1 /* 2 * Copyright (c) 2016-present, 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 * Compiler Options 14 ***************************************/ 15 #ifdef _MSC_VER /* Visual */ 16 # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ 17 # pragma warning(disable : 4204) /* non-constant aggregate initializer */ 18 #endif 19 #if defined(__MINGW32__) && !defined(_POSIX_SOURCE) 20 # define _POSIX_SOURCE 1 /* disable %llu warnings with MinGW on Windows */ 21 #endif 22 23 /*-************************************* 24 * Includes 25 ***************************************/ 26 #include "platform.h" /* Large Files support, SET_BINARY_MODE */ 27 #include "util.h" /* UTIL_getFileSize, UTIL_isRegularFile */ 28 #include <stdio.h> /* fprintf, fopen, fread, _fileno, stdin, stdout */ 29 #include <stdlib.h> /* malloc, free */ 30 #include <string.h> /* strcmp, strlen */ 31 #include <assert.h> 32 #include <errno.h> /* errno */ 33 #include <signal.h> 34 35 #if defined (_MSC_VER) 36 # include <sys/stat.h> 37 # include <io.h> 38 #endif 39 40 #include "mem.h" /* U32, U64 */ 41 #include "fileio.h" 42 43 #define ZSTD_STATIC_LINKING_ONLY /* ZSTD_magicNumber, ZSTD_frameHeaderSize_max */ 44 #include "zstd.h" 45 #include "zstd_errors.h" /* ZSTD_error_frameParameter_windowTooLarge */ 46 47 #if defined(ZSTD_GZCOMPRESS) || defined(ZSTD_GZDECOMPRESS) 48 # include <zlib.h> 49 # if !defined(z_const) 50 # define z_const 51 # endif 52 #endif 53 54 #if defined(ZSTD_LZMACOMPRESS) || defined(ZSTD_LZMADECOMPRESS) 55 # include <lzma.h> 56 #endif 57 58 #define LZ4_MAGICNUMBER 0x184D2204 59 #if defined(ZSTD_LZ4COMPRESS) || defined(ZSTD_LZ4DECOMPRESS) 60 # define LZ4F_ENABLE_OBSOLETE_ENUMS 61 # include <lz4frame.h> 62 # include <lz4.h> 63 #endif 64 65 66 /*-************************************* 67 * Constants 68 ***************************************/ 69 #define KB *(1<<10) 70 #define MB *(1<<20) 71 #define GB *(1U<<30) 72 73 #define ADAPT_WINDOWLOG_DEFAULT 23 /* 8 MB */ 74 #define DICTSIZE_MAX (32 MB) /* protection against large input (attack scenario) */ 75 76 #define FNSPACE 30 77 78 79 /*-************************************* 80 * Macros 81 ***************************************/ 82 #define DISPLAY(...) fprintf(stderr, __VA_ARGS__) 83 #define DISPLAYOUT(...) fprintf(stdout, __VA_ARGS__) 84 #define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } } 85 static int g_displayLevel = 2; /* 0 : no display; 1: errors; 2: + result + interaction + warnings; 3: + progression; 4: + information */ 86 void FIO_setNotificationLevel(unsigned level) { g_displayLevel=level; } 87 88 static const U64 g_refreshRate = SEC_TO_MICRO / 6; 89 static UTIL_time_t g_displayClock = UTIL_TIME_INITIALIZER; 90 91 #define READY_FOR_UPDATE() (UTIL_clockSpanMicro(g_displayClock) > g_refreshRate) 92 #define DELAY_NEXT_UPDATE() { g_displayClock = UTIL_getTime(); } 93 #define DISPLAYUPDATE(l, ...) { \ 94 if (g_displayLevel>=l) { \ 95 if (READY_FOR_UPDATE() || (g_displayLevel>=4)) { \ 96 DELAY_NEXT_UPDATE(); \ 97 DISPLAY(__VA_ARGS__); \ 98 if (g_displayLevel>=4) fflush(stderr); \ 99 } } } 100 101 #undef MIN /* in case it would be already defined */ 102 #define MIN(a,b) ((a) < (b) ? (a) : (b)) 103 104 105 #define EXM_THROW(error, ...) \ 106 { \ 107 DISPLAYLEVEL(1, "zstd: "); \ 108 DISPLAYLEVEL(5, "Error defined at %s, line %i : \n", __FILE__, __LINE__); \ 109 DISPLAYLEVEL(1, "error %i : ", error); \ 110 DISPLAYLEVEL(1, __VA_ARGS__); \ 111 DISPLAYLEVEL(1, " \n"); \ 112 exit(error); \ 113 } 114 115 #define CHECK_V(v, f) \ 116 v = f; \ 117 if (ZSTD_isError(v)) { \ 118 DISPLAYLEVEL(5, "%s \n", #f); \ 119 EXM_THROW(11, "%s", ZSTD_getErrorName(v)); \ 120 } 121 #define CHECK(f) { size_t err; CHECK_V(err, f); } 122 123 124 /*-************************************ 125 * Signal (Ctrl-C trapping) 126 **************************************/ 127 static const char* g_artefact = NULL; 128 static void INThandler(int sig) 129 { 130 assert(sig==SIGINT); (void)sig; 131 #if !defined(_MSC_VER) 132 signal(sig, SIG_IGN); /* this invocation generates a buggy warning in Visual Studio */ 133 #endif 134 if (g_artefact) { 135 assert(UTIL_isRegularFile(g_artefact)); 136 remove(g_artefact); 137 } 138 DISPLAY("\n"); 139 exit(2); 140 } 141 static void addHandler(char const* dstFileName) 142 { 143 if (UTIL_isRegularFile(dstFileName)) { 144 g_artefact = dstFileName; 145 signal(SIGINT, INThandler); 146 } else { 147 g_artefact = NULL; 148 } 149 } 150 /* Idempotent */ 151 static void clearHandler(void) 152 { 153 if (g_artefact) signal(SIGINT, SIG_DFL); 154 g_artefact = NULL; 155 } 156 157 158 /*-********************************************************* 159 * Termination signal trapping (Print debug stack trace) 160 ***********************************************************/ 161 #if defined(__has_feature) && !defined(BACKTRACE_ENABLE) /* Clang compiler */ 162 # if (__has_feature(address_sanitizer)) 163 # define BACKTRACE_ENABLE 0 164 # endif /* __has_feature(address_sanitizer) */ 165 #elif defined(__SANITIZE_ADDRESS__) && !defined(BACKTRACE_ENABLE) /* GCC compiler */ 166 # define BACKTRACE_ENABLE 0 167 #endif 168 169 #if !defined(BACKTRACE_ENABLE) 170 /* automatic detector : backtrace enabled by default on linux+glibc and osx */ 171 # if (defined(__linux__) && defined(__GLIBC__)) \ 172 || (defined(__APPLE__) && defined(__MACH__)) 173 # define BACKTRACE_ENABLE 1 174 # else 175 # define BACKTRACE_ENABLE 0 176 # endif 177 #endif 178 179 /* note : after this point, BACKTRACE_ENABLE is necessarily defined */ 180 181 182 #if BACKTRACE_ENABLE 183 184 #include <execinfo.h> /* backtrace, backtrace_symbols */ 185 186 #define MAX_STACK_FRAMES 50 187 188 static void ABRThandler(int sig) { 189 const char* name; 190 void* addrlist[MAX_STACK_FRAMES]; 191 char** symbollist; 192 U32 addrlen, i; 193 194 switch (sig) { 195 case SIGABRT: name = "SIGABRT"; break; 196 case SIGFPE: name = "SIGFPE"; break; 197 case SIGILL: name = "SIGILL"; break; 198 case SIGINT: name = "SIGINT"; break; 199 case SIGSEGV: name = "SIGSEGV"; break; 200 default: name = "UNKNOWN"; 201 } 202 203 DISPLAY("Caught %s signal, printing stack:\n", name); 204 /* Retrieve current stack addresses. */ 205 addrlen = backtrace(addrlist, MAX_STACK_FRAMES); 206 if (addrlen == 0) { 207 DISPLAY("\n"); 208 return; 209 } 210 /* Create readable strings to each frame. */ 211 symbollist = backtrace_symbols(addrlist, addrlen); 212 /* Print the stack trace, excluding calls handling the signal. */ 213 for (i = ZSTD_START_SYMBOLLIST_FRAME; i < addrlen; i++) { 214 DISPLAY("%s\n", symbollist[i]); 215 } 216 free(symbollist); 217 /* Reset and raise the signal so default handler runs. */ 218 signal(sig, SIG_DFL); 219 raise(sig); 220 } 221 #endif 222 223 void FIO_addAbortHandler() 224 { 225 #if BACKTRACE_ENABLE 226 signal(SIGABRT, ABRThandler); 227 signal(SIGFPE, ABRThandler); 228 signal(SIGILL, ABRThandler); 229 signal(SIGSEGV, ABRThandler); 230 signal(SIGBUS, ABRThandler); 231 #endif 232 } 233 234 235 /*-************************************************************ 236 * Avoid fseek()'s 2GiB barrier with MSVC, macOS, *BSD, MinGW 237 ***************************************************************/ 238 #if defined(_MSC_VER) && _MSC_VER >= 1400 239 # define LONG_SEEK _fseeki64 240 #elif !defined(__64BIT__) && (PLATFORM_POSIX_VERSION >= 200112L) /* No point defining Large file for 64 bit */ 241 # define LONG_SEEK fseeko 242 #elif defined(__MINGW32__) && !defined(__STRICT_ANSI__) && !defined(__NO_MINGW_LFS) && defined(__MSVCRT__) 243 # define LONG_SEEK fseeko64 244 #elif defined(_WIN32) && !defined(__DJGPP__) 245 # include <windows.h> 246 static int LONG_SEEK(FILE* file, __int64 offset, int origin) { 247 LARGE_INTEGER off; 248 DWORD method; 249 off.QuadPart = offset; 250 if (origin == SEEK_END) 251 method = FILE_END; 252 else if (origin == SEEK_CUR) 253 method = FILE_CURRENT; 254 else 255 method = FILE_BEGIN; 256 257 if (SetFilePointerEx((HANDLE) _get_osfhandle(_fileno(file)), off, NULL, method)) 258 return 0; 259 else 260 return -1; 261 } 262 #else 263 # define LONG_SEEK fseek 264 #endif 265 266 267 /*-************************************* 268 * Local Parameters - Not thread safe 269 ***************************************/ 270 static FIO_compressionType_t g_compressionType = FIO_zstdCompression; 271 void FIO_setCompressionType(FIO_compressionType_t compressionType) { g_compressionType = compressionType; } 272 static U32 g_overwrite = 0; 273 void FIO_overwriteMode(void) { g_overwrite=1; } 274 static U32 g_sparseFileSupport = 1; /* 0: no sparse allowed; 1: auto (file yes, stdout no); 2: force sparse */ 275 void FIO_setSparseWrite(unsigned sparse) { g_sparseFileSupport=sparse; } 276 static U32 g_dictIDFlag = 1; 277 void FIO_setDictIDFlag(unsigned dictIDFlag) { g_dictIDFlag = dictIDFlag; } 278 static U32 g_checksumFlag = 1; 279 void FIO_setChecksumFlag(unsigned checksumFlag) { g_checksumFlag = checksumFlag; } 280 static U32 g_removeSrcFile = 0; 281 void FIO_setRemoveSrcFile(unsigned flag) { g_removeSrcFile = (flag>0); } 282 static U32 g_memLimit = 0; 283 void FIO_setMemLimit(unsigned memLimit) { g_memLimit = memLimit; } 284 static U32 g_nbWorkers = 1; 285 void FIO_setNbWorkers(unsigned nbWorkers) { 286 #ifndef ZSTD_MULTITHREAD 287 if (nbWorkers > 0) DISPLAYLEVEL(2, "Note : multi-threading is disabled \n"); 288 #endif 289 g_nbWorkers = nbWorkers; 290 } 291 static U32 g_blockSize = 0; 292 void FIO_setBlockSize(unsigned blockSize) { 293 if (blockSize && g_nbWorkers==0) 294 DISPLAYLEVEL(2, "Setting block size is useless in single-thread mode \n"); 295 g_blockSize = blockSize; 296 } 297 #define FIO_OVERLAP_LOG_NOTSET 9999 298 static U32 g_overlapLog = FIO_OVERLAP_LOG_NOTSET; 299 void FIO_setOverlapLog(unsigned overlapLog){ 300 if (overlapLog && g_nbWorkers==0) 301 DISPLAYLEVEL(2, "Setting overlapLog is useless in single-thread mode \n"); 302 g_overlapLog = overlapLog; 303 } 304 static U32 g_adaptiveMode = 0; 305 void FIO_setAdaptiveMode(unsigned adapt) { 306 if ((adapt>0) && (g_nbWorkers==0)) 307 EXM_THROW(1, "Adaptive mode is not compatible with single thread mode \n"); 308 g_adaptiveMode = adapt; 309 } 310 static int g_minAdaptLevel = -50; /* initializing this value requires a constant, so ZSTD_minCLevel() doesn't work */ 311 void FIO_setAdaptMin(int minCLevel) 312 { 313 #ifndef ZSTD_NOCOMPRESS 314 assert(minCLevel >= ZSTD_minCLevel()); 315 #endif 316 g_minAdaptLevel = minCLevel; 317 } 318 static int g_maxAdaptLevel = 22; /* initializing this value requires a constant, so ZSTD_maxCLevel() doesn't work */ 319 void FIO_setAdaptMax(int maxCLevel) 320 { 321 g_maxAdaptLevel = maxCLevel; 322 } 323 324 static U32 g_ldmFlag = 0; 325 void FIO_setLdmFlag(unsigned ldmFlag) { 326 g_ldmFlag = (ldmFlag>0); 327 } 328 static U32 g_ldmHashLog = 0; 329 void FIO_setLdmHashLog(unsigned ldmHashLog) { 330 g_ldmHashLog = ldmHashLog; 331 } 332 static U32 g_ldmMinMatch = 0; 333 void FIO_setLdmMinMatch(unsigned ldmMinMatch) { 334 g_ldmMinMatch = ldmMinMatch; 335 } 336 337 #define FIO_LDM_PARAM_NOTSET 9999 338 static U32 g_ldmBucketSizeLog = FIO_LDM_PARAM_NOTSET; 339 void FIO_setLdmBucketSizeLog(unsigned ldmBucketSizeLog) { 340 g_ldmBucketSizeLog = ldmBucketSizeLog; 341 } 342 343 static U32 g_ldmHashEveryLog = FIO_LDM_PARAM_NOTSET; 344 void FIO_setLdmHashEveryLog(unsigned ldmHashEveryLog) { 345 g_ldmHashEveryLog = ldmHashEveryLog; 346 } 347 348 349 350 /*-************************************* 351 * Functions 352 ***************************************/ 353 /** FIO_remove() : 354 * @result : Unlink `fileName`, even if it's read-only */ 355 static int FIO_remove(const char* path) 356 { 357 if (!UTIL_isRegularFile(path)) { 358 DISPLAYLEVEL(2, "zstd: Refusing to remove non-regular file %s\n", path); 359 return 0; 360 } 361 #if defined(_WIN32) || defined(WIN32) 362 /* windows doesn't allow remove read-only files, 363 * so try to make it writable first */ 364 chmod(path, _S_IWRITE); 365 #endif 366 return remove(path); 367 } 368 369 /** FIO_openSrcFile() : 370 * condition : `srcFileName` must be non-NULL. 371 * @result : FILE* to `srcFileName`, or NULL if it fails */ 372 static FILE* FIO_openSrcFile(const char* srcFileName) 373 { 374 assert(srcFileName != NULL); 375 if (!strcmp (srcFileName, stdinmark)) { 376 DISPLAYLEVEL(4,"Using stdin for input\n"); 377 SET_BINARY_MODE(stdin); 378 return stdin; 379 } 380 381 if (!UTIL_isRegularFile(srcFileName)) { 382 DISPLAYLEVEL(1, "zstd: %s is not a regular file -- ignored \n", 383 srcFileName); 384 return NULL; 385 } 386 387 { FILE* const f = fopen(srcFileName, "rb"); 388 if (f == NULL) 389 DISPLAYLEVEL(1, "zstd: %s: %s \n", srcFileName, strerror(errno)); 390 return f; 391 } 392 } 393 394 /** FIO_openDstFile() : 395 * condition : `dstFileName` must be non-NULL. 396 * @result : FILE* to `dstFileName`, or NULL if it fails */ 397 static FILE* FIO_openDstFile(const char* dstFileName) 398 { 399 assert(dstFileName != NULL); 400 if (!strcmp (dstFileName, stdoutmark)) { 401 DISPLAYLEVEL(4,"Using stdout for output\n"); 402 SET_BINARY_MODE(stdout); 403 if (g_sparseFileSupport==1) { 404 g_sparseFileSupport = 0; 405 DISPLAYLEVEL(4, "Sparse File Support is automatically disabled on stdout ; try --sparse \n"); 406 } 407 return stdout; 408 } 409 410 if (g_sparseFileSupport == 1) { 411 g_sparseFileSupport = ZSTD_SPARSE_DEFAULT; 412 } 413 414 if (UTIL_isRegularFile(dstFileName)) { 415 FILE* fCheck; 416 if (!strcmp(dstFileName, nulmark)) { 417 EXM_THROW(40, "%s is unexpectedly a regular file", dstFileName); 418 } 419 /* Check if destination file already exists */ 420 fCheck = fopen( dstFileName, "rb" ); 421 if (fCheck != NULL) { /* dst file exists, authorization prompt */ 422 fclose(fCheck); 423 if (!g_overwrite) { 424 if (g_displayLevel <= 1) { 425 /* No interaction possible */ 426 DISPLAY("zstd: %s already exists; not overwritten \n", 427 dstFileName); 428 return NULL; 429 } 430 DISPLAY("zstd: %s already exists; overwrite (y/N) ? ", 431 dstFileName); 432 { int ch = getchar(); 433 if ((ch!='Y') && (ch!='y')) { 434 DISPLAY(" not overwritten \n"); 435 return NULL; 436 } 437 /* flush rest of input line */ 438 while ((ch!=EOF) && (ch!='\n')) ch = getchar(); 439 } } 440 /* need to unlink */ 441 FIO_remove(dstFileName); 442 } } 443 444 { FILE* const f = fopen( dstFileName, "wb" ); 445 if (f == NULL) 446 DISPLAYLEVEL(1, "zstd: %s: %s\n", dstFileName, strerror(errno)); 447 return f; 448 } 449 } 450 451 452 /*! FIO_createDictBuffer() : 453 * creates a buffer, pointed by `*bufferPtr`, 454 * loads `filename` content into it, up to DICTSIZE_MAX bytes. 455 * @return : loaded size 456 * if fileName==NULL, returns 0 and a NULL pointer 457 */ 458 static size_t FIO_createDictBuffer(void** bufferPtr, const char* fileName) 459 { 460 FILE* fileHandle; 461 U64 fileSize; 462 463 assert(bufferPtr != NULL); 464 *bufferPtr = NULL; 465 if (fileName == NULL) return 0; 466 467 DISPLAYLEVEL(4,"Loading %s as dictionary \n", fileName); 468 fileHandle = fopen(fileName, "rb"); 469 if (fileHandle==NULL) EXM_THROW(31, "%s: %s", fileName, strerror(errno)); 470 fileSize = UTIL_getFileSize(fileName); 471 if (fileSize > DICTSIZE_MAX) { 472 EXM_THROW(32, "Dictionary file %s is too large (> %u MB)", 473 fileName, DICTSIZE_MAX >> 20); /* avoid extreme cases */ 474 } 475 *bufferPtr = malloc((size_t)fileSize); 476 if (*bufferPtr==NULL) EXM_THROW(34, "%s", strerror(errno)); 477 { size_t const readSize = fread(*bufferPtr, 1, (size_t)fileSize, fileHandle); 478 if (readSize!=fileSize) 479 EXM_THROW(35, "Error reading dictionary file %s", fileName); 480 } 481 fclose(fileHandle); 482 return (size_t)fileSize; 483 } 484 485 #ifndef ZSTD_NOCOMPRESS 486 487 /* ********************************************************************** 488 * Compression 489 ************************************************************************/ 490 typedef struct { 491 FILE* srcFile; 492 FILE* dstFile; 493 void* srcBuffer; 494 size_t srcBufferSize; 495 void* dstBuffer; 496 size_t dstBufferSize; 497 ZSTD_CStream* cctx; 498 } cRess_t; 499 500 static cRess_t FIO_createCResources(const char* dictFileName, int cLevel, 501 U64 srcSize, 502 ZSTD_compressionParameters comprParams) { 503 cRess_t ress; 504 memset(&ress, 0, sizeof(ress)); 505 506 DISPLAYLEVEL(6, "FIO_createCResources \n"); 507 ress.cctx = ZSTD_createCCtx(); 508 if (ress.cctx == NULL) 509 EXM_THROW(30, "allocation error : can't create ZSTD_CCtx"); 510 ress.srcBufferSize = ZSTD_CStreamInSize(); 511 ress.srcBuffer = malloc(ress.srcBufferSize); 512 ress.dstBufferSize = ZSTD_CStreamOutSize(); 513 ress.dstBuffer = malloc(ress.dstBufferSize); 514 if (!ress.srcBuffer || !ress.dstBuffer) 515 EXM_THROW(31, "allocation error : not enough memory"); 516 517 /* Advanced parameters, including dictionary */ 518 { void* dictBuffer; 519 size_t const dictBuffSize = FIO_createDictBuffer(&dictBuffer, dictFileName); /* works with dictFileName==NULL */ 520 if (dictFileName && (dictBuffer==NULL)) 521 EXM_THROW(32, "allocation error : can't create dictBuffer"); 522 523 if (g_adaptiveMode && !g_ldmFlag && !comprParams.windowLog) 524 comprParams.windowLog = ADAPT_WINDOWLOG_DEFAULT; 525 526 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_contentSizeFlag, 1) ); /* always enable content size when available (note: supposed to be default) */ 527 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_dictIDFlag, g_dictIDFlag) ); 528 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_checksumFlag, g_checksumFlag) ); 529 /* compression level */ 530 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_compressionLevel, (unsigned)cLevel) ); 531 /* long distance matching */ 532 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_enableLongDistanceMatching, g_ldmFlag) ); 533 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_ldmHashLog, g_ldmHashLog) ); 534 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_ldmMinMatch, g_ldmMinMatch) ); 535 if (g_ldmBucketSizeLog != FIO_LDM_PARAM_NOTSET) { 536 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_ldmBucketSizeLog, g_ldmBucketSizeLog) ); 537 } 538 if (g_ldmHashEveryLog != FIO_LDM_PARAM_NOTSET) { 539 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_ldmHashEveryLog, g_ldmHashEveryLog) ); 540 } 541 /* compression parameters */ 542 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_windowLog, comprParams.windowLog) ); 543 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_chainLog, comprParams.chainLog) ); 544 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_hashLog, comprParams.hashLog) ); 545 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_searchLog, comprParams.searchLog) ); 546 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_minMatch, comprParams.searchLength) ); 547 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_targetLength, comprParams.targetLength) ); 548 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_compressionStrategy, (U32)comprParams.strategy) ); 549 /* multi-threading */ 550 #ifdef ZSTD_MULTITHREAD 551 DISPLAYLEVEL(5,"set nb workers = %u \n", g_nbWorkers); 552 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_nbWorkers, g_nbWorkers) ); 553 if ( (g_overlapLog == FIO_OVERLAP_LOG_NOTSET) 554 && (cLevel == ZSTD_maxCLevel()) ) 555 g_overlapLog = 9; /* full overlap */ 556 if (g_overlapLog != FIO_OVERLAP_LOG_NOTSET) { 557 DISPLAYLEVEL(3,"set overlapLog = %u \n", g_overlapLog); 558 CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_overlapSizeLog, g_overlapLog) ); 559 } 560 #endif 561 /* dictionary */ 562 CHECK( ZSTD_CCtx_setPledgedSrcSize(ress.cctx, srcSize) ); /* set the value temporarily for dictionary loading, to adapt compression parameters */ 563 CHECK( ZSTD_CCtx_loadDictionary(ress.cctx, dictBuffer, dictBuffSize) ); 564 CHECK( ZSTD_CCtx_setPledgedSrcSize(ress.cctx, ZSTD_CONTENTSIZE_UNKNOWN) ); /* reset */ 565 566 free(dictBuffer); 567 } 568 569 return ress; 570 } 571 572 static void FIO_freeCResources(cRess_t ress) 573 { 574 free(ress.srcBuffer); 575 free(ress.dstBuffer); 576 ZSTD_freeCStream(ress.cctx); /* never fails */ 577 } 578 579 580 #ifdef ZSTD_GZCOMPRESS 581 static unsigned long long 582 FIO_compressGzFrame(cRess_t* ress, 583 const char* srcFileName, U64 const srcFileSize, 584 int compressionLevel, U64* readsize) 585 { 586 unsigned long long inFileSize = 0, outFileSize = 0; 587 z_stream strm; 588 int ret; 589 590 if (compressionLevel > Z_BEST_COMPRESSION) 591 compressionLevel = Z_BEST_COMPRESSION; 592 593 strm.zalloc = Z_NULL; 594 strm.zfree = Z_NULL; 595 strm.opaque = Z_NULL; 596 597 ret = deflateInit2(&strm, compressionLevel, Z_DEFLATED, 598 15 /* maxWindowLogSize */ + 16 /* gzip only */, 599 8, Z_DEFAULT_STRATEGY); /* see http://www.zlib.net/manual.html */ 600 if (ret != Z_OK) 601 EXM_THROW(71, "zstd: %s: deflateInit2 error %d \n", srcFileName, ret); 602 603 strm.next_in = 0; 604 strm.avail_in = 0; 605 strm.next_out = (Bytef*)ress->dstBuffer; 606 strm.avail_out = (uInt)ress->dstBufferSize; 607 608 while (1) { 609 if (strm.avail_in == 0) { 610 size_t const inSize = fread(ress->srcBuffer, 1, ress->srcBufferSize, ress->srcFile); 611 if (inSize == 0) break; 612 inFileSize += inSize; 613 strm.next_in = (z_const unsigned char*)ress->srcBuffer; 614 strm.avail_in = (uInt)inSize; 615 } 616 ret = deflate(&strm, Z_NO_FLUSH); 617 if (ret != Z_OK) 618 EXM_THROW(72, "zstd: %s: deflate error %d \n", srcFileName, ret); 619 { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; 620 if (decompBytes) { 621 if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) 622 EXM_THROW(73, "Write error : cannot write to output file"); 623 outFileSize += decompBytes; 624 strm.next_out = (Bytef*)ress->dstBuffer; 625 strm.avail_out = (uInt)ress->dstBufferSize; 626 } 627 } 628 if (srcFileSize == UTIL_FILESIZE_UNKNOWN) 629 DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", 630 (U32)(inFileSize>>20), 631 (double)outFileSize/inFileSize*100) 632 else 633 DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", 634 (U32)(inFileSize>>20), (U32)(srcFileSize>>20), 635 (double)outFileSize/inFileSize*100); 636 } 637 638 while (1) { 639 ret = deflate(&strm, Z_FINISH); 640 { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; 641 if (decompBytes) { 642 if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) 643 EXM_THROW(75, "Write error : cannot write to output file"); 644 outFileSize += decompBytes; 645 strm.next_out = (Bytef*)ress->dstBuffer; 646 strm.avail_out = (uInt)ress->dstBufferSize; 647 } } 648 if (ret == Z_STREAM_END) break; 649 if (ret != Z_BUF_ERROR) 650 EXM_THROW(77, "zstd: %s: deflate error %d \n", srcFileName, ret); 651 } 652 653 ret = deflateEnd(&strm); 654 if (ret != Z_OK) 655 EXM_THROW(79, "zstd: %s: deflateEnd error %d \n", srcFileName, ret); 656 *readsize = inFileSize; 657 658 return outFileSize; 659 } 660 #endif 661 662 663 #ifdef ZSTD_LZMACOMPRESS 664 static unsigned long long 665 FIO_compressLzmaFrame(cRess_t* ress, 666 const char* srcFileName, U64 const srcFileSize, 667 int compressionLevel, U64* readsize, int plain_lzma) 668 { 669 unsigned long long inFileSize = 0, outFileSize = 0; 670 lzma_stream strm = LZMA_STREAM_INIT; 671 lzma_action action = LZMA_RUN; 672 lzma_ret ret; 673 674 if (compressionLevel < 0) compressionLevel = 0; 675 if (compressionLevel > 9) compressionLevel = 9; 676 677 if (plain_lzma) { 678 lzma_options_lzma opt_lzma; 679 if (lzma_lzma_preset(&opt_lzma, compressionLevel)) 680 EXM_THROW(71, "zstd: %s: lzma_lzma_preset error", srcFileName); 681 ret = lzma_alone_encoder(&strm, &opt_lzma); /* LZMA */ 682 if (ret != LZMA_OK) 683 EXM_THROW(71, "zstd: %s: lzma_alone_encoder error %d", srcFileName, ret); 684 } else { 685 ret = lzma_easy_encoder(&strm, compressionLevel, LZMA_CHECK_CRC64); /* XZ */ 686 if (ret != LZMA_OK) 687 EXM_THROW(71, "zstd: %s: lzma_easy_encoder error %d", srcFileName, ret); 688 } 689 690 strm.next_in = 0; 691 strm.avail_in = 0; 692 strm.next_out = (BYTE*)ress->dstBuffer; 693 strm.avail_out = ress->dstBufferSize; 694 695 while (1) { 696 if (strm.avail_in == 0) { 697 size_t const inSize = fread(ress->srcBuffer, 1, ress->srcBufferSize, ress->srcFile); 698 if (inSize == 0) action = LZMA_FINISH; 699 inFileSize += inSize; 700 strm.next_in = (BYTE const*)ress->srcBuffer; 701 strm.avail_in = inSize; 702 } 703 704 ret = lzma_code(&strm, action); 705 706 if (ret != LZMA_OK && ret != LZMA_STREAM_END) 707 EXM_THROW(72, "zstd: %s: lzma_code encoding error %d", srcFileName, ret); 708 { size_t const compBytes = ress->dstBufferSize - strm.avail_out; 709 if (compBytes) { 710 if (fwrite(ress->dstBuffer, 1, compBytes, ress->dstFile) != compBytes) 711 EXM_THROW(73, "Write error : cannot write to output file"); 712 outFileSize += compBytes; 713 strm.next_out = (BYTE*)ress->dstBuffer; 714 strm.avail_out = ress->dstBufferSize; 715 } } 716 if (srcFileSize == UTIL_FILESIZE_UNKNOWN) 717 DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", 718 (U32)(inFileSize>>20), 719 (double)outFileSize/inFileSize*100) 720 else 721 DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", 722 (U32)(inFileSize>>20), (U32)(srcFileSize>>20), 723 (double)outFileSize/inFileSize*100); 724 if (ret == LZMA_STREAM_END) break; 725 } 726 727 lzma_end(&strm); 728 *readsize = inFileSize; 729 730 return outFileSize; 731 } 732 #endif 733 734 #ifdef ZSTD_LZ4COMPRESS 735 #if LZ4_VERSION_NUMBER <= 10600 736 #define LZ4F_blockLinked blockLinked 737 #define LZ4F_max64KB max64KB 738 #endif 739 static int FIO_LZ4_GetBlockSize_FromBlockId (int id) { return (1 << (8 + (2 * id))); } 740 static unsigned long long 741 FIO_compressLz4Frame(cRess_t* ress, 742 const char* srcFileName, U64 const srcFileSize, 743 int compressionLevel, U64* readsize) 744 { 745 const size_t blockSize = FIO_LZ4_GetBlockSize_FromBlockId(LZ4F_max64KB); 746 unsigned long long inFileSize = 0, outFileSize = 0; 747 748 LZ4F_preferences_t prefs; 749 LZ4F_compressionContext_t ctx; 750 751 LZ4F_errorCode_t const errorCode = LZ4F_createCompressionContext(&ctx, LZ4F_VERSION); 752 if (LZ4F_isError(errorCode)) 753 EXM_THROW(31, "zstd: failed to create lz4 compression context"); 754 755 memset(&prefs, 0, sizeof(prefs)); 756 757 assert(blockSize <= ress->srcBufferSize); 758 759 prefs.autoFlush = 1; 760 prefs.compressionLevel = compressionLevel; 761 prefs.frameInfo.blockMode = LZ4F_blockLinked; 762 prefs.frameInfo.blockSizeID = LZ4F_max64KB; 763 prefs.frameInfo.contentChecksumFlag = (contentChecksum_t)g_checksumFlag; 764 #if LZ4_VERSION_NUMBER >= 10600 765 prefs.frameInfo.contentSize = (srcFileSize==UTIL_FILESIZE_UNKNOWN) ? 0 : srcFileSize; 766 #endif 767 assert(LZ4F_compressBound(blockSize, &prefs) <= ress->dstBufferSize); 768 769 { 770 size_t readSize; 771 size_t headerSize = LZ4F_compressBegin(ctx, ress->dstBuffer, ress->dstBufferSize, &prefs); 772 if (LZ4F_isError(headerSize)) 773 EXM_THROW(33, "File header generation failed : %s", 774 LZ4F_getErrorName(headerSize)); 775 if (fwrite(ress->dstBuffer, 1, headerSize, ress->dstFile) != headerSize) 776 EXM_THROW(34, "Write error : cannot write header"); 777 outFileSize += headerSize; 778 779 /* Read first block */ 780 readSize = fread(ress->srcBuffer, (size_t)1, (size_t)blockSize, ress->srcFile); 781 inFileSize += readSize; 782 783 /* Main Loop */ 784 while (readSize>0) { 785 size_t outSize; 786 787 /* Compress Block */ 788 outSize = LZ4F_compressUpdate(ctx, ress->dstBuffer, ress->dstBufferSize, ress->srcBuffer, readSize, NULL); 789 if (LZ4F_isError(outSize)) 790 EXM_THROW(35, "zstd: %s: lz4 compression failed : %s", 791 srcFileName, LZ4F_getErrorName(outSize)); 792 outFileSize += outSize; 793 if (srcFileSize == UTIL_FILESIZE_UNKNOWN) 794 DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", 795 (U32)(inFileSize>>20), 796 (double)outFileSize/inFileSize*100) 797 else 798 DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", 799 (U32)(inFileSize>>20), (U32)(srcFileSize>>20), 800 (double)outFileSize/inFileSize*100); 801 802 /* Write Block */ 803 { size_t const sizeCheck = fwrite(ress->dstBuffer, 1, outSize, ress->dstFile); 804 if (sizeCheck!=outSize) EXM_THROW(36, "Write error : cannot write compressed block"); } 805 806 /* Read next block */ 807 readSize = fread(ress->srcBuffer, (size_t)1, (size_t)blockSize, ress->srcFile); 808 inFileSize += readSize; 809 } 810 if (ferror(ress->srcFile)) EXM_THROW(37, "Error reading %s ", srcFileName); 811 812 /* End of Stream mark */ 813 headerSize = LZ4F_compressEnd(ctx, ress->dstBuffer, ress->dstBufferSize, NULL); 814 if (LZ4F_isError(headerSize)) 815 EXM_THROW(38, "zstd: %s: lz4 end of file generation failed : %s", 816 srcFileName, LZ4F_getErrorName(headerSize)); 817 818 { size_t const sizeCheck = fwrite(ress->dstBuffer, 1, headerSize, ress->dstFile); 819 if (sizeCheck!=headerSize) EXM_THROW(39, "Write error : cannot write end of stream"); } 820 outFileSize += headerSize; 821 } 822 823 *readsize = inFileSize; 824 LZ4F_freeCompressionContext(ctx); 825 826 return outFileSize; 827 } 828 #endif 829 830 831 static unsigned long long 832 FIO_compressZstdFrame(const cRess_t* ressPtr, 833 const char* srcFileName, U64 fileSize, 834 int compressionLevel, U64* readsize) 835 { 836 cRess_t const ress = *ressPtr; 837 FILE* const srcFile = ress.srcFile; 838 FILE* const dstFile = ress.dstFile; 839 U64 compressedfilesize = 0; 840 ZSTD_EndDirective directive = ZSTD_e_continue; 841 842 /* stats */ 843 ZSTD_frameProgression previous_zfp_update = { 0, 0, 0, 0, 0, 0 }; 844 ZSTD_frameProgression previous_zfp_correction = { 0, 0, 0, 0, 0, 0 }; 845 typedef enum { noChange, slower, faster } speedChange_e; 846 speedChange_e speedChange = noChange; 847 unsigned flushWaiting = 0; 848 unsigned inputPresented = 0; 849 unsigned inputBlocked = 0; 850 unsigned lastJobID = 0; 851 852 DISPLAYLEVEL(6, "compression using zstd format \n"); 853 854 /* init */ 855 if (fileSize != UTIL_FILESIZE_UNKNOWN) { 856 CHECK(ZSTD_CCtx_setPledgedSrcSize(ress.cctx, fileSize)); 857 } 858 (void)srcFileName; 859 860 /* Main compression loop */ 861 do { 862 size_t stillToFlush; 863 /* Fill input Buffer */ 864 size_t const inSize = fread(ress.srcBuffer, (size_t)1, ress.srcBufferSize, srcFile); 865 ZSTD_inBuffer inBuff = { ress.srcBuffer, inSize, 0 }; 866 DISPLAYLEVEL(6, "fread %u bytes from source \n", (U32)inSize); 867 *readsize += inSize; 868 869 if ((inSize == 0) || (*readsize == fileSize)) 870 directive = ZSTD_e_end; 871 872 stillToFlush = 1; 873 while ((inBuff.pos != inBuff.size) /* input buffer must be entirely ingested */ 874 || (directive == ZSTD_e_end && stillToFlush != 0) ) { 875 876 size_t const oldIPos = inBuff.pos; 877 ZSTD_outBuffer outBuff = { ress.dstBuffer, ress.dstBufferSize, 0 }; 878 size_t const toFlushNow = ZSTD_toFlushNow(ress.cctx); 879 CHECK_V(stillToFlush, ZSTD_compress_generic(ress.cctx, &outBuff, &inBuff, directive)); 880 881 /* count stats */ 882 inputPresented++; 883 if (oldIPos == inBuff.pos) inputBlocked++; /* input buffer is full and can't take any more : input speed is faster than consumption rate */ 884 if (!toFlushNow) flushWaiting = 1; 885 886 /* Write compressed stream */ 887 DISPLAYLEVEL(6, "ZSTD_compress_generic(end:%u) => input pos(%u)<=(%u)size ; output generated %u bytes \n", 888 (U32)directive, (U32)inBuff.pos, (U32)inBuff.size, (U32)outBuff.pos); 889 if (outBuff.pos) { 890 size_t const sizeCheck = fwrite(ress.dstBuffer, 1, outBuff.pos, dstFile); 891 if (sizeCheck != outBuff.pos) 892 EXM_THROW(25, "Write error : cannot write compressed block"); 893 compressedfilesize += outBuff.pos; 894 } 895 896 /* display notification; and adapt compression level */ 897 if (READY_FOR_UPDATE()) { 898 ZSTD_frameProgression const zfp = ZSTD_getFrameProgression(ress.cctx); 899 double const cShare = (double)zfp.produced / (zfp.consumed + !zfp.consumed/*avoid div0*/) * 100; 900 901 /* display progress notifications */ 902 if (g_displayLevel >= 3) { 903 DISPLAYUPDATE(3, "\r(L%i) Buffered :%4u MB - Consumed :%4u MB - Compressed :%4u MB => %.2f%% ", 904 compressionLevel, 905 (U32)((zfp.ingested - zfp.consumed) >> 20), 906 (U32)(zfp.consumed >> 20), 907 (U32)(zfp.produced >> 20), 908 cShare ); 909 } else { /* summarized notifications if == 2; */ 910 DISPLAYLEVEL(2, "\rRead : %u ", (U32)(zfp.consumed >> 20)); 911 if (fileSize != UTIL_FILESIZE_UNKNOWN) 912 DISPLAYLEVEL(2, "/ %u ", (U32)(fileSize >> 20)); 913 DISPLAYLEVEL(2, "MB ==> %2.f%% ", cShare); 914 DELAY_NEXT_UPDATE(); 915 } 916 917 /* adaptive mode : statistics measurement and speed correction */ 918 if (g_adaptiveMode) { 919 920 /* check output speed */ 921 if (zfp.currentJobID > 1) { /* only possible if nbWorkers >= 1 */ 922 923 unsigned long long newlyProduced = zfp.produced - previous_zfp_update.produced; 924 unsigned long long newlyFlushed = zfp.flushed - previous_zfp_update.flushed; 925 assert(zfp.produced >= previous_zfp_update.produced); 926 assert(g_nbWorkers >= 1); 927 928 /* test if compression is blocked 929 * either because output is slow and all buffers are full 930 * or because input is slow and no job can start while waiting for at least one buffer to be filled. 931 * note : excluse starting part, since currentJobID > 1 */ 932 if ( (zfp.consumed == previous_zfp_update.consumed) /* no data compressed : no data available, or no more buffer to compress to, OR compression is really slow (compression of a single block is slower than update rate)*/ 933 && (zfp.nbActiveWorkers == 0) /* confirmed : no compression ongoing */ 934 ) { 935 DISPLAYLEVEL(6, "all buffers full : compression stopped => slow down \n") 936 speedChange = slower; 937 } 938 939 previous_zfp_update = zfp; 940 941 if ( (newlyProduced > (newlyFlushed * 9 / 8)) /* compression produces more data than output can flush (though production can be spiky, due to work unit : (N==4)*block sizes) */ 942 && (flushWaiting == 0) /* flush speed was never slowed by lack of production, so it's operating at max capacity */ 943 ) { 944 DISPLAYLEVEL(6, "compression faster than flush (%llu > %llu), and flushed was never slowed down by lack of production => slow down \n", newlyProduced, newlyFlushed); 945 speedChange = slower; 946 } 947 flushWaiting = 0; 948 } 949 950 /* course correct only if there is at least one new job completed */ 951 if (zfp.currentJobID > lastJobID) { 952 DISPLAYLEVEL(6, "compression level adaptation check \n") 953 954 /* check input speed */ 955 if (zfp.currentJobID > g_nbWorkers+1) { /* warm up period, to fill all workers */ 956 if (inputBlocked <= 0) { 957 DISPLAYLEVEL(6, "input is never blocked => input is slower than ingestion \n"); 958 speedChange = slower; 959 } else if (speedChange == noChange) { 960 unsigned long long newlyIngested = zfp.ingested - previous_zfp_correction.ingested; 961 unsigned long long newlyConsumed = zfp.consumed - previous_zfp_correction.consumed; 962 unsigned long long newlyProduced = zfp.produced - previous_zfp_correction.produced; 963 unsigned long long newlyFlushed = zfp.flushed - previous_zfp_correction.flushed; 964 previous_zfp_correction = zfp; 965 assert(inputPresented > 0); 966 DISPLAYLEVEL(6, "input blocked %u/%u(%.2f) - ingested:%u vs %u:consumed - flushed:%u vs %u:produced \n", 967 inputBlocked, inputPresented, (double)inputBlocked/inputPresented*100, 968 (U32)newlyIngested, (U32)newlyConsumed, 969 (U32)newlyFlushed, (U32)newlyProduced); 970 if ( (inputBlocked > inputPresented / 8) /* input is waiting often, because input buffers is full : compression or output too slow */ 971 && (newlyFlushed * 33 / 32 > newlyProduced) /* flush everything that is produced */ 972 && (newlyIngested * 33 / 32 > newlyConsumed) /* input speed as fast or faster than compression speed */ 973 ) { 974 DISPLAYLEVEL(6, "recommend faster as in(%llu) >= (%llu)comp(%llu) <= out(%llu) \n", 975 newlyIngested, newlyConsumed, newlyProduced, newlyFlushed); 976 speedChange = faster; 977 } 978 } 979 inputBlocked = 0; 980 inputPresented = 0; 981 } 982 983 if (speedChange == slower) { 984 DISPLAYLEVEL(6, "slower speed , higher compression \n") 985 compressionLevel ++; 986 if (compressionLevel > ZSTD_maxCLevel()) compressionLevel = ZSTD_maxCLevel(); 987 if (compressionLevel > g_maxAdaptLevel) compressionLevel = g_maxAdaptLevel; 988 compressionLevel += (compressionLevel == 0); /* skip 0 */ 989 ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_compressionLevel, (unsigned)compressionLevel); 990 } 991 if (speedChange == faster) { 992 DISPLAYLEVEL(6, "faster speed , lighter compression \n") 993 compressionLevel --; 994 if (compressionLevel < g_minAdaptLevel) compressionLevel = g_minAdaptLevel; 995 compressionLevel -= (compressionLevel == 0); /* skip 0 */ 996 ZSTD_CCtx_setParameter(ress.cctx, ZSTD_p_compressionLevel, (unsigned)compressionLevel); 997 } 998 speedChange = noChange; 999 1000 lastJobID = zfp.currentJobID; 1001 } /* if (zfp.currentJobID > lastJobID) */ 1002 } /* if (g_adaptiveMode) */ 1003 } /* if (READY_FOR_UPDATE()) */ 1004 } /* while ((inBuff.pos != inBuff.size) */ 1005 } while (directive != ZSTD_e_end); 1006 1007 if (ferror(srcFile)) { 1008 EXM_THROW(26, "Read error : I/O error"); 1009 } 1010 if (fileSize != UTIL_FILESIZE_UNKNOWN && *readsize != fileSize) { 1011 EXM_THROW(27, "Read error : Incomplete read : %llu / %llu B", 1012 (unsigned long long)*readsize, (unsigned long long)fileSize); 1013 } 1014 1015 return compressedfilesize; 1016 } 1017 1018 /*! FIO_compressFilename_internal() : 1019 * same as FIO_compressFilename_extRess(), with `ress.desFile` already opened. 1020 * @return : 0 : compression completed correctly, 1021 * 1 : missing or pb opening srcFileName 1022 */ 1023 static int 1024 FIO_compressFilename_internal(cRess_t ress, 1025 const char* dstFileName, const char* srcFileName, 1026 int compressionLevel) 1027 { 1028 U64 readsize = 0; 1029 U64 compressedfilesize = 0; 1030 U64 const fileSize = UTIL_getFileSize(srcFileName); 1031 DISPLAYLEVEL(5, "%s: %u bytes \n", srcFileName, (U32)fileSize); 1032 1033 /* compression format selection */ 1034 switch (g_compressionType) { 1035 default: 1036 case FIO_zstdCompression: 1037 compressedfilesize = FIO_compressZstdFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize); 1038 break; 1039 1040 case FIO_gzipCompression: 1041 #ifdef ZSTD_GZCOMPRESS 1042 compressedfilesize = FIO_compressGzFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize); 1043 #else 1044 (void)compressionLevel; 1045 EXM_THROW(20, "zstd: %s: file cannot be compressed as gzip (zstd compiled without ZSTD_GZCOMPRESS) -- ignored \n", 1046 srcFileName); 1047 #endif 1048 break; 1049 1050 case FIO_xzCompression: 1051 case FIO_lzmaCompression: 1052 #ifdef ZSTD_LZMACOMPRESS 1053 compressedfilesize = FIO_compressLzmaFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize, g_compressionType==FIO_lzmaCompression); 1054 #else 1055 (void)compressionLevel; 1056 EXM_THROW(20, "zstd: %s: file cannot be compressed as xz/lzma (zstd compiled without ZSTD_LZMACOMPRESS) -- ignored \n", 1057 srcFileName); 1058 #endif 1059 break; 1060 1061 case FIO_lz4Compression: 1062 #ifdef ZSTD_LZ4COMPRESS 1063 compressedfilesize = FIO_compressLz4Frame(&ress, srcFileName, fileSize, compressionLevel, &readsize); 1064 #else 1065 (void)compressionLevel; 1066 EXM_THROW(20, "zstd: %s: file cannot be compressed as lz4 (zstd compiled without ZSTD_LZ4COMPRESS) -- ignored \n", 1067 srcFileName); 1068 #endif 1069 break; 1070 } 1071 1072 /* Status */ 1073 DISPLAYLEVEL(2, "\r%79s\r", ""); 1074 DISPLAYLEVEL(2,"%-20s :%6.2f%% (%6llu => %6llu bytes, %s) \n", 1075 srcFileName, 1076 (double)compressedfilesize / (readsize+(!readsize)/*avoid div by zero*/) * 100, 1077 (unsigned long long)readsize, (unsigned long long) compressedfilesize, 1078 dstFileName); 1079 1080 return 0; 1081 } 1082 1083 1084 /*! FIO_compressFilename_dstFile() : 1085 * open dstFileName, or pass-through if ress.dstFile != NULL, 1086 * then start compression with FIO_compressFilename_internal(). 1087 * Manages source removal (--rm) and file permissions transfer. 1088 * note : ress.srcFile must be != NULL, 1089 * so reach this function through FIO_compressFilename_srcFile(). 1090 * @return : 0 : compression completed correctly, 1091 * 1 : pb 1092 */ 1093 static int FIO_compressFilename_dstFile(cRess_t ress, 1094 const char* dstFileName, 1095 const char* srcFileName, 1096 int compressionLevel) 1097 { 1098 int closeDstFile = 0; 1099 int result; 1100 stat_t statbuf; 1101 int transfer_permissions = 0; 1102 1103 assert(ress.srcFile != NULL); 1104 1105 if (ress.dstFile == NULL) { 1106 closeDstFile = 1; 1107 DISPLAYLEVEL(6, "FIO_compressFilename_dstFile: opening dst: %s", dstFileName); 1108 ress.dstFile = FIO_openDstFile(dstFileName); 1109 if (ress.dstFile==NULL) return 1; /* could not open dstFileName */ 1110 /* Must only be added after FIO_openDstFile() succeeds. 1111 * Otherwise we may delete the destination file if it already exists, 1112 * and the user presses Ctrl-C when asked if they wish to overwrite. 1113 */ 1114 addHandler(dstFileName); 1115 1116 if ( strcmp (srcFileName, stdinmark) 1117 && UTIL_getFileStat(srcFileName, &statbuf)) 1118 transfer_permissions = 1; 1119 } 1120 1121 result = FIO_compressFilename_internal(ress, dstFileName, srcFileName, compressionLevel); 1122 1123 if (closeDstFile) { 1124 FILE* const dstFile = ress.dstFile; 1125 ress.dstFile = NULL; 1126 1127 clearHandler(); 1128 1129 if (fclose(dstFile)) { /* error closing dstFile */ 1130 DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno)); 1131 result=1; 1132 } 1133 if ( (result != 0) /* operation failure */ 1134 && strcmp(dstFileName, nulmark) /* special case : don't remove() /dev/null */ 1135 && strcmp(dstFileName, stdoutmark) /* special case : don't remove() stdout */ 1136 ) { 1137 FIO_remove(dstFileName); /* remove compression artefact; note don't do anything special if remove() fails */ 1138 } else if ( strcmp(dstFileName, stdoutmark) 1139 && strcmp(dstFileName, nulmark) 1140 && transfer_permissions) { 1141 UTIL_setFileStat(dstFileName, &statbuf); 1142 } 1143 } 1144 1145 return result; 1146 } 1147 1148 1149 /*! FIO_compressFilename_srcFile() : 1150 * @return : 0 : compression completed correctly, 1151 * 1 : missing or pb opening srcFileName 1152 */ 1153 static int 1154 FIO_compressFilename_srcFile(cRess_t ress, 1155 const char* dstFileName, 1156 const char* srcFileName, 1157 int compressionLevel) 1158 { 1159 int result; 1160 1161 /* File check */ 1162 if (UTIL_isDirectory(srcFileName)) { 1163 DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName); 1164 return 1; 1165 } 1166 1167 ress.srcFile = FIO_openSrcFile(srcFileName); 1168 if (ress.srcFile == NULL) return 1; /* srcFile could not be opened */ 1169 1170 result = FIO_compressFilename_dstFile(ress, dstFileName, srcFileName, compressionLevel); 1171 1172 fclose(ress.srcFile); 1173 ress.srcFile = NULL; 1174 if ( g_removeSrcFile /* --rm */ 1175 && result == 0 /* success */ 1176 && strcmp(srcFileName, stdinmark) /* exception : don't erase stdin */ 1177 ) { 1178 /* We must clear the handler, since after this point calling it would 1179 * delete both the source and destination files. 1180 */ 1181 clearHandler(); 1182 if (FIO_remove(srcFileName)) 1183 EXM_THROW(1, "zstd: %s: %s", srcFileName, strerror(errno)); 1184 } 1185 return result; 1186 } 1187 1188 1189 int FIO_compressFilename(const char* dstFileName, const char* srcFileName, 1190 const char* dictFileName, int compressionLevel, 1191 ZSTD_compressionParameters comprParams) 1192 { 1193 clock_t const start = clock(); 1194 U64 const fileSize = UTIL_getFileSize(srcFileName); 1195 U64 const srcSize = (fileSize == UTIL_FILESIZE_UNKNOWN) ? ZSTD_CONTENTSIZE_UNKNOWN : fileSize; 1196 1197 cRess_t const ress = FIO_createCResources(dictFileName, compressionLevel, srcSize, comprParams); 1198 int const result = FIO_compressFilename_srcFile(ress, dstFileName, srcFileName, compressionLevel); 1199 1200 double const seconds = (double)(clock() - start) / CLOCKS_PER_SEC; 1201 DISPLAYLEVEL(4, "Completed in %.2f sec \n", seconds); 1202 1203 FIO_freeCResources(ress); 1204 return result; 1205 } 1206 1207 1208 /* FIO_determineCompressedName() : 1209 * create a destination filename for compressed srcFileName. 1210 * @return a pointer to it. 1211 * This function never returns an error (it may abort() in case of pb) 1212 */ 1213 static const char* 1214 FIO_determineCompressedName(const char* srcFileName, const char* suffix) 1215 { 1216 static size_t dfnbCapacity = 0; 1217 static char* dstFileNameBuffer = NULL; /* using static allocation : this function cannot be multi-threaded */ 1218 1219 size_t const sfnSize = strlen(srcFileName); 1220 size_t const suffixSize = strlen(suffix); 1221 1222 if (dfnbCapacity <= sfnSize+suffixSize+1) { 1223 /* resize buffer for dstName */ 1224 free(dstFileNameBuffer); 1225 dfnbCapacity = sfnSize + suffixSize + 30; 1226 dstFileNameBuffer = (char*)malloc(dfnbCapacity); 1227 if (!dstFileNameBuffer) { 1228 EXM_THROW(30, "zstd: %s", strerror(errno)); 1229 } } 1230 assert(dstFileNameBuffer != NULL); 1231 memcpy(dstFileNameBuffer, srcFileName, sfnSize); 1232 memcpy(dstFileNameBuffer+sfnSize, suffix, suffixSize+1 /* Include terminating null */); 1233 1234 return dstFileNameBuffer; 1235 } 1236 1237 1238 /* FIO_compressMultipleFilenames() : 1239 * compress nbFiles files 1240 * into one destination (outFileName) 1241 * or into one file each (outFileName == NULL, but suffix != NULL). 1242 */ 1243 int FIO_compressMultipleFilenames(const char** inFileNamesTable, unsigned nbFiles, 1244 const char* outFileName, const char* suffix, 1245 const char* dictFileName, int compressionLevel, 1246 ZSTD_compressionParameters comprParams) 1247 { 1248 int error = 0; 1249 U64 const firstFileSize = UTIL_getFileSize(inFileNamesTable[0]); 1250 U64 const firstSrcSize = (firstFileSize == UTIL_FILESIZE_UNKNOWN) ? ZSTD_CONTENTSIZE_UNKNOWN : firstFileSize; 1251 U64 const srcSize = (nbFiles != 1) ? ZSTD_CONTENTSIZE_UNKNOWN : firstSrcSize ; 1252 cRess_t ress = FIO_createCResources(dictFileName, compressionLevel, srcSize, comprParams); 1253 1254 /* init */ 1255 assert(outFileName != NULL || suffix != NULL); 1256 1257 if (outFileName != NULL) { /* output into a single destination (stdout typically) */ 1258 ress.dstFile = FIO_openDstFile(outFileName); 1259 if (ress.dstFile == NULL) { /* could not open outFileName */ 1260 error = 1; 1261 } else { 1262 unsigned u; 1263 for (u=0; u<nbFiles; u++) 1264 error |= FIO_compressFilename_srcFile(ress, outFileName, inFileNamesTable[u], compressionLevel); 1265 if (fclose(ress.dstFile)) 1266 EXM_THROW(29, "Write error : cannot properly close %s", outFileName); 1267 ress.dstFile = NULL; 1268 } 1269 } else { 1270 unsigned u; 1271 for (u=0; u<nbFiles; u++) { 1272 const char* const srcFileName = inFileNamesTable[u]; 1273 const char* const dstFileName = FIO_determineCompressedName(srcFileName, suffix); /* cannot fail */ 1274 error |= FIO_compressFilename_srcFile(ress, dstFileName, srcFileName, compressionLevel); 1275 } } 1276 1277 FIO_freeCResources(ress); 1278 return error; 1279 } 1280 1281 #endif /* #ifndef ZSTD_NOCOMPRESS */ 1282 1283 1284 1285 #ifndef ZSTD_NODECOMPRESS 1286 1287 /* ************************************************************************** 1288 * Decompression 1289 ***************************************************************************/ 1290 typedef struct { 1291 void* srcBuffer; 1292 size_t srcBufferSize; 1293 size_t srcBufferLoaded; 1294 void* dstBuffer; 1295 size_t dstBufferSize; 1296 ZSTD_DStream* dctx; 1297 FILE* dstFile; 1298 } dRess_t; 1299 1300 static dRess_t FIO_createDResources(const char* dictFileName) 1301 { 1302 dRess_t ress; 1303 memset(&ress, 0, sizeof(ress)); 1304 1305 /* Allocation */ 1306 ress.dctx = ZSTD_createDStream(); 1307 if (ress.dctx==NULL) EXM_THROW(60, "Can't create ZSTD_DStream"); 1308 CHECK( ZSTD_setDStreamParameter(ress.dctx, DStream_p_maxWindowSize, g_memLimit) ); 1309 ress.srcBufferSize = ZSTD_DStreamInSize(); 1310 ress.srcBuffer = malloc(ress.srcBufferSize); 1311 ress.dstBufferSize = ZSTD_DStreamOutSize(); 1312 ress.dstBuffer = malloc(ress.dstBufferSize); 1313 if (!ress.srcBuffer || !ress.dstBuffer) 1314 EXM_THROW(61, "Allocation error : not enough memory"); 1315 1316 /* dictionary */ 1317 { void* dictBuffer; 1318 size_t const dictBufferSize = FIO_createDictBuffer(&dictBuffer, dictFileName); 1319 CHECK( ZSTD_initDStream_usingDict(ress.dctx, dictBuffer, dictBufferSize) ); 1320 free(dictBuffer); 1321 } 1322 1323 return ress; 1324 } 1325 1326 static void FIO_freeDResources(dRess_t ress) 1327 { 1328 CHECK( ZSTD_freeDStream(ress.dctx) ); 1329 free(ress.srcBuffer); 1330 free(ress.dstBuffer); 1331 } 1332 1333 1334 /** FIO_fwriteSparse() : 1335 * @return : storedSkips, to be provided to next call to FIO_fwriteSparse() of LZ4IO_fwriteSparseEnd() */ 1336 static unsigned FIO_fwriteSparse(FILE* file, const void* buffer, size_t bufferSize, unsigned storedSkips) 1337 { 1338 const size_t* const bufferT = (const size_t*)buffer; /* Buffer is supposed malloc'ed, hence aligned on size_t */ 1339 size_t bufferSizeT = bufferSize / sizeof(size_t); 1340 const size_t* const bufferTEnd = bufferT + bufferSizeT; 1341 const size_t* ptrT = bufferT; 1342 static const size_t segmentSizeT = (32 KB) / sizeof(size_t); /* 0-test re-attempted every 32 KB */ 1343 1344 if (!g_sparseFileSupport) { /* normal write */ 1345 size_t const sizeCheck = fwrite(buffer, 1, bufferSize, file); 1346 if (sizeCheck != bufferSize) EXM_THROW(70, "Write error : cannot write decoded block"); 1347 return 0; 1348 } 1349 1350 /* avoid int overflow */ 1351 if (storedSkips > 1 GB) { 1352 int const seekResult = LONG_SEEK(file, 1 GB, SEEK_CUR); 1353 if (seekResult != 0) EXM_THROW(71, "1 GB skip error (sparse file support)"); 1354 storedSkips -= 1 GB; 1355 } 1356 1357 while (ptrT < bufferTEnd) { 1358 size_t seg0SizeT = segmentSizeT; 1359 size_t nb0T; 1360 1361 /* count leading zeros */ 1362 if (seg0SizeT > bufferSizeT) seg0SizeT = bufferSizeT; 1363 bufferSizeT -= seg0SizeT; 1364 for (nb0T=0; (nb0T < seg0SizeT) && (ptrT[nb0T] == 0); nb0T++) ; 1365 storedSkips += (unsigned)(nb0T * sizeof(size_t)); 1366 1367 if (nb0T != seg0SizeT) { /* not all 0s */ 1368 int const seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); 1369 if (seekResult) EXM_THROW(72, "Sparse skip error ; try --no-sparse"); 1370 storedSkips = 0; 1371 seg0SizeT -= nb0T; 1372 ptrT += nb0T; 1373 { size_t const sizeCheck = fwrite(ptrT, sizeof(size_t), seg0SizeT, file); 1374 if (sizeCheck != seg0SizeT) 1375 EXM_THROW(73, "Write error : cannot write decoded block"); 1376 } } 1377 ptrT += seg0SizeT; 1378 } 1379 1380 { static size_t const maskT = sizeof(size_t)-1; 1381 if (bufferSize & maskT) { 1382 /* size not multiple of sizeof(size_t) : implies end of block */ 1383 const char* const restStart = (const char*)bufferTEnd; 1384 const char* restPtr = restStart; 1385 size_t restSize = bufferSize & maskT; 1386 const char* const restEnd = restStart + restSize; 1387 for ( ; (restPtr < restEnd) && (*restPtr == 0); restPtr++) ; 1388 storedSkips += (unsigned) (restPtr - restStart); 1389 if (restPtr != restEnd) { 1390 int seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); 1391 if (seekResult) 1392 EXM_THROW(74, "Sparse skip error ; try --no-sparse"); 1393 storedSkips = 0; 1394 { size_t const sizeCheck = fwrite(restPtr, 1, restEnd - restPtr, file); 1395 if (sizeCheck != (size_t)(restEnd - restPtr)) 1396 EXM_THROW(75, "Write error : cannot write decoded end of block"); 1397 } } } } 1398 1399 return storedSkips; 1400 } 1401 1402 static void FIO_fwriteSparseEnd(FILE* file, unsigned storedSkips) 1403 { 1404 if (storedSkips-->0) { /* implies g_sparseFileSupport>0 */ 1405 int const seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); 1406 if (seekResult != 0) EXM_THROW(69, "Final skip error (sparse file)"); 1407 { const char lastZeroByte[1] = { 0 }; 1408 size_t const sizeCheck = fwrite(lastZeroByte, 1, 1, file); 1409 if (sizeCheck != 1) 1410 EXM_THROW(69, "Write error : cannot write last zero"); 1411 } } 1412 } 1413 1414 1415 /** FIO_passThrough() : just copy input into output, for compatibility with gzip -df mode 1416 @return : 0 (no error) */ 1417 static unsigned FIO_passThrough(FILE* foutput, FILE* finput, void* buffer, size_t bufferSize, size_t alreadyLoaded) 1418 { 1419 size_t const blockSize = MIN(64 KB, bufferSize); 1420 size_t readFromInput = 1; 1421 unsigned storedSkips = 0; 1422 1423 /* assumption : ress->srcBufferLoaded bytes already loaded and stored within buffer */ 1424 { size_t const sizeCheck = fwrite(buffer, 1, alreadyLoaded, foutput); 1425 if (sizeCheck != alreadyLoaded) { 1426 DISPLAYLEVEL(1, "Pass-through write error \n"); 1427 return 1; 1428 } } 1429 1430 while (readFromInput) { 1431 readFromInput = fread(buffer, 1, blockSize, finput); 1432 storedSkips = FIO_fwriteSparse(foutput, buffer, readFromInput, storedSkips); 1433 } 1434 1435 FIO_fwriteSparseEnd(foutput, storedSkips); 1436 return 0; 1437 } 1438 1439 /* FIO_highbit64() : 1440 * gives position of highest bit. 1441 * note : only works for v > 0 ! 1442 */ 1443 static unsigned FIO_highbit64(unsigned long long v) 1444 { 1445 unsigned count = 0; 1446 assert(v != 0); 1447 v >>= 1; 1448 while (v) { v >>= 1; count++; } 1449 return count; 1450 } 1451 1452 /* FIO_zstdErrorHelp() : 1453 * detailed error message when requested window size is too large */ 1454 static void FIO_zstdErrorHelp(dRess_t* ress, size_t err, char const* srcFileName) 1455 { 1456 ZSTD_frameHeader header; 1457 1458 /* Help message only for one specific error */ 1459 if (ZSTD_getErrorCode(err) != ZSTD_error_frameParameter_windowTooLarge) 1460 return; 1461 1462 /* Try to decode the frame header */ 1463 err = ZSTD_getFrameHeader(&header, ress->srcBuffer, ress->srcBufferLoaded); 1464 if (err == 0) { 1465 unsigned long long const windowSize = header.windowSize; 1466 U32 const windowLog = FIO_highbit64(windowSize) + ((windowSize & (windowSize - 1)) != 0); 1467 assert(g_memLimit > 0); 1468 DISPLAYLEVEL(1, "%s : Window size larger than maximum : %llu > %u\n", 1469 srcFileName, windowSize, g_memLimit); 1470 if (windowLog <= ZSTD_WINDOWLOG_MAX) { 1471 U32 const windowMB = (U32)((windowSize >> 20) + ((windowSize & ((1 MB) - 1)) != 0)); 1472 assert(windowSize < (U64)(1ULL << 52)); /* ensure now overflow for windowMB */ 1473 DISPLAYLEVEL(1, "%s : Use --long=%u or --memory=%uMB\n", 1474 srcFileName, windowLog, windowMB); 1475 return; 1476 } 1477 } 1478 DISPLAYLEVEL(1, "%s : Window log larger than ZSTD_WINDOWLOG_MAX=%u; not supported\n", 1479 srcFileName, ZSTD_WINDOWLOG_MAX); 1480 } 1481 1482 /** FIO_decompressFrame() : 1483 * @return : size of decoded zstd frame, or an error code 1484 */ 1485 #define FIO_ERROR_FRAME_DECODING ((unsigned long long)(-2)) 1486 static unsigned long long FIO_decompressZstdFrame(dRess_t* ress, 1487 FILE* finput, 1488 const char* srcFileName, 1489 U64 alreadyDecoded) 1490 { 1491 U64 frameSize = 0; 1492 U32 storedSkips = 0; 1493 1494 size_t const srcFileLength = strlen(srcFileName); 1495 if (srcFileLength>20) srcFileName += srcFileLength-20; /* display last 20 characters only */ 1496 1497 ZSTD_resetDStream(ress->dctx); 1498 1499 /* Header loading : ensures ZSTD_getFrameHeader() will succeed */ 1500 { size_t const toDecode = ZSTD_FRAMEHEADERSIZE_MAX; 1501 if (ress->srcBufferLoaded < toDecode) { 1502 size_t const toRead = toDecode - ress->srcBufferLoaded; 1503 void* const startPosition = (char*)ress->srcBuffer + ress->srcBufferLoaded; 1504 ress->srcBufferLoaded += fread(startPosition, 1, toRead, finput); 1505 } } 1506 1507 /* Main decompression Loop */ 1508 while (1) { 1509 ZSTD_inBuffer inBuff = { ress->srcBuffer, ress->srcBufferLoaded, 0 }; 1510 ZSTD_outBuffer outBuff= { ress->dstBuffer, ress->dstBufferSize, 0 }; 1511 size_t const readSizeHint = ZSTD_decompressStream(ress->dctx, &outBuff, &inBuff); 1512 if (ZSTD_isError(readSizeHint)) { 1513 DISPLAYLEVEL(1, "%s : Decoding error (36) : %s \n", 1514 srcFileName, ZSTD_getErrorName(readSizeHint)); 1515 FIO_zstdErrorHelp(ress, readSizeHint, srcFileName); 1516 return FIO_ERROR_FRAME_DECODING; 1517 } 1518 1519 /* Write block */ 1520 storedSkips = FIO_fwriteSparse(ress->dstFile, ress->dstBuffer, outBuff.pos, storedSkips); 1521 frameSize += outBuff.pos; 1522 DISPLAYUPDATE(2, "\r%-20.20s : %u MB... ", 1523 srcFileName, (U32)((alreadyDecoded+frameSize)>>20) ); 1524 1525 if (inBuff.pos > 0) { 1526 memmove(ress->srcBuffer, (char*)ress->srcBuffer + inBuff.pos, inBuff.size - inBuff.pos); 1527 ress->srcBufferLoaded -= inBuff.pos; 1528 } 1529 1530 if (readSizeHint == 0) break; /* end of frame */ 1531 if (inBuff.size != inBuff.pos) { 1532 DISPLAYLEVEL(1, "%s : Decoding error (37) : should consume entire input \n", 1533 srcFileName); 1534 return FIO_ERROR_FRAME_DECODING; 1535 } 1536 1537 /* Fill input buffer */ 1538 { size_t const toDecode = MIN(readSizeHint, ress->srcBufferSize); /* support large skippable frames */ 1539 if (ress->srcBufferLoaded < toDecode) { 1540 size_t const toRead = toDecode - ress->srcBufferLoaded; /* > 0 */ 1541 void* const startPosition = (char*)ress->srcBuffer + ress->srcBufferLoaded; 1542 size_t const readSize = fread(startPosition, 1, toRead, finput); 1543 if (readSize==0) { 1544 DISPLAYLEVEL(1, "%s : Read error (39) : premature end \n", 1545 srcFileName); 1546 return FIO_ERROR_FRAME_DECODING; 1547 } 1548 ress->srcBufferLoaded += readSize; 1549 } } } 1550 1551 FIO_fwriteSparseEnd(ress->dstFile, storedSkips); 1552 1553 return frameSize; 1554 } 1555 1556 1557 #ifdef ZSTD_GZDECOMPRESS 1558 static unsigned long long FIO_decompressGzFrame(dRess_t* ress, 1559 FILE* srcFile, const char* srcFileName) 1560 { 1561 unsigned long long outFileSize = 0; 1562 z_stream strm; 1563 int flush = Z_NO_FLUSH; 1564 int decodingError = 0; 1565 1566 strm.zalloc = Z_NULL; 1567 strm.zfree = Z_NULL; 1568 strm.opaque = Z_NULL; 1569 strm.next_in = 0; 1570 strm.avail_in = 0; 1571 /* see http://www.zlib.net/manual.html */ 1572 if (inflateInit2(&strm, 15 /* maxWindowLogSize */ + 16 /* gzip only */) != Z_OK) 1573 return FIO_ERROR_FRAME_DECODING; 1574 1575 strm.next_out = (Bytef*)ress->dstBuffer; 1576 strm.avail_out = (uInt)ress->dstBufferSize; 1577 strm.avail_in = (uInt)ress->srcBufferLoaded; 1578 strm.next_in = (z_const unsigned char*)ress->srcBuffer; 1579 1580 for ( ; ; ) { 1581 int ret; 1582 if (strm.avail_in == 0) { 1583 ress->srcBufferLoaded = fread(ress->srcBuffer, 1, ress->srcBufferSize, srcFile); 1584 if (ress->srcBufferLoaded == 0) flush = Z_FINISH; 1585 strm.next_in = (z_const unsigned char*)ress->srcBuffer; 1586 strm.avail_in = (uInt)ress->srcBufferLoaded; 1587 } 1588 ret = inflate(&strm, flush); 1589 if (ret == Z_BUF_ERROR) { 1590 DISPLAYLEVEL(1, "zstd: %s: premature gz end \n", srcFileName); 1591 decodingError = 1; break; 1592 } 1593 if (ret != Z_OK && ret != Z_STREAM_END) { 1594 DISPLAYLEVEL(1, "zstd: %s: inflate error %d \n", srcFileName, ret); 1595 decodingError = 1; break; 1596 } 1597 { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; 1598 if (decompBytes) { 1599 if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) { 1600 DISPLAYLEVEL(1, "zstd: %s \n", strerror(errno)); 1601 decodingError = 1; break; 1602 } 1603 outFileSize += decompBytes; 1604 strm.next_out = (Bytef*)ress->dstBuffer; 1605 strm.avail_out = (uInt)ress->dstBufferSize; 1606 } 1607 } 1608 if (ret == Z_STREAM_END) break; 1609 } 1610 1611 if (strm.avail_in > 0) 1612 memmove(ress->srcBuffer, strm.next_in, strm.avail_in); 1613 ress->srcBufferLoaded = strm.avail_in; 1614 if ( (inflateEnd(&strm) != Z_OK) /* release resources ; error detected */ 1615 && (decodingError==0) ) { 1616 DISPLAYLEVEL(1, "zstd: %s: inflateEnd error \n", srcFileName); 1617 decodingError = 1; 1618 } 1619 return decodingError ? FIO_ERROR_FRAME_DECODING : outFileSize; 1620 } 1621 #endif 1622 1623 1624 #ifdef ZSTD_LZMADECOMPRESS 1625 static unsigned long long FIO_decompressLzmaFrame(dRess_t* ress, FILE* srcFile, const char* srcFileName, int plain_lzma) 1626 { 1627 unsigned long long outFileSize = 0; 1628 lzma_stream strm = LZMA_STREAM_INIT; 1629 lzma_action action = LZMA_RUN; 1630 lzma_ret initRet; 1631 int decodingError = 0; 1632 1633 strm.next_in = 0; 1634 strm.avail_in = 0; 1635 if (plain_lzma) { 1636 initRet = lzma_alone_decoder(&strm, UINT64_MAX); /* LZMA */ 1637 } else { 1638 initRet = lzma_stream_decoder(&strm, UINT64_MAX, 0); /* XZ */ 1639 } 1640 1641 if (initRet != LZMA_OK) { 1642 DISPLAYLEVEL(1, "zstd: %s: %s error %d \n", 1643 plain_lzma ? "lzma_alone_decoder" : "lzma_stream_decoder", 1644 srcFileName, initRet); 1645 return FIO_ERROR_FRAME_DECODING; 1646 } 1647 1648 strm.next_out = (BYTE*)ress->dstBuffer; 1649 strm.avail_out = ress->dstBufferSize; 1650 strm.next_in = (BYTE const*)ress->srcBuffer; 1651 strm.avail_in = ress->srcBufferLoaded; 1652 1653 for ( ; ; ) { 1654 lzma_ret ret; 1655 if (strm.avail_in == 0) { 1656 ress->srcBufferLoaded = fread(ress->srcBuffer, 1, ress->srcBufferSize, srcFile); 1657 if (ress->srcBufferLoaded == 0) action = LZMA_FINISH; 1658 strm.next_in = (BYTE const*)ress->srcBuffer; 1659 strm.avail_in = ress->srcBufferLoaded; 1660 } 1661 ret = lzma_code(&strm, action); 1662 1663 if (ret == LZMA_BUF_ERROR) { 1664 DISPLAYLEVEL(1, "zstd: %s: premature lzma end \n", srcFileName); 1665 decodingError = 1; break; 1666 } 1667 if (ret != LZMA_OK && ret != LZMA_STREAM_END) { 1668 DISPLAYLEVEL(1, "zstd: %s: lzma_code decoding error %d \n", 1669 srcFileName, ret); 1670 decodingError = 1; break; 1671 } 1672 { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; 1673 if (decompBytes) { 1674 if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) { 1675 DISPLAYLEVEL(1, "zstd: %s \n", strerror(errno)); 1676 decodingError = 1; break; 1677 } 1678 outFileSize += decompBytes; 1679 strm.next_out = (BYTE*)ress->dstBuffer; 1680 strm.avail_out = ress->dstBufferSize; 1681 } } 1682 if (ret == LZMA_STREAM_END) break; 1683 } 1684 1685 if (strm.avail_in > 0) 1686 memmove(ress->srcBuffer, strm.next_in, strm.avail_in); 1687 ress->srcBufferLoaded = strm.avail_in; 1688 lzma_end(&strm); 1689 return decodingError ? FIO_ERROR_FRAME_DECODING : outFileSize; 1690 } 1691 #endif 1692 1693 #ifdef ZSTD_LZ4DECOMPRESS 1694 static unsigned long long FIO_decompressLz4Frame(dRess_t* ress, 1695 FILE* srcFile, const char* srcFileName) 1696 { 1697 unsigned long long filesize = 0; 1698 LZ4F_errorCode_t nextToLoad; 1699 LZ4F_decompressionContext_t dCtx; 1700 LZ4F_errorCode_t const errorCode = LZ4F_createDecompressionContext(&dCtx, LZ4F_VERSION); 1701 int decodingError = 0; 1702 1703 if (LZ4F_isError(errorCode)) { 1704 DISPLAYLEVEL(1, "zstd: failed to create lz4 decompression context \n"); 1705 return FIO_ERROR_FRAME_DECODING; 1706 } 1707 1708 /* Init feed with magic number (already consumed from FILE* sFile) */ 1709 { size_t inSize = 4; 1710 size_t outSize= 0; 1711 MEM_writeLE32(ress->srcBuffer, LZ4_MAGICNUMBER); 1712 nextToLoad = LZ4F_decompress(dCtx, ress->dstBuffer, &outSize, ress->srcBuffer, &inSize, NULL); 1713 if (LZ4F_isError(nextToLoad)) { 1714 DISPLAYLEVEL(1, "zstd: %s: lz4 header error : %s \n", 1715 srcFileName, LZ4F_getErrorName(nextToLoad)); 1716 LZ4F_freeDecompressionContext(dCtx); 1717 return FIO_ERROR_FRAME_DECODING; 1718 } } 1719 1720 /* Main Loop */ 1721 for (;nextToLoad;) { 1722 size_t readSize; 1723 size_t pos = 0; 1724 size_t decodedBytes = ress->dstBufferSize; 1725 1726 /* Read input */ 1727 if (nextToLoad > ress->srcBufferSize) nextToLoad = ress->srcBufferSize; 1728 readSize = fread(ress->srcBuffer, 1, nextToLoad, srcFile); 1729 if (!readSize) break; /* reached end of file or stream */ 1730 1731 while ((pos < readSize) || (decodedBytes == ress->dstBufferSize)) { /* still to read, or still to flush */ 1732 /* Decode Input (at least partially) */ 1733 size_t remaining = readSize - pos; 1734 decodedBytes = ress->dstBufferSize; 1735 nextToLoad = LZ4F_decompress(dCtx, ress->dstBuffer, &decodedBytes, (char*)(ress->srcBuffer)+pos, &remaining, NULL); 1736 if (LZ4F_isError(nextToLoad)) { 1737 DISPLAYLEVEL(1, "zstd: %s: lz4 decompression error : %s \n", 1738 srcFileName, LZ4F_getErrorName(nextToLoad)); 1739 decodingError = 1; nextToLoad = 0; break; 1740 } 1741 pos += remaining; 1742 1743 /* Write Block */ 1744 if (decodedBytes) { 1745 if (fwrite(ress->dstBuffer, 1, decodedBytes, ress->dstFile) != decodedBytes) { 1746 DISPLAYLEVEL(1, "zstd: %s \n", strerror(errno)); 1747 decodingError = 1; nextToLoad = 0; break; 1748 } 1749 filesize += decodedBytes; 1750 DISPLAYUPDATE(2, "\rDecompressed : %u MB ", (unsigned)(filesize>>20)); 1751 } 1752 1753 if (!nextToLoad) break; 1754 } 1755 } 1756 /* can be out because readSize == 0, which could be an fread() error */ 1757 if (ferror(srcFile)) { 1758 DISPLAYLEVEL(1, "zstd: %s: read error \n", srcFileName); 1759 decodingError=1; 1760 } 1761 1762 if (nextToLoad!=0) { 1763 DISPLAYLEVEL(1, "zstd: %s: unfinished lz4 stream \n", srcFileName); 1764 decodingError=1; 1765 } 1766 1767 LZ4F_freeDecompressionContext(dCtx); 1768 ress->srcBufferLoaded = 0; /* LZ4F will reach exact frame boundary */ 1769 1770 return decodingError ? FIO_ERROR_FRAME_DECODING : filesize; 1771 } 1772 #endif 1773 1774 1775 1776 /** FIO_decompressFrames() : 1777 * Find and decode frames inside srcFile 1778 * srcFile presumed opened and valid 1779 * @return : 0 : OK 1780 * 1 : error 1781 */ 1782 static int FIO_decompressFrames(dRess_t ress, FILE* srcFile, 1783 const char* dstFileName, const char* srcFileName) 1784 { 1785 unsigned readSomething = 0; 1786 unsigned long long filesize = 0; 1787 assert(srcFile != NULL); 1788 1789 /* for each frame */ 1790 for ( ; ; ) { 1791 /* check magic number -> version */ 1792 size_t const toRead = 4; 1793 const BYTE* const buf = (const BYTE*)ress.srcBuffer; 1794 if (ress.srcBufferLoaded < toRead) /* load up to 4 bytes for header */ 1795 ress.srcBufferLoaded += fread((char*)ress.srcBuffer + ress.srcBufferLoaded, 1796 (size_t)1, toRead - ress.srcBufferLoaded, srcFile); 1797 if (ress.srcBufferLoaded==0) { 1798 if (readSomething==0) { /* srcFile is empty (which is invalid) */ 1799 DISPLAYLEVEL(1, "zstd: %s: unexpected end of file \n", srcFileName); 1800 return 1; 1801 } /* else, just reached frame boundary */ 1802 break; /* no more input */ 1803 } 1804 readSomething = 1; /* there is at least 1 byte in srcFile */ 1805 if (ress.srcBufferLoaded < toRead) { 1806 DISPLAYLEVEL(1, "zstd: %s: unknown header \n", srcFileName); 1807 return 1; 1808 } 1809 if (ZSTD_isFrame(buf, ress.srcBufferLoaded)) { 1810 unsigned long long const frameSize = FIO_decompressZstdFrame(&ress, srcFile, srcFileName, filesize); 1811 if (frameSize == FIO_ERROR_FRAME_DECODING) return 1; 1812 filesize += frameSize; 1813 } else if (buf[0] == 31 && buf[1] == 139) { /* gz magic number */ 1814 #ifdef ZSTD_GZDECOMPRESS 1815 unsigned long long const frameSize = FIO_decompressGzFrame(&ress, srcFile, srcFileName); 1816 if (frameSize == FIO_ERROR_FRAME_DECODING) return 1; 1817 filesize += frameSize; 1818 #else 1819 DISPLAYLEVEL(1, "zstd: %s: gzip file cannot be uncompressed (zstd compiled without HAVE_ZLIB) -- ignored \n", srcFileName); 1820 return 1; 1821 #endif 1822 } else if ((buf[0] == 0xFD && buf[1] == 0x37) /* xz magic number */ 1823 || (buf[0] == 0x5D && buf[1] == 0x00)) { /* lzma header (no magic number) */ 1824 #ifdef ZSTD_LZMADECOMPRESS 1825 unsigned long long const frameSize = FIO_decompressLzmaFrame(&ress, srcFile, srcFileName, buf[0] != 0xFD); 1826 if (frameSize == FIO_ERROR_FRAME_DECODING) return 1; 1827 filesize += frameSize; 1828 #else 1829 DISPLAYLEVEL(1, "zstd: %s: xz/lzma file cannot be uncompressed (zstd compiled without HAVE_LZMA) -- ignored \n", srcFileName); 1830 return 1; 1831 #endif 1832 } else if (MEM_readLE32(buf) == LZ4_MAGICNUMBER) { 1833 #ifdef ZSTD_LZ4DECOMPRESS 1834 unsigned long long const frameSize = FIO_decompressLz4Frame(&ress, srcFile, srcFileName); 1835 if (frameSize == FIO_ERROR_FRAME_DECODING) return 1; 1836 filesize += frameSize; 1837 #else 1838 DISPLAYLEVEL(1, "zstd: %s: lz4 file cannot be uncompressed (zstd compiled without HAVE_LZ4) -- ignored \n", srcFileName); 1839 return 1; 1840 #endif 1841 } else if ((g_overwrite) && !strcmp (dstFileName, stdoutmark)) { /* pass-through mode */ 1842 return FIO_passThrough(ress.dstFile, srcFile, 1843 ress.srcBuffer, ress.srcBufferSize, ress.srcBufferLoaded); 1844 } else { 1845 DISPLAYLEVEL(1, "zstd: %s: unsupported format \n", srcFileName); 1846 return 1; 1847 } } /* for each frame */ 1848 1849 /* Final Status */ 1850 DISPLAYLEVEL(2, "\r%79s\r", ""); 1851 DISPLAYLEVEL(2, "%-20s: %llu bytes \n", srcFileName, filesize); 1852 1853 return 0; 1854 } 1855 1856 /** FIO_decompressDstFile() : 1857 open `dstFileName`, 1858 or path-through if ress.dstFile is already != 0, 1859 then start decompression process (FIO_decompressFrames()). 1860 @return : 0 : OK 1861 1 : operation aborted 1862 */ 1863 static int FIO_decompressDstFile(dRess_t ress, FILE* srcFile, 1864 const char* dstFileName, const char* srcFileName) 1865 { 1866 int result; 1867 stat_t statbuf; 1868 int transfer_permissions = 0; 1869 int releaseDstFile = 0; 1870 1871 if (ress.dstFile == NULL) { 1872 releaseDstFile = 1; 1873 1874 ress.dstFile = FIO_openDstFile(dstFileName); 1875 if (ress.dstFile==0) return 1; 1876 1877 /* Must only be added after FIO_openDstFile() succeeds. 1878 * Otherwise we may delete the destination file if it already exists, 1879 * and the user presses Ctrl-C when asked if they wish to overwrite. 1880 */ 1881 addHandler(dstFileName); 1882 1883 if ( strcmp(srcFileName, stdinmark) /* special case : don't transfer permissions from stdin */ 1884 && UTIL_getFileStat(srcFileName, &statbuf) ) 1885 transfer_permissions = 1; 1886 } 1887 1888 1889 result = FIO_decompressFrames(ress, srcFile, dstFileName, srcFileName); 1890 1891 if (releaseDstFile) { 1892 FILE* const dstFile = ress.dstFile; 1893 clearHandler(); 1894 ress.dstFile = NULL; 1895 if (fclose(dstFile)) { 1896 DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno)); 1897 result = 1; 1898 } 1899 1900 if ( (result != 0) /* operation failure */ 1901 && strcmp(dstFileName, nulmark) /* special case : don't remove() /dev/null (#316) */ 1902 && strcmp(dstFileName, stdoutmark) /* special case : don't remove() stdout */ 1903 ) { 1904 FIO_remove(dstFileName); /* remove decompression artefact; note: don't do anything special if remove() fails */ 1905 } else { /* operation success */ 1906 if ( strcmp(dstFileName, stdoutmark) /* special case : don't chmod stdout */ 1907 && strcmp(dstFileName, nulmark) /* special case : don't chmod /dev/null */ 1908 && transfer_permissions ) /* file permissions correctly extracted from src */ 1909 UTIL_setFileStat(dstFileName, &statbuf); /* transfer file permissions from src into dst */ 1910 } 1911 } 1912 1913 return result; 1914 } 1915 1916 1917 /** FIO_decompressSrcFile() : 1918 Open `srcFileName`, transfer control to decompressDstFile() 1919 @return : 0 : OK 1920 1 : error 1921 */ 1922 static int FIO_decompressSrcFile(dRess_t ress, const char* dstFileName, const char* srcFileName) 1923 { 1924 FILE* srcFile; 1925 int result; 1926 1927 if (UTIL_isDirectory(srcFileName)) { 1928 DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName); 1929 return 1; 1930 } 1931 1932 srcFile = FIO_openSrcFile(srcFileName); 1933 if (srcFile==NULL) return 1; 1934 ress.srcBufferLoaded = 0; 1935 1936 result = FIO_decompressDstFile(ress, srcFile, dstFileName, srcFileName); 1937 1938 /* Close file */ 1939 if (fclose(srcFile)) { 1940 DISPLAYLEVEL(1, "zstd: %s: %s \n", srcFileName, strerror(errno)); /* error should not happen */ 1941 return 1; 1942 } 1943 if ( g_removeSrcFile /* --rm */ 1944 && (result==0) /* decompression successful */ 1945 && strcmp(srcFileName, stdinmark) ) /* not stdin */ { 1946 /* We must clear the handler, since after this point calling it would 1947 * delete both the source and destination files. 1948 */ 1949 clearHandler(); 1950 if (FIO_remove(srcFileName)) { 1951 /* failed to remove src file */ 1952 DISPLAYLEVEL(1, "zstd: %s: %s \n", srcFileName, strerror(errno)); 1953 return 1; 1954 } } 1955 return result; 1956 } 1957 1958 1959 1960 int FIO_decompressFilename(const char* dstFileName, const char* srcFileName, 1961 const char* dictFileName) 1962 { 1963 dRess_t const ress = FIO_createDResources(dictFileName); 1964 1965 int const decodingError = FIO_decompressSrcFile(ress, dstFileName, srcFileName); 1966 1967 FIO_freeDResources(ress); 1968 return decodingError; 1969 } 1970 1971 1972 /* FIO_determineDstName() : 1973 * create a destination filename from a srcFileName. 1974 * @return a pointer to it. 1975 * @return == NULL if there is an error */ 1976 static const char* 1977 FIO_determineDstName(const char* srcFileName) 1978 { 1979 static size_t dfnbCapacity = 0; 1980 static char* dstFileNameBuffer = NULL; /* using static allocation : this function cannot be multi-threaded */ 1981 1982 size_t const sfnSize = strlen(srcFileName); 1983 size_t suffixSize; 1984 const char* const suffixPtr = strrchr(srcFileName, '.'); 1985 if (suffixPtr == NULL) { 1986 DISPLAYLEVEL(1, "zstd: %s: unknown suffix -- ignored \n", 1987 srcFileName); 1988 return NULL; 1989 } 1990 suffixSize = strlen(suffixPtr); 1991 1992 /* check suffix is authorized */ 1993 if (sfnSize <= suffixSize 1994 || ( strcmp(suffixPtr, ZSTD_EXTENSION) 1995 #ifdef ZSTD_GZDECOMPRESS 1996 && strcmp(suffixPtr, GZ_EXTENSION) 1997 #endif 1998 #ifdef ZSTD_LZMADECOMPRESS 1999 && strcmp(suffixPtr, XZ_EXTENSION) 2000 && strcmp(suffixPtr, LZMA_EXTENSION) 2001 #endif 2002 #ifdef ZSTD_LZ4DECOMPRESS 2003 && strcmp(suffixPtr, LZ4_EXTENSION) 2004 #endif 2005 ) ) { 2006 const char* suffixlist = ZSTD_EXTENSION 2007 #ifdef ZSTD_GZDECOMPRESS 2008 "/" GZ_EXTENSION 2009 #endif 2010 #ifdef ZSTD_LZMADECOMPRESS 2011 "/" XZ_EXTENSION "/" LZMA_EXTENSION 2012 #endif 2013 #ifdef ZSTD_LZ4DECOMPRESS 2014 "/" LZ4_EXTENSION 2015 #endif 2016 ; 2017 DISPLAYLEVEL(1, "zstd: %s: unknown suffix (%s expected) -- ignored \n", 2018 srcFileName, suffixlist); 2019 return NULL; 2020 } 2021 2022 /* allocate enough space to write dstFilename into it */ 2023 if (dfnbCapacity+suffixSize <= sfnSize+1) { 2024 free(dstFileNameBuffer); 2025 dfnbCapacity = sfnSize + 20; 2026 dstFileNameBuffer = (char*)malloc(dfnbCapacity); 2027 if (dstFileNameBuffer==NULL) 2028 EXM_THROW(74, "not enough memory for dstFileName"); 2029 } 2030 2031 /* return dst name == src name truncated from suffix */ 2032 assert(dstFileNameBuffer != NULL); 2033 memcpy(dstFileNameBuffer, srcFileName, sfnSize - suffixSize); 2034 dstFileNameBuffer[sfnSize-suffixSize] = '\0'; 2035 return dstFileNameBuffer; 2036 2037 /* note : dstFileNameBuffer memory is not going to be free */ 2038 } 2039 2040 2041 int 2042 FIO_decompressMultipleFilenames(const char* srcNamesTable[], unsigned nbFiles, 2043 const char* outFileName, 2044 const char* dictFileName) 2045 { 2046 int error = 0; 2047 dRess_t ress = FIO_createDResources(dictFileName); 2048 2049 if (outFileName) { 2050 unsigned u; 2051 ress.dstFile = FIO_openDstFile(outFileName); 2052 if (ress.dstFile == 0) EXM_THROW(71, "cannot open %s", outFileName); 2053 for (u=0; u<nbFiles; u++) 2054 error |= FIO_decompressSrcFile(ress, outFileName, srcNamesTable[u]); 2055 if (fclose(ress.dstFile)) 2056 EXM_THROW(72, "Write error : cannot properly close output file"); 2057 } else { 2058 unsigned u; 2059 for (u=0; u<nbFiles; u++) { /* create dstFileName */ 2060 const char* const srcFileName = srcNamesTable[u]; 2061 const char* const dstFileName = FIO_determineDstName(srcFileName); 2062 if (dstFileName == NULL) { error=1; continue; } 2063 2064 error |= FIO_decompressSrcFile(ress, dstFileName, srcFileName); 2065 } 2066 } 2067 2068 FIO_freeDResources(ress); 2069 return error; 2070 } 2071 2072 2073 2074 /* ************************************************************************** 2075 * .zst file info (--list command) 2076 ***************************************************************************/ 2077 2078 typedef struct { 2079 U64 decompressedSize; 2080 U64 compressedSize; 2081 U64 windowSize; 2082 int numActualFrames; 2083 int numSkippableFrames; 2084 int decompUnavailable; 2085 int usesCheck; 2086 U32 nbFiles; 2087 } fileInfo_t; 2088 2089 typedef enum { info_success=0, info_frame_error=1, info_not_zstd=2, info_file_error=3 } InfoError; 2090 2091 #define ERROR_IF(c,n,...) { \ 2092 if (c) { \ 2093 DISPLAYLEVEL(1, __VA_ARGS__); \ 2094 DISPLAYLEVEL(1, " \n"); \ 2095 return n; \ 2096 } \ 2097 } 2098 2099 static InfoError 2100 FIO_analyzeFrames(fileInfo_t* info, FILE* const srcFile) 2101 { 2102 /* begin analyzing frame */ 2103 for ( ; ; ) { 2104 BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; 2105 size_t const numBytesRead = fread(headerBuffer, 1, sizeof(headerBuffer), srcFile); 2106 if (numBytesRead < ZSTD_frameHeaderSize_min) { 2107 if ( feof(srcFile) 2108 && (numBytesRead == 0) 2109 && (info->compressedSize > 0) 2110 && (info->compressedSize != UTIL_FILESIZE_UNKNOWN) ) { 2111 break; /* correct end of file => success */ 2112 } 2113 ERROR_IF(feof(srcFile), info_not_zstd, "Error: reached end of file with incomplete frame"); 2114 ERROR_IF(1, info_frame_error, "Error: did not reach end of file but ran out of frames"); 2115 } 2116 { U32 const magicNumber = MEM_readLE32(headerBuffer); 2117 /* Zstandard frame */ 2118 if (magicNumber == ZSTD_MAGICNUMBER) { 2119 ZSTD_frameHeader header; 2120 U64 const frameContentSize = ZSTD_getFrameContentSize(headerBuffer, numBytesRead); 2121 if ( frameContentSize == ZSTD_CONTENTSIZE_ERROR 2122 || frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN ) { 2123 info->decompUnavailable = 1; 2124 } else { 2125 info->decompressedSize += frameContentSize; 2126 } 2127 ERROR_IF(ZSTD_getFrameHeader(&header, headerBuffer, numBytesRead) != 0, 2128 info_frame_error, "Error: could not decode frame header"); 2129 info->windowSize = header.windowSize; 2130 /* move to the end of the frame header */ 2131 { size_t const headerSize = ZSTD_frameHeaderSize(headerBuffer, numBytesRead); 2132 ERROR_IF(ZSTD_isError(headerSize), info_frame_error, "Error: could not determine frame header size"); 2133 ERROR_IF(fseek(srcFile, ((long)headerSize)-((long)numBytesRead), SEEK_CUR) != 0, 2134 info_frame_error, "Error: could not move to end of frame header"); 2135 } 2136 2137 /* skip all blocks in the frame */ 2138 { int lastBlock = 0; 2139 do { 2140 BYTE blockHeaderBuffer[3]; 2141 ERROR_IF(fread(blockHeaderBuffer, 1, 3, srcFile) != 3, 2142 info_frame_error, "Error while reading block header"); 2143 { U32 const blockHeader = MEM_readLE24(blockHeaderBuffer); 2144 U32 const blockTypeID = (blockHeader >> 1) & 3; 2145 U32 const isRLE = (blockTypeID == 1); 2146 U32 const isWrongBlock = (blockTypeID == 3); 2147 long const blockSize = isRLE ? 1 : (long)(blockHeader >> 3); 2148 ERROR_IF(isWrongBlock, info_frame_error, "Error: unsupported block type"); 2149 lastBlock = blockHeader & 1; 2150 ERROR_IF(fseek(srcFile, blockSize, SEEK_CUR) != 0, 2151 info_frame_error, "Error: could not skip to end of block"); 2152 } 2153 } while (lastBlock != 1); 2154 } 2155 2156 /* check if checksum is used */ 2157 { BYTE const frameHeaderDescriptor = headerBuffer[4]; 2158 int const contentChecksumFlag = (frameHeaderDescriptor & (1 << 2)) >> 2; 2159 if (contentChecksumFlag) { 2160 info->usesCheck = 1; 2161 ERROR_IF(fseek(srcFile, 4, SEEK_CUR) != 0, 2162 info_frame_error, "Error: could not skip past checksum"); 2163 } } 2164 info->numActualFrames++; 2165 } 2166 /* Skippable frame */ 2167 else if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { 2168 U32 const frameSize = MEM_readLE32(headerBuffer + 4); 2169 long const seek = (long)(8 + frameSize - numBytesRead); 2170 ERROR_IF(LONG_SEEK(srcFile, seek, SEEK_CUR) != 0, 2171 info_frame_error, "Error: could not find end of skippable frame"); 2172 info->numSkippableFrames++; 2173 } 2174 /* unknown content */ 2175 else { 2176 return info_not_zstd; 2177 } 2178 } /* magic number analysis */ 2179 } /* end analyzing frames */ 2180 return info_success; 2181 } 2182 2183 2184 static InfoError 2185 getFileInfo_fileConfirmed(fileInfo_t* info, const char* inFileName) 2186 { 2187 InfoError status; 2188 FILE* const srcFile = FIO_openSrcFile(inFileName); 2189 ERROR_IF(srcFile == NULL, info_file_error, "Error: could not open source file %s", inFileName); 2190 2191 info->compressedSize = UTIL_getFileSize(inFileName); 2192 status = FIO_analyzeFrames(info, srcFile); 2193 2194 fclose(srcFile); 2195 info->nbFiles = 1; 2196 return status; 2197 } 2198 2199 2200 /** getFileInfo() : 2201 * Reads information from file, stores in *info 2202 * @return : InfoError status 2203 */ 2204 static InfoError 2205 getFileInfo(fileInfo_t* info, const char* srcFileName) 2206 { 2207 ERROR_IF(!UTIL_isRegularFile(srcFileName), 2208 info_file_error, "Error : %s is not a file", srcFileName); 2209 return getFileInfo_fileConfirmed(info, srcFileName); 2210 } 2211 2212 2213 static void 2214 displayInfo(const char* inFileName, const fileInfo_t* info, int displayLevel) 2215 { 2216 unsigned const unit = info->compressedSize < (1 MB) ? (1 KB) : (1 MB); 2217 const char* const unitStr = info->compressedSize < (1 MB) ? "KB" : "MB"; 2218 double const windowSizeUnit = (double)info->windowSize / unit; 2219 double const compressedSizeUnit = (double)info->compressedSize / unit; 2220 double const decompressedSizeUnit = (double)info->decompressedSize / unit; 2221 double const ratio = (info->compressedSize == 0) ? 0 : ((double)info->decompressedSize)/info->compressedSize; 2222 const char* const checkString = (info->usesCheck ? "XXH64" : "None"); 2223 if (displayLevel <= 2) { 2224 if (!info->decompUnavailable) { 2225 DISPLAYOUT("%6d %5d %7.2f %2s %9.2f %2s %5.3f %5s %s\n", 2226 info->numSkippableFrames + info->numActualFrames, 2227 info->numSkippableFrames, 2228 compressedSizeUnit, unitStr, decompressedSizeUnit, unitStr, 2229 ratio, checkString, inFileName); 2230 } else { 2231 DISPLAYOUT("%6d %5d %7.2f %2s %5s %s\n", 2232 info->numSkippableFrames + info->numActualFrames, 2233 info->numSkippableFrames, 2234 compressedSizeUnit, unitStr, 2235 checkString, inFileName); 2236 } 2237 } else { 2238 DISPLAYOUT("%s \n", inFileName); 2239 DISPLAYOUT("# Zstandard Frames: %d\n", info->numActualFrames); 2240 if (info->numSkippableFrames) 2241 DISPLAYOUT("# Skippable Frames: %d\n", info->numSkippableFrames); 2242 DISPLAYOUT("Window Size: %.2f %2s (%llu B)\n", 2243 windowSizeUnit, unitStr, 2244 (unsigned long long)info->windowSize); 2245 DISPLAYOUT("Compressed Size: %.2f %2s (%llu B)\n", 2246 compressedSizeUnit, unitStr, 2247 (unsigned long long)info->compressedSize); 2248 if (!info->decompUnavailable) { 2249 DISPLAYOUT("Decompressed Size: %.2f %2s (%llu B)\n", 2250 decompressedSizeUnit, unitStr, 2251 (unsigned long long)info->decompressedSize); 2252 DISPLAYOUT("Ratio: %.4f\n", ratio); 2253 } 2254 DISPLAYOUT("Check: %s\n", checkString); 2255 DISPLAYOUT("\n"); 2256 } 2257 } 2258 2259 static fileInfo_t FIO_addFInfo(fileInfo_t fi1, fileInfo_t fi2) 2260 { 2261 fileInfo_t total; 2262 memset(&total, 0, sizeof(total)); 2263 total.numActualFrames = fi1.numActualFrames + fi2.numActualFrames; 2264 total.numSkippableFrames = fi1.numSkippableFrames + fi2.numSkippableFrames; 2265 total.compressedSize = fi1.compressedSize + fi2.compressedSize; 2266 total.decompressedSize = fi1.decompressedSize + fi2.decompressedSize; 2267 total.decompUnavailable = fi1.decompUnavailable | fi2.decompUnavailable; 2268 total.usesCheck = fi1.usesCheck & fi2.usesCheck; 2269 total.nbFiles = fi1.nbFiles + fi2.nbFiles; 2270 return total; 2271 } 2272 2273 static int 2274 FIO_listFile(fileInfo_t* total, const char* inFileName, int displayLevel) 2275 { 2276 fileInfo_t info; 2277 memset(&info, 0, sizeof(info)); 2278 { InfoError const error = getFileInfo(&info, inFileName); 2279 if (error == info_frame_error) { 2280 /* display error, but provide output */ 2281 DISPLAYLEVEL(1, "Error while parsing %s \n", inFileName); 2282 } 2283 else if (error == info_not_zstd) { 2284 DISPLAYOUT("File %s not compressed by zstd \n", inFileName); 2285 if (displayLevel > 2) DISPLAYOUT("\n"); 2286 return 1; 2287 } 2288 else if (error == info_file_error) { 2289 /* error occurred while opening the file */ 2290 if (displayLevel > 2) DISPLAYOUT("\n"); 2291 return 1; 2292 } 2293 displayInfo(inFileName, &info, displayLevel); 2294 *total = FIO_addFInfo(*total, info); 2295 assert(error>=0 || error<=1); 2296 return error; 2297 } 2298 } 2299 2300 int FIO_listMultipleFiles(unsigned numFiles, const char** filenameTable, int displayLevel) 2301 { 2302 /* ensure no specified input is stdin (needs fseek() capability) */ 2303 { unsigned u; 2304 for (u=0; u<numFiles;u++) { 2305 ERROR_IF(!strcmp (filenameTable[u], stdinmark), 2306 1, "zstd: --list does not support reading from standard input"); 2307 } } 2308 2309 if (numFiles == 0) { 2310 if (!IS_CONSOLE(stdin)) { 2311 DISPLAYLEVEL(1, "zstd: --list does not support reading from standard input \n"); 2312 } 2313 DISPLAYLEVEL(1, "No files given \n"); 2314 return 1; 2315 } 2316 2317 if (displayLevel <= 2) { 2318 DISPLAYOUT("Frames Skips Compressed Uncompressed Ratio Check Filename\n"); 2319 } 2320 { int error = 0; 2321 fileInfo_t total; 2322 memset(&total, 0, sizeof(total)); 2323 total.usesCheck = 1; 2324 /* --list each file, and check for any error */ 2325 { unsigned u; 2326 for (u=0; u<numFiles;u++) { 2327 error |= FIO_listFile(&total, filenameTable[u], displayLevel); 2328 } } 2329 if (numFiles > 1 && displayLevel <= 2) { /* display total */ 2330 unsigned const unit = total.compressedSize < (1 MB) ? (1 KB) : (1 MB); 2331 const char* const unitStr = total.compressedSize < (1 MB) ? "KB" : "MB"; 2332 double const compressedSizeUnit = (double)total.compressedSize / unit; 2333 double const decompressedSizeUnit = (double)total.decompressedSize / unit; 2334 double const ratio = (total.compressedSize == 0) ? 0 : ((double)total.decompressedSize)/total.compressedSize; 2335 const char* const checkString = (total.usesCheck ? "XXH64" : ""); 2336 DISPLAYOUT("----------------------------------------------------------------- \n"); 2337 if (total.decompUnavailable) { 2338 DISPLAYOUT("%6d %5d %7.2f %2s %5s %u files\n", 2339 total.numSkippableFrames + total.numActualFrames, 2340 total.numSkippableFrames, 2341 compressedSizeUnit, unitStr, 2342 checkString, total.nbFiles); 2343 } else { 2344 DISPLAYOUT("%6d %5d %7.2f %2s %9.2f %2s %5.3f %5s %u files\n", 2345 total.numSkippableFrames + total.numActualFrames, 2346 total.numSkippableFrames, 2347 compressedSizeUnit, unitStr, decompressedSizeUnit, unitStr, 2348 ratio, checkString, total.nbFiles); 2349 } } 2350 return error; 2351 } 2352 } 2353 2354 2355 #endif /* #ifndef ZSTD_NODECOMPRESS */ 2356