDict"
#define DEFAULT_DICTID 0 #define DEFAULT_DICTID 0
#define SAMPLESIZE_MAX (128 KB)
#define RANDOM_MAX_SAMPLES_SIZE (sizeof(size_t) == 8 ? ((U32)-1) : ((U32)1 GB))
#define RANDOM_MEMMULT 9
static const size_t g_maxMemory = (sizeof(size_t) == 4) ? (2 GB - 64 MB) : ((size_t)(512 MB) << sizeof(size_t));
#define NOISELENGTH 32
/*-*************************************
* Structs
***************************************/
typedef struct {
U64 totalSizeToLoad;
unsigned oneSampleTooLarge;
unsigned nbSamples;
} fileStats;
typedef struct {
const void* srcBuffer;
const size_t *samplesSizes;
size_t nbSamples;
}sampleInfo;
/*-*************************************
* Commandline related functions
***************************************/
static unsigned readU32FromChar(const char** stringPtr) static unsigned readU32FromChar(const char** stringPtr)
{ {
const char errorMsg[] = "error: numeric value too large"; const char errorMsg[] = "error: numeric value too large";
@ -42,7 +100,6 @@ static unsigned readU32FromChar(const char** stringPtr)
return result; return result;
} }
/** longCommandWArg() : /** longCommandWArg() :
* check if *stringPtr is the same as longCommand. * check if *stringPtr is the same as longCommand.
* If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand. * If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand.
@ -56,6 +113,225 @@ static unsigned longCommandWArg(const char** stringPtr, const char* longCommand)
return result; return result;
} }
/* ********************************************************
* File related operations
**********************************************************/
/** loadFiles() :
* load samples from files listed in fileNamesTable into buffer.
* works even if buffer is too small to load all samples.
* Also provides the size of each sample into sampleSizes table
* which must be sized correctly, using DiB_fileStats().
* @return : nb of samples effectively loaded into `buffer`
* *bufferSizePtr is modified, it provides the amount data loaded within buffer.
* sampleSizes is filled with the size of each sample.
*/
static unsigned loadFiles(void* buffer, size_t* bufferSizePtr,
size_t* sampleSizes, unsigned sstSize,
const char** fileNamesTable, unsigned nbFiles, size_t targetChunkSize,
unsigned displayLevel)
{
char* const buff = (char*)buffer;
size_t pos = 0;
unsigned nbLoadedChunks = 0, fileIndex;
for (fileIndex=0; fileIndex<nbFiles; fileIndex++) {
const char* const fileName = fileNamesTable[fileIndex];
unsigned long long const fs64 = UTIL_getFileSize(fileName);
unsigned long long remainingToLoad = (fs64 == UTIL_FILESIZE_UNKNOWN) ? 0 : fs64;
U32 const nbChunks = targetChunkSize ? (U32)((fs64 + (targetChunkSize-1)) / targetChunkSize) : 1;
U64 const chunkSize = targetChunkSize ? MIN(targetChunkSize, fs64) : fs64;
size_t const maxChunkSize = (size_t)MIN(chunkSize, SAMPLESIZE_MAX);
U32 cnb;
FILE* const f = fopen(fileName, "rb");
if (f==NULL) EXM_THROW(10, "zstd: dictBuilder: %s %s ", fileName, strerror(errno));
DISPLAYUPDATE(2, "Loading %s... \r", fileName);
for (cnb=0; cnb<nbChunks; cnb++) {
size_t const toLoad = (size_t)MIN(maxChunkSize, remainingToLoad);
if (toLoad > *bufferSizePtr-pos) break;
{ size_t const readSize = fread(buff+pos, 1, toLoad, f);
if (readSize != toLoad) EXM_THROW(11, "Pb reading %s", fileName);
pos += readSize;
sampleSizes[nbLoadedChunks++] = toLoad;
remainingToLoad -= targetChunkSize;
if (nbLoadedChunks == sstSize) { /* no more space left in sampleSizes table */
fileIndex = nbFiles; /* stop there */
break;
}
if (toLoad < targetChunkSize) {
fseek(f, (long)(targetChunkSize - toLoad), SEEK_CUR);
} } }
fclose(f);
}
DISPLAYLEVEL(2, "\r%79s\r", "");
*bufferSizePtr = pos;
DISPLAYLEVEL(4, "loaded : %u KB \n", (U32)(pos >> 10))
return nbLoadedChunks;
}
#define rotl32(x,r) ((x << r) | (x >> (32 - r)))
static U32 getRand(U32* src)
{
static const U32 prime1 = 2654435761U;
static const U32 prime2 = 2246822519U;
U32 rand32 = *src;
rand32 *= prime1;
rand32 ^= prime2;
rand32 = rotl32(rand32, 13);
*src = rand32;
return rand32 >> 5;
}
/* shuffle() :
* shuffle a table of file names in a semi-random way
* It improves dictionary quality by reducing "locality" impact, so if sample set is very large,
* it will load random elements from it, instead of just the first ones. */
static void shuffle(const char** fileNamesTable, unsigned nbFiles) {
U32 seed = 0xFD2FB528;
unsigned i;
for (i = nbFiles - 1; i > 0; --i) {
unsigned const j = getRand(&seed) % (i + 1);
const char* const tmp = fileNamesTable[j];
fileNamesTable[j] = fileNamesTable[i];
fileNamesTable[i] = tmp;
}
}
/*-********************************************************
* Dictionary training functions
**********************************************************/
static size_t findMaxMem(unsigned long long requiredMem)
{
size_t const step = 8 MB;
void* testmem = NULL;
requiredMem = (((requiredMem >> 23) + 1) << 23);
requiredMem += step;
if (requiredMem > g_maxMemory) requiredMem = g_maxMemory;
while (!testmem) {
testmem = malloc((size_t)requiredMem);
requiredMem -= step;
}
free(testmem);
return (size_t)requiredMem;
}
static void saveDict(const char* dictFileName,
const void* buff, size_t buffSize)
{
FILE* const f = fopen(dictFileName, "wb");
if (f==NULL) EXM_THROW(3, "cannot open %s ", dictFileName);
{ size_t const n = fwrite(buff, 1, buffSize, f);
if (n!=buffSize) EXM_THROW(4, "%s : write error", dictFileName) }
{ size_t const n = (size_t)fclose(f);
if (n!=0) EXM_THROW(5, "%s : flush error", dictFileName) }
}
/*! getFileStats() :
* Given a list of files, and a chunkSize (0 == no chunk, whole files)
* provides the amount of data to be loaded and the resulting nb of samples.
* This is useful primarily for allocation purpose => sample buffer, and sample sizes table.
*/
static fileStats getFileStats(const char** fileNamesTable, unsigned nbFiles, size_t chunkSize, unsigned displayLevel)
{
fileStats fs;
unsigned n;
memset(&fs, 0, sizeof(fs));
for (n=0; n<nbFiles; n++) {
U64 const fileSize = UTIL_getFileSize(fileNamesTable[n]);
U64 const srcSize = (fileSize == UTIL_FILESIZE_UNKNOWN) ? 0 : fileSize;
U32 const nbSamples = (U32)(chunkSize ? (srcSize + (chunkSize-1)) / chunkSize : 1);
U64 const chunkToLoad = chunkSize ? MIN(chunkSize, srcSize) : srcSize;
size_t const cappedChunkSize = (size_t)MIN(chunkToLoad, SAMPLESIZE_MAX);
fs.totalSizeToLoad += cappedChunkSize * nbSamples;
fs.oneSampleTooLarge |= (chunkSize > 2*SAMPLESIZE_MAX);
fs.nbSamples += nbSamples;
}
DISPLAYLEVEL(4, "Preparing to load : %u KB \n", (U32)(fs.totalSizeToLoad >> 10));
return fs;
}
int RANDOM_trainFromFiles(const char* dictFileName, sampleInfo *info, unsigned maxDictSize,
ZDICT_random_params_t *params){
unsigned const displayLevel = params->zParams.notificationLevel;
void* const dictBuffer = malloc(maxDictSize);
int result = 0;
/* Checks */
if (!dictBuffer)
EXM_THROW(12, "not enough memory for trainFromFiles"); /* should not happen */
{ size_t dictSize;
dictSize = ZDICT_trainFromBuffer_random(dictBuffer, maxDictSize, info->srcBuffer,
info->samplesSizes, info->nbSamples, *params);
DISPLAYLEVEL(2, "k=%u\n", params->k);
if (ZDICT_isError(dictSize)) {
DISPLAYLEVEL(1, "dictionary training failed : %s \n", ZDICT_getErrorName(dictSize)); /* should not happen */
result = 1;
free(dictBuffer);
}
/* save dict */
DISPLAYLEVEL(2, "Save dictionary of size %u into file %s \n", (U32)dictSize, dictFileName);
saveDict(dictFileName, dictBuffer, dictSize);
}
/* clean up */
free(dictBuffer);
return result;
}
sampleInfo* getSampleInfo(const char** fileNamesTable,
unsigned nbFiles, size_t chunkSize, unsigned maxDictSize, const unsigned displayLevel){
fileStats const fs = getFileStats(fileNamesTable, nbFiles, chunkSize, displayLevel);
size_t* const sampleSizes = (size_t*)malloc(fs.nbSamples * sizeof(size_t));
size_t const memMult = RANDOM_MEMMULT;
size_t const maxMem = findMaxMem(fs.totalSizeToLoad * memMult) / memMult;
size_t loadedSize = (size_t) MIN ((unsigned long long)maxMem, fs.totalSizeToLoad);
void* const srcBuffer = malloc(loadedSize+NOISELENGTH);
/* Checks */
if ((!sampleSizes) || (!srcBuffer))
EXM_THROW(12, "not enough memory for trainFromFiles"); /* should not happen */
if (fs.oneSampleTooLarge) {
DISPLAYLEVEL(2, "! Warning : some sample(s) are very large \n");
DISPLAYLEVEL(2, "! Note that dictionary is only useful for small samples. \n");
DISPLAYLEVEL(2, "! As a consequence, only the first %u bytes of each sample are loaded \n", SAMPLESIZE_MAX);
}
if (fs.nbSamples < 5) {
DISPLAYLEVEL(2, "! Warning : nb of samples too low for proper processing ! \n");
DISPLAYLEVEL(2, "! Please provide _one file per sample_. \n");
DISPLAYLEVEL(2, "! Alternatively, split files into fixed-size blocks representative of samples, with -B# \n");
EXM_THROW(14, "nb of samples too low"); /* we now clearly forbid this case */
}
if (fs.totalSizeToLoad < (unsigned long long)(8 * maxDictSize)) {
DISPLAYLEVEL(2, "! Warning : data size of samples too small for target dictionary size \n");
DISPLAYLEVEL(2, "! Samples should be about 100x larger than target dictionary size \n");
}
/* init */
if (loadedSize < fs.totalSizeToLoad)
DISPLAYLEVEL(1, "Not enough memory; training on %u MB only...\n", (unsigned)(loadedSize >> 20));
/* Load input buffer */
DISPLAYLEVEL(3, "Shuffling input files\n");
shuffle(fileNamesTable, nbFiles);
nbFiles = loadFiles(srcBuffer, &loadedSize, sampleSizes, fs.nbSamples, fileNamesTable, nbFiles, chunkSize, displayLevel);
sampleInfo *info = (sampleInfo *)malloc(sizeof(sampleInfo));
info->nbSamples = fs.nbSamples;
info->samplesSizes = sampleSizes;
info->srcBuffer = srcBuffer;
return info;
}
int main(int argCount, const char* argv[]) int main(int argCount, const char* argv[])
{ {
@ -63,7 +339,7 @@ int main(int argCount, const char* argv[])
const char* programName = argv[0]; const char* programName = argv[0];
int operationResult = 0; int operationResult = 0;
/* Initialize parameters with default value */ unsigned cLevel = DEFAULT_CLEVEL;
char* inputFile = DEFAULT_INPUTFILE; char* inputFile = DEFAULT_INPUTFILE;
unsigned k = DEFAULT_k; unsigned k = DEFAULT_k;
char* outputFile = DEFAULT_OUTPUTFILE; char* outputFile = DEFAULT_OUTPUTFILE;
@ -76,10 +352,10 @@ int main(int argCount, const char* argv[])
for (int i = 1; i < argCount; i++) { for (int i = 1; i < argCount; i++) {
const char* argument = argv[i]; const char* argument = argv[i];
if (longCommandWArg(&argument, "k=")) { k = readU32FromChar(&argument); continue; } if (longCommandWArg(&argument, "k=")) { k = readU32FromChar(&argument); continue; }
if (longCommandWArg(&argument, "c=")) { cLevel = readU32FromChar(&argument); continue; }
if (longCommandWArg(&argument, "dictID=")) { dictID = readU32FromChar(&argument); continue; } if (longCommandWArg(&argument, "dictID=")) { dictID = readU32FromChar(&argument); continue; }
if (longCommandWArg(&argument, "maxdict=")) { maxDictSize = readU32FromChar(&argument); continue; } if (longCommandWArg(&argument, "maxdict=")) { maxDictSize = readU32FromChar(&argument); continue; }
if (longCommandWArg(&argument, "in=")) { if (longCommandWArg(&argument, "in=")) {
/* Allow multiple input files */
inputFile = malloc(strlen(argument) + 1); inputFile = malloc(strlen(argument) + 1);
strcpy(inputFile, argument); strcpy(inputFile, argument);
filenameTable[filenameIdx] = inputFile; filenameTable[filenameIdx] = inputFile;
@ -96,12 +372,6 @@ int main(int argCount, const char* argv[])
return operationResult; return operationResult;
} }
if (maxDictSize == 0) {
DISPLAYLEVEL(1, "maxDictSize should not be 0.\n");
operationResult = 1;
return operationResult;
}
char* fileNamesBuf = NULL; char* fileNamesBuf = NULL;
unsigned fileNamesNb = filenameIdx; unsigned fileNamesNb = filenameIdx;
int followLinks = 0; int followLinks = 0;
@ -119,12 +389,15 @@ int main(int argCount, const char* argv[])
ZDICT_random_params_t params; ZDICT_random_params_t params;
ZDICT_params_t zParams; ZDICT_params_t zParams;
zParams.compressionLevel = DEFAULT_CLEVEL; zParams.compressionLevel = cLevel;
zParams.notificationLevel = displayLevel; zParams.notificationLevel = displayLevel;
zParams.dictID = dictID; zParams.dictID = dictID;
params.zParams = zParams; params.zParams = zParams;
params.k = k; params.k = k;
operationResult = RANDOM_trainFromFiles(outputFile, maxDictSize, filenameTable, filenameIdx, blockSize, &params); sampleInfo* info= getSampleInfo(filenameTable,
filenameIdx, blockSize, maxDictSize, zParams.notificationLevel);
operationResult = RANDOM_trainFromFiles(outputFile, info, maxDictSize, &params);
return operationResult; return operationResult;
} }

343
contrib/randomDictBuilder/random.c
View File

@ -5,24 +5,12 @@
#include <stdlib.h> /* malloc, free, qsort */ #include <stdlib.h> /* malloc, free, qsort */
#include <string.h> /* memset */ #include <string.h> /* memset */
#include <time.h> /* clock */ #include <time.h> /* clock */
#include "zstd_internal.h" /* includes zstd.h */ #include "random.h"
#include "util.h" /* UTIL_getFileSize, UTIL_getTotalFileSize */
#ifndef ZDICT_STATIC_LINKING_ONLY #ifndef ZDICT_STATIC_LINKING_ONLY
#define ZDICT_STATIC_LINKING_ONLY #define ZDICT_STATIC_LINKING_ONLY
#endif #endif
#include "random.h" #include "zdict.h"
#include "platform.h" /* Large Files support */
#include "util.h" /* UTIL_getFileSize, UTIL_getTotalFileSize */
/*-*************************************
* Constants
***************************************/
#define SAMPLESIZE_MAX (128 KB)
#define RANDOM_MAX_SAMPLES_SIZE (sizeof(size_t) == 8 ? ((U32)-1) : ((U32)1 GB))
#define RANDOM_MEMMULT 9
static const size_t g_maxMemory = (sizeof(size_t) == 4) ? (2 GB - 64 MB) : ((size_t)(512 MB) << sizeof(size_t));
#define NOISELENGTH 32
#define DEFAULT_K 200
/*-************************************* /*-*************************************
* Console display * Console display
@ -30,179 +18,16 @@ static const size_t g_maxMemory = (sizeof(size_t) == 4) ? (2 GB - 64 MB) : ((siz
#define DISPLAY(...) fprintf(stderr, __VA_ARGS__) #define DISPLAY(...) fprintf(stderr, __VA_ARGS__)
#define DISPLAYLEVEL(l, ...) if (displayLevel>=l) { DISPLAY(__VA_ARGS__); } #define DISPLAYLEVEL(l, ...) if (displayLevel>=l) { DISPLAY(__VA_ARGS__); }
static const U64 g_refreshRate = SEC_TO_MICRO / 6; #define LOCALDISPLAYUPDATE(displayLevel, l, ...) \
static UTIL_time_t g_displayClock = UTIL_TIME_INITIALIZER; if (displayLevel >= l) { \
if ((clock() - g_time > refreshRate) || (displayLevel >= 4)) { \
#define DISPLAYUPDATE(l, ...) { if (displayLevel>=l) { \ g_time = clock(); \
if ((UTIL_clockSpanMicro(g_displayClock) > g_refreshRate) || (displayLevel>=4)) \ DISPLAY(__VA_ARGS__); \
{ g_displayClock = UTIL_getTime(); DISPLAY(__VA_ARGS__); \ } \
if (displayLevel>=4) fflush(stderr); } } } }
#define DISPLAYUPDATE(l, ...) LOCALDISPLAYUPDATE(displayLevel, l, __VA_ARGS__)
static const clock_t refreshRate = CLOCKS_PER_SEC * 15 / 100;
/*-************************************* static clock_t g_time = 0;
* 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__); \
DISPLAY("Error %i : ", error); \
DISPLAY(__VA_ARGS__); \
DISPLAY("\n"); \
exit(error); \
}
/* ********************************************************
* File related operations
**********************************************************/
/** loadFiles() :
* load samples from files listed in fileNamesTable into buffer.
* works even if buffer is too small to load all samples.
* Also provides the size of each sample into sampleSizes table
* which must be sized correctly, using DiB_fileStats().
* @return : nb of samples effectively loaded into `buffer`
* *bufferSizePtr is modified, it provides the amount data loaded within buffer.
* sampleSizes is filled with the size of each sample.
*/
static unsigned loadFiles(void* buffer, size_t* bufferSizePtr,
size_t* sampleSizes, unsigned sstSize,
const char** fileNamesTable, unsigned nbFiles, size_t targetChunkSize,
unsigned displayLevel)
{
char* const buff = (char*)buffer;
size_t pos = 0;
unsigned nbLoadedChunks = 0, fileIndex;
for (fileIndex=0; fileIndex<nbFiles; fileIndex++) {
const char* const fileName = fileNamesTable[fileIndex];
unsigned long long const fs64 = UTIL_getFileSize(fileName);
unsigned long long remainingToLoad = (fs64 == UTIL_FILESIZE_UNKNOWN) ? 0 : fs64;
U32 const nbChunks = targetChunkSize ? (U32)((fs64 + (targetChunkSize-1)) / targetChunkSize) : 1;
U64 const chunkSize = targetChunkSize ? MIN(targetChunkSize, fs64) : fs64;
size_t const maxChunkSize = (size_t)MIN(chunkSize, SAMPLESIZE_MAX);
U32 cnb;
FILE* const f = fopen(fileName, "rb");
if (f==NULL) EXM_THROW(10, "zstd: dictBuilder: %s %s ", fileName, strerror(errno));
DISPLAYUPDATE(2, "Loading %s... \r", fileName);
for (cnb=0; cnb<nbChunks; cnb++) {
size_t const toLoad = (size_t)MIN(maxChunkSize, remainingToLoad);
if (toLoad > *bufferSizePtr-pos) break;
{ size_t const readSize = fread(buff+pos, 1, toLoad, f);
if (readSize != toLoad) EXM_THROW(11, "Pb reading %s", fileName);
pos += readSize;
sampleSizes[nbLoadedChunks++] = toLoad;
remainingToLoad -= targetChunkSize;
if (nbLoadedChunks == sstSize) { /* no more space left in sampleSizes table */
fileIndex = nbFiles; /* stop there */
break;
}
if (toLoad < targetChunkSize) {
fseek(f, (long)(targetChunkSize - toLoad), SEEK_CUR);
} } }
fclose(f);
}
DISPLAYLEVEL(2, "\r%79s\r", "");
*bufferSizePtr = pos;
DISPLAYLEVEL(4, "loaded : %u KB \n", (U32)(pos >> 10))
return nbLoadedChunks;
}
#define rotl32(x,r) ((x << r) | (x >> (32 - r)))
static U32 getRand(U32* src)
{
static const U32 prime1 = 2654435761U;
static const U32 prime2 = 2246822519U;
U32 rand32 = *src;
rand32 *= prime1;
rand32 ^= prime2;
rand32 = rotl32(rand32, 13);
*src = rand32;
return rand32 >> 5;
}
/* shuffle() :
* shuffle a table of file names in a semi-random way
* It improves dictionary quality by reducing "locality" impact, so if sample set is very large,
* it will load random elements from it, instead of just the first ones. */
static void shuffle(const char** fileNamesTable, unsigned nbFiles) {
U32 seed = 0xFD2FB528;
unsigned i;
for (i = nbFiles - 1; i > 0; --i) {
unsigned const j = getRand(&seed) % (i + 1);
const char* const tmp = fileNamesTable[j];
fileNamesTable[j] = fileNamesTable[i];
fileNamesTable[i] = tmp;
}
}
/*-********************************************************
* Dictionary training functions
**********************************************************/
static size_t findMaxMem(unsigned long long requiredMem)
{
size_t const step = 8 MB;
void* testmem = NULL;
requiredMem = (((requiredMem >> 23) + 1) << 23);
requiredMem += step;
if (requiredMem > g_maxMemory) requiredMem = g_maxMemory;
while (!testmem) {
testmem = malloc((size_t)requiredMem);
requiredMem -= step;
}
free(testmem);
return (size_t)requiredMem;
}
static void saveDict(const char* dictFileName,
const void* buff, size_t buffSize)
{
FILE* const f = fopen(dictFileName, "wb");
if (f==NULL) EXM_THROW(3, "cannot open %s ", dictFileName);
{ size_t const n = fwrite(buff, 1, buffSize, f);
if (n!=buffSize) EXM_THROW(4, "%s : write error", dictFileName) }
{ size_t const n = (size_t)fclose(f);
if (n!=0) EXM_THROW(5, "%s : flush error", dictFileName) }
}
/*! getFileStats() :
* Given a list of files, and a chunkSize (0 == no chunk, whole files)
* provides the amount of data to be loaded and the resulting nb of samples.
* This is useful primarily for allocation purpose => sample buffer, and sample sizes table.
*/
static fileStats getFileStats(const char** fileNamesTable, unsigned nbFiles, size_t chunkSize, unsigned displayLevel)
{
fileStats fs;
unsigned n;
memset(&fs, 0, sizeof(fs));
for (n=0; n<nbFiles; n++) {
U64 const fileSize = UTIL_getFileSize(fileNamesTable[n]);
U64 const srcSize = (fileSize == UTIL_FILESIZE_UNKNOWN) ? 0 : fileSize;
U32 const nbSamples = (U32)(chunkSize ? (srcSize + (chunkSize-1)) / chunkSize : 1);
U64 const chunkToLoad = chunkSize ? MIN(chunkSize, srcSize) : srcSize;
size_t const cappedChunkSize = (size_t)MIN(chunkToLoad, SAMPLESIZE_MAX);
fs.totalSizeToLoad += cappedChunkSize * nbSamples;
fs.oneSampleTooLarge |= (chunkSize > 2*SAMPLESIZE_MAX);
fs.nbSamples += nbSamples;
}
DISPLAYLEVEL(4, "Preparing to load : %u KB \n", (U32)(fs.totalSizeToLoad >> 10));
return fs;
}
@ -225,16 +50,14 @@ static size_t RANDOM_sum(const size_t *samplesSizes, unsigned nbSamples) {
/** /**
* Selects a random segment from totalSamplesSize - k + 1 possible segments * Selects a random segment from totalSamplesSize - k + 1 possible segments
*/ */
static RANDOM_segment_t RANDOM_selectSegment(const RANDOM_ctx_t *ctx, static RANDOM_segment_t RANDOM_selectSegment(const size_t totalSamplesSize,
ZDICT_random_params_t parameters) { ZDICT_random_params_t parameters) {
const U32 k = parameters.k; const U32 k = parameters.k;
RANDOM_segment_t segment; RANDOM_segment_t segment;
unsigned index; unsigned index;
/* Seed random number generator */
srand((unsigned)time(NULL));
/* Randomly generate a number from 0 to sampleSizes - k */ /* Randomly generate a number from 0 to sampleSizes - k */
index = rand()%(ctx->totalSamplesSize - k + 1); index = rand()%(totalSamplesSize - k + 1);
/* inclusive */ /* inclusive */
segment.begin = index; segment.begin = index;
@ -261,65 +84,11 @@ static int RANDOM_checkParameters(ZDICT_random_params_t parameters, size_t maxDi
} }
/**
* Clean up a context initialized with `RANDOM_ctx_init()`.
*/
static void RANDOM_ctx_destroy(RANDOM_ctx_t *ctx) {
if (!ctx) {
return;
}
if (ctx->offsets) {
free(ctx->offsets);
ctx->offsets = NULL;
}
}
/**
* Prepare a context for dictionary building.
* Returns 1 on success or zero on error.
* The context must be destroyed with `RANDOM_ctx_destroy()`.
*/
static int RANDOM_ctx_init(RANDOM_ctx_t *ctx, const void *samplesBuffer,
const size_t *samplesSizes, unsigned nbSamples) {
const BYTE *const samples = (const BYTE *)samplesBuffer;
const size_t totalSamplesSize = RANDOM_sum(samplesSizes, nbSamples);
const int displayLevel = 2;
/* Checks */
if (totalSamplesSize >= (size_t)RANDOM_MAX_SAMPLES_SIZE) {
DISPLAYLEVEL(1, "Total samples size is too large (%u MB), maximum size is %u MB\n",
(U32)(totalSamplesSize>>20), (RANDOM_MAX_SAMPLES_SIZE >> 20));
return 0;
}
memset(ctx, 0, sizeof(*ctx));
DISPLAYLEVEL(1, "Building dictionary from %u samples of total size %u\n", nbSamples,
(U32)totalSamplesSize);
ctx->samples = samples;
ctx->samplesSizes = samplesSizes;
ctx->nbSamples = nbSamples;
ctx->offsets = (size_t *)malloc((nbSamples + 1) * sizeof(size_t));
ctx->totalSamplesSize = (U32)totalSamplesSize;
if (!ctx->offsets) {
DISPLAYLEVEL(1, "Failed to allocate buffer for offsets\n");
RANDOM_ctx_destroy(ctx);
return 0;
}
{
U32 i;
ctx->offsets[0] = 0;
for (i = 1; i <= nbSamples; ++i) {
ctx->offsets[i] = ctx->offsets[i - 1] + samplesSizes[i - 1];
}
}
return 1;
}
/** /**
* Given the prepared context build the dictionary. * Given the prepared context build the dictionary.
*/ */
static size_t RANDOM_buildDictionary(const RANDOM_ctx_t *ctx, void *dictBuffer, static size_t RANDOM_buildDictionary(const size_t totalSamplesSize, const BYTE *samples,
size_t dictBufferCapacity, void *dictBuffer, size_t dictBufferCapacity,
ZDICT_random_params_t parameters) { ZDICT_random_params_t parameters) {
BYTE *const dict = (BYTE *)dictBuffer; BYTE *const dict = (BYTE *)dictBuffer;
size_t tail = dictBufferCapacity; size_t tail = dictBufferCapacity;
@ -327,13 +96,13 @@ static size_t RANDOM_buildDictionary(const RANDOM_ctx_t *ctx, void *dictBuffer,
while (tail > 0) { while (tail > 0) {
/* Select a segment */ /* Select a segment */
RANDOM_segment_t segment = RANDOM_selectSegment(ctx, parameters); RANDOM_segment_t segment = RANDOM_selectSegment(totalSamplesSize, parameters);
size_t segmentSize; size_t segmentSize;
segmentSize = MIN(segment.end - segment.begin + 1, tail); segmentSize = MIN(segment.end - segment.begin + 1, tail);
tail -= segmentSize; tail -= segmentSize;
memcpy(dict + tail, ctx->samples + segment.begin, segmentSize); memcpy(dict + tail, samples + segment.begin, segmentSize);
DISPLAYUPDATE( DISPLAYUPDATE(
2, "\r%u%% ", 2, "\r%u%% ",
(U32)(((dictBufferCapacity - tail) * 100) / dictBufferCapacity)); (U32)(((dictBufferCapacity - tail) * 100) / dictBufferCapacity));
@ -342,6 +111,7 @@ static size_t RANDOM_buildDictionary(const RANDOM_ctx_t *ctx, void *dictBuffer,
return tail; return tail;
} }
/*! ZDICT_trainFromBuffer_random(): /*! ZDICT_trainFromBuffer_random():
* Train a dictionary from an array of samples using the RANDOM algorithm. * Train a dictionary from an array of samples using the RANDOM algorithm.
* Samples must be stored concatenated in a single flat buffer `samplesBuffer`, * Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
@ -356,7 +126,6 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_random(
ZDICT_random_params_t parameters) { ZDICT_random_params_t parameters) {
const int displayLevel = parameters.zParams.notificationLevel; const int displayLevel = parameters.zParams.notificationLevel;
BYTE* const dict = (BYTE*)dictBuffer; BYTE* const dict = (BYTE*)dictBuffer;
RANDOM_ctx_t ctx;
/* Checks */ /* Checks */
if (!RANDOM_checkParameters(parameters, dictBufferCapacity)) { if (!RANDOM_checkParameters(parameters, dictBufferCapacity)) {
DISPLAYLEVEL(1, "k is incorrect\n"); DISPLAYLEVEL(1, "k is incorrect\n");
@ -371,13 +140,12 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_random(
ZDICT_DICTSIZE_MIN); ZDICT_DICTSIZE_MIN);
return ERROR(dstSize_tooSmall); return ERROR(dstSize_tooSmall);
} }
const size_t totalSamplesSize = RANDOM_sum(samplesSizes, nbSamples);
const BYTE *const samples = (const BYTE *)samplesBuffer;
if (!RANDOM_ctx_init(&ctx, samplesBuffer, samplesSizes, nbSamples)) {
return ERROR(GENERIC);
}
DISPLAYLEVEL(2, "Building dictionary\n"); DISPLAYLEVEL(2, "Building dictionary\n");
{ {
const size_t tail = RANDOM_buildDictionary(&ctx, dictBuffer, dictBufferCapacity, parameters); const size_t tail = RANDOM_buildDictionary(totalSamplesSize, samples, dictBuffer, dictBufferCapacity, parameters);
const size_t dictSize = ZDICT_finalizeDictionary( const size_t dictSize = ZDICT_finalizeDictionary(
dict, dictBufferCapacity, dict + tail, dictBufferCapacity - tail, dict, dictBufferCapacity, dict + tail, dictBufferCapacity - tail,
samplesBuffer, samplesSizes, nbSamples, parameters.zParams); samplesBuffer, samplesSizes, nbSamples, parameters.zParams);
@ -385,71 +153,6 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_random(
DISPLAYLEVEL(2, "Constructed dictionary of size %u\n", DISPLAYLEVEL(2, "Constructed dictionary of size %u\n",
(U32)dictSize); (U32)dictSize);
} }
RANDOM_ctx_destroy(&ctx);
return dictSize; return dictSize;
} }
} }
int RANDOM_trainFromFiles(const char* dictFileName, unsigned maxDictSize,
const char** fileNamesTable, unsigned nbFiles,
size_t chunkSize, ZDICT_random_params_t *params){
unsigned const displayLevel = params->zParams.notificationLevel;
void* const dictBuffer = malloc(maxDictSize);
fileStats const fs = getFileStats(fileNamesTable, nbFiles, chunkSize, displayLevel);
size_t* const sampleSizes = (size_t*)malloc(fs.nbSamples * sizeof(size_t));
size_t const memMult = RANDOM_MEMMULT;
size_t const maxMem = findMaxMem(fs.totalSizeToLoad * memMult) / memMult;
size_t loadedSize = (size_t) MIN ((unsigned long long)maxMem, fs.totalSizeToLoad);
void* const srcBuffer = malloc(loadedSize+NOISELENGTH);
int result = 0;
/* Checks */
if ((!sampleSizes) || (!srcBuffer) || (!dictBuffer))
EXM_THROW(12, "not enough memory for DiB_trainFiles"); /* should not happen */
if (fs.oneSampleTooLarge) {
DISPLAYLEVEL(2, "! Warning : some sample(s) are very large \n");
DISPLAYLEVEL(2, "! Note that dictionary is only useful for small samples. \n");
DISPLAYLEVEL(2, "! As a consequence, only the first %u bytes of each sample are loaded \n", SAMPLESIZE_MAX);
}
if (fs.nbSamples < 5) {
DISPLAYLEVEL(2, "! Warning : nb of samples too low for proper processing ! \n");
DISPLAYLEVEL(2, "! Please provide _one file per sample_. \n");
DISPLAYLEVEL(2, "! Alternatively, split files into fixed-size blocks representative of samples, with -B# \n");
EXM_THROW(14, "nb of samples too low"); /* we now clearly forbid this case */
}
if (fs.totalSizeToLoad < (unsigned long long)(8 * maxDictSize)) {
DISPLAYLEVEL(2, "! Warning : data size of samples too small for target dictionary size \n");
DISPLAYLEVEL(2, "! Samples should be about 100x larger than target dictionary size \n");
}
/* init */
if (loadedSize < fs.totalSizeToLoad)
DISPLAYLEVEL(1, "Not enough memory; training on %u MB only...\n", (unsigned)(loadedSize >> 20));
/* Load input buffer */
DISPLAYLEVEL(3, "Shuffling input files\n");
shuffle(fileNamesTable, nbFiles);
nbFiles = loadFiles(srcBuffer, &loadedSize, sampleSizes, fs.nbSamples, fileNamesTable, nbFiles, chunkSize, displayLevel);
{ size_t dictSize;
dictSize = ZDICT_trainFromBuffer_random(dictBuffer, maxDictSize, srcBuffer,
sampleSizes, fs.nbSamples, *params);
DISPLAYLEVEL(2, "k=%u\n", params->k);
if (ZDICT_isError(dictSize)) {
DISPLAYLEVEL(1, "dictionary training failed : %s \n", ZDICT_getErrorName(dictSize)); /* should not happen */
result = 1;
goto _cleanup;
}
/* save dict */
DISPLAYLEVEL(2, "Save dictionary of size %u into file %s \n", (U32)dictSize, dictFileName);
saveDict(dictFileName, dictBuffer, dictSize);
}
/* clean up */
_cleanup:
free(srcBuffer);
free(sampleSizes);
free(dictBuffer);
return result;
}

23
contrib/randomDictBuilder/random.h
View File

@ -8,18 +8,6 @@
#endif #endif
#include "zdict.h" #include "zdict.h"
/**************************************
* Context
***************************************/
typedef struct {
const BYTE *samples;
size_t *offsets;
const size_t *samplesSizes;
size_t nbSamples;
U32 totalSamplesSize;
} RANDOM_ctx_t;
/** /**
* A segment is an inclusive range in the source. * A segment is an inclusive range in the source.
*/ */
@ -35,19 +23,8 @@ typedef struct {
} ZDICT_random_params_t; } ZDICT_random_params_t;
typedef struct {
U64 totalSizeToLoad;
unsigned oneSampleTooLarge;
unsigned nbSamples;
} fileStats;
ZDICTLIB_API size_t ZDICT_trainFromBuffer_random( ZDICTLIB_API size_t ZDICT_trainFromBuffer_random(
void *dictBuffer, size_t dictBufferCapacity, void *dictBuffer, size_t dictBufferCapacity,
const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples, const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples,
ZDICT_random_params_t parameters); ZDICT_random_params_t parameters);
int RANDOM_trainFromFiles(const char* dictFileName, unsigned maxDictSize,
const char** fileNamesTable, unsigned nbFiles,
size_t chunkSize, ZDICT_random_params_t *params);