1 /* 2 * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc. 3 * All rights reserved. 4 * 5 * This source code is licensed under both the BSD-style license (found in the 6 * LICENSE file in the root directory of this source tree) and the GPLv2 (found 7 * in the COPYING file in the root directory of this source tree). 8 * You may select, at your option, one of the above-listed licenses. 9 */ 10 11 12 /*-************************************ 13 * Tuning parameters 14 **************************************/ 15 #ifndef ZSTDCLI_CLEVEL_DEFAULT 16 # define ZSTDCLI_CLEVEL_DEFAULT 3 17 #endif 18 19 #ifndef ZSTDCLI_CLEVEL_MAX 20 # define ZSTDCLI_CLEVEL_MAX 19 /* without using --ultra */ 21 #endif 22 23 24 25 /*-************************************ 26 * Dependencies 27 **************************************/ 28 #include "platform.h" /* IS_CONSOLE, PLATFORM_POSIX_VERSION */ 29 #include "util.h" /* UTIL_HAS_CREATEFILELIST, UTIL_createFileList */ 30 #include <stdlib.h> /* getenv */ 31 #include <string.h> /* strcmp, strlen */ 32 #include <stdio.h> /* fprintf(), stdin, stdout, stderr */ 33 #include <errno.h> /* errno */ 34 #include <assert.h> /* assert */ 35 36 #include "fileio.h" /* stdinmark, stdoutmark, ZSTD_EXTENSION */ 37 #ifndef ZSTD_NOBENCH 38 # include "benchzstd.h" /* BMK_benchFiles */ 39 #endif 40 #ifndef ZSTD_NODICT 41 # include "dibio.h" /* ZDICT_cover_params_t, DiB_trainFromFiles() */ 42 #endif 43 #include "../lib/zstd.h" /* ZSTD_VERSION_STRING, ZSTD_minCLevel, ZSTD_maxCLevel */ 44 45 46 /*-************************************ 47 * Constants 48 **************************************/ 49 #define COMPRESSOR_NAME "zstd command line interface" 50 #ifndef ZSTD_VERSION 51 # define ZSTD_VERSION "v" ZSTD_VERSION_STRING 52 #endif 53 #define AUTHOR "Yann Collet" 54 #define WELCOME_MESSAGE "*** %s %i-bits %s, by %s ***\n", COMPRESSOR_NAME, (int)(sizeof(size_t)*8), ZSTD_VERSION, AUTHOR 55 56 #define ZSTD_ZSTDMT "zstdmt" 57 #define ZSTD_UNZSTD "unzstd" 58 #define ZSTD_CAT "zstdcat" 59 #define ZSTD_ZCAT "zcat" 60 #define ZSTD_GZ "gzip" 61 #define ZSTD_GUNZIP "gunzip" 62 #define ZSTD_GZCAT "gzcat" 63 #define ZSTD_LZMA "lzma" 64 #define ZSTD_UNLZMA "unlzma" 65 #define ZSTD_XZ "xz" 66 #define ZSTD_UNXZ "unxz" 67 #define ZSTD_LZ4 "lz4" 68 #define ZSTD_UNLZ4 "unlz4" 69 70 #define KB *(1 <<10) 71 #define MB *(1 <<20) 72 #define GB *(1U<<30) 73 74 #define DISPLAY_LEVEL_DEFAULT 2 75 76 static const char* g_defaultDictName = "dictionary"; 77 static const unsigned g_defaultMaxDictSize = 110 KB; 78 static const int g_defaultDictCLevel = 3; 79 static const unsigned g_defaultSelectivityLevel = 9; 80 static const unsigned g_defaultMaxWindowLog = 27; 81 #define OVERLAP_LOG_DEFAULT 9999 82 #define LDM_PARAM_DEFAULT 9999 /* Default for parameters where 0 is valid */ 83 static U32 g_overlapLog = OVERLAP_LOG_DEFAULT; 84 static U32 g_ldmHashLog = 0; 85 static U32 g_ldmMinMatch = 0; 86 static U32 g_ldmHashRateLog = LDM_PARAM_DEFAULT; 87 static U32 g_ldmBucketSizeLog = LDM_PARAM_DEFAULT; 88 89 90 #define DEFAULT_ACCEL 1 91 92 typedef enum { cover, fastCover, legacy } dictType; 93 94 /*-************************************ 95 * Display Macros 96 **************************************/ 97 #define DISPLAY_F(f, ...) fprintf((f), __VA_ARGS__) 98 #define DISPLAYOUT(...) DISPLAY_F(stdout, __VA_ARGS__) 99 #define DISPLAY(...) DISPLAY_F(stderr, __VA_ARGS__) 100 #define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } } 101 static int g_displayLevel = DISPLAY_LEVEL_DEFAULT; /* 0 : no display, 1: errors, 2 : + result + interaction + warnings, 3 : + progression, 4 : + information */ 102 103 104 /*-************************************ 105 * Command Line 106 **************************************/ 107 /* print help either in `stderr` or `stdout` depending on originating request 108 * error (badusage) => stderr 109 * help (usage_advanced) => stdout 110 */ 111 static void usage(FILE* f, const char* programName) 112 { 113 DISPLAY_F(f, "Usage : \n"); 114 DISPLAY_F(f, " %s [args] [FILE(s)] [-o file] \n", programName); 115 DISPLAY_F(f, "\n"); 116 DISPLAY_F(f, "FILE : a filename \n"); 117 DISPLAY_F(f, " with no FILE, or when FILE is - , read standard input\n"); 118 DISPLAY_F(f, "Arguments : \n"); 119 #ifndef ZSTD_NOCOMPRESS 120 DISPLAY_F(f, " -# : # compression level (1-%d, default: %d) \n", ZSTDCLI_CLEVEL_MAX, ZSTDCLI_CLEVEL_DEFAULT); 121 #endif 122 #ifndef ZSTD_NODECOMPRESS 123 DISPLAY_F(f, " -d : decompression \n"); 124 #endif 125 DISPLAY_F(f, " -D DICT: use DICT as Dictionary for compression or decompression \n"); 126 DISPLAY_F(f, " -o file: result stored into `file` (only 1 output file) \n"); 127 DISPLAY_F(f, " -f : overwrite output without prompting, also (de)compress links \n"); 128 DISPLAY_F(f, "--rm : remove source file(s) after successful de/compression \n"); 129 DISPLAY_F(f, " -k : preserve source file(s) (default) \n"); 130 DISPLAY_F(f, " -h/-H : display help/long help and exit \n"); 131 } 132 133 static void usage_advanced(const char* programName) 134 { 135 DISPLAYOUT(WELCOME_MESSAGE); 136 usage(stdout, programName); 137 DISPLAYOUT( "\n"); 138 DISPLAYOUT( "Advanced arguments : \n"); 139 DISPLAYOUT( " -V : display Version number and exit \n"); 140 141 DISPLAYOUT( " -c : force write to standard output, even if it is the console \n"); 142 143 DISPLAYOUT( " -v : verbose mode; specify multiple times to increase verbosity \n"); 144 DISPLAYOUT( " -q : suppress warnings; specify twice to suppress errors too \n"); 145 DISPLAYOUT( "--no-progress : do not display the progress counter \n"); 146 147 #ifdef UTIL_HAS_CREATEFILELIST 148 DISPLAYOUT( " -r : operate recursively on directories \n"); 149 DISPLAYOUT( "--filelist=FILE : read list of files to operate upon from FILE \n"); 150 DISPLAYOUT( "--output-dir-flat=DIR : all resulting files are stored into DIR \n"); 151 #endif 152 153 DISPLAYOUT( "-- : All arguments after \"--\" are treated as files \n"); 154 155 #ifndef ZSTD_NOCOMPRESS 156 DISPLAYOUT( "\n"); 157 DISPLAYOUT( "Advanced compression arguments : \n"); 158 DISPLAYOUT( "--ultra : enable levels beyond %i, up to %i (requires more memory) \n", ZSTDCLI_CLEVEL_MAX, ZSTD_maxCLevel()); 159 DISPLAYOUT( "--long[=#]: enable long distance matching with given window log (default: %u) \n", g_defaultMaxWindowLog); 160 DISPLAYOUT( "--fast[=#]: switch to very fast compression levels (default: %u) \n", 1); 161 DISPLAYOUT( "--adapt : dynamically adapt compression level to I/O conditions \n"); 162 # ifdef ZSTD_MULTITHREAD 163 DISPLAYOUT( " -T# : spawns # compression threads (default: 1, 0==# cores) \n"); 164 DISPLAYOUT( " -B# : select size of each job (default: 0==automatic) \n"); 165 DISPLAYOUT( "--single-thread : use a single thread for both I/O and compression (result slightly different than -T1) \n"); 166 DISPLAYOUT( "--rsyncable : compress using a rsync-friendly method (-B sets block size) \n"); 167 # endif 168 DISPLAYOUT( "--exclude-compressed: only compress files that are not already compressed \n"); 169 DISPLAYOUT( "--stream-size=# : specify size of streaming input from `stdin` \n"); 170 DISPLAYOUT( "--size-hint=# optimize compression parameters for streaming input of approximately this size \n"); 171 DISPLAYOUT( "--target-compressed-block-size=# : generate compressed block of approximately targeted size \n"); 172 DISPLAYOUT( "--no-dictID : don't write dictID into header (dictionary compression only) \n"); 173 DISPLAYOUT( "--[no-]check : add XXH64 integrity checksum to frame (default: enabled) \n"); 174 DISPLAYOUT( "--[no-]compress-literals : force (un)compressed literals \n"); 175 176 DISPLAYOUT( "--format=zstd : compress files to the .zst format (default) \n"); 177 #ifdef ZSTD_GZCOMPRESS 178 DISPLAYOUT( "--format=gzip : compress files to the .gz format \n"); 179 #endif 180 #ifdef ZSTD_LZMACOMPRESS 181 DISPLAYOUT( "--format=xz : compress files to the .xz format \n"); 182 DISPLAYOUT( "--format=lzma : compress files to the .lzma format \n"); 183 #endif 184 #ifdef ZSTD_LZ4COMPRESS 185 DISPLAYOUT( "--format=lz4 : compress files to the .lz4 format \n"); 186 #endif 187 #endif /* !ZSTD_NOCOMPRESS */ 188 189 #ifndef ZSTD_NODECOMPRESS 190 DISPLAYOUT( "\n"); 191 DISPLAYOUT( "Advanced decompression arguments : \n"); 192 DISPLAYOUT( " -l : print information about zstd compressed files \n"); 193 DISPLAYOUT( "--test : test compressed file integrity \n"); 194 DISPLAYOUT( " -M# : Set a memory usage limit for decompression \n"); 195 # if ZSTD_SPARSE_DEFAULT 196 DISPLAYOUT( "--[no-]sparse : sparse mode (default: enabled on file, disabled on stdout) \n"); 197 # else 198 DISPLAYOUT( "--[no-]sparse : sparse mode (default: disabled) \n"); 199 # endif 200 #endif /* ZSTD_NODECOMPRESS */ 201 202 #ifndef ZSTD_NODICT 203 DISPLAYOUT( "\n"); 204 DISPLAYOUT( "Dictionary builder : \n"); 205 DISPLAYOUT( "--train ## : create a dictionary from a training set of files \n"); 206 DISPLAYOUT( "--train-cover[=k=#,d=#,steps=#,split=#,shrink[=#]] : use the cover algorithm with optional args \n"); 207 DISPLAYOUT( "--train-fastcover[=k=#,d=#,f=#,steps=#,split=#,accel=#,shrink[=#]] : use the fast cover algorithm with optional args \n"); 208 DISPLAYOUT( "--train-legacy[=s=#] : use the legacy algorithm with selectivity (default: %u) \n", g_defaultSelectivityLevel); 209 DISPLAYOUT( " -o DICT : DICT is dictionary name (default: %s) \n", g_defaultDictName); 210 DISPLAYOUT( "--maxdict=# : limit dictionary to specified size (default: %u) \n", g_defaultMaxDictSize); 211 DISPLAYOUT( "--dictID=# : force dictionary ID to specified value (default: random) \n"); 212 #endif 213 214 #ifndef ZSTD_NOBENCH 215 DISPLAYOUT( "\n"); 216 DISPLAYOUT( "Benchmark arguments : \n"); 217 DISPLAYOUT( " -b# : benchmark file(s), using # compression level (default: %d) \n", ZSTDCLI_CLEVEL_DEFAULT); 218 DISPLAYOUT( " -e# : test all compression levels successively from -b# to -e# (default: 1) \n"); 219 DISPLAYOUT( " -i# : minimum evaluation time in seconds (default: 3s) \n"); 220 DISPLAYOUT( " -B# : cut file into independent blocks of size # (default: no block) \n"); 221 DISPLAYOUT( " -S : output one benchmark result per input file (default: consolidated result) \n"); 222 DISPLAYOUT( "--priority=rt : set process priority to real-time \n"); 223 #endif 224 225 } 226 227 static void badusage(const char* programName) 228 { 229 DISPLAYLEVEL(1, "Incorrect parameters \n"); 230 if (g_displayLevel >= 2) usage(stderr, programName); 231 } 232 233 static void waitEnter(void) 234 { 235 int unused; 236 DISPLAY("Press enter to continue... \n"); 237 unused = getchar(); 238 (void)unused; 239 } 240 241 static const char* lastNameFromPath(const char* path) 242 { 243 const char* name = path; 244 if (strrchr(name, '/')) name = strrchr(name, '/') + 1; 245 if (strrchr(name, '\\')) name = strrchr(name, '\\') + 1; /* windows */ 246 return name; 247 } 248 249 /*! exeNameMatch() : 250 @return : a non-zero value if exeName matches test, excluding the extension 251 */ 252 static int exeNameMatch(const char* exeName, const char* test) 253 { 254 return !strncmp(exeName, test, strlen(test)) && 255 (exeName[strlen(test)] == '\0' || exeName[strlen(test)] == '.'); 256 } 257 258 static void errorOut(const char* msg) 259 { 260 DISPLAY("%s \n", msg); exit(1); 261 } 262 263 /*! readU32FromCharChecked() : 264 * @return 0 if success, and store the result in *value. 265 * allows and interprets K, KB, KiB, M, MB and MiB suffix. 266 * Will also modify `*stringPtr`, advancing it to position where it stopped reading. 267 * @return 1 if an overflow error occurs */ 268 static int readU32FromCharChecked(const char** stringPtr, unsigned* value) 269 { 270 unsigned result = 0; 271 while ((**stringPtr >='0') && (**stringPtr <='9')) { 272 unsigned const max = ((unsigned)(-1)) / 10; 273 unsigned last = result; 274 if (result > max) return 1; /* overflow error */ 275 result *= 10; 276 result += (unsigned)(**stringPtr - '0'); 277 if (result < last) return 1; /* overflow error */ 278 (*stringPtr)++ ; 279 } 280 if ((**stringPtr=='K') || (**stringPtr=='M')) { 281 unsigned const maxK = ((unsigned)(-1)) >> 10; 282 if (result > maxK) return 1; /* overflow error */ 283 result <<= 10; 284 if (**stringPtr=='M') { 285 if (result > maxK) return 1; /* overflow error */ 286 result <<= 10; 287 } 288 (*stringPtr)++; /* skip `K` or `M` */ 289 if (**stringPtr=='i') (*stringPtr)++; 290 if (**stringPtr=='B') (*stringPtr)++; 291 } 292 *value = result; 293 return 0; 294 } 295 296 /*! readU32FromChar() : 297 * @return : unsigned integer value read from input in `char` format. 298 * allows and interprets K, KB, KiB, M, MB and MiB suffix. 299 * Will also modify `*stringPtr`, advancing it to position where it stopped reading. 300 * Note : function will exit() program if digit sequence overflows */ 301 static unsigned readU32FromChar(const char** stringPtr) { 302 static const char errorMsg[] = "error: numeric value overflows 32-bit unsigned int"; 303 unsigned result; 304 if (readU32FromCharChecked(stringPtr, &result)) { errorOut(errorMsg); } 305 return result; 306 } 307 308 /*! readSizeTFromCharChecked() : 309 * @return 0 if success, and store the result in *value. 310 * allows and interprets K, KB, KiB, M, MB and MiB suffix. 311 * Will also modify `*stringPtr`, advancing it to position where it stopped reading. 312 * @return 1 if an overflow error occurs */ 313 static int readSizeTFromCharChecked(const char** stringPtr, size_t* value) 314 { 315 size_t result = 0; 316 while ((**stringPtr >='0') && (**stringPtr <='9')) { 317 size_t const max = ((size_t)(-1)) / 10; 318 size_t last = result; 319 if (result > max) return 1; /* overflow error */ 320 result *= 10; 321 result += (size_t)(**stringPtr - '0'); 322 if (result < last) return 1; /* overflow error */ 323 (*stringPtr)++ ; 324 } 325 if ((**stringPtr=='K') || (**stringPtr=='M')) { 326 size_t const maxK = ((size_t)(-1)) >> 10; 327 if (result > maxK) return 1; /* overflow error */ 328 result <<= 10; 329 if (**stringPtr=='M') { 330 if (result > maxK) return 1; /* overflow error */ 331 result <<= 10; 332 } 333 (*stringPtr)++; /* skip `K` or `M` */ 334 if (**stringPtr=='i') (*stringPtr)++; 335 if (**stringPtr=='B') (*stringPtr)++; 336 } 337 *value = result; 338 return 0; 339 } 340 341 /*! readSizeTFromChar() : 342 * @return : size_t value read from input in `char` format. 343 * allows and interprets K, KB, KiB, M, MB and MiB suffix. 344 * Will also modify `*stringPtr`, advancing it to position where it stopped reading. 345 * Note : function will exit() program if digit sequence overflows */ 346 static size_t readSizeTFromChar(const char** stringPtr) { 347 static const char errorMsg[] = "error: numeric value overflows size_t"; 348 size_t result; 349 if (readSizeTFromCharChecked(stringPtr, &result)) { errorOut(errorMsg); } 350 return result; 351 } 352 353 /** longCommandWArg() : 354 * check if *stringPtr is the same as longCommand. 355 * If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand. 356 * @return 0 and doesn't modify *stringPtr otherwise. 357 */ 358 static int longCommandWArg(const char** stringPtr, const char* longCommand) 359 { 360 size_t const comSize = strlen(longCommand); 361 int const result = !strncmp(*stringPtr, longCommand, comSize); 362 if (result) *stringPtr += comSize; 363 return result; 364 } 365 366 367 #ifndef ZSTD_NODICT 368 369 static const unsigned kDefaultRegression = 1; 370 /** 371 * parseCoverParameters() : 372 * reads cover parameters from *stringPtr (e.g. "--train-cover=k=48,d=8,steps=32") into *params 373 * @return 1 means that cover parameters were correct 374 * @return 0 in case of malformed parameters 375 */ 376 static unsigned parseCoverParameters(const char* stringPtr, ZDICT_cover_params_t* params) 377 { 378 memset(params, 0, sizeof(*params)); 379 for (; ;) { 380 if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 381 if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 382 if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 383 if (longCommandWArg(&stringPtr, "split=")) { 384 unsigned splitPercentage = readU32FromChar(&stringPtr); 385 params->splitPoint = (double)splitPercentage / 100.0; 386 if (stringPtr[0]==',') { stringPtr++; continue; } else break; 387 } 388 if (longCommandWArg(&stringPtr, "shrink")) { 389 params->shrinkDictMaxRegression = kDefaultRegression; 390 params->shrinkDict = 1; 391 if (stringPtr[0]=='=') { 392 stringPtr++; 393 params->shrinkDictMaxRegression = readU32FromChar(&stringPtr); 394 } 395 if (stringPtr[0]==',') { 396 stringPtr++; 397 continue; 398 } 399 else break; 400 } 401 return 0; 402 } 403 if (stringPtr[0] != 0) return 0; 404 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); 405 return 1; 406 } 407 408 /** 409 * parseFastCoverParameters() : 410 * reads fastcover parameters from *stringPtr (e.g. "--train-fastcover=k=48,d=8,f=20,steps=32,accel=2") into *params 411 * @return 1 means that fastcover parameters were correct 412 * @return 0 in case of malformed parameters 413 */ 414 static unsigned parseFastCoverParameters(const char* stringPtr, ZDICT_fastCover_params_t* params) 415 { 416 memset(params, 0, sizeof(*params)); 417 for (; ;) { 418 if (longCommandWArg(&stringPtr, "k=")) { params->k = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 419 if (longCommandWArg(&stringPtr, "d=")) { params->d = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 420 if (longCommandWArg(&stringPtr, "f=")) { params->f = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 421 if (longCommandWArg(&stringPtr, "steps=")) { params->steps = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 422 if (longCommandWArg(&stringPtr, "accel=")) { params->accel = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 423 if (longCommandWArg(&stringPtr, "split=")) { 424 unsigned splitPercentage = readU32FromChar(&stringPtr); 425 params->splitPoint = (double)splitPercentage / 100.0; 426 if (stringPtr[0]==',') { stringPtr++; continue; } else break; 427 } 428 if (longCommandWArg(&stringPtr, "shrink")) { 429 params->shrinkDictMaxRegression = kDefaultRegression; 430 params->shrinkDict = 1; 431 if (stringPtr[0]=='=') { 432 stringPtr++; 433 params->shrinkDictMaxRegression = readU32FromChar(&stringPtr); 434 } 435 if (stringPtr[0]==',') { 436 stringPtr++; 437 continue; 438 } 439 else break; 440 } 441 return 0; 442 } 443 if (stringPtr[0] != 0) return 0; 444 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); 445 return 1; 446 } 447 448 /** 449 * parseLegacyParameters() : 450 * reads legacy dictionary builder parameters from *stringPtr (e.g. "--train-legacy=selectivity=8") into *selectivity 451 * @return 1 means that legacy dictionary builder parameters were correct 452 * @return 0 in case of malformed parameters 453 */ 454 static unsigned parseLegacyParameters(const char* stringPtr, unsigned* selectivity) 455 { 456 if (!longCommandWArg(&stringPtr, "s=") && !longCommandWArg(&stringPtr, "selectivity=")) { return 0; } 457 *selectivity = readU32FromChar(&stringPtr); 458 if (stringPtr[0] != 0) return 0; 459 DISPLAYLEVEL(4, "legacy: selectivity=%u\n", *selectivity); 460 return 1; 461 } 462 463 static ZDICT_cover_params_t defaultCoverParams(void) 464 { 465 ZDICT_cover_params_t params; 466 memset(¶ms, 0, sizeof(params)); 467 params.d = 8; 468 params.steps = 4; 469 params.splitPoint = 1.0; 470 params.shrinkDict = 0; 471 params.shrinkDictMaxRegression = kDefaultRegression; 472 return params; 473 } 474 475 static ZDICT_fastCover_params_t defaultFastCoverParams(void) 476 { 477 ZDICT_fastCover_params_t params; 478 memset(¶ms, 0, sizeof(params)); 479 params.d = 8; 480 params.f = 20; 481 params.steps = 4; 482 params.splitPoint = 0.75; /* different from default splitPoint of cover */ 483 params.accel = DEFAULT_ACCEL; 484 params.shrinkDict = 0; 485 params.shrinkDictMaxRegression = kDefaultRegression; 486 return params; 487 } 488 #endif 489 490 491 /** parseAdaptParameters() : 492 * reads adapt parameters from *stringPtr (e.g. "--zstd=min=1,max=19) and store them into adaptMinPtr and adaptMaxPtr. 493 * Both adaptMinPtr and adaptMaxPtr must be already allocated and correctly initialized. 494 * There is no guarantee that any of these values will be updated. 495 * @return 1 means that parsing was successful, 496 * @return 0 in case of malformed parameters 497 */ 498 static unsigned parseAdaptParameters(const char* stringPtr, int* adaptMinPtr, int* adaptMaxPtr) 499 { 500 for ( ; ;) { 501 if (longCommandWArg(&stringPtr, "min=")) { *adaptMinPtr = (int)readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 502 if (longCommandWArg(&stringPtr, "max=")) { *adaptMaxPtr = (int)readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 503 DISPLAYLEVEL(4, "invalid compression parameter \n"); 504 return 0; 505 } 506 if (stringPtr[0] != 0) return 0; /* check the end of string */ 507 if (*adaptMinPtr > *adaptMaxPtr) { 508 DISPLAYLEVEL(4, "incoherent adaptation limits \n"); 509 return 0; 510 } 511 return 1; 512 } 513 514 515 /** parseCompressionParameters() : 516 * reads compression parameters from *stringPtr (e.g. "--zstd=wlog=23,clog=23,hlog=22,slog=6,mml=3,tlen=48,strat=6") into *params 517 * @return 1 means that compression parameters were correct 518 * @return 0 in case of malformed parameters 519 */ 520 static unsigned parseCompressionParameters(const char* stringPtr, ZSTD_compressionParameters* params) 521 { 522 for ( ; ;) { 523 if (longCommandWArg(&stringPtr, "windowLog=") || longCommandWArg(&stringPtr, "wlog=")) { params->windowLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 524 if (longCommandWArg(&stringPtr, "chainLog=") || longCommandWArg(&stringPtr, "clog=")) { params->chainLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 525 if (longCommandWArg(&stringPtr, "hashLog=") || longCommandWArg(&stringPtr, "hlog=")) { params->hashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 526 if (longCommandWArg(&stringPtr, "searchLog=") || longCommandWArg(&stringPtr, "slog=")) { params->searchLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 527 if (longCommandWArg(&stringPtr, "minMatch=") || longCommandWArg(&stringPtr, "mml=")) { params->minMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 528 if (longCommandWArg(&stringPtr, "targetLength=") || longCommandWArg(&stringPtr, "tlen=")) { params->targetLength = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 529 if (longCommandWArg(&stringPtr, "strategy=") || longCommandWArg(&stringPtr, "strat=")) { params->strategy = (ZSTD_strategy)(readU32FromChar(&stringPtr)); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 530 if (longCommandWArg(&stringPtr, "overlapLog=") || longCommandWArg(&stringPtr, "ovlog=")) { g_overlapLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 531 if (longCommandWArg(&stringPtr, "ldmHashLog=") || longCommandWArg(&stringPtr, "lhlog=")) { g_ldmHashLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 532 if (longCommandWArg(&stringPtr, "ldmMinMatch=") || longCommandWArg(&stringPtr, "lmml=")) { g_ldmMinMatch = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 533 if (longCommandWArg(&stringPtr, "ldmBucketSizeLog=") || longCommandWArg(&stringPtr, "lblog=")) { g_ldmBucketSizeLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 534 if (longCommandWArg(&stringPtr, "ldmHashRateLog=") || longCommandWArg(&stringPtr, "lhrlog=")) { g_ldmHashRateLog = readU32FromChar(&stringPtr); if (stringPtr[0]==',') { stringPtr++; continue; } else break; } 535 DISPLAYLEVEL(4, "invalid compression parameter \n"); 536 return 0; 537 } 538 539 DISPLAYLEVEL(4, "windowLog=%d, chainLog=%d, hashLog=%d, searchLog=%d \n", params->windowLog, params->chainLog, params->hashLog, params->searchLog); 540 DISPLAYLEVEL(4, "minMatch=%d, targetLength=%d, strategy=%d \n", params->minMatch, params->targetLength, params->strategy); 541 if (stringPtr[0] != 0) return 0; /* check the end of string */ 542 return 1; 543 } 544 545 static void printVersion(void) 546 { 547 DISPLAYOUT(WELCOME_MESSAGE); 548 if (g_displayLevel >= 3) { 549 /* format support */ 550 DISPLAYOUT("*** supports: zstd"); 551 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>0) && (ZSTD_LEGACY_SUPPORT<8) 552 DISPLAYOUT(", zstd legacy v0.%d+", ZSTD_LEGACY_SUPPORT); 553 #endif 554 #ifdef ZSTD_GZCOMPRESS 555 DISPLAYOUT(", gzip"); 556 #endif 557 #ifdef ZSTD_LZ4COMPRESS 558 DISPLAYOUT(", lz4"); 559 #endif 560 #ifdef ZSTD_LZMACOMPRESS 561 DISPLAYOUT(", lzma, xz "); 562 #endif 563 DISPLAYOUT("\n"); 564 if (g_displayLevel >= 4) { 565 /* posix support */ 566 #ifdef _POSIX_C_SOURCE 567 DISPLAYOUT("_POSIX_C_SOURCE defined: %ldL\n", (long) _POSIX_C_SOURCE); 568 #endif 569 #ifdef _POSIX_VERSION 570 DISPLAYOUT("_POSIX_VERSION defined: %ldL \n", (long) _POSIX_VERSION); 571 #endif 572 #ifdef PLATFORM_POSIX_VERSION 573 DISPLAYOUT("PLATFORM_POSIX_VERSION defined: %ldL\n", (long) PLATFORM_POSIX_VERSION); 574 #endif 575 } } 576 } 577 578 /* Environment variables for parameter setting */ 579 #define ENV_CLEVEL "ZSTD_CLEVEL" 580 581 /* pick up environment variable */ 582 static int init_cLevel(void) { 583 const char* const env = getenv(ENV_CLEVEL); 584 if (env != NULL) { 585 const char* ptr = env; 586 int sign = 1; 587 if (*ptr == '-') { 588 sign = -1; 589 ptr++; 590 } else if (*ptr == '+') { 591 ptr++; 592 } 593 594 if ((*ptr>='0') && (*ptr<='9')) { 595 unsigned absLevel; 596 if (readU32FromCharChecked(&ptr, &absLevel)) { 597 DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: numeric value too large \n", ENV_CLEVEL, env); 598 return ZSTDCLI_CLEVEL_DEFAULT; 599 } else if (*ptr == 0) { 600 return sign * (int)absLevel; 601 } } 602 603 DISPLAYLEVEL(2, "Ignore environment variable setting %s=%s: not a valid integer value \n", ENV_CLEVEL, env); 604 } 605 606 return ZSTDCLI_CLEVEL_DEFAULT; 607 } 608 609 #define ZSTD_NB_STRATEGIES 9 610 611 static const char* ZSTD_strategyMap[ZSTD_NB_STRATEGIES + 1] = { "", "ZSTD_fast", 612 "ZSTD_dfast", "ZSTD_greedy", "ZSTD_lazy", "ZSTD_lazy2", "ZSTD_btlazy2", 613 "ZSTD_btopt", "ZSTD_btultra", "ZSTD_btultra2"}; 614 615 typedef enum { zom_compress, zom_decompress, zom_test, zom_bench, zom_train, zom_list } zstd_operation_mode; 616 617 #define CLEAN_RETURN(i) { operationResult = (i); goto _end; } 618 619 #ifdef ZSTD_NOCOMPRESS 620 /* symbols from compression library are not defined and should not be invoked */ 621 # define MINCLEVEL -99 622 # define MAXCLEVEL 22 623 #else 624 # define MINCLEVEL ZSTD_minCLevel() 625 # define MAXCLEVEL ZSTD_maxCLevel() 626 #endif 627 628 int main(int const argCount, const char* argv[]) 629 { 630 int argNb, 631 followLinks = 0, 632 forceStdout = 0, 633 lastCommand = 0, 634 ldmFlag = 0, 635 main_pause = 0, 636 nbWorkers = 0, 637 adapt = 0, 638 adaptMin = MINCLEVEL, 639 adaptMax = MAXCLEVEL, 640 rsyncable = 0, 641 nextArgumentIsOutFileName = 0, 642 nextArgumentIsOutDirName = 0, 643 nextArgumentIsMaxDict = 0, 644 nextArgumentIsDictID = 0, 645 nextArgumentsAreFiles = 0, 646 nextEntryIsDictionary = 0, 647 operationResult = 0, 648 separateFiles = 0, 649 setRealTimePrio = 0, 650 singleThread = 0, 651 showDefaultCParams = 0, 652 ultra=0, 653 contentSize=1; 654 double compressibility = 0.5; 655 unsigned bench_nbSeconds = 3; /* would be better if this value was synchronized from bench */ 656 size_t blockSize = 0; 657 658 FIO_prefs_t* const prefs = FIO_createPreferences(); 659 zstd_operation_mode operation = zom_compress; 660 ZSTD_compressionParameters compressionParams; 661 int cLevel = init_cLevel(); 662 int cLevelLast = MINCLEVEL - 1; /* lower than minimum */ 663 unsigned recursive = 0; 664 unsigned memLimit = 0; 665 FileNamesTable* filenames = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */ 666 FileNamesTable* file_of_names = UTIL_allocateFileNamesTable((size_t)argCount); /* argCount >= 1 */ 667 const char* programName = argv[0]; 668 const char* outFileName = NULL; 669 const char* outDirName = NULL; 670 const char* dictFileName = NULL; 671 const char* patchFromDictFileName = NULL; 672 const char* suffix = ZSTD_EXTENSION; 673 unsigned maxDictSize = g_defaultMaxDictSize; 674 unsigned dictID = 0; 675 size_t streamSrcSize = 0; 676 size_t targetCBlockSize = 0; 677 size_t srcSizeHint = 0; 678 int dictCLevel = g_defaultDictCLevel; 679 unsigned dictSelect = g_defaultSelectivityLevel; 680 #ifndef ZSTD_NODICT 681 ZDICT_cover_params_t coverParams = defaultCoverParams(); 682 ZDICT_fastCover_params_t fastCoverParams = defaultFastCoverParams(); 683 dictType dict = fastCover; 684 #endif 685 #ifndef ZSTD_NOBENCH 686 BMK_advancedParams_t benchParams = BMK_initAdvancedParams(); 687 #endif 688 ZSTD_literalCompressionMode_e literalCompressionMode = ZSTD_lcm_auto; 689 690 691 /* init */ 692 (void)recursive; (void)cLevelLast; /* not used when ZSTD_NOBENCH set */ 693 (void)memLimit; 694 assert(argCount >= 1); 695 if ((filenames==NULL) || (file_of_names==NULL)) { DISPLAY("zstd: allocation error \n"); exit(1); } 696 programName = lastNameFromPath(programName); 697 #ifdef ZSTD_MULTITHREAD 698 nbWorkers = 1; 699 #endif 700 701 /* preset behaviors */ 702 if (exeNameMatch(programName, ZSTD_ZSTDMT)) nbWorkers=0, singleThread=0; 703 if (exeNameMatch(programName, ZSTD_UNZSTD)) operation=zom_decompress; 704 if (exeNameMatch(programName, ZSTD_CAT)) { operation=zom_decompress; FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; outFileName=stdoutmark; g_displayLevel=1; } /* supports multiple formats */ 705 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 */ 706 if (exeNameMatch(programName, ZSTD_GZ)) { suffix = GZ_EXTENSION; FIO_setCompressionType(prefs, FIO_gzipCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like gzip */ 707 if (exeNameMatch(programName, ZSTD_GUNZIP)) { operation=zom_decompress; FIO_setRemoveSrcFile(prefs, 1); } /* behave like gunzip, also supports multiple formats */ 708 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 */ 709 if (exeNameMatch(programName, ZSTD_LZMA)) { suffix = LZMA_EXTENSION; FIO_setCompressionType(prefs, FIO_lzmaCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like lzma */ 710 if (exeNameMatch(programName, ZSTD_UNLZMA)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_lzmaCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like unlzma, also supports multiple formats */ 711 if (exeNameMatch(programName, ZSTD_XZ)) { suffix = XZ_EXTENSION; FIO_setCompressionType(prefs, FIO_xzCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like xz */ 712 if (exeNameMatch(programName, ZSTD_UNXZ)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_xzCompression); FIO_setRemoveSrcFile(prefs, 1); } /* behave like unxz, also supports multiple formats */ 713 if (exeNameMatch(programName, ZSTD_LZ4)) { suffix = LZ4_EXTENSION; FIO_setCompressionType(prefs, FIO_lz4Compression); } /* behave like lz4 */ 714 if (exeNameMatch(programName, ZSTD_UNLZ4)) { operation=zom_decompress; FIO_setCompressionType(prefs, FIO_lz4Compression); } /* behave like unlz4, also supports multiple formats */ 715 memset(&compressionParams, 0, sizeof(compressionParams)); 716 717 /* init crash handler */ 718 FIO_addAbortHandler(); 719 720 /* command switches */ 721 for (argNb=1; argNb<argCount; argNb++) { 722 const char* argument = argv[argNb]; 723 if (!argument) continue; /* Protection if argument empty */ 724 725 if (nextArgumentsAreFiles) { 726 UTIL_refFilename(filenames, argument); 727 continue; 728 } 729 730 /* "-" means stdin/stdout */ 731 if (!strcmp(argument, "-")){ 732 UTIL_refFilename(filenames, stdinmark); 733 continue; 734 } 735 736 /* Decode commands (note : aggregated commands are allowed) */ 737 if (argument[0]=='-') { 738 739 if (argument[1]=='-') { 740 /* long commands (--long-word) */ 741 if (!strcmp(argument, "--")) { nextArgumentsAreFiles=1; continue; } /* only file names allowed from now on */ 742 if (!strcmp(argument, "--list")) { operation=zom_list; continue; } 743 if (!strcmp(argument, "--compress")) { operation=zom_compress; continue; } 744 if (!strcmp(argument, "--decompress")) { operation=zom_decompress; continue; } 745 if (!strcmp(argument, "--uncompress")) { operation=zom_decompress; continue; } 746 if (!strcmp(argument, "--force")) { FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; continue; } 747 if (!strcmp(argument, "--version")) { printVersion(); CLEAN_RETURN(0); } 748 if (!strcmp(argument, "--help")) { usage_advanced(programName); CLEAN_RETURN(0); } 749 if (!strcmp(argument, "--verbose")) { g_displayLevel++; continue; } 750 if (!strcmp(argument, "--quiet")) { g_displayLevel--; continue; } 751 if (!strcmp(argument, "--stdout")) { forceStdout=1; outFileName=stdoutmark; g_displayLevel-=(g_displayLevel==2); continue; } 752 if (!strcmp(argument, "--ultra")) { ultra=1; continue; } 753 if (!strcmp(argument, "--check")) { FIO_setChecksumFlag(prefs, 2); continue; } 754 if (!strcmp(argument, "--no-check")) { FIO_setChecksumFlag(prefs, 0); continue; } 755 if (!strcmp(argument, "--sparse")) { FIO_setSparseWrite(prefs, 2); continue; } 756 if (!strcmp(argument, "--no-sparse")) { FIO_setSparseWrite(prefs, 0); continue; } 757 if (!strcmp(argument, "--test")) { operation=zom_test; continue; } 758 if (!strcmp(argument, "--train")) { operation=zom_train; if (outFileName==NULL) outFileName=g_defaultDictName; continue; } 759 if (!strcmp(argument, "--maxdict")) { nextArgumentIsMaxDict=1; lastCommand=1; continue; } /* kept available for compatibility with old syntax ; will be removed one day */ 760 if (!strcmp(argument, "--dictID")) { nextArgumentIsDictID=1; lastCommand=1; continue; } /* kept available for compatibility with old syntax ; will be removed one day */ 761 if (!strcmp(argument, "--no-dictID")) { FIO_setDictIDFlag(prefs, 0); continue; } 762 if (!strcmp(argument, "--keep")) { FIO_setRemoveSrcFile(prefs, 0); continue; } 763 if (!strcmp(argument, "--rm")) { FIO_setRemoveSrcFile(prefs, 1); continue; } 764 if (!strcmp(argument, "--priority=rt")) { setRealTimePrio = 1; continue; } 765 if (!strcmp(argument, "--output-dir-flat")) {nextArgumentIsOutDirName=1; lastCommand=1; continue; } 766 if (!strcmp(argument, "--show-default-cparams")) { showDefaultCParams = 1; continue; } 767 if (!strcmp(argument, "--content-size")) { contentSize = 1; continue; } 768 if (!strcmp(argument, "--no-content-size")) { contentSize = 0; continue; } 769 if (!strcmp(argument, "--adapt")) { adapt = 1; continue; } 770 if (longCommandWArg(&argument, "--adapt=")) { adapt = 1; if (!parseAdaptParameters(argument, &adaptMin, &adaptMax)) { badusage(programName); CLEAN_RETURN(1); } continue; } 771 if (!strcmp(argument, "--single-thread")) { nbWorkers = 0; singleThread = 1; continue; } 772 if (!strcmp(argument, "--format=zstd")) { suffix = ZSTD_EXTENSION; FIO_setCompressionType(prefs, FIO_zstdCompression); continue; } 773 #ifdef ZSTD_GZCOMPRESS 774 if (!strcmp(argument, "--format=gzip")) { suffix = GZ_EXTENSION; FIO_setCompressionType(prefs, FIO_gzipCompression); continue; } 775 #endif 776 #ifdef ZSTD_LZMACOMPRESS 777 if (!strcmp(argument, "--format=lzma")) { suffix = LZMA_EXTENSION; FIO_setCompressionType(prefs, FIO_lzmaCompression); continue; } 778 if (!strcmp(argument, "--format=xz")) { suffix = XZ_EXTENSION; FIO_setCompressionType(prefs, FIO_xzCompression); continue; } 779 #endif 780 #ifdef ZSTD_LZ4COMPRESS 781 if (!strcmp(argument, "--format=lz4")) { suffix = LZ4_EXTENSION; FIO_setCompressionType(prefs, FIO_lz4Compression); continue; } 782 #endif 783 if (!strcmp(argument, "--rsyncable")) { rsyncable = 1; continue; } 784 if (!strcmp(argument, "--compress-literals")) { literalCompressionMode = ZSTD_lcm_huffman; continue; } 785 if (!strcmp(argument, "--no-compress-literals")) { literalCompressionMode = ZSTD_lcm_uncompressed; continue; } 786 if (!strcmp(argument, "--no-progress")) { FIO_setNoProgress(1); continue; } 787 if (!strcmp(argument, "--exclude-compressed")) { FIO_setExcludeCompressedFile(prefs, 1); continue; } 788 /* long commands with arguments */ 789 #ifndef ZSTD_NODICT 790 if (longCommandWArg(&argument, "--train-cover")) { 791 operation = zom_train; 792 if (outFileName == NULL) 793 outFileName = g_defaultDictName; 794 dict = cover; 795 /* Allow optional arguments following an = */ 796 if (*argument == 0) { memset(&coverParams, 0, sizeof(coverParams)); } 797 else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); } 798 else if (!parseCoverParameters(argument, &coverParams)) { badusage(programName); CLEAN_RETURN(1); } 799 continue; 800 } 801 if (longCommandWArg(&argument, "--train-fastcover")) { 802 operation = zom_train; 803 if (outFileName == NULL) 804 outFileName = g_defaultDictName; 805 dict = fastCover; 806 /* Allow optional arguments following an = */ 807 if (*argument == 0) { memset(&fastCoverParams, 0, sizeof(fastCoverParams)); } 808 else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); } 809 else if (!parseFastCoverParameters(argument, &fastCoverParams)) { badusage(programName); CLEAN_RETURN(1); } 810 continue; 811 } 812 if (longCommandWArg(&argument, "--train-legacy")) { 813 operation = zom_train; 814 if (outFileName == NULL) 815 outFileName = g_defaultDictName; 816 dict = legacy; 817 /* Allow optional arguments following an = */ 818 if (*argument == 0) { continue; } 819 else if (*argument++ != '=') { badusage(programName); CLEAN_RETURN(1); } 820 else if (!parseLegacyParameters(argument, &dictSelect)) { badusage(programName); CLEAN_RETURN(1); } 821 continue; 822 } 823 #endif 824 if (longCommandWArg(&argument, "--threads=")) { nbWorkers = (int)readU32FromChar(&argument); continue; } 825 if (longCommandWArg(&argument, "--memlimit=")) { memLimit = readU32FromChar(&argument); continue; } 826 if (longCommandWArg(&argument, "--memory=")) { memLimit = readU32FromChar(&argument); continue; } 827 if (longCommandWArg(&argument, "--memlimit-decompress=")) { memLimit = readU32FromChar(&argument); continue; } 828 if (longCommandWArg(&argument, "--block-size=")) { blockSize = readSizeTFromChar(&argument); continue; } 829 if (longCommandWArg(&argument, "--maxdict=")) { maxDictSize = readU32FromChar(&argument); continue; } 830 if (longCommandWArg(&argument, "--dictID=")) { dictID = readU32FromChar(&argument); continue; } 831 if (longCommandWArg(&argument, "--zstd=")) { if (!parseCompressionParameters(argument, &compressionParams)) { badusage(programName); CLEAN_RETURN(1); } continue; } 832 if (longCommandWArg(&argument, "--stream-size=")) { streamSrcSize = readSizeTFromChar(&argument); continue; } 833 if (longCommandWArg(&argument, "--target-compressed-block-size=")) { targetCBlockSize = readSizeTFromChar(&argument); continue; } 834 if (longCommandWArg(&argument, "--size-hint=")) { srcSizeHint = readSizeTFromChar(&argument); continue; } 835 if (longCommandWArg(&argument, "--output-dir-flat=")) { outDirName = argument; continue; } 836 if (longCommandWArg(&argument, "--patch-from=")) { patchFromDictFileName = argument; continue; } 837 if (longCommandWArg(&argument, "--long")) { 838 unsigned ldmWindowLog = 0; 839 ldmFlag = 1; 840 /* Parse optional window log */ 841 if (*argument == '=') { 842 ++argument; 843 ldmWindowLog = readU32FromChar(&argument); 844 } else if (*argument != 0) { 845 /* Invalid character following --long */ 846 badusage(programName); 847 CLEAN_RETURN(1); 848 } 849 /* Only set windowLog if not already set by --zstd */ 850 if (compressionParams.windowLog == 0) 851 compressionParams.windowLog = ldmWindowLog; 852 continue; 853 } 854 #ifndef ZSTD_NOCOMPRESS /* linking ZSTD_minCLevel() requires compression support */ 855 if (longCommandWArg(&argument, "--fast")) { 856 /* Parse optional acceleration factor */ 857 if (*argument == '=') { 858 U32 const maxFast = (U32)-ZSTD_minCLevel(); 859 U32 fastLevel; 860 ++argument; 861 fastLevel = readU32FromChar(&argument); 862 if (fastLevel > maxFast) fastLevel = maxFast; 863 if (fastLevel) { 864 dictCLevel = cLevel = -(int)fastLevel; 865 } else { 866 badusage(programName); 867 CLEAN_RETURN(1); 868 } 869 } else if (*argument != 0) { 870 /* Invalid character following --fast */ 871 badusage(programName); 872 CLEAN_RETURN(1); 873 } else { 874 cLevel = -1; /* default for --fast */ 875 } 876 continue; 877 } 878 #endif 879 880 if (longCommandWArg(&argument, "--filelist=")) { 881 UTIL_refFilename(file_of_names, argument); 882 continue; 883 } 884 885 /* fall-through, will trigger bad_usage() later on */ 886 } 887 888 argument++; 889 while (argument[0]!=0) { 890 if (lastCommand) { 891 DISPLAY("error : command must be followed by argument \n"); 892 CLEAN_RETURN(1); 893 } 894 #ifndef ZSTD_NOCOMPRESS 895 /* compression Level */ 896 if ((*argument>='0') && (*argument<='9')) { 897 dictCLevel = cLevel = (int)readU32FromChar(&argument); 898 continue; 899 } 900 #endif 901 902 switch(argument[0]) 903 { 904 /* Display help */ 905 case 'V': printVersion(); CLEAN_RETURN(0); /* Version Only */ 906 case 'H': 907 case 'h': usage_advanced(programName); CLEAN_RETURN(0); 908 909 /* Compress */ 910 case 'z': operation=zom_compress; argument++; break; 911 912 /* Decoding */ 913 case 'd': 914 #ifndef ZSTD_NOBENCH 915 benchParams.mode = BMK_decodeOnly; 916 if (operation==zom_bench) { argument++; break; } /* benchmark decode (hidden option) */ 917 #endif 918 operation=zom_decompress; argument++; break; 919 920 /* Force stdout, even if stdout==console */ 921 case 'c': forceStdout=1; outFileName=stdoutmark; argument++; break; 922 923 /* Use file content as dictionary */ 924 case 'D': nextEntryIsDictionary = 1; lastCommand = 1; argument++; break; 925 926 /* Overwrite */ 927 case 'f': FIO_overwriteMode(prefs); forceStdout=1; followLinks=1; argument++; break; 928 929 /* Verbose mode */ 930 case 'v': g_displayLevel++; argument++; break; 931 932 /* Quiet mode */ 933 case 'q': g_displayLevel--; argument++; break; 934 935 /* keep source file (default) */ 936 case 'k': FIO_setRemoveSrcFile(prefs, 0); argument++; break; 937 938 /* Checksum */ 939 case 'C': FIO_setChecksumFlag(prefs, 2); argument++; break; 940 941 /* test compressed file */ 942 case 't': operation=zom_test; argument++; break; 943 944 /* destination file name */ 945 case 'o': nextArgumentIsOutFileName=1; lastCommand=1; argument++; break; 946 947 /* limit memory */ 948 case 'M': 949 argument++; 950 memLimit = readU32FromChar(&argument); 951 break; 952 case 'l': operation=zom_list; argument++; break; 953 #ifdef UTIL_HAS_CREATEFILELIST 954 /* recursive */ 955 case 'r': recursive=1; argument++; break; 956 #endif 957 958 #ifndef ZSTD_NOBENCH 959 /* Benchmark */ 960 case 'b': 961 operation=zom_bench; 962 argument++; 963 break; 964 965 /* range bench (benchmark only) */ 966 case 'e': 967 /* compression Level */ 968 argument++; 969 cLevelLast = (int)readU32FromChar(&argument); 970 break; 971 972 /* Modify Nb Iterations (benchmark only) */ 973 case 'i': 974 argument++; 975 bench_nbSeconds = readU32FromChar(&argument); 976 break; 977 978 /* cut input into blocks (benchmark only) */ 979 case 'B': 980 argument++; 981 blockSize = readU32FromChar(&argument); 982 break; 983 984 /* benchmark files separately (hidden option) */ 985 case 'S': 986 argument++; 987 separateFiles = 1; 988 break; 989 990 #endif /* ZSTD_NOBENCH */ 991 992 /* nb of threads (hidden option) */ 993 case 'T': 994 argument++; 995 nbWorkers = (int)readU32FromChar(&argument); 996 break; 997 998 /* Dictionary Selection level */ 999 case 's': 1000 argument++; 1001 dictSelect = readU32FromChar(&argument); 1002 break; 1003 1004 /* Pause at the end (-p) or set an additional param (-p#) (hidden option) */ 1005 case 'p': argument++; 1006 #ifndef ZSTD_NOBENCH 1007 if ((*argument>='0') && (*argument<='9')) { 1008 benchParams.additionalParam = (int)readU32FromChar(&argument); 1009 } else 1010 #endif 1011 main_pause=1; 1012 break; 1013 1014 /* Select compressibility of synthetic sample */ 1015 case 'P': 1016 { argument++; 1017 compressibility = (double)readU32FromChar(&argument) / 100; 1018 } 1019 break; 1020 1021 /* unknown command */ 1022 default : badusage(programName); CLEAN_RETURN(1); 1023 } 1024 } 1025 continue; 1026 } /* if (argument[0]=='-') */ 1027 1028 if (nextArgumentIsMaxDict) { /* kept available for compatibility with old syntax ; will be removed one day */ 1029 nextArgumentIsMaxDict = 0; 1030 lastCommand = 0; 1031 maxDictSize = readU32FromChar(&argument); 1032 continue; 1033 } 1034 1035 if (nextArgumentIsDictID) { /* kept available for compatibility with old syntax ; will be removed one day */ 1036 nextArgumentIsDictID = 0; 1037 lastCommand = 0; 1038 dictID = readU32FromChar(&argument); 1039 continue; 1040 } 1041 1042 if (nextEntryIsDictionary) { 1043 nextEntryIsDictionary = 0; 1044 lastCommand = 0; 1045 dictFileName = argument; 1046 continue; 1047 } 1048 1049 if (nextArgumentIsOutFileName) { 1050 nextArgumentIsOutFileName = 0; 1051 lastCommand = 0; 1052 outFileName = argument; 1053 if (!strcmp(outFileName, "-")) outFileName = stdoutmark; 1054 continue; 1055 } 1056 1057 if (nextArgumentIsOutDirName) { 1058 nextArgumentIsOutDirName = 0; 1059 lastCommand = 0; 1060 outDirName = argument; 1061 continue; 1062 } 1063 1064 /* none of the above : add filename to list */ 1065 UTIL_refFilename(filenames, argument); 1066 } 1067 1068 if (lastCommand) { /* forgotten argument */ 1069 DISPLAY("error : command must be followed by argument \n"); 1070 CLEAN_RETURN(1); 1071 } 1072 1073 /* Welcome message (if verbose) */ 1074 DISPLAYLEVEL(3, WELCOME_MESSAGE); 1075 1076 #ifdef ZSTD_MULTITHREAD 1077 if ((nbWorkers==0) && (!singleThread)) { 1078 /* automatically set # workers based on # of reported cpus */ 1079 nbWorkers = UTIL_countPhysicalCores(); 1080 DISPLAYLEVEL(3, "Note: %d physical core(s) detected \n", nbWorkers); 1081 } 1082 #else 1083 (void)singleThread; (void)nbWorkers; 1084 #endif 1085 1086 #ifdef UTIL_HAS_CREATEFILELIST 1087 g_utilDisplayLevel = g_displayLevel; 1088 if (!followLinks) { 1089 unsigned u, fileNamesNb; 1090 unsigned const nbFilenames = (unsigned)filenames->tableSize; 1091 for (u=0, fileNamesNb=0; u<nbFilenames; u++) { 1092 if ( UTIL_isLink(filenames->fileNames[u]) 1093 && !UTIL_isFIFO(filenames->fileNames[u]) 1094 ) { 1095 DISPLAYLEVEL(2, "Warning : %s is a symbolic link, ignoring \n", filenames->fileNames[u]); 1096 } else { 1097 filenames->fileNames[fileNamesNb++] = filenames->fileNames[u]; 1098 } } 1099 if (fileNamesNb == 0 && nbFilenames > 0) /* all names are eliminated */ 1100 CLEAN_RETURN(1); 1101 filenames->tableSize = fileNamesNb; 1102 } /* if (!followLinks) */ 1103 1104 /* read names from a file */ 1105 if (file_of_names->tableSize) { 1106 size_t const nbFileLists = file_of_names->tableSize; 1107 size_t flNb; 1108 for (flNb=0; flNb < nbFileLists; flNb++) { 1109 FileNamesTable* const fnt = UTIL_createFileNamesTable_fromFileName(file_of_names->fileNames[flNb]); 1110 if (fnt==NULL) { 1111 DISPLAYLEVEL(1, "zstd: error reading %s \n", file_of_names->fileNames[flNb]); 1112 CLEAN_RETURN(1); 1113 } 1114 filenames = UTIL_mergeFileNamesTable(filenames, fnt); 1115 } 1116 } 1117 1118 if (recursive) { /* at this stage, filenameTable is a list of paths, which can contain both files and directories */ 1119 UTIL_expandFNT(&filenames, followLinks); 1120 } 1121 #else 1122 (void)followLinks; 1123 #endif 1124 1125 if (operation == zom_list) { 1126 #ifndef ZSTD_NODECOMPRESS 1127 int const ret = FIO_listMultipleFiles((unsigned)filenames->tableSize, filenames->fileNames, g_displayLevel); 1128 CLEAN_RETURN(ret); 1129 #else 1130 DISPLAY("file information is not supported \n"); 1131 CLEAN_RETURN(1); 1132 #endif 1133 } 1134 1135 /* Check if benchmark is selected */ 1136 if (operation==zom_bench) { 1137 #ifndef ZSTD_NOBENCH 1138 benchParams.blockSize = blockSize; 1139 benchParams.nbWorkers = nbWorkers; 1140 benchParams.realTime = (unsigned)setRealTimePrio; 1141 benchParams.nbSeconds = bench_nbSeconds; 1142 benchParams.ldmFlag = ldmFlag; 1143 benchParams.ldmMinMatch = (int)g_ldmMinMatch; 1144 benchParams.ldmHashLog = (int)g_ldmHashLog; 1145 if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) { 1146 benchParams.ldmBucketSizeLog = (int)g_ldmBucketSizeLog; 1147 } 1148 if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) { 1149 benchParams.ldmHashRateLog = (int)g_ldmHashRateLog; 1150 } 1151 benchParams.literalCompressionMode = literalCompressionMode; 1152 1153 if (cLevel > ZSTD_maxCLevel()) cLevel = ZSTD_maxCLevel(); 1154 if (cLevelLast > ZSTD_maxCLevel()) cLevelLast = ZSTD_maxCLevel(); 1155 if (cLevelLast < cLevel) cLevelLast = cLevel; 1156 if (cLevelLast > cLevel) 1157 DISPLAYLEVEL(3, "Benchmarking levels from %d to %d\n", cLevel, cLevelLast); 1158 if (filenames->tableSize > 0) { 1159 if(separateFiles) { 1160 unsigned i; 1161 for(i = 0; i < filenames->tableSize; i++) { 1162 int c; 1163 DISPLAYLEVEL(3, "Benchmarking %s \n", filenames->fileNames[i]); 1164 for(c = cLevel; c <= cLevelLast; c++) { 1165 BMK_benchFilesAdvanced(&filenames->fileNames[i], 1, dictFileName, c, &compressionParams, g_displayLevel, &benchParams); 1166 } } 1167 } else { 1168 for(; cLevel <= cLevelLast; cLevel++) { 1169 BMK_benchFilesAdvanced(filenames->fileNames, (unsigned)filenames->tableSize, dictFileName, cLevel, &compressionParams, g_displayLevel, &benchParams); 1170 } } 1171 } else { 1172 for(; cLevel <= cLevelLast; cLevel++) { 1173 BMK_syntheticTest(cLevel, compressibility, &compressionParams, g_displayLevel, &benchParams); 1174 } } 1175 1176 #else 1177 (void)bench_nbSeconds; (void)blockSize; (void)setRealTimePrio; (void)separateFiles; (void)compressibility; 1178 #endif 1179 goto _end; 1180 } 1181 1182 /* Check if dictionary builder is selected */ 1183 if (operation==zom_train) { 1184 #ifndef ZSTD_NODICT 1185 ZDICT_params_t zParams; 1186 zParams.compressionLevel = dictCLevel; 1187 zParams.notificationLevel = (unsigned)g_displayLevel; 1188 zParams.dictID = dictID; 1189 if (dict == cover) { 1190 int const optimize = !coverParams.k || !coverParams.d; 1191 coverParams.nbThreads = (unsigned)nbWorkers; 1192 coverParams.zParams = zParams; 1193 operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, NULL, &coverParams, NULL, optimize); 1194 } else if (dict == fastCover) { 1195 int const optimize = !fastCoverParams.k || !fastCoverParams.d; 1196 fastCoverParams.nbThreads = (unsigned)nbWorkers; 1197 fastCoverParams.zParams = zParams; 1198 operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, NULL, NULL, &fastCoverParams, optimize); 1199 } else { 1200 ZDICT_legacy_params_t dictParams; 1201 memset(&dictParams, 0, sizeof(dictParams)); 1202 dictParams.selectivityLevel = dictSelect; 1203 dictParams.zParams = zParams; 1204 operationResult = DiB_trainFromFiles(outFileName, maxDictSize, filenames->fileNames, (unsigned)filenames->tableSize, blockSize, &dictParams, NULL, NULL, 0); 1205 } 1206 #else 1207 (void)dictCLevel; (void)dictSelect; (void)dictID; (void)maxDictSize; /* not used when ZSTD_NODICT set */ 1208 DISPLAYLEVEL(1, "training mode not available \n"); 1209 operationResult = 1; 1210 #endif 1211 goto _end; 1212 } 1213 1214 #ifndef ZSTD_NODECOMPRESS 1215 if (operation==zom_test) { FIO_setTestMode(prefs, 1); outFileName=nulmark; FIO_setRemoveSrcFile(prefs, 0); } /* test mode */ 1216 #endif 1217 1218 /* No input filename ==> use stdin and stdout */ 1219 if (filenames->tableSize == 0) UTIL_refFilename(filenames, stdinmark); 1220 if (!strcmp(filenames->fileNames[0], stdinmark) && !outFileName) 1221 outFileName = stdoutmark; /* when input is stdin, default output is stdout */ 1222 1223 /* Check if input/output defined as console; trigger an error in this case */ 1224 if (!strcmp(filenames->fileNames[0], stdinmark) && IS_CONSOLE(stdin) ) { 1225 badusage(programName); 1226 CLEAN_RETURN(1); 1227 } 1228 if ( outFileName && !strcmp(outFileName, stdoutmark) 1229 && IS_CONSOLE(stdout) 1230 && !strcmp(filenames->fileNames[0], stdinmark) 1231 && !forceStdout 1232 && operation!=zom_decompress ) { 1233 badusage(programName); 1234 CLEAN_RETURN(1); 1235 } 1236 1237 #ifndef ZSTD_NOCOMPRESS 1238 /* check compression level limits */ 1239 { int const maxCLevel = ultra ? ZSTD_maxCLevel() : ZSTDCLI_CLEVEL_MAX; 1240 if (cLevel > maxCLevel) { 1241 DISPLAYLEVEL(2, "Warning : compression level higher than max, reduced to %i \n", maxCLevel); 1242 cLevel = maxCLevel; 1243 } } 1244 #endif 1245 1246 if (showDefaultCParams) { 1247 if (operation == zom_decompress) { 1248 DISPLAY("error : can't use --show-default-cparams in decomrpession mode \n"); 1249 CLEAN_RETURN(1); 1250 } 1251 } 1252 1253 if (dictFileName != NULL && patchFromDictFileName != NULL) { 1254 DISPLAY("error : can't use -D and --patch-from=# at the same time \n"); 1255 CLEAN_RETURN(1); 1256 } 1257 1258 if (patchFromDictFileName != NULL && filenames->tableSize > 1) { 1259 DISPLAY("error : can't use --patch-from=# on multiple files \n"); 1260 CLEAN_RETURN(1); 1261 } 1262 1263 /* No status message in pipe mode (stdin - stdout) or multi-files mode */ 1264 if (!strcmp(filenames->fileNames[0], stdinmark) && outFileName && !strcmp(outFileName,stdoutmark) && (g_displayLevel==2)) g_displayLevel=1; 1265 if ((filenames->tableSize > 1) & (g_displayLevel==2)) g_displayLevel=1; 1266 1267 /* IO Stream/File */ 1268 FIO_setNotificationLevel(g_displayLevel); 1269 FIO_setPatchFromMode(prefs, patchFromDictFileName != NULL); 1270 if (memLimit == 0) { 1271 if (compressionParams.windowLog == 0) { 1272 memLimit = (U32)1 << g_defaultMaxWindowLog; 1273 } else { 1274 memLimit = (U32)1 << (compressionParams.windowLog & 31); 1275 } } 1276 if (patchFromDictFileName != NULL) 1277 dictFileName = patchFromDictFileName; 1278 FIO_setMemLimit(prefs, memLimit); 1279 if (operation==zom_compress) { 1280 #ifndef ZSTD_NOCOMPRESS 1281 FIO_setContentSize(prefs, contentSize); 1282 FIO_setNbWorkers(prefs, nbWorkers); 1283 FIO_setBlockSize(prefs, (int)blockSize); 1284 if (g_overlapLog!=OVERLAP_LOG_DEFAULT) FIO_setOverlapLog(prefs, (int)g_overlapLog); 1285 FIO_setLdmFlag(prefs, (unsigned)ldmFlag); 1286 FIO_setLdmHashLog(prefs, (int)g_ldmHashLog); 1287 FIO_setLdmMinMatch(prefs, (int)g_ldmMinMatch); 1288 if (g_ldmBucketSizeLog != LDM_PARAM_DEFAULT) FIO_setLdmBucketSizeLog(prefs, (int)g_ldmBucketSizeLog); 1289 if (g_ldmHashRateLog != LDM_PARAM_DEFAULT) FIO_setLdmHashRateLog(prefs, (int)g_ldmHashRateLog); 1290 FIO_setAdaptiveMode(prefs, (unsigned)adapt); 1291 FIO_setAdaptMin(prefs, adaptMin); 1292 FIO_setAdaptMax(prefs, adaptMax); 1293 FIO_setRsyncable(prefs, rsyncable); 1294 FIO_setStreamSrcSize(prefs, streamSrcSize); 1295 FIO_setTargetCBlockSize(prefs, targetCBlockSize); 1296 FIO_setSrcSizeHint(prefs, srcSizeHint); 1297 FIO_setLiteralCompressionMode(prefs, literalCompressionMode); 1298 if (adaptMin > cLevel) cLevel = adaptMin; 1299 if (adaptMax < cLevel) cLevel = adaptMax; 1300 1301 /* Compare strategies constant with the ground truth */ 1302 { ZSTD_bounds strategyBounds = ZSTD_cParam_getBounds(ZSTD_c_strategy); 1303 assert(ZSTD_NB_STRATEGIES == strategyBounds.upperBound); 1304 (void)strategyBounds; } 1305 1306 if (showDefaultCParams) { 1307 size_t fileNb; 1308 for (fileNb = 0; fileNb < (size_t)filenames->tableSize; fileNb++) { 1309 unsigned long long fileSize = UTIL_getFileSize(filenames->fileNames[fileNb]); 1310 const size_t dictSize = dictFileName != NULL ? (size_t)UTIL_getFileSize(dictFileName) : 0; 1311 const ZSTD_compressionParameters cParams = ZSTD_getCParams(cLevel, fileSize, dictSize); 1312 if (fileSize != UTIL_FILESIZE_UNKNOWN) DISPLAY("%s (%u bytes)\n", filenames->fileNames[fileNb], (unsigned)fileSize); 1313 else DISPLAY("%s (src size unknown)\n", filenames->fileNames[fileNb]); 1314 DISPLAY(" - windowLog : %u\n", cParams.windowLog); 1315 DISPLAY(" - chainLog : %u\n", cParams.chainLog); 1316 DISPLAY(" - hashLog : %u\n", cParams.hashLog); 1317 DISPLAY(" - searchLog : %u\n", cParams.searchLog); 1318 DISPLAY(" - minMatch : %u\n", cParams.minMatch); 1319 DISPLAY(" - targetLength : %u\n", cParams.targetLength); 1320 assert(cParams.strategy < ZSTD_NB_STRATEGIES + 1); 1321 DISPLAY(" - strategy : %s (%u)\n", ZSTD_strategyMap[(int)cParams.strategy], (unsigned)cParams.strategy); 1322 } 1323 } 1324 1325 if ((filenames->tableSize==1) && outFileName) 1326 operationResult = FIO_compressFilename(prefs, outFileName, filenames->fileNames[0], dictFileName, cLevel, compressionParams); 1327 else 1328 operationResult = FIO_compressMultipleFilenames(prefs, filenames->fileNames, (unsigned)filenames->tableSize, outDirName, outFileName, suffix, dictFileName, cLevel, compressionParams); 1329 #else 1330 (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 */ 1331 DISPLAY("Compression not supported \n"); 1332 #endif 1333 } else { /* decompression or test */ 1334 #ifndef ZSTD_NODECOMPRESS 1335 if (filenames->tableSize == 1 && outFileName) { 1336 operationResult = FIO_decompressFilename(prefs, outFileName, filenames->fileNames[0], dictFileName); 1337 } else { 1338 operationResult = FIO_decompressMultipleFilenames(prefs, filenames->fileNames, (unsigned)filenames->tableSize, outDirName, outFileName, dictFileName); 1339 } 1340 #else 1341 DISPLAY("Decompression not supported \n"); 1342 #endif 1343 } 1344 1345 _end: 1346 FIO_freePreferences(prefs); 1347 if (main_pause) waitEnter(); 1348 UTIL_freeFileNamesTable(filenames); 1349 UTIL_freeFileNamesTable(file_of_names); 1350 1351 return operationResult; 1352 } 1353