zstd/programs/zstdcli.c

1338 lines
61 KiB
C
Raw Normal View History

/*
* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
2016-08-30 17:04:33 +00:00
* All rights reserved.
*
* This source code is licensed under both the BSD-style license (found in the
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
* in the COPYING file in the root directory of this source tree).
* You may select, at your option, one of the above-listed licenses.
2016-08-30 17:04:33 +00:00
*/
2016-08-28 17:00:49 +00:00
2016-05-29 03:16:05 +00:00
/*-************************************
* Tuning parameters
**************************************/
#ifndef ZSTDCLI_CLEVEL_DEFAULT
# define ZSTDCLI_CLEVEL_DEFAULT 3
#endif
2016-08-12 16:04:15 +00:00
#ifndef ZSTDCLI_CLEVEL_MAX
# define ZSTDCLI_CLEVEL_MAX 19 /* without using --ultra */
2016-08-12 16:04:15 +00:00
#endif
2016-05-29 03:16:05 +00:00
/*-************************************
* Dependencies
2016-05-29 03:16:05 +00:00
**************************************/
#include "platform.h" /* IS_CONSOLE, PLATFORM_POSIX_VERSION */
#include "util.h" /* UTIL_HAS_CREATEFILELIST, UTIL_createFileList */
#include <stdlib.h> /* getenv */
2016-05-29 03:16:05 +00:00
#include <string.h> /* strcmp, strlen */
#include <stdio.h> /* fprintf(), stdin, stdout, stderr */
#include <errno.h> /* errno */
#include <assert.h> /* assert */
#include "fileio.h" /* stdinmark, stdoutmark, ZSTD_EXTENSION */
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOBENCH
# include "benchzstd.h" /* BMK_benchFiles */
2016-05-29 03:16:05 +00:00
#endif
#ifndef ZSTD_NODICT
# include "dibio.h" /* ZDICT_cover_params_t, DiB_trainFromFiles() */
2016-05-29 03:16:05 +00:00
#endif
#include "../lib/zstd.h" /* ZSTD_VERSION_STRING, ZSTD_minCLevel, ZSTD_maxCLevel */
2016-05-29 03:16:05 +00:00
/*-************************************
* Constants
**************************************/
#define COMPRESSOR_NAME "zstd command line interface"
#ifndef ZSTD_VERSION
# define ZSTD_VERSION "v" ZSTD_VERSION_STRING
#endif
#define AUTHOR "Yann Collet"
#define WELCOME_MESSAGE "*** %s %i-bits %s, by %s ***\n", COMPRESSOR_NAME, (int)(sizeof(size_t)*8), ZSTD_VERSION, AUTHOR
2017-04-14 18:33:04 +00:00
#define ZSTD_ZSTDMT "zstdmt"
2016-05-29 03:16:05 +00:00
#define ZSTD_UNZSTD "unzstd"
#define ZSTD_CAT "zstdcat"
#define ZSTD_ZCAT "zcat"
#define ZSTD_GZ "gzip"
#define ZSTD_GUNZIP "gunzip"
#define ZSTD_GZCAT "gzcat"
#define ZSTD_LZMA "lzma"
2017-06-26 18:24:36 +00:00
#define ZSTD_UNLZMA "unlzma"
#define ZSTD_XZ "xz"
2017-06-26 18:24:36 +00:00
#define ZSTD_UNXZ "unxz"
2017-09-28 23:18:15 +00:00
#define ZSTD_LZ4 "lz4"
#define ZSTD_UNLZ4 "unlz4"
2016-05-29 03:16:05 +00:00
#define KB *(1 <<10)
#define MB *(1 <<20)
#define GB *(1U<<30)
#define DISPLAY_LEVEL_DEFAULT 2
2016-05-29 03:16:05 +00:00
static const char* g_defaultDictName = "dictionary";
static const unsigned g_defaultMaxDictSize = 110 KB;
static const int g_defaultDictCLevel = 3;
2016-05-29 03:16:05 +00:00
static const unsigned g_defaultSelectivityLevel = 9;
static const unsigned g_defaultMaxWindowLog = 27;
2017-01-30 22:37:08 +00:00
#define OVERLAP_LOG_DEFAULT 9999
2017-09-03 04:10:36 +00:00
#define LDM_PARAM_DEFAULT 9999 /* Default for parameters where 0 is valid */
2017-01-30 22:37:08 +00:00
static U32 g_overlapLog = OVERLAP_LOG_DEFAULT;
static U32 g_ldmHashLog = 0;
static U32 g_ldmMinMatch = 0;
2018-11-21 22:36:57 +00:00
static U32 g_ldmHashRateLog = LDM_PARAM_DEFAULT;
2017-09-03 04:10:36 +00:00
static U32 g_ldmBucketSizeLog = LDM_PARAM_DEFAULT;
2016-05-29 03:16:05 +00:00
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
#define DEFAULT_ACCEL 1
typedef enum { cover, fastCover, legacy } dictType;
2016-05-29 03:16:05 +00:00
/*-************************************
* Display Macros
**************************************/
#define DISPLAY_F(f, ...) fprintf((f), __VA_ARGS__)
#define DISPLAYOUT(...) DISPLAY_F(stdout, __VA_ARGS__)
#define DISPLAY(...) DISPLAY_F(stderr, __VA_ARGS__)
#define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } }
static int g_displayLevel = DISPLAY_LEVEL_DEFAULT; /* 0 : no display, 1: errors, 2 : + result + interaction + warnings, 3 : + progression, 4 : + information */
2016-05-29 03:16:05 +00:00
/*-************************************
* Command Line
**************************************/
/* print help either in `stderr` or `stdout` depending on originating request
* error (badusage) => stderr
* help (usage_advanced) => stdout
*/
static void usage(FILE* f, const char* programName)
2016-05-29 03:16:05 +00:00
{
DISPLAY_F(f, "Usage : \n");
DISPLAY_F(f, " %s [args] [FILE(s)] [-o file] \n", programName);
DISPLAY_F(f, "\n");
DISPLAY_F(f, "FILE : a filename \n");
DISPLAY_F(f, " with no FILE, or when FILE is - , read standard input\n");
DISPLAY_F(f, "Arguments : \n");
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOCOMPRESS
DISPLAY_F(f, " -# : # compression level (1-%d, default: %d) \n", ZSTDCLI_CLEVEL_MAX, ZSTDCLI_CLEVEL_DEFAULT);
2016-05-29 03:16:05 +00:00
#endif
#ifndef ZSTD_NODECOMPRESS
DISPLAY_F(f, " -d : decompression \n");
2016-05-29 03:16:05 +00:00
#endif
DISPLAY_F(f, " -D DICT: use DICT as Dictionary for compression or decompression \n");
2020-05-08 21:20:47 +00:00
DISPLAY_F(f, " -o file: result stored into `file` (only 1 output file) \n");
DISPLAY_F(f, " -f : overwrite output without prompting, also (de)compress links \n");
DISPLAY_F(f, "--rm : remove source file(s) after successful de/compression \n");
DISPLAY_F(f, " -k : preserve source file(s) (default) \n");
DISPLAY_F(f, " -h/-H : display help/long help and exit \n");
2016-05-29 03:16:05 +00:00
}
static void usage_advanced(const char* programName)
2016-05-29 03:16:05 +00:00
{
DISPLAYOUT(WELCOME_MESSAGE);
usage(stdout, programName);
DISPLAYOUT( "\n");
DISPLAYOUT( "Advanced arguments : \n");
DISPLAYOUT( " -V : display Version number and exit \n");
DISPLAYOUT( " -c : force write to standard output, even if it is the console \n");
DISPLAYOUT( " -v : verbose mode; specify multiple times to increase verbosity \n");
DISPLAYOUT( " -q : suppress warnings; specify twice to suppress errors too \n");
DISPLAYOUT( "--no-progress : do not display the progress counter \n");
#ifdef UTIL_HAS_CREATEFILELIST
DISPLAYOUT( " -r : operate recursively on directories \n");
2020-07-23 19:10:57 +00:00
DISPLAYOUT( "--filelist FILE : read list of files to operate upon from FILE \n");
DISPLAYOUT( "--output-dir-flat DIR : processed files are stored into DIR \n");
#endif
2020-06-20 02:35:51 +00:00
#ifdef UTIL_HAS_MIRRORFILELIST
2020-07-23 19:10:57 +00:00
DISPLAYOUT( "--output-dir-mirror DIR : processed files are stored into DIR respecting original directory structure \n");
2020-06-20 02:35:51 +00:00
#endif
DISPLAYOUT( "-- : All arguments after \"--\" are treated as files \n");
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOCOMPRESS
DISPLAYOUT( "\n");
DISPLAYOUT( "Advanced compression arguments : \n");
DISPLAYOUT( "--ultra : enable levels beyond %i, up to %i (requires more memory) \n", ZSTDCLI_CLEVEL_MAX, ZSTD_maxCLevel());
DISPLAYOUT( "--long[=#]: enable long distance matching with given window log (default: %u) \n", g_defaultMaxWindowLog);
DISPLAYOUT( "--fast[=#]: switch to very fast compression levels (default: %u) \n", 1);
DISPLAYOUT( "--adapt : dynamically adapt compression level to I/O conditions \n");
# ifdef ZSTD_MULTITHREAD
DISPLAYOUT( " -T# : spawns # compression threads (default: 1, 0==# cores) \n");
DISPLAYOUT( " -B# : select size of each job (default: 0==automatic) \n");
DISPLAYOUT( "--single-thread : use a single thread for both I/O and compression (result slightly different than -T1) \n");
DISPLAYOUT( "--rsyncable : compress using a rsync-friendly method (-B sets block size) \n");
# endif
DISPLAYOUT( "--exclude-compressed: only compress files that are not already compressed \n");
DISPLAYOUT( "--stream-size=# : specify size of streaming input from `stdin` \n");
DISPLAYOUT( "--size-hint=# optimize compression parameters for streaming input of approximately this size \n");
DISPLAYOUT( "--target-compressed-block-size=# : generate compressed block of approximately targeted size \n");
DISPLAYOUT( "--no-dictID : don't write dictID into header (dictionary compression only) \n");
DISPLAYOUT( "--[no-]check : add XXH64 integrity checksum to frame (default: enabled) \n");
DISPLAYOUT( "--[no-]compress-literals : force (un)compressed literals \n");
DISPLAYOUT( "--format=zstd : compress files to the .zst format (default) \n");
2017-02-08 14:17:55 +00:00
#ifdef ZSTD_GZCOMPRESS
DISPLAYOUT( "--format=gzip : compress files to the .gz format \n");
2017-02-08 14:17:55 +00:00
#endif
#ifdef ZSTD_LZMACOMPRESS
DISPLAYOUT( "--format=xz : compress files to the .xz format \n");
DISPLAYOUT( "--format=lzma : compress files to the .lzma format \n");
#endif
#ifdef ZSTD_LZ4COMPRESS
DISPLAYOUT( "--format=lz4 : compress files to the .lz4 format \n");
#endif
#endif /* !ZSTD_NOCOMPRESS */
#ifndef ZSTD_NODECOMPRESS
DISPLAYOUT( "\n");
DISPLAYOUT( "Advanced decompression arguments : \n");
DISPLAYOUT( " -l : print information about zstd compressed files \n");
DISPLAYOUT( "--test : test compressed file integrity \n");
DISPLAYOUT( " -M# : Set a memory usage limit for decompression \n");
DISPLAYOUT( "--no-check : disable validation of checksums in compressed frame \n");
# if ZSTD_SPARSE_DEFAULT
DISPLAYOUT( "--[no-]sparse : sparse mode (default: enabled on file, disabled on stdout) \n");
# else
DISPLAYOUT( "--[no-]sparse : sparse mode (default: disabled) \n");
# endif
#endif /* ZSTD_NODECOMPRESS */
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NODICT
DISPLAYOUT( "\n");
DISPLAYOUT( "Dictionary builder : \n");
DISPLAYOUT( "--train ## : create a dictionary from a training set of files \n");
DISPLAYOUT( "--train-cover[=k=#,d=#,steps=#,split=#,shrink[=#]] : use the cover algorithm with optional args \n");
DISPLAYOUT( "--train-fastcover[=k=#,d=#,f=#,steps=#,split=#,accel=#,shrink[=#]] : use the fast cover algorithm with optional args \n");
DISPLAYOUT( "--train-legacy[=s=#] : use the legacy algorithm with selectivity (default: %u) \n", g_defaultSelectivityLevel);
DISPLAYOUT( " -o DICT : DICT is dictionary name (default: %s) \n", g_defaultDictName);
DISPLAYOUT( "--maxdict=# : limit dictionary to specified size (default: %u) \n", g_defaultMaxDictSize);
DISPLAYOUT( "--dictID=# : force dictionary ID to specified value (default: random) \n");
2016-05-29 03:16:05 +00:00
#endif
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOBENCH
DISPLAYOUT( "\n");
DISPLAYOUT( "Benchmark arguments : \n");
DISPLAYOUT( " -b# : benchmark file(s), using # compression level (default: %d) \n", ZSTDCLI_CLEVEL_DEFAULT);
DISPLAYOUT( " -e# : test all compression levels successively from -b# to -e# (default: 1) \n");
DISPLAYOUT( " -i# : minimum evaluation time in seconds (default: 3s) \n");
DISPLAYOUT( " -B# : cut file into independent blocks of size # (default: no block) \n");
DISPLAYOUT( " -S : output one benchmark result per input file (default: consolidated result) \n");
DISPLAYOUT( "--priority=rt : set process priority to real-time \n");
2016-05-29 03:16:05 +00:00
#endif
2016-05-29 03:16:05 +00:00
}
static void badusage(const char* programName)
2016-05-29 03:16:05 +00:00
{
DISPLAYLEVEL(1, "Incorrect parameters \n");
if (g_displayLevel >= 2) usage(stderr, programName);
2016-05-29 03:16:05 +00:00
}
static void waitEnter(void)
{
int unused;
DISPLAY("Press enter to continue... \n");
2016-05-29 03:16:05 +00:00
unused = getchar();
(void)unused;
}
static const char* lastNameFromPath(const char* path)
{
const char* name = path;
if (strrchr(name, '/')) name = strrchr(name, '/') + 1;
if (strrchr(name, '\\')) name = strrchr(name, '\\') + 1; /* windows */
return name;
}
/*! exeNameMatch() :
@return : a non-zero value if exeName matches test, excluding the extension
*/
static int exeNameMatch(const char* exeName, const char* test)
{
return !strncmp(exeName, test, strlen(test)) &&
(exeName[strlen(test)] == '\0' || exeName[strlen(test)] == '.');
}
static void errorOut(const char* msg)
{
DISPLAY("%s \n", msg); exit(1);
}
/*! readU32FromCharChecked() :
* @return 0 if success, and store the result in *value.
* allows and interprets K, KB, KiB, M, MB and MiB suffix.
* Will also modify `*stringPtr`, advancing it to position where it stopped reading.
* @return 1 if an overflow error occurs */
static int readU32FromCharChecked(const char** stringPtr, unsigned* value)
2016-06-03 13:14:09 +00:00
{
unsigned result = 0;
while ((**stringPtr >='0') && (**stringPtr <='9')) {
2020-02-18 23:30:59 +00:00
unsigned const max = ((unsigned)(-1)) / 10;
unsigned last = result;
if (result > max) return 1; /* overflow error */
result *= 10;
result += (unsigned)(**stringPtr - '0');
2020-02-18 23:30:59 +00:00
if (result < last) return 1; /* overflow error */
(*stringPtr)++ ;
}
2016-10-18 00:48:48 +00:00
if ((**stringPtr=='K') || (**stringPtr=='M')) {
unsigned const maxK = ((unsigned)(-1)) >> 10;
if (result > maxK) return 1; /* overflow error */
2016-10-18 00:48:48 +00:00
result <<= 10;
if (**stringPtr=='M') {
if (result > maxK) return 1; /* overflow error */
result <<= 10;
}
(*stringPtr)++; /* skip `K` or `M` */
2016-10-18 00:48:48 +00:00
if (**stringPtr=='i') (*stringPtr)++;
if (**stringPtr=='B') (*stringPtr)++;
}
*value = result;
return 0;
}
/*! readU32FromChar() :
* @return : unsigned integer value read from input in `char` format.
* allows and interprets K, KB, KiB, M, MB and MiB suffix.
* Will also modify `*stringPtr`, advancing it to position where it stopped reading.
* Note : function will exit() program if digit sequence overflows */
static unsigned readU32FromChar(const char** stringPtr) {
static const char errorMsg[] = "error: numeric value overflows 32-bit unsigned int";
unsigned result;
if (readU32FromCharChecked(stringPtr, &result)) { errorOut(errorMsg); }
2016-06-03 13:14:09 +00:00
return result;
}
/*! readSizeTFromCharChecked() :
* @return 0 if success, and store the result in *value.
* allows and interprets K, KB, KiB, M, MB and MiB suffix.
* Will also modify `*stringPtr`, advancing it to position where it stopped reading.
* @return 1 if an overflow error occurs */
static int readSizeTFromCharChecked(const char** stringPtr, size_t* value)
{
size_t result = 0;
while ((**stringPtr >='0') && (**stringPtr <='9')) {
size_t const max = ((size_t)(-1)) / 10;
size_t last = result;
if (result > max) return 1; /* overflow error */
result *= 10;
result += (size_t)(**stringPtr - '0');
if (result < last) return 1; /* overflow error */
(*stringPtr)++ ;
}
if ((**stringPtr=='K') || (**stringPtr=='M')) {
size_t const maxK = ((size_t)(-1)) >> 10;
if (result > maxK) return 1; /* overflow error */
result <<= 10;
if (**stringPtr=='M') {
if (result > maxK) return 1; /* overflow error */
result <<= 10;
}
(*stringPtr)++; /* skip `K` or `M` */
if (**stringPtr=='i') (*stringPtr)++;
if (**stringPtr=='B') (*stringPtr)++;
}
*value = result;
return 0;
}
/*! readSizeTFromChar() :
* @return : size_t value read from input in `char` format.
* allows and interprets K, KB, KiB, M, MB and MiB suffix.
* Will also modify `*stringPtr`, advancing it to position where it stopped reading.
* Note : function will exit() program if digit sequence overflows */
static size_t readSizeTFromChar(const char** stringPtr) {
static const char errorMsg[] = "error: numeric value overflows size_t";
size_t result;
if (readSizeTFromCharChecked(stringPtr, &result)) { errorOut(errorMsg); }
return result;
}
2016-12-02 23:18:57 +00:00
/** longCommandWArg() :
* check if *stringPtr is the same as longCommand.
2016-12-02 23:18:57 +00:00
* If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand.
* @return 0 and doesn't modify *stringPtr otherwise.
2016-12-02 23:18:57 +00:00
*/
2019-10-17 22:27:25 +00:00
static int longCommandWArg(const char** stringPtr, const char* longCommand)
2016-10-14 21:41:17 +00:00
{
size_t const comSize = strlen(longCommand);
2016-12-02 23:18:57 +00:00
int const result = !strncmp(*stringPtr, longCommand, comSize);
2016-10-14 21:41:17 +00:00
if (result) *stringPtr += comSize;
return result;
}
2016-12-13 11:18:07 +00:00
2017-01-01 05:08:24 +00:00
#ifndef ZSTD_NODICT
static const unsigned kDefaultRegression = 1;
2017-01-01 05:08:24 +00:00
/**
* parseCoverParameters() :
* reads cover parameters from *stringPtr (e.g. "--train-cover=k=48,d=8,steps=32") into *params
2017-01-01 05:08:24 +00:00
* @return 1 means that cover parameters were correct
* @return 0 in case of malformed parameters
*/
static unsigned parseCoverParameters(const char* stringPtr, ZDICT_cover_params_t* params)
2017-01-01 05:08:24 +00:00
{
memset(params, 0, sizeof(*params));
for (; ;) {
2017-01-02 20:40:43 +00:00
if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
2017-01-01 05:08:24 +00:00
if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
2017-01-02 20:40:43 +00:00
if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
2018-06-30 00:54:41 +00:00
if (longCommandWArg(&stringPtr, "split=")) {
unsigned splitPercentage = readU32FromChar(&stringPtr);
2018-06-30 00:54:41 +00:00
params->splitPoint = (double)splitPercentage / 100.0;
if (stringPtr[0]==',') { stringPtr++; continue; } else break;
}
if (longCommandWArg(&stringPtr, "shrink")) {
params->shrinkDictMaxRegression = kDefaultRegression;
params->shrinkDict = 1;
if (stringPtr[0]=='=') {
stringPtr++;
params->shrinkDictMaxRegression = readU32FromChar(&stringPtr);
}
if (stringPtr[0]==',') {
stringPtr++;
continue;
}
else break;
}
2017-01-01 05:08:24 +00:00
return 0;
}
if (stringPtr[0] != 0) return 0;
DISPLAYLEVEL(4, "cover: k=%u\nd=%u\nsteps=%u\nsplit=%u\nshrink%u\n", params->k, params->d, params->steps, (unsigned)(params->splitPoint * 100), params->shrinkDictMaxRegression);
2017-01-01 05:08:24 +00:00
return 1;
}
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
/**
* parseFastCoverParameters() :
* reads fastcover parameters from *stringPtr (e.g. "--train-fastcover=k=48,d=8,f=20,steps=32,accel=2") into *params
* @return 1 means that fastcover parameters were correct
* @return 0 in case of malformed parameters
*/
static unsigned parseFastCoverParameters(const char* stringPtr, ZDICT_fastCover_params_t* params)
{
memset(params, 0, sizeof(*params));
for (; ;) {
if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "f=")) { params->f = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "accel=")) { params->accel = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "split=")) {
unsigned splitPercentage = readU32FromChar(&stringPtr);
params->splitPoint = (double)splitPercentage / 100.0;
if (stringPtr[0]==',') { stringPtr++; continue; } else break;
}
if (longCommandWArg(&stringPtr, "shrink")) {
params->shrinkDictMaxRegression = kDefaultRegression;
params->shrinkDict = 1;
if (stringPtr[0]=='=') {
stringPtr++;
params->shrinkDictMaxRegression = readU32FromChar(&stringPtr);
}
if (stringPtr[0]==',') {
stringPtr++;
continue;
}
else break;
}
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
return 0;
}
if (stringPtr[0] != 0) return 0;
DISPLAYLEVEL(4, "cover: k=%u\nd=%u\nf=%u\nsteps=%u\nsplit=%u\naccel=%u\nshrink=%u\n", params->k, params->d, params->f, params->steps, (unsigned)(params->splitPoint * 100), params->accel, params->shrinkDictMaxRegression);
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
return 1;
}
/**
* parseLegacyParameters() :
Spelling (#1582) * spelling: accidentally * spelling: across * spelling: additionally * spelling: addresses * spelling: appropriate * spelling: assumed * spelling: available * spelling: builder * spelling: capacity * spelling: compiler * spelling: compressibility * spelling: compressor * spelling: compression * spelling: contract * spelling: convenience * spelling: decompress * spelling: description * spelling: deflate * spelling: deterministically * spelling: dictionary * spelling: display * spelling: eliminate * spelling: preemptively * spelling: exclude * spelling: failure * spelling: independence * spelling: independent * spelling: intentionally * spelling: matching * spelling: maximum * spelling: meaning * spelling: mishandled * spelling: memory * spelling: occasionally * spelling: occurrence * spelling: official * spelling: offsets * spelling: original * spelling: output * spelling: overflow * spelling: overridden * spelling: parameter * spelling: performance * spelling: probability * spelling: receives * spelling: redundant * spelling: recompression * spelling: resources * spelling: sanity * spelling: segment * spelling: series * spelling: specified * spelling: specify * spelling: subtracted * spelling: successful * spelling: return * spelling: translation * spelling: update * spelling: unrelated * spelling: useless * spelling: variables * spelling: variety * spelling: verbatim * spelling: verification * spelling: visited * spelling: warming * spelling: workers * spelling: with
2019-04-12 18:18:11 +00:00
* reads legacy dictionary builder parameters from *stringPtr (e.g. "--train-legacy=selectivity=8") into *selectivity
* @return 1 means that legacy dictionary builder parameters were correct
* @return 0 in case of malformed parameters
*/
static unsigned parseLegacyParameters(const char* stringPtr, unsigned* selectivity)
{
if (!longCommandWArg(&stringPtr, "s=") && !longCommandWArg(&stringPtr, "selectivity=")) { return 0; }
*selectivity = readU32FromChar(&stringPtr);
if (stringPtr[0] != 0) return 0;
DISPLAYLEVEL(4, "legacy: selectivity=%u\n", *selectivity);
return 1;
}
static ZDICT_cover_params_t defaultCoverParams(void)
{
ZDICT_cover_params_t params;
memset(&params, 0, sizeof(params));
params.d = 8;
params.steps = 4;
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
params.splitPoint = 1.0;
params.shrinkDict = 0;
params.shrinkDictMaxRegression = kDefaultRegression;
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
return params;
}
static ZDICT_fastCover_params_t defaultFastCoverParams(void)
{
ZDICT_fastCover_params_t params;
memset(&params, 0, sizeof(params));
params.d = 8;
2018-09-05 00:12:35 +00:00
params.f = 20;
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
params.steps = 4;
params.splitPoint = 0.75; /* different from default splitPoint of cover */
params.accel = DEFAULT_ACCEL;
params.shrinkDict = 0;
params.shrinkDictMaxRegression = kDefaultRegression;
return params;
}
2017-01-01 05:08:24 +00:00
#endif
/** parseAdaptParameters() :
* reads adapt parameters from *stringPtr (e.g. "--zstd=min=1,max=19) and store them into adaptMinPtr and adaptMaxPtr.
* Both adaptMinPtr and adaptMaxPtr must be already allocated and correctly initialized.
* There is no guarantee that any of these values will be updated.
* @return 1 means that parsing was successful,
* @return 0 in case of malformed parameters
*/
static unsigned parseAdaptParameters(const char* stringPtr, int* adaptMinPtr, int* adaptMaxPtr)
{
for ( ; ;) {
2019-10-17 22:27:25 +00:00
if (longCommandWArg(&stringPtr, "min=")) { *adaptMinPtr = (int)readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "max=")) { *adaptMaxPtr = (int)readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
DISPLAYLEVEL(4, "invalid compression parameter \n");
return 0;
}
if (stringPtr[0] != 0) return 0; /* check the end of string */
if (*adaptMinPtr > *adaptMaxPtr) {
DISPLAYLEVEL(4, "incoherent adaptation limits \n");
return 0;
}
return 1;
}
2016-12-13 11:18:07 +00:00
/** parseCompressionParameters() :
* reads compression parameters from *stringPtr (e.g. "--zstd=wlog=23,clog=23,hlog=22,slog=6,mml=3,tlen=48,strat=6") into *params
2016-12-13 11:18:07 +00:00
* @return 1 means that compression parameters were correct
* @return 0 in case of malformed parameters
*/
static unsigned parseCompressionParameters(const char* stringPtr, ZSTD_compressionParameters* params)
{
for ( ; ;) {
if (longCommandWArg(&stringPtr, "windowLog=") || longCommandWArg(&stringPtr, "wlog=")) { params->windowLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "chainLog=") || longCommandWArg(&stringPtr, "clog=")) { params->chainLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "hashLog=") || longCommandWArg(&stringPtr, "hlog=")) { params->hashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "searchLog=") || longCommandWArg(&stringPtr, "slog=")) { params->searchLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "minMatch=") || longCommandWArg(&stringPtr, "mml=")) { params->minMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
2016-12-13 11:18:07 +00:00
if (longCommandWArg(&stringPtr, "targetLength=") || longCommandWArg(&stringPtr, "tlen=")) { params->targetLength = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "strategy=") || longCommandWArg(&stringPtr, "strat=")) { params->strategy = (ZSTD_strategy)(readU32FromChar(&stringPtr)); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "overlapLog=") || longCommandWArg(&stringPtr, "ovlog=")) { g_overlapLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "ldmHashLog=") || longCommandWArg(&stringPtr, "lhlog=")) { g_ldmHashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "ldmMinMatch=") || longCommandWArg(&stringPtr, "lmml=")) { g_ldmMinMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
if (longCommandWArg(&stringPtr, "ldmBucketSizeLog=") || longCommandWArg(&stringPtr, "lblog=")) { g_ldmBucketSizeLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
2018-11-21 22:36:57 +00:00
if (longCommandWArg(&stringPtr, "ldmHashRateLog=") || longCommandWArg(&stringPtr, "lhrlog=")) { g_ldmHashRateLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; }
DISPLAYLEVEL(4, "invalid compression parameter \n");
2016-12-13 11:18:07 +00:00
return 0;
}
DISPLAYLEVEL(4, "windowLog=%d, chainLog=%d, hashLog=%d, searchLog=%d \n", params->windowLog, params->chainLog, params->hashLog, params->searchLog);
DISPLAYLEVEL(4, "minMatch=%d, targetLength=%d, strategy=%d \n", params->minMatch, params->targetLength, params->strategy);
2016-12-13 11:18:07 +00:00
if (stringPtr[0] != 0) return 0; /* check the end of string */
return 1;
}
static void printVersion(void)
{
if (g_displayLevel < DISPLAY_LEVEL_DEFAULT) {
DISPLAYOUT("%s\n", ZSTD_VERSION_STRING);
return;
}
DISPLAYOUT(WELCOME_MESSAGE);
if (g_displayLevel >= 3) {
/* format support */
DISPLAYOUT("*** supports: zstd");
#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>0) && (ZSTD_LEGACY_SUPPORT<8)
DISPLAYOUT(", zstd legacy v0.%d+", ZSTD_LEGACY_SUPPORT);
#endif
#ifdef ZSTD_GZCOMPRESS
DISPLAYOUT(", gzip");
#endif
#ifdef ZSTD_LZ4COMPRESS
DISPLAYOUT(", lz4");
#endif
#ifdef ZSTD_LZMACOMPRESS
DISPLAYOUT(", lzma, xz ");
#endif
DISPLAYOUT("\n");
if (g_displayLevel >= 4) {
/* posix support */
#ifdef _POSIX_C_SOURCE
DISPLAYOUT("_POSIX_C_SOURCE defined: %ldL\n", (long) _POSIX_C_SOURCE);
#endif
#ifdef _POSIX_VERSION
DISPLAYOUT("_POSIX_VERSION defined: %ldL \n", (long) _POSIX_VERSION);
#endif
#ifdef PLATFORM_POSIX_VERSION
DISPLAYOUT("PLATFORM_POSIX_VERSION defined: %ldL\n", (long) PLATFORM_POSIX_VERSION);
#endif
} }
}
2016-12-13 11:18:07 +00:00
/* Environment variables for parameter setting */
#define ENV_CLEVEL "ZSTD_CLEVEL"
/* pick up environment variable */
2018-12-20 01:56:45 +00:00
static int init_cLevel(void) {
const char* const env = getenv(ENV_CLEVEL);
if (env != NULL) {
const char* ptr = env;
int sign = 1;
if (*ptr == '-') {
sign = -1;
ptr++;
} else if (*ptr == '+') {
ptr++;
}
if ((*ptr>='0') && (*ptr<='9')) {
unsigned absLevel;
if (readU32FromCharChecked(&ptr, &absLevel)) {
DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: numeric value too large \n", ENV_CLEVEL, env);
return ZSTDCLI_CLEVEL_DEFAULT;
} else if (*ptr == 0) {
2019-10-17 22:27:25 +00:00
return sign * (int)absLevel;
} }
DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: not a valid integer value \n", ENV_CLEVEL, env);
}
return ZSTDCLI_CLEVEL_DEFAULT;
}
#define NEXT_FIELD(ptr) { \
if (*argument == '=') { \
ptr = ++argument; \
argument += strlen(ptr); \
} else { \
argNb++; \
if (argNb >= argCount) { \
DISPLAY("error: missing command argument \n"); \
CLEAN_RETURN(1); \
} \
ptr = argv[argNb]; \
assert(ptr != NULL); \
if (ptr[0]=='-') { \
DISPLAY("error: command cannot be separated from its argument by another command \n"); \
CLEAN_RETURN(1); \
} } }
#define NEXT_UINT32(val32) { \
const char* __nb; \
NEXT_FIELD(__nb); \
val32 = readU32FromChar(&__nb); \
}
#define ZSTD_NB_STRATEGIES 9
static const char* ZSTD_strategyMap[ZSTD_NB_STRATEGIES + 1] = { "", "ZSTD_fast",
"ZSTD_dfast", "ZSTD_greedy", "ZSTD_lazy", "ZSTD_lazy2", "ZSTD_btlazy2",
"ZSTD_btopt", "ZSTD_btultra", "ZSTD_btultra2"};
typedef enum { zom_compress, zom_decompress, zom_test, zom_bench, zom_train, zom_list } zstd_operation_mode;
2016-05-29 03:16:05 +00:00
#define CLEAN_RETURN(i) { operationResult = (i); goto _end; }
#ifdef ZSTD_NOCOMPRESS
/* symbols from compression library are not defined and should not be invoked */
# define MINCLEVEL -99
# define MAXCLEVEL 22
#else
# define MINCLEVEL ZSTD_minCLevel()
# define MAXCLEVEL ZSTD_maxCLevel()
#endif
int main(int const argCount, const char* argv[])
2016-05-29 03:16:05 +00:00
{
int argNb,
followLinks = 0,
forceStdout = 0,
2018-02-09 23:53:27 +00:00
ldmFlag = 0,
main_pause = 0,
nbWorkers = 0,
2018-08-12 03:48:06 +00:00
adapt = 0,
adaptMin = MINCLEVEL,
adaptMax = MAXCLEVEL,
2018-11-13 03:59:42 +00:00
rsyncable = 0,
nextArgumentsAreFiles = 0,
operationResult = 0,
separateFiles = 0,
2018-02-09 23:53:27 +00:00
setRealTimePrio = 0,
singleThread = 0,
showDefaultCParams = 0,
2020-03-09 18:07:29 +00:00
ultra=0,
2020-03-09 19:12:52 +00:00
contentSize=1;
double compressibility = 0.5;
unsigned bench_nbSeconds = 3; /* would be better if this value was synchronized from bench */
size_t blockSize = 0;
FIO_prefs_t* const prefs = FIO_createPreferences();
2016-10-28 20:58:31 +00:00
zstd_operation_mode operation = zom_compress;
2016-12-13 11:18:07 +00:00
ZSTD_compressionParameters compressionParams;
int cLevel = init_cLevel();
int cLevelLast = MINCLEVEL - 1; /* lower than minimum */
2016-05-29 03:16:05 +00:00
unsigned recursive = 0;
unsigned memLimit = 0;
FileNamesTable* filenames = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */
FileNamesTable* file_of_names = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */
2016-05-29 03:16:05 +00:00
const char* programName = argv[0];
const char* outFileName = NULL;
const char* outDirName = NULL;
2020-06-20 02:35:51 +00:00
const char* outMirroredDirName = NULL;
2016-05-29 03:16:05 +00:00
const char* dictFileName = NULL;
const char* patchFromDictFileName = NULL;
2017-02-08 16:37:14 +00:00
const char* suffix = ZSTD_EXTENSION;
2016-05-29 03:16:05 +00:00
unsigned maxDictSize = g_defaultMaxDictSize;
unsigned dictID = 0;
2019-08-16 06:57:55 +00:00
size_t streamSrcSize = 0;
2019-06-24 20:40:52 +00:00
size_t targetCBlockSize = 0;
2019-08-19 15:52:08 +00:00
size_t srcSizeHint = 0;
int dictCLevel = g_defaultDictCLevel;
2016-05-29 03:16:05 +00:00
unsigned dictSelect = g_defaultSelectivityLevel;
2017-01-01 05:08:24 +00:00
#ifndef ZSTD_NODICT
ZDICT_cover_params_t coverParams = defaultCoverParams();
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
ZDICT_fastCover_params_t fastCoverParams = defaultFastCoverParams();
dictType dict = fastCover;
2017-01-01 05:08:24 +00:00
#endif
#ifndef ZSTD_NOBENCH
BMK_advancedParams_t benchParams = BMK_initAdvancedParams();
#endif
ZSTD_literalCompressionMode_e literalCompressionMode = ZSTD_lcm_auto;
2016-05-29 03:16:05 +00:00
2016-05-29 03:16:05 +00:00
/* init */
2016-06-04 23:05:01 +00:00
(void)recursive; (void)cLevelLast; /* not used when ZSTD_NOBENCH set */
(void)memLimit;
assert(argCount >= 1);
if ((filenames==NULL) || (file_of_names==NULL)) { DISPLAY("zstd: allocation error \n"); exit(1); }
programName = lastNameFromPath(programName);
#ifdef ZSTD_MULTITHREAD
nbWorkers = 1;
#endif
2016-05-29 03:16:05 +00:00
/* preset behaviors */
if (exeNameMatch(programName, ZSTD_ZSTDMT)) nbWorkers=0, singleThread=0;
if (exeNameMatch(programName, ZSTD_UNZSTD)) operation=zom_decompress;
if (exeNameMatch(programName, ZSTD_CAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; outFileName=stdoutmark; g_displayLevel=1; } /* supports multiple formats */
if (exeNameMatch(programName, ZSTD_ZCAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; outFileName=stdoutmark; g_displayLevel=1; } /* behave like zcat, also supports multiple formats */
if (exeNameMatch(programName, ZSTD_GZ)) { suffix = GZ_EXTENSION; FIO_setCompressionType(prefs, FIO_gzipCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like gzip */
if (exeNameMatch(programName, ZSTD_GUNZIP)) { operation=zom_decompress; FIO_setRemoveSrcFile(prefs, 1); } /* behave like gunzip, also supports multiple formats */
if (exeNameMatch(programName, ZSTD_GZCAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; outFileName=stdoutmark; g_displayLevel=1; } /* behave like gzcat, also supports multiple formats */
if (exeNameMatch(programName, ZSTD_LZMA)) { suffix = LZMA_EXTENSION; FIO_setCompressionType(prefs, FIO_lzmaCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like lzma */
if (exeNameMatch(programName, ZSTD_UNLZMA)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_lzmaCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like unlzma, also supports multiple formats */
if (exeNameMatch(programName, ZSTD_XZ)) { suffix = XZ_EXTENSION; FIO_setCompressionType(prefs, FIO_xzCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like xz */
if (exeNameMatch(programName, ZSTD_UNXZ)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_xzCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like unxz, also supports multiple formats */
if (exeNameMatch(programName, ZSTD_LZ4)) { suffix = LZ4_EXTENSION; FIO_setCompressionType(prefs, FIO_lz4Compression); } /* behave like lz4 */
if (exeNameMatch(programName, ZSTD_UNLZ4)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_lz4Compression); } /* behave like unlz4, also supports multiple formats */
2016-12-13 11:18:07 +00:00
memset(&compressionParams, 0, sizeof(compressionParams));
2016-05-29 03:16:05 +00:00
/* init crash handler */
2018-09-11 18:56:50 +00:00
FIO_addAbortHandler();
2016-05-29 03:16:05 +00:00
/* command switches */
2016-10-14 21:07:11 +00:00
for (argNb=1; argNb<argCount; argNb++) {
2016-05-29 03:16:05 +00:00
const char* argument = argv[argNb];
if (!argument) continue; /* Protection if argument empty */
if (nextArgumentsAreFiles) {
UTIL_refFilename(filenames, argument);
continue;
}
/* "-" means stdin/stdout */
if (!strcmp(argument, "-")){
UTIL_refFilename(filenames, stdinmark);
continue;
}
2016-05-29 03:16:05 +00:00
/* Decode commands (note : aggregated commands are allowed) */
if (argument[0]=='-') {
if (argument[1]=='-') {
/* long commands (--long-word) */
if (!strcmp(argument, "--")) { nextArgumentsAreFiles=1; continue; } /* only file names allowed from now on */
if (!strcmp(argument, "--list")) { operation=zom_list; continue; }
if (!strcmp(argument, "--compress")) { operation=zom_compress; continue; }
if (!strcmp(argument, "--decompress")) { operation=zom_decompress; continue; }
if (!strcmp(argument, "--uncompress")) { operation=zom_decompress; continue; }
if (!strcmp(argument, "--force")) { FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; continue; }
if (!strcmp(argument, "--version")) { printVersion(); CLEAN_RETURN(0); }
if (!strcmp(argument, "--help")) { usage_advanced(programName); CLEAN_RETURN(0); }
if (!strcmp(argument, "--verbose")) { g_displayLevel++; continue; }
if (!strcmp(argument, "--quiet")) { g_displayLevel--; continue; }
if (!strcmp(argument, "--stdout")) { forceStdout=1; outFileName=stdoutmark; g_displayLevel-=(g_displayLevel==2); continue; }
if (!strcmp(argument, "--ultra")) { ultra=1; continue; }
if (!strcmp(argument, "--check")) { FIO_setChecksumFlag(prefs, 2); continue; }
if (!strcmp(argument, "--no-check")) { FIO_setChecksumFlag(prefs, 0); continue; }
if (!strcmp(argument, "--sparse")) { FIO_setSparseWrite(prefs, 2); continue; }
if (!strcmp(argument, "--no-sparse")) { FIO_setSparseWrite(prefs, 0); continue; }
if (!strcmp(argument, "--test")) { operation=zom_test; continue; }
if (!strcmp(argument, "--train")) { operation=zom_train; if (outFileName==NULL) outFileName=g_defaultDictName; continue; }
if (!strcmp(argument, "--no-dictID")) { FIO_setDictIDFlag(prefs, 0); continue; }
if (!strcmp(argument, "--keep")) { FIO_setRemoveSrcFile(prefs, 0); continue; }
if (!strcmp(argument, "--rm")) { FIO_setRemoveSrcFile(prefs, 1); continue; }
if (!strcmp(argument, "--priority=rt")) { setRealTimePrio = 1; continue; }
if (!strcmp(argument, "--show-default-cparams")) { showDefaultCParams = 1; continue; }
2020-03-09 19:19:05 +00:00
if (!strcmp(argument, "--content-size")) { contentSize = 1; continue; }
2020-03-09 19:12:52 +00:00
if (!strcmp(argument, "--no-content-size")) { contentSize = 0; continue; }
if (!strcmp(argument, "--adapt")) { adapt = 1; continue; }
if (longCommandWArg(&argument, "--adapt=")) { adapt = 1; if (!parseAdaptParameters(argument, &adaptMin, &adaptMax)) { badusage(programName); CLEAN_RETURN(1); } continue; }
if (!strcmp(argument, "--single-thread")) { nbWorkers = 0; singleThread = 1; continue; }
if (!strcmp(argument, "--format=zstd")) { suffix = ZSTD_EXTENSION; FIO_setCompressionType(prefs, FIO_zstdCompression); continue; }
#ifdef ZSTD_GZCOMPRESS
if (!strcmp(argument, "--format=gzip")) { suffix = GZ_EXTENSION; FIO_setCompressionType(prefs, FIO_gzipCompression); continue; }
#endif
#ifdef ZSTD_LZMACOMPRESS
if (!strcmp(argument, "--format=lzma")) { suffix = LZMA_EXTENSION; FIO_setCompressionType(prefs, FIO_lzmaCompression); continue; }
if (!strcmp(argument, "--format=xz")) { suffix = XZ_EXTENSION; FIO_setCompressionType(prefs, FIO_xzCompression); continue; }
#endif
#ifdef ZSTD_LZ4COMPRESS
if (!strcmp(argument, "--format=lz4")) { suffix = LZ4_EXTENSION; FIO_setCompressionType(prefs, FIO_lz4Compression); continue; }
#endif
if (!strcmp(argument, "--rsyncable")) { rsyncable = 1; continue; }
if (!strcmp(argument, "--compress-literals")) { literalCompressionMode = ZSTD_lcm_huffman; continue; }
if (!strcmp(argument, "--no-compress-literals")) { literalCompressionMode = ZSTD_lcm_uncompressed; continue; }
if (!strcmp(argument, "--no-progress")) { FIO_setNoProgress(1); continue; }
if (!strcmp(argument, "--exclude-compressed")) { FIO_setExcludeCompressedFile(prefs, 1); continue; }
/* long commands with arguments */
#ifndef ZSTD_NODICT
if (longCommandWArg(&argument, "--train-cover")) {
operation = zom_train;
if (outFileName == NULL)
outFileName = g_defaultDictName;
dict = cover;
/* Allow optional arguments following an = */
if (*argument == 0) { memset(&coverParams, 0, sizeof(coverParams)); }
else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); }
else if (!parseCoverParameters(argument, &coverParams)) { badusage(programName); CLEAN_RETURN(1); }
continue;
}
if (longCommandWArg(&argument, "--train-fastcover")) {
operation = zom_train;
if (outFileName == NULL)
outFileName = g_defaultDictName;
dict = fastCover;
/* Allow optional arguments following an = */
if (*argument == 0) { memset(&fastCoverParams, 0, sizeof(fastCoverParams)); }
else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); }
else if (!parseFastCoverParameters(argument, &fastCoverParams)) { badusage(programName); CLEAN_RETURN(1); }
continue;
}
if (longCommandWArg(&argument, "--train-legacy")) {
operation = zom_train;
if (outFileName == NULL)
outFileName = g_defaultDictName;
dict = legacy;
/* Allow optional arguments following an = */
if (*argument == 0) { continue; }
else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); }
else if (!parseLegacyParameters(argument, &dictSelect)) { badusage(programName); CLEAN_RETURN(1); }
continue;
}
2017-01-01 05:08:24 +00:00
#endif
if (longCommandWArg(&argument, "--threads")) { NEXT_UINT32(nbWorkers); continue; }
if (longCommandWArg(&argument, "--memlimit")) { NEXT_UINT32(memLimit); continue; }
if (longCommandWArg(&argument, "--memory")) { NEXT_UINT32(memLimit); continue; }
if (longCommandWArg(&argument, "--memlimit-decompress")) { NEXT_UINT32(memLimit); continue; }
if (longCommandWArg(&argument, "--block-size=")) { blockSize = readSizeTFromChar(&argument); continue; }
if (longCommandWArg(&argument, "--maxdict")) { NEXT_UINT32(maxDictSize); continue; }
if (longCommandWArg(&argument, "--dictID")) { NEXT_UINT32(dictID); continue; }
if (longCommandWArg(&argument, "--zstd=")) { if (!parseCompressionParameters(argument, &compressionParams)) { badusage(programName); CLEAN_RETURN(1); } continue; }
if (longCommandWArg(&argument, "--stream-size=")) { streamSrcSize = readSizeTFromChar(&argument); continue; }
if (longCommandWArg(&argument, "--target-compressed-block-size=")) { targetCBlockSize = readSizeTFromChar(&argument); continue; }
if (longCommandWArg(&argument, "--size-hint=")) { srcSizeHint = readSizeTFromChar(&argument); continue; }
if (longCommandWArg(&argument, "--output-dir-flat")) { NEXT_FIELD(outDirName); continue; }
#ifdef UTIL_HAS_MIRRORFILELIST
if (longCommandWArg(&argument, "--output-dir-mirror")) { NEXT_FIELD(outMirroredDirName); continue; }
#endif
if (longCommandWArg(&argument, "--patch-from")) { NEXT_FIELD(patchFromDictFileName); continue; }
if (longCommandWArg(&argument, "--long")) {
unsigned ldmWindowLog = 0;
ldmFlag = 1;
/* Parse optional window log */
if (*argument == '=') {
++argument;
ldmWindowLog = readU32FromChar(&argument);
} else if (*argument != 0) {
/* Invalid character following --long */
badusage(programName);
CLEAN_RETURN(1);
}
/* Only set windowLog if not already set by --zstd */
if (compressionParams.windowLog == 0)
compressionParams.windowLog = ldmWindowLog;
continue;
}
#ifndef ZSTD_NOCOMPRESS /* linking ZSTD_minCLevel() requires compression support */
if (longCommandWArg(&argument, "--fast")) {
/* Parse optional acceleration factor */
if (*argument == '=') {
U32 const maxFast = (U32)-ZSTD_minCLevel();
U32 fastLevel;
++argument;
fastLevel = readU32FromChar(&argument);
if (fastLevel > maxFast) fastLevel = maxFast;
if (fastLevel) {
dictCLevel = cLevel = -(int)fastLevel;
} else {
badusage(programName);
CLEAN_RETURN(1);
}
} else if (*argument != 0) {
/* Invalid character following --fast */
badusage(programName);
CLEAN_RETURN(1);
} else {
cLevel = -1; /* default for --fast */
}
continue;
}
#endif
2019-10-15 06:49:13 +00:00
if (longCommandWArg(&argument, "--filelist")) {
const char* listName;
NEXT_FIELD(listName);
UTIL_refFilename(file_of_names, listName);
continue;
}
2019-10-15 06:49:13 +00:00
/* fall-through, will trigger bad_usage() later on */
}
2019-10-15 06:49:13 +00:00
argument++;
while (argument[0]!=0) {
#ifndef ZSTD_NOCOMPRESS
/* compression Level */
if ((*argument>='0') && (*argument<='9')) {
dictCLevel = cLevel = (int)readU32FromChar(&argument);
continue;
}
#endif
2016-05-29 03:16:05 +00:00
switch(argument[0])
{
/* Display help */
case 'V': printVersion(); CLEAN_RETURN(0); /* Version Only */
case 'H':
case 'h': usage_advanced(programName); CLEAN_RETURN(0);
2016-05-29 03:16:05 +00:00
/* Compress */
case 'z': operation=zom_compress; argument++; break;
2016-10-28 20:58:31 +00:00
/* Decoding */
case 'd':
#ifndef ZSTD_NOBENCH
benchParams.mode = BMK_decodeOnly;
if (operation==zom_bench) { argument++; break; } /* benchmark decode (hidden option) */
#endif
operation=zom_decompress; argument++; break;
2016-05-29 03:16:05 +00:00
/* Force stdout, even if stdout==console */
case 'c': forceStdout=1; outFileName=stdoutmark; argument++; break;
2016-05-29 03:16:05 +00:00
/* Use file content as dictionary */
case 'D': argument++; NEXT_FIELD(dictFileName); break;
2016-05-29 03:16:05 +00:00
/* Overwrite */
case 'f': FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; argument++; break;
2016-05-29 03:16:05 +00:00
/* Verbose mode */
case 'v': g_displayLevel++; argument++; break;
2016-05-29 03:16:05 +00:00
/* Quiet mode */
case 'q': g_displayLevel--; argument++; break;
2016-05-29 03:16:05 +00:00
/* keep source file (default) */
case 'k': FIO_setRemoveSrcFile(prefs, 0); argument++; break;
2016-05-29 03:16:05 +00:00
/* Checksum */
case 'C': FIO_setChecksumFlag(prefs, 2); argument++; break;
/* test compressed file */
case 't': operation=zom_test; argument++; break;
2016-05-29 03:16:05 +00:00
/* destination file name */
case 'o': argument++; NEXT_FIELD(outFileName); break;
2019-10-17 22:27:25 +00:00
/* limit memory */
case 'M':
argument++;
memLimit = readU32FromChar(&argument);
break;
case 'l': operation=zom_list; argument++; break;
#ifdef UTIL_HAS_CREATEFILELIST
/* recursive */
case 'r': recursive=1; argument++; break;
#endif
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOBENCH
/* Benchmark */
case 'b':
operation=zom_bench;
argument++;
break;
/* range bench (benchmark only) */
case 'e':
/* compression Level */
argument++;
cLevelLast = (int)readU32FromChar(&argument);
break;
/* Modify Nb Iterations (benchmark only) */
case 'i':
argument++;
bench_nbSeconds = readU32FromChar(&argument);
break;
/* cut input into blocks (benchmark only) */
case 'B':
argument++;
blockSize = readU32FromChar(&argument);
break;
/* benchmark files separately (hidden option) */
case 'S':
argument++;
separateFiles = 1;
break;
#endif /* ZSTD_NOBENCH */
/* nb of threads (hidden option) */
case 'T':
argument++;
nbWorkers = (int)readU32FromChar(&argument);
break;
2016-05-29 03:16:05 +00:00
/* Dictionary Selection level */
case 's':
argument++;
dictSelect = readU32FromChar(&argument);
break;
2016-05-29 03:16:05 +00:00
/* Pause at the end (-p) or set an additional param (-p#) (hidden option) */
case 'p': argument++;
#ifndef ZSTD_NOBENCH
if ((*argument>='0') && (*argument<='9')) {
benchParams.additionalParam = (int)readU32FromChar(&argument);
} else
#endif
main_pause=1;
break;
/* Select compressibility of synthetic sample */
case 'P':
argument++;
compressibility = (double)readU32FromChar(&argument) / 100;
break;
/* unknown command */
default : badusage(programName); CLEAN_RETURN(1);
}
}
continue;
} /* if (argument[0]=='-') */
/* none of the above : add filename to list */
UTIL_refFilename(filenames, argument);
2016-05-29 03:16:05 +00:00
}
/* Welcome message (if verbose) */
DISPLAYLEVEL(3, WELCOME_MESSAGE);
#ifdef ZSTD_MULTITHREAD
if ((nbWorkers==0) && (!singleThread)) {
/* automatically set # workers based on # of reported cpus */
nbWorkers = UTIL_countPhysicalCores();
DISPLAYLEVEL(3, "Note: %d physical core(s) detected \n", nbWorkers);
2017-04-13 19:28:28 +00:00
}
2018-06-26 08:22:45 +00:00
#else
(void)singleThread; (void)nbWorkers;
#endif
2018-06-26 08:22:45 +00:00
#ifdef UTIL_HAS_CREATEFILELIST
2017-03-23 19:09:35 +00:00
g_utilDisplayLevel = g_displayLevel;
if (!followLinks) {
unsigned u, fileNamesNb;
unsigned const nbFilenames = (unsigned)filenames->tableSize;
for (u=0, fileNamesNb=0; u<nbFilenames; u++) {
2019-11-26 19:20:26 +00:00
if ( UTIL_isLink(filenames->fileNames[u])
&& !UTIL_isFIFO(filenames->fileNames[u])
) {
2019-11-26 19:20:26 +00:00
DISPLAYLEVEL(2, "Warning : %s is a symbolic link, ignoring \n", filenames->fileNames[u]);
} else {
filenames->fileNames[fileNamesNb++] = filenames->fileNames[u];
2019-10-26 01:16:45 +00:00
} }
if (fileNamesNb == 0 && nbFilenames > 0) /* all names are eliminated */
CLEAN_RETURN(1);
filenames->tableSize = fileNamesNb;
} /* if (!followLinks) */
/* read names from a file */
if (file_of_names->tableSize) {
size_t const nbFileLists = file_of_names->tableSize;
size_t flNb;
for (flNb=0; flNb < nbFileLists; flNb++) {
FileNamesTable* const fnt = UTIL_createFileNamesTable_fromFileName(file_of_names->fileNames[flNb]);
if (fnt==NULL) {
DISPLAYLEVEL(1, "zstd: error reading %s \n", file_of_names->fileNames[flNb]);
CLEAN_RETURN(1);
}
filenames = UTIL_mergeFileNamesTable(filenames, fnt);
}
}
2016-09-21 10:24:43 +00:00
if (recursive) { /* at this stage, filenameTable is a list of paths, which can contain both files and directories */
UTIL_expandFNT(&filenames, followLinks);
}
2018-06-26 08:22:45 +00:00
#else
(void)followLinks;
2016-05-29 03:16:05 +00:00
#endif
if (operation == zom_list) {
#ifndef ZSTD_NODECOMPRESS
int const ret = FIO_listMultipleFiles((unsigned)filenames->tableSize, filenames->fileNames, g_displayLevel);
CLEAN_RETURN(ret);
#else
DISPLAY("file information is not supported \n");
CLEAN_RETURN(1);
#endif
}
2016-05-29 03:16:05 +00:00
/* Check if benchmark is selected */
2016-10-28 20:58:31 +00:00
if (operation==zom_bench) {
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOBENCH
benchParams.blockSize = blockSize;
benchParams.nbWorkers = nbWorkers;
2019-10-17 22:27:25 +00:00
benchParams.realTime = (unsigned)setRealTimePrio;
benchParams.nbSeconds = bench_nbSeconds;
benchParams.ldmFlag = ldmFlag;
2019-10-17 22:27:25 +00:00
benchParams.ldmMinMatch = (int)g_ldmMinMatch;
benchParams.ldmHashLog = (int)g_ldmHashLog;
2017-09-03 04:10:36 +00:00
if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) {
2019-10-17 22:27:25 +00:00
benchParams.ldmBucketSizeLog = (int)g_ldmBucketSizeLog;
2017-09-03 04:10:36 +00:00
}
2018-11-21 22:36:57 +00:00
if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) {
2019-10-17 22:27:25 +00:00
benchParams.ldmHashRateLog = (int)g_ldmHashRateLog;
2017-09-03 04:10:36 +00:00
}
benchParams.literalCompressionMode = literalCompressionMode;
if (cLevel > ZSTD_maxCLevel()) cLevel = ZSTD_maxCLevel();
if (cLevelLast > ZSTD_maxCLevel()) cLevelLast = ZSTD_maxCLevel();
if (cLevelLast < cLevel) cLevelLast = cLevel;
if (cLevelLast > cLevel)
DISPLAYLEVEL(3, "Benchmarking levels from %d to %d\n", cLevel, cLevelLast);
if (filenames->tableSize > 0) {
if(separateFiles) {
unsigned i;
for(i = 0; i < filenames->tableSize; i++) {
int c;
DISPLAYLEVEL(3, "Benchmarking %s \n", filenames->fileNames[i]);
for(c = cLevel; c <= cLevelLast; c++) {
BMK_benchFilesAdvanced(&filenames->fileNames[i], 1, dictFileName, c, &compressionParams, g_displayLevel, &benchParams);
} }
} else {
for(; cLevel <= cLevelLast; cLevel++) {
BMK_benchFilesAdvanced(filenames->fileNames, (unsigned)filenames->tableSize, dictFileName, cLevel, &compressionParams, g_displayLevel, &benchParams);
} }
} else {
for(; cLevel <= cLevelLast; cLevel++) {
BMK_syntheticTest(cLevel, compressibility, &compressionParams, g_displayLevel, &benchParams);
} }
#else
2018-08-03 15:30:01 +00:00
(void)bench_nbSeconds; (void)blockSize; (void)setRealTimePrio; (void)separateFiles; (void)compressibility;
2016-05-29 03:16:05 +00:00
#endif
goto _end;
}
/* Check if dictionary builder is selected */
2016-10-28 20:58:31 +00:00
if (operation==zom_train) {
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NODICT
ZDICT_params_t zParams;
zParams.compressionLevel = dictCLevel;
2019-10-17 22:27:25 +00:00
zParams.notificationLevel = (unsigned)g_displayLevel;
zParams.dictID = dictID;
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
if (dict == cover) {
int const optimize = !coverParams.k || !coverParams.d;
2019-10-17 22:27:25 +00:00
coverParams.nbThreads = (unsigned)nbWorkers;
coverParams.zParams = zParams;
operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, NULL, &coverParams, NULL, optimize);
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
} else if (dict == fastCover) {
int const optimize = !fastCoverParams.k || !fastCoverParams.d;
2019-10-17 22:27:25 +00:00
fastCoverParams.nbThreads = (unsigned)nbWorkers;
Merge fastCover into DictBuilder (#1274) * Minor fix * Run non-optimize FASTCOVER 5 times in benchmark * Merge fastCover into dictBuilder * Fix mixed declaration issue * Add fastcover to symbol.c * Add fastCover.c and cover.h to build * Change fastCover.c to fastcover.c * Update benchmark to run FASTCOVER in dictBuilder * Undo spliting fastcover_param into cover_param and f * Remove convert param functions * Assign f to parameter * Add zdict.h to Makefile in lib * Add cover.h to BUCK * Cast 1 to U64 before shifting * Remove trimming of zero freq head and tail in selectSegment and rebenchmark * Remove f as a separate parameter of tryParam * Read 8 bytes when d is 6 * Add trimming off zero frequency head and tail * Use best functions from COVER and remove trimming part(which leads to worse compression ratio after previous bugs were fixed) * Add finalize= argument to FASTCOVER to specify percentage of training samples passed to ZDICT_finalizeDictionary * Change nbDmer to always read 8 bytes even when d=6 * Add skip=# argument to allow skipping dmers in computeFrequency in FASTCOVER * Update comments and benchmarking result * Change default method of ZDICT_trainFromBuffer to ZDICT_optimizeTrainFromBuffer_fastCover * Add dictType enum and fix bug about passing zParam when converting to coverParam * Combine finalize and skip into a single parameter * Update acceleration parameters and benchmark on 3 sample sets * Change default splitPoint of FASTCOVER to 0.75 and benchmark first 3 sample sets * Initialize variables outside of for loop in benchmark.c * Update benchmark result for hg-manifest * Remove cover.h from install-includes * Add explanation of f * Set default compression level for trainFromBuffer to 3 * Add assertion of fastCoverParams in DiB_trainFromFiles * Add checkTotalCompressedSize function + some minor fixes * Add test for multithreading fastCovr * Initialize segmentFreqs in every FASTCOVER_selectSegment and move mutex_unnlock to end of COVER_best_finish * Free segmentFreqs * Initialize segmentFreqs before calling FASTCOVER_buildDictionary instead of in FASTCOVER_selectSegment * Add FASTCOVER_MEMMULT * Minor fix * Update benchmarking result
2018-08-23 19:06:20 +00:00
fastCoverParams.zParams = zParams;
operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, NULL, NULL, &fastCoverParams, optimize);
2017-01-01 05:08:24 +00:00
} else {
ZDICT_legacy_params_t dictParams;
2017-01-01 05:08:24 +00:00
memset(&dictParams, 0, sizeof(dictParams));
dictParams.selectivityLevel = dictSelect;
dictParams.zParams = zParams;
operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, &dictParams, NULL, NULL, 0);
2017-01-01 05:08:24 +00:00
}
2018-09-24 07:52:19 +00:00
#else
(void)dictCLevel; (void)dictSelect; (void)dictID; (void)maxDictSize; /* not used when ZSTD_NODICT set */
DISPLAYLEVEL(1, "training mode not available \n");
operationResult = 1;
2016-05-29 03:16:05 +00:00
#endif
goto _end;
}
2017-07-18 21:45:49 +00:00
#ifndef ZSTD_NODECOMPRESS
if (operation==zom_test) { FIO_setTestMode(prefs, 1); outFileName=nulmark; FIO_setRemoveSrcFile(prefs, 0); } /* test mode */
2017-07-18 21:45:49 +00:00
#endif
2016-05-29 03:16:05 +00:00
/* No input filename ==> use stdin and stdout */
if (filenames->tableSize == 0) UTIL_refFilename(filenames, stdinmark);
if (!strcmp(filenames->fileNames[0], stdinmark) && !outFileName)
outFileName = stdoutmark; /* when input is stdin, default output is stdout */
2016-05-29 03:16:05 +00:00
/* Check if input/output defined as console; trigger an error in this case */
if (!strcmp(filenames->fileNames[0], stdinmark) && IS_CONSOLE(stdin) ) {
badusage(programName);
CLEAN_RETURN(1);
}
if ( outFileName && !strcmp(outFileName, stdoutmark)
&& IS_CONSOLE(stdout)
&& !strcmp(filenames->fileNames[0], stdinmark)
&& !forceStdout
&& operation!=zom_decompress ) {
badusage(programName);
CLEAN_RETURN(1);
}
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NOCOMPRESS
2016-08-12 16:04:15 +00:00
/* check compression level limits */
{ int const maxCLevel = ultra ? ZSTD_maxCLevel() : ZSTDCLI_CLEVEL_MAX;
if (cLevel > maxCLevel) {
DISPLAYLEVEL(2, "Warning : compression level higher than max, reduced to %i \n", maxCLevel);
cLevel = maxCLevel;
} }
#endif
2016-08-12 16:04:15 +00:00
if (showDefaultCParams) {
if (operation == zom_decompress) {
DISPLAY("error : can't use --show-default-cparams in decomrpession mode \n");
CLEAN_RETURN(1);
}
}
if (dictFileName != NULL && patchFromDictFileName != NULL) {
DISPLAY("error : can't use -D and --patch-from=# at the same time \n");
CLEAN_RETURN(1);
}
if (patchFromDictFileName != NULL && filenames->tableSize > 1) {
DISPLAY("error : can't use --patch-from=# on multiple files \n");
CLEAN_RETURN(1);
}
/* No status message in pipe mode (stdin - stdout) or multi-files mode */
if (!strcmp(filenames->fileNames[0], stdinmark) && outFileName && !strcmp(outFileName,stdoutmark) && (g_displayLevel==2)) g_displayLevel=1;
if ((filenames->tableSize > 1) & (g_displayLevel==2)) g_displayLevel=1;
2016-05-29 03:16:05 +00:00
/* IO Stream/File */
FIO_setNotificationLevel(g_displayLevel);
FIO_setPatchFromMode(prefs, patchFromDictFileName != NULL);
if (memLimit == 0) {
if (compressionParams.windowLog == 0) {
memLimit = (U32)1 << g_defaultMaxWindowLog;
} else {
memLimit = (U32)1 << (compressionParams.windowLog & 31);
} }
if (patchFromDictFileName != NULL)
dictFileName = patchFromDictFileName;
FIO_setMemLimit(prefs, memLimit);
2016-10-28 20:58:31 +00:00
if (operation==zom_compress) {
2016-09-21 10:24:43 +00:00
#ifndef ZSTD_NOCOMPRESS
2020-03-09 19:12:52 +00:00
FIO_setContentSize(prefs, contentSize);
FIO_setNbWorkers(prefs, nbWorkers);
2019-10-17 22:27:25 +00:00
FIO_setBlockSize(prefs, (int)blockSize);
if (g_overlapLog!=OVERLAP_LOG_DEFAULT) FIO_setOverlapLog(prefs, (int)g_overlapLog);
FIO_setLdmFlag(prefs, (unsigned)ldmFlag);
FIO_setLdmHashLog(prefs, (int)g_ldmHashLog);
FIO_setLdmMinMatch(prefs, (int)g_ldmMinMatch);
if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) FIO_setLdmBucketSizeLog(prefs, (int)g_ldmBucketSizeLog);
if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) FIO_setLdmHashRateLog(prefs, (int)g_ldmHashRateLog);
FIO_setAdaptiveMode(prefs, (unsigned)adapt);
FIO_setAdaptMin(prefs, adaptMin);
FIO_setAdaptMax(prefs, adaptMax);
FIO_setRsyncable(prefs, rsyncable);
2019-08-16 06:57:55 +00:00
FIO_setStreamSrcSize(prefs, streamSrcSize);
2019-06-24 20:40:52 +00:00
FIO_setTargetCBlockSize(prefs, targetCBlockSize);
2019-08-19 15:52:08 +00:00
FIO_setSrcSizeHint(prefs, srcSizeHint);
FIO_setLiteralCompressionMode(prefs, literalCompressionMode);
if (adaptMin > cLevel) cLevel = adaptMin;
if (adaptMax < cLevel) cLevel = adaptMax;
/* Compare strategies constant with the ground truth */
{ ZSTD_bounds strategyBounds = ZSTD_cParam_getBounds(ZSTD_c_strategy);
2020-05-08 20:42:15 +00:00
assert(ZSTD_NB_STRATEGIES == strategyBounds.upperBound);
(void)strategyBounds; }
if (showDefaultCParams) {
size_t fileNb;
for (fileNb = 0; fileNb < (size_t)filenames->tableSize; fileNb++) {
unsigned long long fileSize = UTIL_getFileSize(filenames->fileNames[fileNb]);
const size_t dictSize = dictFileName != NULL ? (size_t)UTIL_getFileSize(dictFileName) : 0;
2020-01-31 01:05:33 +00:00
const ZSTD_compressionParameters cParams = ZSTD_getCParams(cLevel, fileSize, dictSize);
if (fileSize != UTIL_FILESIZE_UNKNOWN) DISPLAY("%s (%u bytes)\n", filenames->fileNames[fileNb], (unsigned)fileSize);
else DISPLAY("%s (src size unknown)\n", filenames->fileNames[fileNb]);
DISPLAY(" - windowLog : %u\n", cParams.windowLog);
DISPLAY(" - chainLog : %u\n", cParams.chainLog);
DISPLAY(" - hashLog : %u\n", cParams.hashLog);
DISPLAY(" - searchLog : %u\n", cParams.searchLog);
DISPLAY(" - minMatch : %u\n", cParams.minMatch);
DISPLAY(" - targetLength : %u\n", cParams.targetLength);
assert(cParams.strategy < ZSTD_NB_STRATEGIES + 1);
DISPLAY(" - strategy : %s (%u)\n", ZSTD_strategyMap[(int)cParams.strategy], (unsigned)cParams.strategy);
}
}
if ((filenames->tableSize==1) && outFileName)
operationResult = FIO_compressFilename(prefs, outFileName, filenames->fileNames[0], dictFileName, cLevel, compressionParams);
2016-05-29 03:16:05 +00:00
else
2020-06-20 02:35:51 +00:00
operationResult = FIO_compressMultipleFilenames(prefs, filenames->fileNames, (unsigned)filenames->tableSize, outMirroredDirName, outDirName, outFileName, suffix, dictFileName, cLevel, compressionParams);
2016-09-21 10:24:43 +00:00
#else
2020-03-09 19:12:52 +00:00
(void)contentSize; (void)suffix; (void)adapt; (void)rsyncable; (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; (void)targetCBlockSize; (void)streamSrcSize; (void)srcSizeHint; (void)ZSTD_strategyMap; /* not used when ZSTD_NOCOMPRESS set */
2018-09-24 23:56:45 +00:00
DISPLAY("Compression not supported \n");
2016-05-29 03:16:05 +00:00
#endif
2016-10-28 20:58:31 +00:00
} else { /* decompression or test */
2016-05-29 03:16:05 +00:00
#ifndef ZSTD_NODECOMPRESS
if (filenames->tableSize == 1 && outFileName) {
operationResult = FIO_decompressFilename(prefs, outFileName, filenames->fileNames[0], dictFileName);
} else {
2020-06-20 02:35:51 +00:00
operationResult = FIO_decompressMultipleFilenames(prefs, filenames->fileNames, (unsigned)filenames->tableSize, outMirroredDirName, outDirName, outFileName, dictFileName);
}
2016-05-29 03:16:05 +00:00
#else
2018-09-24 23:56:45 +00:00
DISPLAY("Decompression not supported \n");
2016-05-29 03:16:05 +00:00
#endif
}
_end:
FIO_freePreferences(prefs);
2016-05-29 03:16:05 +00:00
if (main_pause) waitEnter();
UTIL_freeFileNamesTable(filenames);
UTIL_freeFileNamesTable(file_of_names);
2019-10-26 01:16:45 +00:00
2016-05-29 03:16:05 +00:00
return operationResult;
}