Yann Collet | 4ded9e5 | 2016-08-30 10:04:33 -0700 | [diff] [blame] | 1 | /** |
| 2 | * Copyright (c) 2016-present, Yann Collet, Facebook, Inc. |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * This source code is licensed under the BSD-style license found in the |
| 6 | * LICENSE file in the root directory of this source tree. An additional grant |
| 7 | * of patent rights can be found in the PATENTS file in the same directory. |
| 8 | */ |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 9 | |
Yann Collet | 8dafb1a | 2017-01-25 17:01:13 -0800 | [diff] [blame] | 10 | |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 11 | /* ************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 12 | * Compiler Options |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 13 | ***************************************/ |
Yann Collet | 94ca85d | 2016-08-14 01:19:12 +0200 | [diff] [blame] | 14 | #ifdef _MSC_VER /* Visual */ |
Przemyslaw Skibinski | e679741 | 2016-12-21 13:47:11 +0100 | [diff] [blame] | 15 | # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ |
Yann Collet | 94ca85d | 2016-08-14 01:19:12 +0200 | [diff] [blame] | 16 | # pragma warning(disable : 4204) /* non-constant aggregate initializer */ |
| 17 | #endif |
Przemyslaw Skibinski | 2f6ccee | 2016-12-21 13:23:34 +0100 | [diff] [blame] | 18 | #if defined(__MINGW32__) && !defined(_POSIX_SOURCE) |
| 19 | # define _POSIX_SOURCE 1 /* disable %llu warnings with MinGW on Windows */ |
| 20 | #endif |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 21 | |
Yann Collet | 179b197 | 2016-11-02 17:30:49 -0700 | [diff] [blame] | 22 | |
Yann Collet | 6f3acba | 2016-02-12 20:19:48 +0100 | [diff] [blame] | 23 | /*-************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 24 | * Includes |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 25 | ***************************************/ |
Przemyslaw Skibinski | 7a8a03c | 2016-12-21 15:08:44 +0100 | [diff] [blame] | 26 | #include "platform.h" /* Large Files support, SET_BINARY_MODE */ |
| 27 | #include "util.h" /* UTIL_getFileSize */ |
Yann Collet | 9f43292 | 2015-11-09 17:42:17 +0100 | [diff] [blame] | 28 | #include <stdio.h> /* fprintf, fopen, fread, _fileno, stdin, stdout */ |
| 29 | #include <stdlib.h> /* malloc, free */ |
| 30 | #include <string.h> /* strcmp, strlen */ |
| 31 | #include <time.h> /* clock */ |
| 32 | #include <errno.h> /* errno */ |
inikep | d5ff2c3 | 2016-04-28 14:40:45 +0200 | [diff] [blame] | 33 | |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 34 | #if defined (_MSC_VER) |
| 35 | # include <sys/stat.h> |
| 36 | # include <io.h> |
| 37 | #endif |
| 38 | |
Przemyslaw Skibinski | 2f6ccee | 2016-12-21 13:23:34 +0100 | [diff] [blame] | 39 | #include "mem.h" |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 40 | #include "fileio.h" |
Yann Collet | d3b7f8d | 2016-06-04 19:47:02 +0200 | [diff] [blame] | 41 | #define ZSTD_STATIC_LINKING_ONLY /* ZSTD_magicNumber, ZSTD_frameHeaderSize_max */ |
| 42 | #include "zstd.h" |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 43 | #ifdef ZSTD_MULTITHREAD |
| 44 | # include "zstdmt_compress.h" |
Przemyslaw Skibinski | 6b508b1 | 2016-12-05 18:02:40 +0100 | [diff] [blame] | 45 | #endif |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 46 | #if defined(ZSTD_GZCOMPRESS) || defined(ZSTD_GZDECOMPRESS) |
Nick Terrell | 2cb8ee8 | 2017-02-06 11:32:13 -0800 | [diff] [blame] | 47 | # include <zlib.h> |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 48 | # if !defined(z_const) |
| 49 | # define z_const |
| 50 | # endif |
Przemyslaw Skibinski | abfb51f | 2016-11-30 15:05:54 +0100 | [diff] [blame] | 51 | #endif |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 52 | #if defined(ZSTD_LZMACOMPRESS) || defined(ZSTD_LZMADECOMPRESS) |
| 53 | # include <lzma.h> |
| 54 | #endif |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 55 | |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 56 | #define LZ4_MAGICNUMBER 0x184D2204 |
| 57 | #if defined(ZSTD_LZ4COMPRESS) || defined(ZSTD_LZ4DECOMPRESS) |
| 58 | # include <lz4frame.h> |
Sean Purcell | 2c4b6fe | 2017-04-25 11:00:54 -0700 | [diff] [blame] | 59 | # include <lz4.h> |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 60 | #endif |
| 61 | |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 62 | |
Yann Collet | 6f3acba | 2016-02-12 20:19:48 +0100 | [diff] [blame] | 63 | /*-************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 64 | * Constants |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 65 | ***************************************/ |
Yann Collet | 92d7566 | 2016-07-03 01:10:53 +0200 | [diff] [blame] | 66 | #define KB *(1<<10) |
| 67 | #define MB *(1<<20) |
| 68 | #define GB *(1U<<30) |
| 69 | |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 70 | #define _1BIT 0x01 |
| 71 | #define _2BITS 0x03 |
| 72 | #define _3BITS 0x07 |
| 73 | #define _4BITS 0x0F |
| 74 | #define _6BITS 0x3F |
| 75 | #define _8BITS 0xFF |
| 76 | |
Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 77 | #define BLOCKSIZE (128 KB) |
| 78 | #define ROLLBUFFERSIZE (BLOCKSIZE*8*64) |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 79 | |
Yann Collet | 6f3acba | 2016-02-12 20:19:48 +0100 | [diff] [blame] | 80 | #define FIO_FRAMEHEADERSIZE 5 /* as a define, because needed to allocated table on stack */ |
| 81 | #define FSE_CHECKSUM_SEED 0 |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 82 | |
| 83 | #define CACHELINE 64 |
| 84 | |
Yann Collet | 0e30059 | 2017-04-11 14:41:02 -0700 | [diff] [blame] | 85 | #define DICTSIZE_MAX (32 MB) /* protection against large input (attack scenario) */ |
Yann Collet | 6f3acba | 2016-02-12 20:19:48 +0100 | [diff] [blame] | 86 | |
inikep | 3c7c352 | 2016-04-22 13:59:05 +0200 | [diff] [blame] | 87 | #define FNSPACE 30 |
| 88 | |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 89 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 90 | /*-************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 91 | * Macros |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 92 | ***************************************/ |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 93 | #define DISPLAY(...) fprintf(stderr, __VA_ARGS__) |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 94 | #define DISPLAYOUT(...) fprintf(stdout, __VA_ARGS__) |
Yann Collet | 5d9b894 | 2017-01-27 13:30:18 -0800 | [diff] [blame] | 95 | #define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } } |
Sean Purcell | 042ba12 | 2017-03-23 11:13:52 -0700 | [diff] [blame] | 96 | static int g_displayLevel = 2; /* 0 : no display; 1: errors; 2 : + result + interaction + warnings; 3 : + progression; 4 : + information */ |
Yann Collet | 8a1d1a6 | 2016-03-10 21:02:25 +0100 | [diff] [blame] | 97 | void FIO_setNotificationLevel(unsigned level) { g_displayLevel=level; } |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 98 | |
Yann Collet | 5d9b894 | 2017-01-27 13:30:18 -0800 | [diff] [blame] | 99 | #define DISPLAYUPDATE(l, ...) { if (g_displayLevel>=l) { \ |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 100 | if ((clock() - g_time > refreshRate) || (g_displayLevel>=4)) \ |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 101 | { g_time = clock(); DISPLAY(__VA_ARGS__); \ |
Sean Purcell | 042ba12 | 2017-03-23 11:13:52 -0700 | [diff] [blame] | 102 | if (g_displayLevel>=4) fflush(stderr); } } } |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 103 | static const clock_t refreshRate = CLOCKS_PER_SEC * 15 / 100; |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 104 | static clock_t g_time = 0; |
| 105 | |
Sean Purcell | 42bac7f | 2017-04-13 15:35:05 -0700 | [diff] [blame] | 106 | #undef MIN |
| 107 | #define MIN(a,b) ((a) < (b) ? (a) : (b)) |
Yann Collet | 92d7566 | 2016-07-03 01:10:53 +0200 | [diff] [blame] | 108 | |
Yann Collet | 31533ba | 2017-04-27 00:29:04 -0700 | [diff] [blame] | 109 | |
Przemyslaw Skibinski | bf33657 | 2017-02-14 09:45:33 +0100 | [diff] [blame] | 110 | /* ************************************************************ |
| 111 | * Avoid fseek()'s 2GiB barrier with MSVC, MacOS, *BSD, MinGW |
| 112 | ***************************************************************/ |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 113 | #if defined(_MSC_VER) && _MSC_VER >= 1400 |
| 114 | # define LONG_SEEK _fseeki64 |
Przemyslaw Skibinski | bf33657 | 2017-02-14 09:45:33 +0100 | [diff] [blame] | 115 | #elif !defined(__64BIT__) && (PLATFORM_POSIX_VERSION >= 200112L) /* No point defining Large file for 64 bit */ |
Przemyslaw Skibinski | ce13d08 | 2017-02-14 09:52:52 +0100 | [diff] [blame] | 116 | # define LONG_SEEK fseeko |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 117 | #elif defined(__MINGW32__) && !defined(__STRICT_ANSI__) && !defined(__NO_MINGW_LFS) && defined(__MSVCRT__) |
| 118 | # define LONG_SEEK fseeko64 |
| 119 | #elif defined(_WIN32) && !defined(__DJGPP__) |
| 120 | # include <windows.h> |
| 121 | static int LONG_SEEK(FILE* file, __int64 offset, int origin) { |
| 122 | LARGE_INTEGER off; |
| 123 | DWORD method; |
| 124 | off.QuadPart = offset; |
| 125 | if (origin == SEEK_END) |
| 126 | method = FILE_END; |
| 127 | else if (origin == SEEK_CUR) |
| 128 | method = FILE_CURRENT; |
| 129 | else |
| 130 | method = FILE_BEGIN; |
| 131 | |
| 132 | if (SetFilePointerEx((HANDLE) _get_osfhandle(_fileno(file)), off, NULL, method)) |
ds77 | 6220bfc | 2017-02-13 12:00:59 +0100 | [diff] [blame] | 133 | return 0; |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 134 | else |
ds77 | 6220bfc | 2017-02-13 12:00:59 +0100 | [diff] [blame] | 135 | return -1; |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 136 | } |
| 137 | #else |
| 138 | # define LONG_SEEK fseek |
| 139 | #endif |
| 140 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 141 | |
| 142 | /*-************************************* |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 143 | * Local Parameters - Not thread safe |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 144 | ***************************************/ |
Przemyslaw Skibinski | 98509a7 | 2017-02-14 09:23:32 +0100 | [diff] [blame] | 145 | static FIO_compressionType_t g_compressionType = FIO_zstdCompression; |
| 146 | void FIO_setCompressionType(FIO_compressionType_t compressionType) { g_compressionType = compressionType; } |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 147 | static U32 g_overwrite = 0; |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 148 | void FIO_overwriteMode(void) { g_overwrite=1; } |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 149 | static U32 g_sparseFileSupport = 1; /* 0 : no sparse allowed; 1: auto (file yes, stdout no); 2: force sparse */ |
| 150 | void FIO_setSparseWrite(unsigned sparse) { g_sparseFileSupport=sparse; } |
Yann Collet | 6381e99 | 2016-05-31 02:29:45 +0200 | [diff] [blame] | 151 | static U32 g_dictIDFlag = 1; |
| 152 | void FIO_setDictIDFlag(unsigned dictIDFlag) { g_dictIDFlag = dictIDFlag; } |
Yann Collet | c98f8e7 | 2016-06-20 16:31:24 +0200 | [diff] [blame] | 153 | static U32 g_checksumFlag = 1; |
Yann Collet | 87cfbe3 | 2016-06-01 19:22:15 +0200 | [diff] [blame] | 154 | void FIO_setChecksumFlag(unsigned checksumFlag) { g_checksumFlag = checksumFlag; } |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 155 | static U32 g_removeSrcFile = 0; |
| 156 | void FIO_setRemoveSrcFile(unsigned flag) { g_removeSrcFile = (flag>0); } |
Yann Collet | d4cda27 | 2016-10-14 13:13:13 -0700 | [diff] [blame] | 157 | static U32 g_memLimit = 0; |
| 158 | void FIO_setMemLimit(unsigned memLimit) { g_memLimit = memLimit; } |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 159 | static U32 g_nbThreads = 1; |
| 160 | void FIO_setNbThreads(unsigned nbThreads) { |
| 161 | #ifndef ZSTD_MULTITHREAD |
| 162 | if (nbThreads > 1) DISPLAYLEVEL(2, "Note : multi-threading is disabled \n"); |
| 163 | #endif |
| 164 | g_nbThreads = nbThreads; |
| 165 | } |
Yann Collet | 512cbe8 | 2017-01-24 17:02:26 -0800 | [diff] [blame] | 166 | static U32 g_blockSize = 0; |
| 167 | void FIO_setBlockSize(unsigned blockSize) { |
| 168 | if (blockSize && g_nbThreads==1) |
| 169 | DISPLAYLEVEL(2, "Setting block size is useless in single-thread mode \n"); |
| 170 | #ifdef ZSTD_MULTITHREAD |
| 171 | if (blockSize-1 < ZSTDMT_SECTION_SIZE_MIN-1) /* intentional underflow */ |
| 172 | DISPLAYLEVEL(2, "Note : minimum block size is %u KB \n", (ZSTDMT_SECTION_SIZE_MIN>>10)); |
| 173 | #endif |
| 174 | g_blockSize = blockSize; |
| 175 | } |
Yann Collet | 8d8513f | 2017-01-30 14:37:08 -0800 | [diff] [blame] | 176 | #define FIO_OVERLAP_LOG_NOTSET 9999 |
| 177 | static U32 g_overlapLog = FIO_OVERLAP_LOG_NOTSET; |
Yann Collet | 6be2337 | 2017-01-30 11:17:26 -0800 | [diff] [blame] | 178 | void FIO_setOverlapLog(unsigned overlapLog){ |
| 179 | if (overlapLog && g_nbThreads==1) |
| 180 | DISPLAYLEVEL(2, "Setting overlapLog is useless in single-thread mode \n"); |
| 181 | g_overlapLog = overlapLog; |
| 182 | } |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 183 | |
| 184 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 185 | /*-************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 186 | * Exceptions |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 187 | ***************************************/ |
| 188 | #ifndef DEBUG |
| 189 | # define DEBUG 0 |
| 190 | #endif |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 191 | #define DEBUGOUTPUT(...) if (DEBUG) DISPLAY(__VA_ARGS__); |
| 192 | #define EXM_THROW(error, ...) \ |
| 193 | { \ |
| 194 | DEBUGOUTPUT("Error defined at %s, line %i : \n", __FILE__, __LINE__); \ |
| 195 | DISPLAYLEVEL(1, "Error %i : ", error); \ |
| 196 | DISPLAYLEVEL(1, __VA_ARGS__); \ |
Yann Collet | cdff19c | 2016-11-11 17:26:54 -0800 | [diff] [blame] | 197 | DISPLAYLEVEL(1, " \n"); \ |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 198 | exit(error); \ |
| 199 | } |
| 200 | |
| 201 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 202 | /*-************************************* |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 203 | * Functions |
Yann Collet | eeb8ba1 | 2015-10-22 16:55:40 +0100 | [diff] [blame] | 204 | ***************************************/ |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 205 | /** FIO_remove() : |
| 206 | * @result : Unlink `fileName`, even if it's read-only */ |
| 207 | static int FIO_remove(const char* path) |
| 208 | { |
| 209 | #if defined(_WIN32) || defined(WIN32) |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 210 | /* windows doesn't allow remove read-only files, |
| 211 | * so try to make it writable first */ |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 212 | chmod(path, _S_IWRITE); |
| 213 | #endif |
| 214 | return remove(path); |
| 215 | } |
| 216 | |
Yann Collet | cdff19c | 2016-11-11 17:26:54 -0800 | [diff] [blame] | 217 | /** FIO_openSrcFile() : |
| 218 | * condition : `dstFileName` must be non-NULL. |
| 219 | * @result : FILE* to `dstFileName`, or NULL if it fails */ |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 220 | static FILE* FIO_openSrcFile(const char* srcFileName) |
| 221 | { |
| 222 | FILE* f; |
| 223 | |
| 224 | if (!strcmp (srcFileName, stdinmark)) { |
| 225 | DISPLAYLEVEL(4,"Using stdin for input\n"); |
| 226 | f = stdin; |
| 227 | SET_BINARY_MODE(stdin); |
| 228 | } else { |
Przemyslaw Skibinski | 442c75f | 2017-02-14 09:38:51 +0100 | [diff] [blame] | 229 | if (!UTIL_isRegFile(srcFileName)) { |
Przemyslaw Skibinski | 5022a18 | 2017-01-25 13:11:26 +0100 | [diff] [blame] | 230 | DISPLAYLEVEL(1, "zstd: %s is not a regular file -- ignored \n", srcFileName); |
| 231 | return NULL; |
| 232 | } |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 233 | f = fopen(srcFileName, "rb"); |
Yann Collet | fb5e385 | 2016-08-13 20:49:47 +0200 | [diff] [blame] | 234 | if ( f==NULL ) DISPLAYLEVEL(1, "zstd: %s: %s \n", srcFileName, strerror(errno)); |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 235 | } |
| 236 | |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 237 | return f; |
| 238 | } |
| 239 | |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 240 | /** FIO_openDstFile() : |
| 241 | * condition : `dstFileName` must be non-NULL. |
| 242 | * @result : FILE* to `dstFileName`, or NULL if it fails */ |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 243 | static FILE* FIO_openDstFile(const char* dstFileName) |
| 244 | { |
| 245 | FILE* f; |
| 246 | |
| 247 | if (!strcmp (dstFileName, stdoutmark)) { |
| 248 | DISPLAYLEVEL(4,"Using stdout for output\n"); |
| 249 | f = stdout; |
| 250 | SET_BINARY_MODE(stdout); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 251 | if (g_sparseFileSupport==1) { |
| 252 | g_sparseFileSupport = 0; |
| 253 | DISPLAYLEVEL(4, "Sparse File Support is automatically disabled on stdout ; try --sparse \n"); |
| 254 | } |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 255 | } else { |
Nick Terrell | eb38617 | 2017-03-31 15:20:50 -0700 | [diff] [blame] | 256 | if (g_sparseFileSupport == 1) { |
| 257 | g_sparseFileSupport = ZSTD_SPARSE_DEFAULT; |
Nick Terrell | 96fe545 | 2017-03-31 15:16:43 -0700 | [diff] [blame] | 258 | } |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 259 | if (strcmp (dstFileName, nulmark)) { /* Check if destination file already exists */ |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 260 | f = fopen( dstFileName, "rb" ); |
| 261 | if (f != 0) { /* dest file exists, prompt for overwrite authorization */ |
| 262 | fclose(f); |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 263 | if (!g_overwrite) { |
| 264 | if (g_displayLevel <= 1) { |
| 265 | /* No interaction possible */ |
| 266 | DISPLAY("zstd: %s already exists; not overwritten \n", dstFileName); |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 267 | return NULL; |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 268 | } |
Sean Purcell | 279be20 | 2017-04-06 12:56:40 -0700 | [diff] [blame] | 269 | DISPLAY("zstd: %s already exists; do you wish to overwrite (y/N) ? ", dstFileName); |
| 270 | { int ch = getchar(); |
| 271 | if ((ch!='Y') && (ch!='y')) { |
| 272 | DISPLAY(" not overwritten \n"); |
| 273 | return NULL; |
| 274 | } |
| 275 | while ((ch!=EOF) && (ch!='\n')) ch = getchar(); /* flush rest of input line */ |
| 276 | } |
| 277 | } |
| 278 | |
| 279 | /* need to unlink */ |
| 280 | FIO_remove(dstFileName); |
| 281 | } } |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 282 | f = fopen( dstFileName, "wb" ); |
Yann Collet | fb5e385 | 2016-08-13 20:49:47 +0200 | [diff] [blame] | 283 | if (f==NULL) DISPLAYLEVEL(1, "zstd: %s: %s\n", dstFileName, strerror(errno)); |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 284 | } |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 285 | |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 286 | return f; |
| 287 | } |
| 288 | |
| 289 | |
Yann Collet | 0e30059 | 2017-04-11 14:41:02 -0700 | [diff] [blame] | 290 | /*! FIO_createDictBuffer() : |
| 291 | * creates a buffer, pointed by `*bufferPtr`, |
| 292 | * loads `filename` content into it, up to DICTSIZE_MAX bytes. |
| 293 | * @return : loaded size |
| 294 | * if fileName==NULL, returns 0 and a NULL pointer |
| 295 | */ |
| 296 | static size_t FIO_createDictBuffer(void** bufferPtr, const char* fileName) |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 297 | { |
| 298 | FILE* fileHandle; |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 299 | U64 fileSize; |
| 300 | |
| 301 | *bufferPtr = NULL; |
Yann Collet | 2ce4923 | 2016-02-02 14:36:49 +0100 | [diff] [blame] | 302 | if (fileName == NULL) return 0; |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 303 | |
| 304 | DISPLAYLEVEL(4,"Loading %s as dictionary \n", fileName); |
| 305 | fileHandle = fopen(fileName, "rb"); |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 306 | if (fileHandle==0) EXM_THROW(31, "zstd: %s: %s", fileName, strerror(errno)); |
inikep | 69fcd7c | 2016-04-28 12:23:33 +0200 | [diff] [blame] | 307 | fileSize = UTIL_getFileSize(fileName); |
Yann Collet | 0e30059 | 2017-04-11 14:41:02 -0700 | [diff] [blame] | 308 | if (fileSize > DICTSIZE_MAX) EXM_THROW(32, "Dictionary file %s is too large (> %u MB)", fileName, DICTSIZE_MAX >> 20); /* avoid extreme cases */ |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 309 | *bufferPtr = malloc((size_t)fileSize); |
Yann Collet | ed7fb84 | 2016-07-02 11:14:30 +0200 | [diff] [blame] | 310 | if (*bufferPtr==NULL) EXM_THROW(34, "zstd: %s", strerror(errno)); |
Yann Collet | 6fca9e7 | 2016-05-31 02:40:42 +0200 | [diff] [blame] | 311 | { size_t const readSize = fread(*bufferPtr, 1, (size_t)fileSize, fileHandle); |
| 312 | if (readSize!=fileSize) EXM_THROW(35, "Error reading dictionary file %s", fileName); } |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 313 | fclose(fileHandle); |
| 314 | return (size_t)fileSize; |
| 315 | } |
| 316 | |
inikep | 3c7c352 | 2016-04-22 13:59:05 +0200 | [diff] [blame] | 317 | #ifndef ZSTD_NOCOMPRESS |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 318 | |
Yann Collet | 8a1d1a6 | 2016-03-10 21:02:25 +0100 | [diff] [blame] | 319 | /*-********************************************************************** |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 320 | * Compression |
| 321 | ************************************************************************/ |
| 322 | typedef struct { |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 323 | FILE* srcFile; |
| 324 | FILE* dstFile; |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 325 | void* srcBuffer; |
| 326 | size_t srcBufferSize; |
| 327 | void* dstBuffer; |
| 328 | size_t dstBufferSize; |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 329 | #ifdef ZSTD_MULTITHREAD |
| 330 | ZSTDMT_CCtx* cctx; |
| 331 | #else |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 332 | ZSTD_CStream* cctx; |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 333 | #endif |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 334 | } cRess_t; |
| 335 | |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 336 | static cRess_t FIO_createCResources(const char* dictFileName, int cLevel, |
Sean Purcell | 0f5c95a | 2017-02-07 16:33:48 -0800 | [diff] [blame] | 337 | U64 srcSize, int srcRegFile, |
| 338 | ZSTD_compressionParameters* comprParams) { |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 339 | cRess_t ress; |
Yann Collet | 5e80dd3 | 2016-07-13 17:38:39 +0200 | [diff] [blame] | 340 | memset(&ress, 0, sizeof(ress)); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 341 | |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 342 | #ifdef ZSTD_MULTITHREAD |
| 343 | ress.cctx = ZSTDMT_createCCtx(g_nbThreads); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 344 | if (ress.cctx == NULL) |
| 345 | EXM_THROW(30, "zstd: allocation error : can't create ZSTD_CStream"); |
Yann Collet | 8d8513f | 2017-01-30 14:37:08 -0800 | [diff] [blame] | 346 | if ((cLevel==ZSTD_maxCLevel()) && (g_overlapLog==FIO_OVERLAP_LOG_NOTSET)) |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 347 | /* use complete window for overlap */ |
| 348 | ZSTDMT_setMTCtxParameter(ress.cctx, ZSTDMT_p_overlapSectionLog, 9); |
Yann Collet | 8d8513f | 2017-01-30 14:37:08 -0800 | [diff] [blame] | 349 | if (g_overlapLog != FIO_OVERLAP_LOG_NOTSET) |
Yann Collet | 6be2337 | 2017-01-30 11:17:26 -0800 | [diff] [blame] | 350 | ZSTDMT_setMTCtxParameter(ress.cctx, ZSTDMT_p_overlapSectionLog, g_overlapLog); |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 351 | #else |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 352 | ress.cctx = ZSTD_createCStream(); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 353 | if (ress.cctx == NULL) |
| 354 | EXM_THROW(30, "zstd: allocation error : can't create ZSTD_CStream"); |
Yann Collet | 8dafb1a | 2017-01-25 17:01:13 -0800 | [diff] [blame] | 355 | #endif |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 356 | ress.srcBufferSize = ZSTD_CStreamInSize(); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 357 | ress.srcBuffer = malloc(ress.srcBufferSize); |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 358 | ress.dstBufferSize = ZSTD_CStreamOutSize(); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 359 | ress.dstBuffer = malloc(ress.dstBufferSize); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 360 | if (!ress.srcBuffer || !ress.dstBuffer) |
| 361 | EXM_THROW(31, "zstd: allocation error : not enough memory"); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 362 | |
| 363 | /* dictionary */ |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 364 | { void* dictBuffer; |
Yann Collet | 5c42d0e | 2017-04-11 16:57:32 -0700 | [diff] [blame] | 365 | size_t const dictBuffSize = FIO_createDictBuffer(&dictBuffer, dictFileName); /* works with dictFileName==NULL */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 366 | if (dictFileName && (dictBuffer==NULL)) |
| 367 | EXM_THROW(32, "zstd: allocation error : can't create dictBuffer"); |
Yann Collet | 993060e | 2016-09-21 16:46:08 +0200 | [diff] [blame] | 368 | { ZSTD_parameters params = ZSTD_getParams(cLevel, srcSize, dictBuffSize); |
Sean Purcell | 0f5c95a | 2017-02-07 16:33:48 -0800 | [diff] [blame] | 369 | params.fParams.contentSizeFlag = srcRegFile; |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 370 | params.fParams.checksumFlag = g_checksumFlag; |
| 371 | params.fParams.noDictIDFlag = !g_dictIDFlag; |
Przemyslaw Skibinski | 8349d67 | 2016-12-13 13:24:59 +0100 | [diff] [blame] | 372 | if (comprParams->windowLog) params.cParams.windowLog = comprParams->windowLog; |
| 373 | if (comprParams->chainLog) params.cParams.chainLog = comprParams->chainLog; |
| 374 | if (comprParams->hashLog) params.cParams.hashLog = comprParams->hashLog; |
| 375 | if (comprParams->searchLog) params.cParams.searchLog = comprParams->searchLog; |
| 376 | if (comprParams->searchLength) params.cParams.searchLength = comprParams->searchLength; |
| 377 | if (comprParams->targetLength) params.cParams.targetLength = comprParams->targetLength; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 378 | if (comprParams->strategy) params.cParams.strategy = (ZSTD_strategy) comprParams->strategy; |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 379 | #ifdef ZSTD_MULTITHREAD |
| 380 | { size_t const errorCode = ZSTDMT_initCStream_advanced(ress.cctx, dictBuffer, dictBuffSize, params, srcSize); |
Yann Collet | 512cbe8 | 2017-01-24 17:02:26 -0800 | [diff] [blame] | 381 | if (ZSTD_isError(errorCode)) EXM_THROW(33, "Error initializing CStream : %s", ZSTD_getErrorName(errorCode)); |
| 382 | ZSTDMT_setMTCtxParameter(ress.cctx, ZSTDMT_p_sectionSize, g_blockSize); |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 383 | #else |
Yann Collet | 993060e | 2016-09-21 16:46:08 +0200 | [diff] [blame] | 384 | { size_t const errorCode = ZSTD_initCStream_advanced(ress.cctx, dictBuffer, dictBuffSize, params, srcSize); |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 385 | if (ZSTD_isError(errorCode)) EXM_THROW(33, "Error initializing CStream : %s", ZSTD_getErrorName(errorCode)); |
Yann Collet | 512cbe8 | 2017-01-24 17:02:26 -0800 | [diff] [blame] | 386 | #endif |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 387 | } } |
| 388 | free(dictBuffer); |
| 389 | } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 390 | |
| 391 | return ress; |
| 392 | } |
| 393 | |
| 394 | static void FIO_freeCResources(cRess_t ress) |
| 395 | { |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 396 | free(ress.srcBuffer); |
| 397 | free(ress.dstBuffer); |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 398 | #ifdef ZSTD_MULTITHREAD |
| 399 | ZSTDMT_freeCCtx(ress.cctx); |
| 400 | #else |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 401 | ZSTD_freeCStream(ress.cctx); /* never fails */ |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 402 | #endif |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 403 | } |
| 404 | |
| 405 | |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 406 | #ifdef ZSTD_GZCOMPRESS |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 407 | static unsigned long long FIO_compressGzFrame(cRess_t* ress, |
| 408 | const char* srcFileName, U64 const srcFileSize, |
| 409 | int compressionLevel, U64* readsize) |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 410 | { |
| 411 | unsigned long long inFileSize = 0, outFileSize = 0; |
| 412 | z_stream strm; |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 413 | int ret; |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 414 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 415 | if (compressionLevel > Z_BEST_COMPRESSION) |
| 416 | compressionLevel = Z_BEST_COMPRESSION; |
Przemyslaw Skibinski | 64f7221 | 2017-02-13 21:00:41 +0100 | [diff] [blame] | 417 | |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 418 | strm.zalloc = Z_NULL; |
| 419 | strm.zfree = Z_NULL; |
| 420 | strm.opaque = Z_NULL; |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 421 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 422 | ret = deflateInit2(&strm, compressionLevel, Z_DEFLATED, |
| 423 | 15 /* maxWindowLogSize */ + 16 /* gzip only */, |
| 424 | 8, Z_DEFAULT_STRATEGY); /* see http://www.zlib.net/manual.html */ |
| 425 | if (ret != Z_OK) |
| 426 | EXM_THROW(71, "zstd: %s: deflateInit2 error %d \n", srcFileName, ret); |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 427 | |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 428 | strm.next_in = 0; |
Przemyslaw Skibinski | 862698f | 2017-02-27 13:21:05 +0100 | [diff] [blame] | 429 | strm.avail_in = 0; |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 430 | strm.next_out = (Bytef*)ress->dstBuffer; |
| 431 | strm.avail_out = (uInt)ress->dstBufferSize; |
| 432 | |
| 433 | while (1) { |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 434 | if (strm.avail_in == 0) { |
| 435 | size_t const inSize = fread(ress->srcBuffer, 1, ress->srcBufferSize, ress->srcFile); |
| 436 | if (inSize == 0) break; |
| 437 | inFileSize += inSize; |
| 438 | strm.next_in = (z_const unsigned char*)ress->srcBuffer; |
| 439 | strm.avail_in = (uInt)inSize; |
| 440 | } |
| 441 | ret = deflate(&strm, Z_NO_FLUSH); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 442 | if (ret != Z_OK) |
| 443 | EXM_THROW(72, "zstd: %s: deflate error %d \n", srcFileName, ret); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 444 | { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; |
| 445 | if (decompBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 446 | if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) |
| 447 | EXM_THROW(73, "Write error : cannot write to output file"); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 448 | outFileSize += decompBytes; |
| 449 | strm.next_out = (Bytef*)ress->dstBuffer; |
| 450 | strm.avail_out = (uInt)ress->dstBufferSize; |
| 451 | } |
| 452 | } |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 453 | if (!srcFileSize) |
| 454 | DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", |
| 455 | (U32)(inFileSize>>20), |
| 456 | (double)outFileSize/inFileSize*100) |
| 457 | else |
| 458 | DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", |
| 459 | (U32)(inFileSize>>20), (U32)(srcFileSize>>20), |
| 460 | (double)outFileSize/inFileSize*100); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 461 | } |
| 462 | |
| 463 | while (1) { |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 464 | ret = deflate(&strm, Z_FINISH); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 465 | { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; |
| 466 | if (decompBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 467 | if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) |
| 468 | EXM_THROW(75, "Write error : cannot write to output file"); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 469 | outFileSize += decompBytes; |
| 470 | strm.next_out = (Bytef*)ress->dstBuffer; |
| 471 | strm.avail_out = (uInt)ress->dstBufferSize; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 472 | } } |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 473 | if (ret == Z_STREAM_END) break; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 474 | if (ret != Z_BUF_ERROR) |
| 475 | EXM_THROW(77, "zstd: %s: deflate error %d \n", srcFileName, ret); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 476 | } |
| 477 | |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 478 | ret = deflateEnd(&strm); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 479 | if (ret != Z_OK) |
| 480 | EXM_THROW(79, "zstd: %s: deflateEnd error %d \n", srcFileName, ret); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 481 | *readsize = inFileSize; |
| 482 | |
| 483 | return outFileSize; |
| 484 | } |
| 485 | #endif |
| 486 | |
| 487 | |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 488 | #ifdef ZSTD_LZMACOMPRESS |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 489 | static unsigned long long FIO_compressLzmaFrame(cRess_t* ress, |
| 490 | const char* srcFileName, U64 const srcFileSize, |
| 491 | int compressionLevel, U64* readsize, int plain_lzma) |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 492 | { |
| 493 | unsigned long long inFileSize = 0, outFileSize = 0; |
| 494 | lzma_stream strm = LZMA_STREAM_INIT; |
| 495 | lzma_action action = LZMA_RUN; |
| 496 | lzma_ret ret; |
| 497 | |
| 498 | if (compressionLevel < 0) compressionLevel = 0; |
| 499 | if (compressionLevel > 9) compressionLevel = 9; |
| 500 | |
| 501 | if (plain_lzma) { |
| 502 | lzma_options_lzma opt_lzma; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 503 | if (lzma_lzma_preset(&opt_lzma, compressionLevel)) |
| 504 | EXM_THROW(71, "zstd: %s: lzma_lzma_preset error", srcFileName); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 505 | ret = lzma_alone_encoder(&strm, &opt_lzma); /* LZMA */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 506 | if (ret != LZMA_OK) |
| 507 | EXM_THROW(71, "zstd: %s: lzma_alone_encoder error %d", srcFileName, ret); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 508 | } else { |
| 509 | ret = lzma_easy_encoder(&strm, compressionLevel, LZMA_CHECK_CRC64); /* XZ */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 510 | if (ret != LZMA_OK) |
| 511 | EXM_THROW(71, "zstd: %s: lzma_easy_encoder error %d", srcFileName, ret); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 512 | } |
| 513 | |
| 514 | strm.next_in = 0; |
| 515 | strm.avail_in = 0; |
| 516 | strm.next_out = ress->dstBuffer; |
| 517 | strm.avail_out = ress->dstBufferSize; |
| 518 | |
| 519 | while (1) { |
| 520 | if (strm.avail_in == 0) { |
| 521 | size_t const inSize = fread(ress->srcBuffer, 1, ress->srcBufferSize, ress->srcFile); |
| 522 | if (inSize == 0) action = LZMA_FINISH; |
| 523 | inFileSize += inSize; |
| 524 | strm.next_in = ress->srcBuffer; |
| 525 | strm.avail_in = inSize; |
| 526 | } |
| 527 | |
| 528 | ret = lzma_code(&strm, action); |
| 529 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 530 | if (ret != LZMA_OK && ret != LZMA_STREAM_END) |
| 531 | EXM_THROW(72, "zstd: %s: lzma_code encoding error %d", srcFileName, ret); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 532 | { size_t const compBytes = ress->dstBufferSize - strm.avail_out; |
| 533 | if (compBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 534 | if (fwrite(ress->dstBuffer, 1, compBytes, ress->dstFile) != compBytes) |
| 535 | EXM_THROW(73, "Write error : cannot write to output file"); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 536 | outFileSize += compBytes; |
| 537 | strm.next_out = ress->dstBuffer; |
| 538 | strm.avail_out = ress->dstBufferSize; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 539 | } } |
| 540 | if (!srcFileSize) |
| 541 | DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", |
| 542 | (U32)(inFileSize>>20), |
| 543 | (double)outFileSize/inFileSize*100) |
| 544 | else |
| 545 | DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", |
| 546 | (U32)(inFileSize>>20), (U32)(srcFileSize>>20), |
| 547 | (double)outFileSize/inFileSize*100); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 548 | if (ret == LZMA_STREAM_END) break; |
| 549 | } |
| 550 | |
| 551 | lzma_end(&strm); |
| 552 | *readsize = inFileSize; |
| 553 | |
| 554 | return outFileSize; |
| 555 | } |
| 556 | #endif |
| 557 | |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 558 | #ifdef ZSTD_LZ4COMPRESS |
| 559 | static int FIO_LZ4_GetBlockSize_FromBlockId (int id) { return (1 << (8 + (2 * id))); } |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 560 | static unsigned long long FIO_compressLz4Frame(cRess_t* ress, |
| 561 | const char* srcFileName, U64 const srcFileSize, |
| 562 | int compressionLevel, U64* readsize) |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 563 | { |
| 564 | unsigned long long inFileSize = 0, outFileSize = 0; |
| 565 | |
| 566 | LZ4F_preferences_t prefs; |
Sean Purcell | 2c4b6fe | 2017-04-25 11:00:54 -0700 | [diff] [blame] | 567 | LZ4F_compressionContext_t ctx; |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 568 | |
| 569 | LZ4F_errorCode_t const errorCode = LZ4F_createCompressionContext(&ctx, LZ4F_VERSION); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 570 | if (LZ4F_isError(errorCode)) |
| 571 | EXM_THROW(31, "zstd: failed to create lz4 compression context"); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 572 | |
| 573 | memset(&prefs, 0, sizeof(prefs)); |
| 574 | |
Sean Purcell | eab41c1 | 2017-04-26 10:17:38 -0700 | [diff] [blame] | 575 | #if LZ4_VERSION_NUMBER <= 10600 |
| 576 | #define LZ4F_blockIndependent blockIndependent |
| 577 | #define LZ4F_max4MB max4MB |
| 578 | #endif |
| 579 | |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 580 | prefs.autoFlush = 1; |
| 581 | prefs.compressionLevel = compressionLevel; |
Sean Purcell | eab41c1 | 2017-04-26 10:17:38 -0700 | [diff] [blame] | 582 | prefs.frameInfo.blockMode = LZ4F_blockIndependent; /* stick to defaults for lz4 cli */ |
| 583 | prefs.frameInfo.blockSizeID = LZ4F_max4MB; |
Sean Purcell | 2c4b6fe | 2017-04-25 11:00:54 -0700 | [diff] [blame] | 584 | prefs.frameInfo.contentChecksumFlag = (contentChecksum_t)g_checksumFlag; |
| 585 | #if LZ4_VERSION_NUMBER >= 10600 |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 586 | prefs.frameInfo.contentSize = srcFileSize; |
Sean Purcell | 2c4b6fe | 2017-04-25 11:00:54 -0700 | [diff] [blame] | 587 | #endif |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 588 | |
| 589 | { |
Sean Purcell | eab41c1 | 2017-04-26 10:17:38 -0700 | [diff] [blame] | 590 | size_t blockSize = FIO_LZ4_GetBlockSize_FromBlockId(LZ4F_max4MB); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 591 | size_t readSize; |
| 592 | size_t headerSize = LZ4F_compressBegin(ctx, ress->dstBuffer, ress->dstBufferSize, &prefs); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 593 | if (LZ4F_isError(headerSize)) |
| 594 | EXM_THROW(33, "File header generation failed : %s", |
| 595 | LZ4F_getErrorName(headerSize)); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 596 | { size_t const sizeCheck = fwrite(ress->dstBuffer, 1, headerSize, ress->dstFile); |
| 597 | if (sizeCheck!=headerSize) EXM_THROW(34, "Write error : cannot write header"); } |
| 598 | outFileSize += headerSize; |
| 599 | |
| 600 | /* Read first block */ |
| 601 | readSize = fread(ress->srcBuffer, (size_t)1, (size_t)blockSize, ress->srcFile); |
| 602 | inFileSize += readSize; |
| 603 | |
| 604 | /* Main Loop */ |
| 605 | while (readSize>0) { |
| 606 | size_t outSize; |
| 607 | |
| 608 | /* Compress Block */ |
| 609 | outSize = LZ4F_compressUpdate(ctx, ress->dstBuffer, ress->dstBufferSize, ress->srcBuffer, readSize, NULL); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 610 | if (LZ4F_isError(outSize)) |
| 611 | EXM_THROW(35, "zstd: %s: lz4 compression failed : %s", |
| 612 | srcFileName, LZ4F_getErrorName(outSize)); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 613 | outFileSize += outSize; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 614 | if (!srcFileSize) |
| 615 | DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", |
| 616 | (U32)(inFileSize>>20), |
| 617 | (double)outFileSize/inFileSize*100) |
| 618 | else |
| 619 | DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", |
| 620 | (U32)(inFileSize>>20), (U32)(srcFileSize>>20), |
| 621 | (double)outFileSize/inFileSize*100); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 622 | |
| 623 | /* Write Block */ |
| 624 | { size_t const sizeCheck = fwrite(ress->dstBuffer, 1, outSize, ress->dstFile); |
| 625 | if (sizeCheck!=outSize) EXM_THROW(36, "Write error : cannot write compressed block"); } |
| 626 | |
| 627 | /* Read next block */ |
| 628 | readSize = fread(ress->srcBuffer, (size_t)1, (size_t)blockSize, ress->srcFile); |
| 629 | inFileSize += readSize; |
| 630 | } |
| 631 | if (ferror(ress->srcFile)) EXM_THROW(37, "Error reading %s ", srcFileName); |
| 632 | |
| 633 | /* End of Stream mark */ |
| 634 | headerSize = LZ4F_compressEnd(ctx, ress->dstBuffer, ress->dstBufferSize, NULL); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 635 | if (LZ4F_isError(headerSize)) |
| 636 | EXM_THROW(38, "zstd: %s: lz4 end of file generation failed : %s", |
| 637 | srcFileName, LZ4F_getErrorName(headerSize)); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 638 | |
| 639 | { size_t const sizeCheck = fwrite(ress->dstBuffer, 1, headerSize, ress->dstFile); |
| 640 | if (sizeCheck!=headerSize) EXM_THROW(39, "Write error : cannot write end of stream"); } |
| 641 | outFileSize += headerSize; |
| 642 | } |
| 643 | |
| 644 | *readsize = inFileSize; |
| 645 | LZ4F_freeCompressionContext(ctx); |
| 646 | |
| 647 | return outFileSize; |
| 648 | } |
| 649 | #endif |
| 650 | |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 651 | |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 652 | /*! FIO_compressFilename_internal() : |
Yann Collet | 8a1d1a6 | 2016-03-10 21:02:25 +0100 | [diff] [blame] | 653 | * same as FIO_compressFilename_extRess(), with `ress.desFile` already opened. |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 654 | * @return : 0 : compression completed correctly, |
| 655 | * 1 : missing or pb opening srcFileName |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 656 | */ |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 657 | static int FIO_compressFilename_internal(cRess_t ress, |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 658 | const char* dstFileName, const char* srcFileName, int compressionLevel) |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 659 | { |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 660 | FILE* const srcFile = ress.srcFile; |
| 661 | FILE* const dstFile = ress.dstFile; |
Yann Collet | b44be74 | 2016-03-26 20:52:14 +0100 | [diff] [blame] | 662 | U64 readsize = 0; |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 663 | U64 compressedfilesize = 0; |
inikep | 69fcd7c | 2016-04-28 12:23:33 +0200 | [diff] [blame] | 664 | U64 const fileSize = UTIL_getFileSize(srcFileName); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 665 | |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 666 | switch (g_compressionType) { |
| 667 | case FIO_zstdCompression: |
| 668 | break; |
Yann Collet | 2e63a87 | 2017-05-02 15:40:42 -0700 | [diff] [blame] | 669 | |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 670 | case FIO_gzipCompression: |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 671 | #ifdef ZSTD_GZCOMPRESS |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 672 | compressedfilesize = FIO_compressGzFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize); |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 673 | #else |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 674 | (void)compressionLevel; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 675 | EXM_THROW(20, "zstd: %s: file cannot be compressed as gzip (zstd compiled without ZSTD_GZCOMPRESS) -- ignored \n", |
| 676 | srcFileName); |
Przemyslaw Skibinski | cb56306 | 2017-02-08 17:37:14 +0100 | [diff] [blame] | 677 | #endif |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 678 | goto finish; |
Yann Collet | 2e63a87 | 2017-05-02 15:40:42 -0700 | [diff] [blame] | 679 | |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 680 | case FIO_xzCompression: |
| 681 | case FIO_lzmaCompression: |
| 682 | #ifdef ZSTD_LZMACOMPRESS |
| 683 | compressedfilesize = FIO_compressLzmaFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize, g_compressionType==FIO_lzmaCompression); |
| 684 | #else |
| 685 | (void)compressionLevel; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 686 | EXM_THROW(20, "zstd: %s: file cannot be compressed as xz/lzma (zstd compiled without ZSTD_LZMACOMPRESS) -- ignored \n", |
| 687 | srcFileName); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 688 | #endif |
Yann Collet | 2e63a87 | 2017-05-02 15:40:42 -0700 | [diff] [blame] | 689 | goto finish; |
| 690 | |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 691 | case FIO_lz4Compression: |
| 692 | #ifdef ZSTD_LZ4COMPRESS |
| 693 | compressedfilesize = FIO_compressLz4Frame(&ress, srcFileName, fileSize, compressionLevel, &readsize); |
| 694 | #else |
| 695 | (void)compressionLevel; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 696 | EXM_THROW(20, "zstd: %s: file cannot be compressed as lz4 (zstd compiled without ZSTD_LZ4COMPRESS) -- ignored \n", |
| 697 | srcFileName); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 698 | #endif |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 699 | goto finish; |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 700 | } |
| 701 | |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 702 | /* init */ |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 703 | #ifdef ZSTD_MULTITHREAD |
| 704 | { size_t const resetError = ZSTDMT_resetCStream(ress.cctx, fileSize); |
| 705 | #else |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 706 | { size_t const resetError = ZSTD_resetCStream(ress.cctx, fileSize); |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 707 | #endif |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 708 | if (ZSTD_isError(resetError)) |
| 709 | EXM_THROW(21, "Error initializing compression : %s", |
| 710 | ZSTD_getErrorName(resetError)); |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 711 | } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 712 | |
| 713 | /* Main compression loop */ |
Yann Collet | 1c8e194 | 2016-01-26 16:31:22 +0100 | [diff] [blame] | 714 | while (1) { |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 715 | /* Fill input Buffer */ |
Yann Collet | b44be74 | 2016-03-26 20:52:14 +0100 | [diff] [blame] | 716 | size_t const inSize = fread(ress.srcBuffer, (size_t)1, ress.srcBufferSize, srcFile); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 717 | if (inSize==0) break; |
Yann Collet | b44be74 | 2016-03-26 20:52:14 +0100 | [diff] [blame] | 718 | readsize += inSize; |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 719 | |
Yann Collet | 53e17fb | 2016-08-17 01:39:22 +0200 | [diff] [blame] | 720 | { ZSTD_inBuffer inBuff = { ress.srcBuffer, inSize, 0 }; |
Yann Collet | 5c42d0e | 2017-04-11 16:57:32 -0700 | [diff] [blame] | 721 | while (inBuff.pos != inBuff.size) { |
| 722 | ZSTD_outBuffer outBuff = { ress.dstBuffer, ress.dstBufferSize, 0 }; |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 723 | #ifdef ZSTD_MULTITHREAD |
| 724 | size_t const result = ZSTDMT_compressStream(ress.cctx, &outBuff, &inBuff); |
| 725 | #else |
| 726 | size_t const result = ZSTD_compressStream(ress.cctx, &outBuff, &inBuff); |
| 727 | #endif |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 728 | if (ZSTD_isError(result)) |
| 729 | EXM_THROW(23, "Compression error : %s ", ZSTD_getErrorName(result)); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 730 | |
Yann Collet | 943cff9 | 2017-01-25 12:31:07 -0800 | [diff] [blame] | 731 | /* Write compressed stream */ |
| 732 | if (outBuff.pos) { |
| 733 | size_t const sizeCheck = fwrite(ress.dstBuffer, 1, outBuff.pos, dstFile); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 734 | if (sizeCheck!=outBuff.pos) |
| 735 | EXM_THROW(25, "Write error : cannot write compressed block into %s", dstFileName); |
Yann Collet | 943cff9 | 2017-01-25 12:31:07 -0800 | [diff] [blame] | 736 | compressedfilesize += outBuff.pos; |
| 737 | } } } |
Yann Collet | 5c42d0e | 2017-04-11 16:57:32 -0700 | [diff] [blame] | 738 | if (g_nbThreads > 1) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 739 | if (!fileSize) |
| 740 | DISPLAYUPDATE(2, "\rRead : %u MB", (U32)(readsize>>20)) |
| 741 | else |
| 742 | DISPLAYUPDATE(2, "\rRead : %u / %u MB", |
| 743 | (U32)(readsize>>20), (U32)(fileSize>>20)); |
Yann Collet | 5c42d0e | 2017-04-11 16:57:32 -0700 | [diff] [blame] | 744 | } else { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 745 | if (!fileSize) |
| 746 | DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", |
| 747 | (U32)(readsize>>20), |
| 748 | (double)compressedfilesize/readsize*100) |
| 749 | else |
| 750 | DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", |
| 751 | (U32)(readsize>>20), (U32)(fileSize>>20), |
| 752 | (double)compressedfilesize/readsize*100); |
Yann Collet | 5c42d0e | 2017-04-11 16:57:32 -0700 | [diff] [blame] | 753 | } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 754 | } |
| 755 | |
| 756 | /* End of Frame */ |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 757 | { size_t result = 1; |
| 758 | while (result!=0) { /* note : is there any possibility of endless loop ? */ |
| 759 | ZSTD_outBuffer outBuff = { ress.dstBuffer, ress.dstBufferSize, 0 }; |
| 760 | #ifdef ZSTD_MULTITHREAD |
| 761 | result = ZSTDMT_endStream(ress.cctx, &outBuff); |
| 762 | #else |
| 763 | result = ZSTD_endStream(ress.cctx, &outBuff); |
| 764 | #endif |
| 765 | if (ZSTD_isError(result)) EXM_THROW(26, "Compression error during frame end : %s", ZSTD_getErrorName(result)); |
| 766 | { size_t const sizeCheck = fwrite(ress.dstBuffer, 1, outBuff.pos, dstFile); |
| 767 | if (sizeCheck!=outBuff.pos) EXM_THROW(27, "Write error : cannot write frame end into %s", dstFileName); } |
| 768 | compressedfilesize += outBuff.pos; |
| 769 | } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 770 | } |
| 771 | |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 772 | finish: |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 773 | /* Status */ |
| 774 | DISPLAYLEVEL(2, "\r%79s\r", ""); |
David Lam | 0f27045 | 2016-08-13 11:26:21 -0700 | [diff] [blame] | 775 | DISPLAYLEVEL(2,"%-20s :%6.2f%% (%6llu => %6llu bytes, %s) \n", srcFileName, |
Yann Collet | 44f684d | 2016-07-13 19:30:40 +0200 | [diff] [blame] | 776 | (double)compressedfilesize/(readsize+(!readsize) /* avoid div by zero */ )*100, |
| 777 | (unsigned long long)readsize, (unsigned long long) compressedfilesize, |
| 778 | dstFileName); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 779 | |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 780 | return 0; |
| 781 | } |
| 782 | |
| 783 | |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 784 | /*! FIO_compressFilename_srcFile() : |
| 785 | * note : ress.destFile already opened |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 786 | * @return : 0 : compression completed correctly, |
| 787 | * 1 : missing or pb opening srcFileName |
| 788 | */ |
| 789 | static int FIO_compressFilename_srcFile(cRess_t ress, |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 790 | const char* dstFileName, const char* srcFileName, |
| 791 | int compressionLevel) |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 792 | { |
| 793 | int result; |
| 794 | |
| 795 | /* File check */ |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 796 | if (UTIL_isDirectory(srcFileName)) { |
| 797 | DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName); |
| 798 | return 1; |
| 799 | } |
Przemyslaw Skibinski | 64fa2db | 2017-01-25 13:02:33 +0100 | [diff] [blame] | 800 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 801 | ress.srcFile = FIO_openSrcFile(srcFileName); |
| 802 | if (!ress.srcFile) return 1; /* srcFile could not be opened */ |
| 803 | |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 804 | result = FIO_compressFilename_internal(ress, dstFileName, srcFileName, compressionLevel); |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 805 | |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 806 | fclose(ress.srcFile); |
Yann Collet | dccd6b6 | 2017-02-27 15:57:50 -0800 | [diff] [blame] | 807 | if (g_removeSrcFile /* --rm */ && !result && strcmp(srcFileName, stdinmark)) { |
| 808 | if (remove(srcFileName)) |
| 809 | EXM_THROW(1, "zstd: %s: %s", srcFileName, strerror(errno)); |
| 810 | } |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 811 | return result; |
| 812 | } |
| 813 | |
| 814 | |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 815 | /*! FIO_compressFilename_dstFile() : |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 816 | * @return : 0 : compression completed correctly, |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 817 | * 1 : pb |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 818 | */ |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 819 | static int FIO_compressFilename_dstFile(cRess_t ress, |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 820 | const char* dstFileName, |
| 821 | const char* srcFileName, |
| 822 | int compressionLevel) |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 823 | { |
| 824 | int result; |
Przemyslaw Skibinski | fcf22e3 | 2016-11-02 14:08:07 +0100 | [diff] [blame] | 825 | stat_t statbuf; |
| 826 | int stat_result = 0; |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 827 | |
| 828 | ress.dstFile = FIO_openDstFile(dstFileName); |
Yann Collet | 43eeea4 | 2016-09-15 15:38:44 +0200 | [diff] [blame] | 829 | if (ress.dstFile==NULL) return 1; /* could not open dstFileName */ |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 830 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 831 | if (strcmp (srcFileName, stdinmark) && UTIL_getFileStat(srcFileName, &statbuf)) |
| 832 | stat_result = 1; |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 833 | result = FIO_compressFilename_srcFile(ress, dstFileName, srcFileName, compressionLevel); |
Chip Turner | 6de382c | 2016-03-13 22:24:46 -0700 | [diff] [blame] | 834 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 835 | if (fclose(ress.dstFile)) { /* error closing dstFile */ |
| 836 | DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno)); |
| 837 | result=1; |
| 838 | } |
| 839 | if (result!=0) { /* remove operation artefact */ |
| 840 | if (remove(dstFileName)) |
| 841 | EXM_THROW(1, "zstd: %s: %s", dstFileName, strerror(errno)); |
| 842 | } |
| 843 | else if (strcmp (dstFileName, stdoutmark) && stat_result) |
| 844 | UTIL_setFileStat(dstFileName, &statbuf); |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 845 | return result; |
| 846 | } |
| 847 | |
| 848 | |
Yann Collet | 9d90922 | 2015-12-17 14:09:55 +0100 | [diff] [blame] | 849 | int FIO_compressFilename(const char* dstFileName, const char* srcFileName, |
Przemyslaw Skibinski | 8349d67 | 2016-12-13 13:24:59 +0100 | [diff] [blame] | 850 | const char* dictFileName, int compressionLevel, ZSTD_compressionParameters* comprParams) |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 851 | { |
Yann Collet | 6381e99 | 2016-05-31 02:29:45 +0200 | [diff] [blame] | 852 | clock_t const start = clock(); |
Yann Collet | 993060e | 2016-09-21 16:46:08 +0200 | [diff] [blame] | 853 | U64 const srcSize = UTIL_getFileSize(srcFileName); |
Sean Purcell | 0f5c95a | 2017-02-07 16:33:48 -0800 | [diff] [blame] | 854 | int const regFile = UTIL_isRegFile(srcFileName); |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 855 | |
Sean Purcell | 0f5c95a | 2017-02-07 16:33:48 -0800 | [diff] [blame] | 856 | cRess_t const ress = FIO_createCResources(dictFileName, compressionLevel, srcSize, regFile, comprParams); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 857 | int const result = FIO_compressFilename_dstFile(ress, dstFileName, srcFileName, compressionLevel); |
Yann Collet | 9d90922 | 2015-12-17 14:09:55 +0100 | [diff] [blame] | 858 | |
Yann Collet | b71adf4 | 2016-07-02 01:05:31 +0200 | [diff] [blame] | 859 | double const seconds = (double)(clock() - start) / CLOCKS_PER_SEC; |
| 860 | DISPLAYLEVEL(4, "Completed in %.2f sec \n", seconds); |
| 861 | |
| 862 | FIO_freeCResources(ress); |
| 863 | return result; |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 864 | } |
| 865 | |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 866 | typedef struct { |
| 867 | int numActualFrames; |
| 868 | int numSkippableFrames; |
Paul Cruz | fc428ab | 2017-06-14 13:26:19 -0700 | [diff] [blame] | 869 | unsigned long long decompressedSize; |
Paul Cruz | 6e02044 | 2017-06-12 15:51:59 -0700 | [diff] [blame] | 870 | int canComputeDecompSize; |
Paul Cruz | fc428ab | 2017-06-14 13:26:19 -0700 | [diff] [blame] | 871 | unsigned long long compressedSize; |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 872 | int usesCheck; |
| 873 | } fileInfo_t; |
| 874 | |
| 875 | /* |
| 876 | * Reads information from file, stores in *info |
| 877 | * if successful, returns 0, otherwise returns 1 |
| 878 | */ |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 879 | static int getFileInfo(fileInfo_t* info, const char* inFileName){ |
| 880 | FILE* const srcFile = FIO_openSrcFile(inFileName); |
| 881 | if (srcFile == NULL) { |
| 882 | DISPLAY("Error: could not open source file %s\n", inFileName); |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 883 | return 1; |
| 884 | } |
Paul Cruz | fc428ab | 2017-06-14 13:26:19 -0700 | [diff] [blame] | 885 | info->compressedSize = (unsigned long long)UTIL_getFileSize(inFileName); |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 886 | info->decompressedSize = 0; |
| 887 | info->numActualFrames = 0; |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 888 | info->numSkippableFrames = 0; |
Paul Cruz | 6e02044 | 2017-06-12 15:51:59 -0700 | [diff] [blame] | 889 | info->canComputeDecompSize = 1; |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 890 | /* begin analyzing frame */ |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 891 | for( ; ; ){ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 892 | BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; |
| 893 | size_t const numBytesRead = fread(headerBuffer, 1, sizeof(headerBuffer), srcFile); |
| 894 | if (numBytesRead < ZSTD_frameHeaderSize_min) { |
| 895 | if(feof(srcFile)){ |
| 896 | DISPLAY("ran out of files\n"); |
| 897 | break; |
| 898 | } |
| 899 | else{ |
| 900 | DISPLAY("Error: did not reach end of file but ran out of frames\n"); |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 901 | fclose(srcFile); |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 902 | return 1; |
Paul Cruz | 618a7b6 | 2017-06-12 16:53:50 -0700 | [diff] [blame] | 903 | } |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 904 | } |
| 905 | U32 const magicNumber = MEM_readLE32(headerBuffer); |
| 906 | if (magicNumber == ZSTD_MAGICNUMBER) { |
| 907 | U64 const frameContentSize = ZSTD_getFrameContentSize(headerBuffer, numBytesRead); |
| 908 | if (frameContentSize == ZSTD_CONTENTSIZE_ERROR || frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN) { |
| 909 | info->canComputeDecompSize = 0; |
| 910 | DISPLAY("could not compute decompressed size\n"); |
| 911 | } |
| 912 | else { |
| 913 | info->decompressedSize += frameContentSize; |
| 914 | } |
| 915 | size_t const headerSize = ZSTD_frameHeaderSize(headerBuffer, numBytesRead); |
| 916 | if (ZSTD_isError(headerSize)) { |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 917 | DISPLAY("Error: could not determine frame header size\n"); |
Paul Cruz | 4495e9a | 2017-06-15 15:02:54 -0700 | [diff] [blame^] | 918 | fclose(srcFile); |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 919 | return 1; |
| 920 | } |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 921 | |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 922 | { |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 923 | /* go back to the beginning of the frame */ |
| 924 | int const ret = fseek(srcFile, -numBytesRead, SEEK_CUR); |
| 925 | if (ret != 0) { |
| 926 | DISPLAY("Error: could not rewind to beginning of frame\n"); |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 927 | fclose(srcFile); |
| 928 | return 1; |
| 929 | } |
| 930 | } |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 931 | |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 932 | { |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 933 | /* skip frame header */ |
| 934 | int const ret = fseek(srcFile, headerSize, SEEK_CUR); |
| 935 | if (ret != 0) { |
| 936 | DISPLAY("Error: could not skip header\n"); |
| 937 | fclose(srcFile); |
| 938 | return 1; |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 939 | } |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 940 | } |
| 941 | |
| 942 | /* skip the rest of the blocks in the frame */ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 943 | { |
| 944 | int lastBlock = 0; |
| 945 | int readBytes = 0; |
| 946 | do{ |
| 947 | BYTE blockHeaderBuffer[3]; |
| 948 | U32 blockHeader; |
| 949 | int blockSize; |
| 950 | readBytes = fread(blockHeaderBuffer, 1, 3, srcFile); |
| 951 | if (readBytes != 3) { |
| 952 | DISPLAY("There was a problem reading the block header\n"); |
| 953 | fclose(srcFile); |
| 954 | return 1; |
| 955 | } |
| 956 | blockHeader = MEM_readLE24(blockHeaderBuffer); |
| 957 | lastBlock = blockHeader & 1; |
| 958 | blockSize = (blockHeader - (blockHeader & 7)) >> 3; |
| 959 | fseek(srcFile, blockSize, SEEK_CUR); |
| 960 | }while (lastBlock != 1); |
| 961 | } |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 962 | { |
| 963 | /* check if checksum is used */ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 964 | BYTE frameHeaderDescriptor = headerBuffer[4]; |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 965 | int contentChecksumFlag = (frameHeaderDescriptor & (1 << 2)) >> 2; |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 966 | if (contentChecksumFlag) { |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 967 | info->usesCheck = 1; |
| 968 | } |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 969 | if (contentChecksumFlag) { |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 970 | fseek(srcFile, 4, SEEK_CUR); |
| 971 | } |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 972 | } |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 973 | info->numActualFrames++; |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 974 | } |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 975 | else if (magicNumber == ZSTD_MAGIC_SKIPPABLE_START) { |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 976 | BYTE frameSizeBuffer[4]; |
Paul Cruz | 28dbf8e | 2017-06-12 17:09:08 -0700 | [diff] [blame] | 977 | long frameSize; |
Paul Cruz | 618a7b6 | 2017-06-12 16:53:50 -0700 | [diff] [blame] | 978 | size_t readBytes = fread(frameSizeBuffer, 1, 4, srcFile); |
Paul Cruz | c828b52 | 2017-06-12 17:19:53 -0700 | [diff] [blame] | 979 | info->numSkippableFrames++; |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 980 | if (readBytes != 4) { |
Paul Cruz | 618a7b6 | 2017-06-12 16:53:50 -0700 | [diff] [blame] | 981 | DISPLAY("There was an error reading skippable frame size"); |
| 982 | exit(1); |
| 983 | } |
Paul Cruz | 28dbf8e | 2017-06-12 17:09:08 -0700 | [diff] [blame] | 984 | frameSize = MEM_readLE32(frameSizeBuffer); |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 985 | { |
| 986 | int const ret = fseek(srcFile, frameSize, SEEK_CUR); |
| 987 | if (ret != 0) { |
| 988 | DISPLAY("Error: could not find end of skippable frame\n"); |
| 989 | fclose(srcFile); |
| 990 | return 1; |
| 991 | } |
| 992 | } |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 993 | fseek(srcFile, frameSize, SEEK_CUR); |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 994 | info->numSkippableFrames++; |
Paul Cruz | 9cb602e | 2017-06-12 15:22:48 -0700 | [diff] [blame] | 995 | } |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 996 | } |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 997 | fclose(srcFile); |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 998 | return 0; |
| 999 | } |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1000 | void displayInfo(const char* inFileName, fileInfo_t* info, int displayLevel){ |
Paul Cruz | e208992 | 2017-06-15 12:27:32 -0700 | [diff] [blame] | 1001 | double const compressedSizeMB = (double)info->compressedSize/(1 MB); |
| 1002 | double const decompressedSizeMB = (double)info->decompressedSize/(1 MB); |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1003 | |
| 1004 | if(displayLevel<=2){ |
| 1005 | if(info->usesCheck && info->canComputeDecompSize){ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1006 | DISPLAYOUT("Skippable Non-Skippable Compressed Uncompressed Ratio Check Filename\n"); |
| 1007 | DISPLAYOUT("%9d %13d %7.2f MB %7.2f MB %5.3f XXH64 %s\n", |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1008 | info->numSkippableFrames, info->numActualFrames, compressedSizeMB, decompressedSizeMB, |
| 1009 | compressedSizeMB/decompressedSizeMB, inFileName); |
| 1010 | } |
| 1011 | else if(!info->usesCheck){ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1012 | DISPLAYOUT("Skippable Non-Skippable Compressed Uncompressed Ratio Check Filename\n"); |
| 1013 | DISPLAYOUT("%9d %13d %7.2f MB %7.2f MB %5.3f %s\n", |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1014 | info->numSkippableFrames, info->numActualFrames, compressedSizeMB, decompressedSizeMB, |
| 1015 | compressedSizeMB/decompressedSizeMB, inFileName); |
| 1016 | } |
| 1017 | else if(!info->canComputeDecompSize){ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1018 | DISPLAYOUT("Skippable Non-Skippable Compressed Uncompressed Ratio Check Filename\n"); |
| 1019 | DISPLAYOUT("%9d %13d %7.2f MB XXH64 %s\n", |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1020 | info->numSkippableFrames, info->numActualFrames, compressedSizeMB, inFileName); |
| 1021 | } |
| 1022 | else{ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1023 | DISPLAYOUT("Skippable Non-Skippable Filename\n"); |
| 1024 | DISPLAYOUT("%9d %13d %7.2f MB %s\n", |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1025 | info->numSkippableFrames, info->numActualFrames, compressedSizeMB, inFileName); |
| 1026 | } |
| 1027 | } |
| 1028 | else{ |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1029 | DISPLAYOUT("# Zstandard Frames: %d\n", info->numActualFrames); |
| 1030 | DISPLAYOUT("# Skippable Frames: %d\n", info->numSkippableFrames); |
| 1031 | DISPLAYOUT("Compressed Size: %.2f MB (%llu B)\n", compressedSizeMB, info->compressedSize); |
| 1032 | if (info->canComputeDecompSize) { |
| 1033 | DISPLAYOUT("Decompressed Size: %.2f MB (%llu B)\n", decompressedSizeMB, info->decompressedSize); |
| 1034 | DISPLAYOUT("Ratio: %.4f\n", compressedSizeMB/decompressedSizeMB); |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1035 | } |
Paul Cruz | a9b77c8 | 2017-06-15 14:13:28 -0700 | [diff] [blame] | 1036 | if (info->usesCheck) { |
| 1037 | DISPLAYOUT("Check: XXH64\n"); |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1038 | } |
| 1039 | } |
Paul Cruz | 4495e9a | 2017-06-15 15:02:54 -0700 | [diff] [blame^] | 1040 | |
| 1041 | DISPLAYOUT("\n"); |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1042 | } |
Paul Cruz | 786b7ca | 2017-06-12 13:46:39 -0700 | [diff] [blame] | 1043 | |
| 1044 | int FIO_listFile(const char* inFileName, int displayLevel){ |
Paul Cruz | 81fa33b | 2017-06-13 10:06:49 -0700 | [diff] [blame] | 1045 | DISPLAY("File: %s\n", inFileName); |
Paul Cruz | 4495e9a | 2017-06-15 15:02:54 -0700 | [diff] [blame^] | 1046 | fileInfo_t* info = (fileInfo_t*)malloc(sizeof(fileInfo_t)); |
| 1047 | int error = getFileInfo(info, inFileName); |
| 1048 | if (error == 1) { |
| 1049 | DISPLAY("An error occurred with getting file info\n"); |
Paul Cruz | a3d54cf | 2017-06-12 10:58:34 -0700 | [diff] [blame] | 1050 | return 1; |
Paul Cruz | 4128f67 | 2017-06-05 15:00:06 -0700 | [diff] [blame] | 1051 | } |
Paul Cruz | 4495e9a | 2017-06-15 15:02:54 -0700 | [diff] [blame^] | 1052 | displayInfo(inFileName, info, displayLevel); |
| 1053 | free(info); |
Paul Cruz | 901435e | 2017-06-05 14:45:31 -0700 | [diff] [blame] | 1054 | return 0; |
| 1055 | } |
Yann Collet | 4856a00 | 2015-01-24 01:58:16 +0100 | [diff] [blame] | 1056 | |
Yann Collet | 9d90922 | 2015-12-17 14:09:55 +0100 | [diff] [blame] | 1057 | int FIO_compressMultipleFilenames(const char** inFileNamesTable, unsigned nbFiles, |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1058 | const char* suffix, |
Przemyslaw Skibinski | 8349d67 | 2016-12-13 13:24:59 +0100 | [diff] [blame] | 1059 | const char* dictFileName, int compressionLevel, |
| 1060 | ZSTD_compressionParameters* comprParams) |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1061 | { |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1062 | int missed_files = 0; |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1063 | size_t dfnSize = FNSPACE; |
Yann Collet | 44f684d | 2016-07-13 19:30:40 +0200 | [diff] [blame] | 1064 | char* dstFileName = (char*)malloc(FNSPACE); |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1065 | size_t const suffixSize = suffix ? strlen(suffix) : 0; |
Yann Collet | 993060e | 2016-09-21 16:46:08 +0200 | [diff] [blame] | 1066 | U64 const srcSize = (nbFiles != 1) ? 0 : UTIL_getFileSize(inFileNamesTable[0]) ; |
Sean Purcell | 0f5c95a | 2017-02-07 16:33:48 -0800 | [diff] [blame] | 1067 | int const regFile = (nbFiles != 1) ? 0 : UTIL_isRegFile(inFileNamesTable[0]); |
| 1068 | cRess_t ress = FIO_createCResources(dictFileName, compressionLevel, srcSize, regFile, comprParams); |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1069 | |
| 1070 | /* init */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1071 | if (dstFileName==NULL) |
| 1072 | EXM_THROW(27, "FIO_compressMultipleFilenames : allocation error for dstFileName"); |
| 1073 | if (suffix == NULL) |
| 1074 | EXM_THROW(28, "FIO_compressMultipleFilenames : dst unknown"); /* should never happen */ |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1075 | |
| 1076 | /* loop on each file */ |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 1077 | if (!strcmp(suffix, stdoutmark)) { |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1078 | unsigned u; |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 1079 | ress.dstFile = stdout; |
inikep | 60af95d | 2016-05-19 10:29:49 +0200 | [diff] [blame] | 1080 | SET_BINARY_MODE(stdout); |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 1081 | for (u=0; u<nbFiles; u++) |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 1082 | missed_files += FIO_compressFilename_srcFile(ress, stdoutmark, inFileNamesTable[u], compressionLevel); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1083 | if (fclose(ress.dstFile)) |
| 1084 | EXM_THROW(29, "Write error : cannot properly close stdout"); |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 1085 | } else { |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1086 | unsigned u; |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 1087 | for (u=0; u<nbFiles; u++) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1088 | size_t const ifnSize = strlen(inFileNamesTable[u]); |
| 1089 | if (dfnSize <= ifnSize+suffixSize+1) { /* resize name buffer */ |
| 1090 | free(dstFileName); |
| 1091 | dfnSize = ifnSize + 20; |
| 1092 | dstFileName = (char*)malloc(dfnSize); |
| 1093 | if (!dstFileName) |
| 1094 | EXM_THROW(30, "zstd: %s", strerror(errno)); |
| 1095 | } |
Yann Collet | f062436 | 2016-02-12 15:56:46 +0100 | [diff] [blame] | 1096 | strcpy(dstFileName, inFileNamesTable[u]); |
| 1097 | strcat(dstFileName, suffix); |
Przemyslaw Skibinski | 02018c8 | 2017-02-08 16:54:23 +0100 | [diff] [blame] | 1098 | missed_files += FIO_compressFilename_dstFile(ress, dstFileName, inFileNamesTable[u], compressionLevel); |
Yann Collet | 459a6b7 | 2016-02-15 20:37:23 +0100 | [diff] [blame] | 1099 | } } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1100 | |
| 1101 | /* Close & Free */ |
| 1102 | FIO_freeCResources(ress); |
| 1103 | free(dstFileName); |
| 1104 | |
| 1105 | return missed_files; |
| 1106 | } |
| 1107 | |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1108 | #endif /* #ifndef ZSTD_NOCOMPRESS */ |
inikep | 3c7c352 | 2016-04-22 13:59:05 +0200 | [diff] [blame] | 1109 | |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1110 | |
inikep | db39643 | 2016-04-22 18:22:30 +0200 | [diff] [blame] | 1111 | |
| 1112 | #ifndef ZSTD_NODECOMPRESS |
| 1113 | |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1114 | /* ************************************************************************** |
| 1115 | * Decompression |
| 1116 | ****************************************************************************/ |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1117 | typedef struct { |
| 1118 | void* srcBuffer; |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1119 | size_t srcBufferLoaded; |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1120 | size_t srcBufferSize; |
| 1121 | void* dstBuffer; |
| 1122 | size_t dstBufferSize; |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 1123 | ZSTD_DStream* dctx; |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1124 | FILE* dstFile; |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1125 | } dRess_t; |
| 1126 | |
| 1127 | static dRess_t FIO_createDResources(const char* dictFileName) |
| 1128 | { |
| 1129 | dRess_t ress; |
Yann Collet | 5e80dd3 | 2016-07-13 17:38:39 +0200 | [diff] [blame] | 1130 | memset(&ress, 0, sizeof(ress)); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1131 | |
Yann Collet | 5e80dd3 | 2016-07-13 17:38:39 +0200 | [diff] [blame] | 1132 | /* Allocation */ |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 1133 | ress.dctx = ZSTD_createDStream(); |
| 1134 | if (ress.dctx==NULL) EXM_THROW(60, "Can't create ZSTD_DStream"); |
Yann Collet | bb00274 | 2017-01-25 16:25:38 -0800 | [diff] [blame] | 1135 | ZSTD_setDStreamParameter(ress.dctx, DStream_p_maxWindowSize, g_memLimit); |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 1136 | ress.srcBufferSize = ZSTD_DStreamInSize(); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1137 | ress.srcBuffer = malloc(ress.srcBufferSize); |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 1138 | ress.dstBufferSize = ZSTD_DStreamOutSize(); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1139 | ress.dstBuffer = malloc(ress.dstBufferSize); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1140 | if (!ress.srcBuffer || !ress.dstBuffer) |
| 1141 | EXM_THROW(61, "Allocation error : not enough memory"); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1142 | |
| 1143 | /* dictionary */ |
Yann Collet | 3ecbe6a | 2016-09-14 17:26:59 +0200 | [diff] [blame] | 1144 | { void* dictBuffer; |
Yann Collet | 0e30059 | 2017-04-11 14:41:02 -0700 | [diff] [blame] | 1145 | size_t const dictBufferSize = FIO_createDictBuffer(&dictBuffer, dictFileName); |
Yann Collet | 3ecbe6a | 2016-09-14 17:26:59 +0200 | [diff] [blame] | 1146 | size_t const initError = ZSTD_initDStream_usingDict(ress.dctx, dictBuffer, dictBufferSize); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1147 | if (ZSTD_isError(initError)) |
| 1148 | EXM_THROW(61, "ZSTD_initDStream_usingDict error : %s", ZSTD_getErrorName(initError)); |
Yann Collet | 3ecbe6a | 2016-09-14 17:26:59 +0200 | [diff] [blame] | 1149 | free(dictBuffer); |
| 1150 | } |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1151 | |
| 1152 | return ress; |
| 1153 | } |
| 1154 | |
| 1155 | static void FIO_freeDResources(dRess_t ress) |
| 1156 | { |
Yann Collet | 6263ba5 | 2016-08-13 23:45:45 +0200 | [diff] [blame] | 1157 | size_t const errorCode = ZSTD_freeDStream(ress.dctx); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1158 | if (ZSTD_isError(errorCode)) |
| 1159 | EXM_THROW(69, "Error : can't free ZSTD_DStream context resource : %s", |
| 1160 | ZSTD_getErrorName(errorCode)); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1161 | free(ress.srcBuffer); |
| 1162 | free(ress.dstBuffer); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1163 | } |
Yann Collet | 4f13703 | 2015-12-17 02:23:58 +0100 | [diff] [blame] | 1164 | |
| 1165 | |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1166 | /** FIO_fwriteSparse() : |
| 1167 | * @return : storedSkips, to be provided to next call to FIO_fwriteSparse() of LZ4IO_fwriteSparseEnd() */ |
| 1168 | static unsigned FIO_fwriteSparse(FILE* file, const void* buffer, size_t bufferSize, unsigned storedSkips) |
| 1169 | { |
| 1170 | const size_t* const bufferT = (const size_t*)buffer; /* Buffer is supposed malloc'ed, hence aligned on size_t */ |
| 1171 | size_t bufferSizeT = bufferSize / sizeof(size_t); |
| 1172 | const size_t* const bufferTEnd = bufferT + bufferSizeT; |
| 1173 | const size_t* ptrT = bufferT; |
| 1174 | static const size_t segmentSizeT = (32 KB) / sizeof(size_t); /* 0-test re-attempted every 32 KB */ |
| 1175 | |
| 1176 | if (!g_sparseFileSupport) { /* normal write */ |
| 1177 | size_t const sizeCheck = fwrite(buffer, 1, bufferSize, file); |
| 1178 | if (sizeCheck != bufferSize) EXM_THROW(70, "Write error : cannot write decoded block"); |
| 1179 | return 0; |
| 1180 | } |
| 1181 | |
| 1182 | /* avoid int overflow */ |
| 1183 | if (storedSkips > 1 GB) { |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 1184 | int const seekResult = LONG_SEEK(file, 1 GB, SEEK_CUR); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1185 | if (seekResult != 0) EXM_THROW(71, "1 GB skip error (sparse file support)"); |
| 1186 | storedSkips -= 1 GB; |
| 1187 | } |
| 1188 | |
| 1189 | while (ptrT < bufferTEnd) { |
| 1190 | size_t seg0SizeT = segmentSizeT; |
| 1191 | size_t nb0T; |
| 1192 | |
| 1193 | /* count leading zeros */ |
| 1194 | if (seg0SizeT > bufferSizeT) seg0SizeT = bufferSizeT; |
| 1195 | bufferSizeT -= seg0SizeT; |
| 1196 | for (nb0T=0; (nb0T < seg0SizeT) && (ptrT[nb0T] == 0); nb0T++) ; |
| 1197 | storedSkips += (unsigned)(nb0T * sizeof(size_t)); |
| 1198 | |
| 1199 | if (nb0T != seg0SizeT) { /* not all 0s */ |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 1200 | int const seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1201 | if (seekResult) EXM_THROW(72, "Sparse skip error ; try --no-sparse"); |
| 1202 | storedSkips = 0; |
| 1203 | seg0SizeT -= nb0T; |
| 1204 | ptrT += nb0T; |
| 1205 | { size_t const sizeCheck = fwrite(ptrT, sizeof(size_t), seg0SizeT, file); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1206 | if (sizeCheck != seg0SizeT) |
| 1207 | EXM_THROW(73, "Write error : cannot write decoded block"); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1208 | } } |
| 1209 | ptrT += seg0SizeT; |
| 1210 | } |
| 1211 | |
| 1212 | { static size_t const maskT = sizeof(size_t)-1; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1213 | if (bufferSize & maskT) { |
| 1214 | /* size not multiple of sizeof(size_t) : implies end of block */ |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1215 | const char* const restStart = (const char*)bufferTEnd; |
| 1216 | const char* restPtr = restStart; |
| 1217 | size_t restSize = bufferSize & maskT; |
| 1218 | const char* const restEnd = restStart + restSize; |
| 1219 | for ( ; (restPtr < restEnd) && (*restPtr == 0); restPtr++) ; |
| 1220 | storedSkips += (unsigned) (restPtr - restStart); |
| 1221 | if (restPtr != restEnd) { |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 1222 | int seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1223 | if (seekResult) |
| 1224 | EXM_THROW(74, "Sparse skip error ; try --no-sparse"); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1225 | storedSkips = 0; |
| 1226 | { size_t const sizeCheck = fwrite(restPtr, 1, restEnd - restPtr, file); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1227 | if (sizeCheck != (size_t)(restEnd - restPtr)) |
| 1228 | EXM_THROW(75, "Write error : cannot write decoded end of block"); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1229 | } } } } |
| 1230 | |
| 1231 | return storedSkips; |
| 1232 | } |
| 1233 | |
| 1234 | static void FIO_fwriteSparseEnd(FILE* file, unsigned storedSkips) |
| 1235 | { |
| 1236 | if (storedSkips-->0) { /* implies g_sparseFileSupport>0 */ |
ds77 | 168d9b8 | 2017-02-12 10:27:18 +0100 | [diff] [blame] | 1237 | int const seekResult = LONG_SEEK(file, storedSkips, SEEK_CUR); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1238 | if (seekResult != 0) EXM_THROW(69, "Final skip error (sparse file)"); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1239 | { const char lastZeroByte[1] = { 0 }; |
| 1240 | size_t const sizeCheck = fwrite(lastZeroByte, 1, 1, file); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1241 | if (sizeCheck != 1) |
| 1242 | EXM_THROW(69, "Write error : cannot write last zero"); |
Yann Collet | 75424d1 | 2016-05-23 16:56:56 +0200 | [diff] [blame] | 1243 | } } |
| 1244 | } |
| 1245 | |
Yann Collet | be50aaa | 2015-09-10 23:26:09 +0100 | [diff] [blame] | 1246 | |
Yann Collet | de95f96 | 2016-05-23 19:46:47 +0200 | [diff] [blame] | 1247 | /** FIO_passThrough() : just copy input into output, for compatibility with gzip -df mode |
| 1248 | @return : 0 (no error) */ |
Przemyslaw Skibinski | 7c6bbc3 | 2016-12-05 18:31:14 +0100 | [diff] [blame] | 1249 | static unsigned FIO_passThrough(FILE* foutput, FILE* finput, void* buffer, size_t bufferSize, size_t alreadyLoaded) |
Yann Collet | de95f96 | 2016-05-23 19:46:47 +0200 | [diff] [blame] | 1250 | { |
Yann Collet | 179b197 | 2016-11-02 17:30:49 -0700 | [diff] [blame] | 1251 | size_t const blockSize = MIN(64 KB, bufferSize); |
Yann Collet | ddbb8e2 | 2016-05-24 00:52:14 +0200 | [diff] [blame] | 1252 | size_t readFromInput = 1; |
Yann Collet | de95f96 | 2016-05-23 19:46:47 +0200 | [diff] [blame] | 1253 | unsigned storedSkips = 0; |
| 1254 | |
Przemyslaw Skibinski | 7c6bbc3 | 2016-12-05 18:31:14 +0100 | [diff] [blame] | 1255 | /* assumption : ress->srcBufferLoaded bytes already loaded and stored within buffer */ |
| 1256 | { size_t const sizeCheck = fwrite(buffer, 1, alreadyLoaded, foutput); |
| 1257 | if (sizeCheck != alreadyLoaded) EXM_THROW(50, "Pass-through write error"); } |
Yann Collet | de95f96 | 2016-05-23 19:46:47 +0200 | [diff] [blame] | 1258 | |
Yann Collet | ddbb8e2 | 2016-05-24 00:52:14 +0200 | [diff] [blame] | 1259 | while (readFromInput) { |
| 1260 | readFromInput = fread(buffer, 1, blockSize, finput); |
| 1261 | storedSkips = FIO_fwriteSparse(foutput, buffer, readFromInput, storedSkips); |
Yann Collet | de95f96 | 2016-05-23 19:46:47 +0200 | [diff] [blame] | 1262 | } |
| 1263 | |
| 1264 | FIO_fwriteSparseEnd(foutput, storedSkips); |
| 1265 | return 0; |
| 1266 | } |
| 1267 | |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1268 | |
| 1269 | /** FIO_decompressFrame() : |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1270 | @return : size of decoded frame, or an error code |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1271 | */ |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1272 | #define FIO_ERROR_ZSTD_DECODING ((unsigned long long)(-2)) |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1273 | unsigned long long FIO_decompressFrame(dRess_t* ress, |
| 1274 | FILE* finput, |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1275 | const char* srcFileName, |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1276 | U64 alreadyDecoded) |
| 1277 | { |
| 1278 | U64 frameSize = 0; |
| 1279 | U32 storedSkips = 0; |
| 1280 | |
| 1281 | ZSTD_resetDStream(ress->dctx); |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1282 | if (strlen(srcFileName)>20) srcFileName += strlen(srcFileName)-20; /* display last 20 characters */ |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1283 | |
| 1284 | /* Header loading (optional, saves one loop) */ |
| 1285 | { size_t const toRead = 9; |
| 1286 | if (ress->srcBufferLoaded < toRead) |
| 1287 | ress->srcBufferLoaded += fread(((char*)ress->srcBuffer) + ress->srcBufferLoaded, 1, toRead - ress->srcBufferLoaded, finput); |
| 1288 | } |
| 1289 | |
| 1290 | /* Main decompression Loop */ |
| 1291 | while (1) { |
| 1292 | ZSTD_inBuffer inBuff = { ress->srcBuffer, ress->srcBufferLoaded, 0 }; |
| 1293 | ZSTD_outBuffer outBuff= { ress->dstBuffer, ress->dstBufferSize, 0 }; |
| 1294 | size_t const readSizeHint = ZSTD_decompressStream(ress->dctx, &outBuff, &inBuff); |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1295 | if (ZSTD_isError(readSizeHint)) { |
| 1296 | DISPLAYLEVEL(1, "%s : Decoding error (36) : %s \n", |
| 1297 | srcFileName, ZSTD_getErrorName(readSizeHint)); |
| 1298 | return FIO_ERROR_ZSTD_DECODING; |
| 1299 | } |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1300 | |
| 1301 | /* Write block */ |
Yann Collet | 500014a | 2017-01-19 16:59:56 -0800 | [diff] [blame] | 1302 | storedSkips = FIO_fwriteSparse(ress->dstFile, ress->dstBuffer, outBuff.pos, storedSkips); |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1303 | frameSize += outBuff.pos; |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1304 | DISPLAYUPDATE(2, "\r%-20.20s : %u MB... ", |
| 1305 | srcFileName, (U32)((alreadyDecoded+frameSize)>>20) ); |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1306 | |
| 1307 | if (inBuff.pos > 0) { |
| 1308 | memmove(ress->srcBuffer, (char*)ress->srcBuffer + inBuff.pos, inBuff.size - inBuff.pos); |
| 1309 | ress->srcBufferLoaded -= inBuff.pos; |
| 1310 | } |
| 1311 | |
| 1312 | if (readSizeHint == 0) break; /* end of frame */ |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1313 | if (inBuff.size != inBuff.pos) { |
| 1314 | DISPLAYLEVEL(1, "%s : Decoding error (37) : should consume entire input \n", |
| 1315 | srcFileName); |
| 1316 | return FIO_ERROR_ZSTD_DECODING; |
| 1317 | } |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1318 | |
| 1319 | /* Fill input buffer */ |
| 1320 | { size_t const toRead = MIN(readSizeHint, ress->srcBufferSize); /* support large skippable frames */ |
| 1321 | if (ress->srcBufferLoaded < toRead) |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1322 | ress->srcBufferLoaded += fread((char*)ress->srcBuffer + ress->srcBufferLoaded, |
| 1323 | 1, toRead - ress->srcBufferLoaded, finput); |
| 1324 | if (ress->srcBufferLoaded < toRead) { |
| 1325 | DISPLAYLEVEL(1, "%s : Read error (39) : premature end \n", |
| 1326 | srcFileName); |
| 1327 | return FIO_ERROR_ZSTD_DECODING; |
| 1328 | } } } |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1329 | |
| 1330 | FIO_fwriteSparseEnd(ress->dstFile, storedSkips); |
| 1331 | |
| 1332 | return frameSize; |
| 1333 | } |
| 1334 | |
| 1335 | |
Przemyslaw Skibinski | 19aad42 | 2016-12-01 11:56:31 +0100 | [diff] [blame] | 1336 | #ifdef ZSTD_GZDECOMPRESS |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1337 | static unsigned long long FIO_decompressGzFrame(dRess_t* ress, |
| 1338 | FILE* srcFile, const char* srcFileName) |
Przemyslaw Skibinski | b0f2ef2 | 2016-12-02 13:50:29 +0100 | [diff] [blame] | 1339 | { |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1340 | unsigned long long outFileSize = 0; |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1341 | z_stream strm; |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1342 | int flush = Z_NO_FLUSH; |
Przemyslaw Skibinski | 862698f | 2017-02-27 13:21:05 +0100 | [diff] [blame] | 1343 | int ret; |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1344 | |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1345 | strm.zalloc = Z_NULL; |
| 1346 | strm.zfree = Z_NULL; |
| 1347 | strm.opaque = Z_NULL; |
| 1348 | strm.next_in = 0; |
Przemyslaw Skibinski | 862698f | 2017-02-27 13:21:05 +0100 | [diff] [blame] | 1349 | strm.avail_in = 0; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1350 | /* see http://www.zlib.net/manual.html */ |
| 1351 | if (inflateInit2(&strm, 15 /* maxWindowLogSize */ + 16 /* gzip only */) != Z_OK) |
| 1352 | return 0; |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1353 | |
Yann Collet | b02ac8d | 2017-02-03 08:43:06 -0800 | [diff] [blame] | 1354 | strm.next_out = (Bytef*)ress->dstBuffer; |
Yann Collet | c3cba9d | 2017-02-02 17:12:50 -0800 | [diff] [blame] | 1355 | strm.avail_out = (uInt)ress->dstBufferSize; |
| 1356 | strm.avail_in = (uInt)ress->srcBufferLoaded; |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1357 | strm.next_in = (z_const unsigned char*)ress->srcBuffer; |
Przemyslaw Skibinski | 4b504f1 | 2016-12-02 13:11:39 +0100 | [diff] [blame] | 1358 | |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1359 | for ( ; ; ) { |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1360 | if (strm.avail_in == 0) { |
| 1361 | ress->srcBufferLoaded = fread(ress->srcBuffer, 1, ress->srcBufferSize, srcFile); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1362 | if (ress->srcBufferLoaded == 0) flush = Z_FINISH; |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1363 | strm.next_in = (z_const unsigned char*)ress->srcBuffer; |
Yann Collet | c3cba9d | 2017-02-02 17:12:50 -0800 | [diff] [blame] | 1364 | strm.avail_in = (uInt)ress->srcBufferLoaded; |
Przemyslaw Skibinski | 4b504f1 | 2016-12-02 13:11:39 +0100 | [diff] [blame] | 1365 | } |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1366 | ret = inflate(&strm, flush); |
| 1367 | if (ret == Z_BUF_ERROR) EXM_THROW(39, "zstd: %s: premature end", srcFileName); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1368 | if (ret != Z_OK && ret != Z_STREAM_END) { |
| 1369 | DISPLAY("zstd: %s: inflate error %d \n", srcFileName, ret); |
| 1370 | return 0; |
| 1371 | } |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1372 | { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1373 | if (decompBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1374 | if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) |
| 1375 | EXM_THROW(31, "Write error : cannot write to output file"); |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1376 | outFileSize += decompBytes; |
Yann Collet | b02ac8d | 2017-02-03 08:43:06 -0800 | [diff] [blame] | 1377 | strm.next_out = (Bytef*)ress->dstBuffer; |
Yann Collet | c3cba9d | 2017-02-02 17:12:50 -0800 | [diff] [blame] | 1378 | strm.avail_out = (uInt)ress->dstBufferSize; |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1379 | } |
| 1380 | } |
| 1381 | if (ret == Z_STREAM_END) break; |
| 1382 | } |
Przemyslaw Skibinski | 19aad42 | 2016-12-01 11:56:31 +0100 | [diff] [blame] | 1383 | |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1384 | if (strm.avail_in > 0) memmove(ress->srcBuffer, strm.next_in, strm.avail_in); |
| 1385 | ress->srcBufferLoaded = strm.avail_in; |
Przemyslaw Skibinski | 862698f | 2017-02-27 13:21:05 +0100 | [diff] [blame] | 1386 | ret = inflateEnd(&strm); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1387 | if (ret != Z_OK) EXM_THROW(32, "zstd: %s: inflateEnd error %d", srcFileName, ret); |
| 1388 | return outFileSize; |
| 1389 | } |
| 1390 | #endif |
| 1391 | |
| 1392 | |
| 1393 | #ifdef ZSTD_LZMADECOMPRESS |
| 1394 | static unsigned long long FIO_decompressLzmaFrame(dRess_t* ress, FILE* srcFile, const char* srcFileName, int plain_lzma) |
| 1395 | { |
| 1396 | unsigned long long outFileSize = 0; |
| 1397 | lzma_stream strm = LZMA_STREAM_INIT; |
| 1398 | lzma_action action = LZMA_RUN; |
| 1399 | lzma_ret ret; |
| 1400 | |
| 1401 | strm.next_in = 0; |
| 1402 | strm.avail_in = 0; |
| 1403 | if (plain_lzma) { |
| 1404 | ret = lzma_alone_decoder(&strm, UINT64_MAX); /* LZMA */ |
| 1405 | } else { |
| 1406 | ret = lzma_stream_decoder(&strm, UINT64_MAX, 0); /* XZ */ |
| 1407 | } |
| 1408 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1409 | if (ret != LZMA_OK) |
| 1410 | EXM_THROW(71, "zstd: %s: lzma_alone_decoder/lzma_stream_decoder error %d", |
| 1411 | srcFileName, ret); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1412 | |
| 1413 | strm.next_out = ress->dstBuffer; |
| 1414 | strm.avail_out = ress->dstBufferSize; |
| 1415 | strm.avail_in = ress->srcBufferLoaded; |
| 1416 | strm.next_in = ress->srcBuffer; |
| 1417 | |
| 1418 | for ( ; ; ) { |
| 1419 | if (strm.avail_in == 0) { |
| 1420 | ress->srcBufferLoaded = fread(ress->srcBuffer, 1, ress->srcBufferSize, srcFile); |
| 1421 | if (ress->srcBufferLoaded == 0) action = LZMA_FINISH; |
| 1422 | strm.next_in = ress->srcBuffer; |
| 1423 | strm.avail_in = ress->srcBufferLoaded; |
| 1424 | } |
| 1425 | ret = lzma_code(&strm, action); |
| 1426 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1427 | if (ret == LZMA_BUF_ERROR) |
| 1428 | EXM_THROW(39, "zstd: %s: premature end", srcFileName); |
| 1429 | if (ret != LZMA_OK && ret != LZMA_STREAM_END) { |
| 1430 | DISPLAY("zstd: %s: lzma_code decoding error %d \n", srcFileName, ret); |
| 1431 | return 0; |
| 1432 | } |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1433 | { size_t const decompBytes = ress->dstBufferSize - strm.avail_out; |
| 1434 | if (decompBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1435 | if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) |
| 1436 | EXM_THROW(31, "Write error : cannot write to output file"); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1437 | outFileSize += decompBytes; |
| 1438 | strm.next_out = ress->dstBuffer; |
| 1439 | strm.avail_out = ress->dstBufferSize; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1440 | } } |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1441 | if (ret == LZMA_STREAM_END) break; |
| 1442 | } |
| 1443 | |
| 1444 | if (strm.avail_in > 0) memmove(ress->srcBuffer, strm.next_in, strm.avail_in); |
| 1445 | ress->srcBufferLoaded = strm.avail_in; |
| 1446 | lzma_end(&strm); |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1447 | return outFileSize; |
Przemyslaw Skibinski | 19aad42 | 2016-12-01 11:56:31 +0100 | [diff] [blame] | 1448 | } |
| 1449 | #endif |
| 1450 | |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1451 | #ifdef ZSTD_LZ4DECOMPRESS |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1452 | static unsigned long long FIO_decompressLz4Frame(dRess_t* ress, |
| 1453 | FILE* srcFile, const char* srcFileName) |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1454 | { |
| 1455 | unsigned long long filesize = 0; |
| 1456 | LZ4F_errorCode_t nextToLoad; |
Sean Purcell | 2c4b6fe | 2017-04-25 11:00:54 -0700 | [diff] [blame] | 1457 | LZ4F_decompressionContext_t dCtx; |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1458 | LZ4F_errorCode_t const errorCode = LZ4F_createDecompressionContext(&dCtx, LZ4F_VERSION); |
| 1459 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1460 | if (LZ4F_isError(errorCode)) |
| 1461 | EXM_THROW(61, "zstd: failed to create lz4 decompression context"); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1462 | |
| 1463 | /* Init feed with magic number (already consumed from FILE* sFile) */ |
| 1464 | { size_t inSize = 4; |
| 1465 | size_t outSize= 0; |
| 1466 | MEM_writeLE32(ress->srcBuffer, LZ4_MAGICNUMBER); |
| 1467 | nextToLoad = LZ4F_decompress(dCtx, ress->dstBuffer, &outSize, ress->srcBuffer, &inSize, NULL); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1468 | if (LZ4F_isError(nextToLoad)) |
| 1469 | EXM_THROW(62, "zstd: %s: lz4 header error : %s", |
| 1470 | srcFileName, LZ4F_getErrorName(nextToLoad)); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1471 | } |
| 1472 | |
| 1473 | /* Main Loop */ |
| 1474 | for (;nextToLoad;) { |
| 1475 | size_t readSize; |
| 1476 | size_t pos = 0; |
| 1477 | size_t decodedBytes = ress->dstBufferSize; |
| 1478 | |
| 1479 | /* Read input */ |
| 1480 | if (nextToLoad > ress->srcBufferSize) nextToLoad = ress->srcBufferSize; |
| 1481 | readSize = fread(ress->srcBuffer, 1, nextToLoad, srcFile); |
| 1482 | if (!readSize) break; /* reached end of file or stream */ |
| 1483 | |
| 1484 | while ((pos < readSize) || (decodedBytes == ress->dstBufferSize)) { /* still to read, or still to flush */ |
| 1485 | /* Decode Input (at least partially) */ |
| 1486 | size_t remaining = readSize - pos; |
| 1487 | decodedBytes = ress->dstBufferSize; |
| 1488 | nextToLoad = LZ4F_decompress(dCtx, ress->dstBuffer, &decodedBytes, (char*)(ress->srcBuffer)+pos, &remaining, NULL); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1489 | if (LZ4F_isError(nextToLoad)) |
| 1490 | EXM_THROW(66, "zstd: %s: decompression error : %s", |
| 1491 | srcFileName, LZ4F_getErrorName(nextToLoad)); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1492 | pos += remaining; |
| 1493 | |
| 1494 | /* Write Block */ |
| 1495 | if (decodedBytes) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1496 | if (fwrite(ress->dstBuffer, 1, decodedBytes, ress->dstFile) != decodedBytes) |
| 1497 | EXM_THROW(63, "Write error : cannot write to output file"); |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1498 | filesize += decodedBytes; |
| 1499 | DISPLAYUPDATE(2, "\rDecompressed : %u MB ", (unsigned)(filesize>>20)); |
| 1500 | } |
| 1501 | |
| 1502 | if (!nextToLoad) break; |
| 1503 | } |
| 1504 | } |
| 1505 | /* can be out because readSize == 0, which could be an fread() error */ |
| 1506 | if (ferror(srcFile)) EXM_THROW(67, "zstd: %s: read error", srcFileName); |
| 1507 | |
| 1508 | if (nextToLoad!=0) EXM_THROW(68, "zstd: %s: unfinished stream", srcFileName); |
| 1509 | |
| 1510 | LZ4F_freeDecompressionContext(dCtx); |
| 1511 | ress->srcBufferLoaded = 0; /* LZ4F will go to the frame boundary */ |
| 1512 | |
| 1513 | return filesize; |
| 1514 | } |
| 1515 | #endif |
| 1516 | |
| 1517 | |
Przemyslaw Skibinski | 19aad42 | 2016-12-01 11:56:31 +0100 | [diff] [blame] | 1518 | |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1519 | /** FIO_decompressSrcFile() : |
| 1520 | Decompression `srcFileName` into `ress.dstFile` |
| 1521 | @return : 0 : OK |
| 1522 | 1 : operation not started |
| 1523 | */ |
Yann Collet | 743b33f | 2016-12-02 15:18:57 -0800 | [diff] [blame] | 1524 | static int FIO_decompressSrcFile(dRess_t ress, const char* dstFileName, const char* srcFileName) |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1525 | { |
Przemyslaw Skibinski | b0f2ef2 | 2016-12-02 13:50:29 +0100 | [diff] [blame] | 1526 | FILE* srcFile; |
Yann Collet | a1dd6b9 | 2016-07-26 16:44:09 +0200 | [diff] [blame] | 1527 | unsigned readSomething = 0; |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1528 | unsigned long long filesize = 0; |
Przemyslaw Skibinski | 0e14675 | 2016-11-30 13:34:21 +0100 | [diff] [blame] | 1529 | |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 1530 | if (UTIL_isDirectory(srcFileName)) { |
| 1531 | DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName); |
| 1532 | return 1; |
| 1533 | } |
Przemyslaw Skibinski | 0e14675 | 2016-11-30 13:34:21 +0100 | [diff] [blame] | 1534 | |
Przemyslaw Skibinski | b0f2ef2 | 2016-12-02 13:50:29 +0100 | [diff] [blame] | 1535 | srcFile = FIO_openSrcFile(srcFileName); |
Yann Collet | 0b9b894 | 2017-02-27 00:27:30 -0800 | [diff] [blame] | 1536 | if (srcFile==NULL) return 1; |
Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1537 | |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1538 | /* for each frame */ |
| 1539 | for ( ; ; ) { |
| 1540 | /* check magic number -> version */ |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1541 | size_t const toRead = 4; |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1542 | const BYTE* buf = (const BYTE*)ress.srcBuffer; |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1543 | if (ress.srcBufferLoaded < toRead) |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1544 | ress.srcBufferLoaded += fread((char*)ress.srcBuffer + ress.srcBufferLoaded, |
| 1545 | (size_t)1, toRead - ress.srcBufferLoaded, srcFile); |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1546 | if (ress.srcBufferLoaded==0) { |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1547 | if (readSomething==0) { |
| 1548 | DISPLAY("zstd: %s: unexpected end of file \n", srcFileName); |
| 1549 | fclose(srcFile); |
| 1550 | return 1; |
| 1551 | } /* srcFileName is empty */ |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1552 | break; /* no more input */ |
| 1553 | } |
| 1554 | readSomething = 1; /* there is at least >= 4 bytes in srcFile */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1555 | if (ress.srcBufferLoaded < toRead) { |
| 1556 | DISPLAY("zstd: %s: unknown header \n", srcFileName); |
| 1557 | fclose(srcFile); |
| 1558 | return 1; |
| 1559 | } |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1560 | if (buf[0] == 31 && buf[1] == 139) { /* gz magic number */ |
Przemyslaw Skibinski | c5eebca | 2016-12-02 15:01:31 +0100 | [diff] [blame] | 1561 | #ifdef ZSTD_GZDECOMPRESS |
Przemyslaw Skibinski | b493e3b | 2016-12-05 17:39:38 +0100 | [diff] [blame] | 1562 | unsigned long long const result = FIO_decompressGzFrame(&ress, srcFile, srcFileName); |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1563 | if (result == 0) return 1; |
| 1564 | filesize += result; |
Przemyslaw Skibinski | abfb51f | 2016-11-30 15:05:54 +0100 | [diff] [blame] | 1565 | #else |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1566 | DISPLAYLEVEL(1, "zstd: %s: gzip file cannot be uncompressed (zstd compiled without ZSTD_GZDECOMPRESS) -- ignored \n", srcFileName); |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1567 | return 1; |
Przemyslaw Skibinski | abfb51f | 2016-11-30 15:05:54 +0100 | [diff] [blame] | 1568 | #endif |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1569 | } else if ((buf[0] == 0xFD && buf[1] == 0x37) /* xz magic number */ |
| 1570 | || (buf[0] == 0x5D && buf[1] == 0x00)) { /* lzma header (no magic number) */ |
| 1571 | #ifdef ZSTD_LZMADECOMPRESS |
| 1572 | unsigned long long const result = FIO_decompressLzmaFrame(&ress, srcFile, srcFileName, buf[0] != 0xFD); |
| 1573 | if (result == 0) return 1; |
| 1574 | filesize += result; |
| 1575 | #else |
| 1576 | DISPLAYLEVEL(1, "zstd: %s: xz/lzma file cannot be uncompressed (zstd compiled without ZSTD_LZMADECOMPRESS) -- ignored \n", srcFileName); |
| 1577 | return 1; |
| 1578 | #endif |
Sean Purcell | 4de8632 | 2017-04-24 16:48:25 -0700 | [diff] [blame] | 1579 | } else if (MEM_readLE32(buf) == LZ4_MAGICNUMBER) { |
| 1580 | #ifdef ZSTD_LZ4DECOMPRESS |
| 1581 | unsigned long long const result = FIO_decompressLz4Frame(&ress, srcFile, srcFileName); |
| 1582 | if (result == 0) return 1; |
| 1583 | filesize += result; |
| 1584 | #else |
| 1585 | DISPLAYLEVEL(1, "zstd: %s: lz4 file cannot be uncompressed (zstd compiled without ZSTD_LZ4DECOMPRESS) -- ignored \n", srcFileName); |
| 1586 | return 1; |
| 1587 | #endif |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1588 | } else { |
| 1589 | if (!ZSTD_isFrame(ress.srcBuffer, toRead)) { |
Yann Collet | 743b33f | 2016-12-02 15:18:57 -0800 | [diff] [blame] | 1590 | if ((g_overwrite) && !strcmp (dstFileName, stdoutmark)) { /* pass-through mode */ |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1591 | unsigned const result = FIO_passThrough(ress.dstFile, srcFile, |
| 1592 | ress.srcBuffer, ress.srcBufferSize, ress.srcBufferLoaded); |
Przemyslaw Skibinski | 690753e | 2016-12-02 16:20:16 +0100 | [diff] [blame] | 1593 | if (fclose(srcFile)) EXM_THROW(32, "zstd: %s close error", srcFileName); /* error should never happen */ |
| 1594 | return result; |
| 1595 | } else { |
| 1596 | DISPLAYLEVEL(1, "zstd: %s: not in zstd format \n", srcFileName); |
| 1597 | fclose(srcFile); |
| 1598 | return 1; |
| 1599 | } } |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1600 | { unsigned long long const frameSize = FIO_decompressFrame(&ress, srcFile, srcFileName, filesize); |
| 1601 | if (frameSize == FIO_ERROR_ZSTD_DECODING) { |
| 1602 | fclose(srcFile); |
| 1603 | return 1; |
| 1604 | } |
| 1605 | filesize += frameSize; |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1606 | } } |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1607 | } /* for each frame */ |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1608 | |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1609 | /* Final Status */ |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1610 | DISPLAYLEVEL(2, "\r%79s\r", ""); |
Yann Collet | 1c69baa | 2016-08-28 12:47:17 -0700 | [diff] [blame] | 1611 | DISPLAYLEVEL(2, "%-20s: %llu bytes \n", srcFileName, filesize); |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1612 | |
Przemyslaw Skibinski | b0f2ef2 | 2016-12-02 13:50:29 +0100 | [diff] [blame] | 1613 | /* Close file */ |
| 1614 | if (fclose(srcFile)) EXM_THROW(33, "zstd: %s close error", srcFileName); /* error should never happen */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1615 | if (g_removeSrcFile /* --rm */ && strcmp(srcFileName, stdinmark)) { |
| 1616 | if (remove(srcFileName)) |
| 1617 | EXM_THROW(34, "zstd: %s: %s", srcFileName, strerror(errno)); |
| 1618 | } |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1619 | return 0; |
| 1620 | } |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1621 | |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1622 | |
| 1623 | /** FIO_decompressFile_extRess() : |
| 1624 | decompress `srcFileName` into `dstFileName` |
| 1625 | @return : 0 : OK |
| 1626 | 1 : operation aborted (src not available, dst already taken, etc.) |
| 1627 | */ |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 1628 | static int FIO_decompressDstFile(dRess_t ress, |
Yann Collet | 5bd4237 | 2016-12-02 12:40:57 -0800 | [diff] [blame] | 1629 | const char* dstFileName, const char* srcFileName) |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1630 | { |
Chip Turner | 6de382c | 2016-03-13 22:24:46 -0700 | [diff] [blame] | 1631 | int result; |
Przemyslaw Skibinski | fcf22e3 | 2016-11-02 14:08:07 +0100 | [diff] [blame] | 1632 | stat_t statbuf; |
| 1633 | int stat_result = 0; |
Przemyslaw Skibinski | a42794d | 2016-11-02 13:08:39 +0100 | [diff] [blame] | 1634 | |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1635 | ress.dstFile = FIO_openDstFile(dstFileName); |
| 1636 | if (ress.dstFile==0) return 1; |
| 1637 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1638 | if (strcmp (srcFileName, stdinmark) && UTIL_getFileStat(srcFileName, &statbuf)) |
| 1639 | stat_result = 1; |
Yann Collet | 743b33f | 2016-12-02 15:18:57 -0800 | [diff] [blame] | 1640 | result = FIO_decompressSrcFile(ress, dstFileName, srcFileName); |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1641 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1642 | if (fclose(ress.dstFile)) |
| 1643 | EXM_THROW(38, "Write error : cannot properly close %s", dstFileName); |
Przemyslaw Skibinski | a42794d | 2016-11-02 13:08:39 +0100 | [diff] [blame] | 1644 | |
Yann Collet | aad9fe5 | 2016-09-07 07:00:08 +0200 | [diff] [blame] | 1645 | if ( (result != 0) |
| 1646 | && strcmp(dstFileName, nulmark) /* special case : don't remove() /dev/null (#316) */ |
| 1647 | && remove(dstFileName) ) |
Yann Collet | 03d3f23 | 2016-09-07 07:01:33 +0200 | [diff] [blame] | 1648 | result=1; /* don't do anything special if remove() fails */ |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1649 | else if (strcmp (dstFileName, stdoutmark) && stat_result) |
| 1650 | UTIL_setFileStat(dstFileName, &statbuf); |
Chip Turner | 6de382c | 2016-03-13 22:24:46 -0700 | [diff] [blame] | 1651 | return result; |
Yann Collet | b1f3f4b | 2015-10-18 22:18:32 +0100 | [diff] [blame] | 1652 | } |
| 1653 | |
| 1654 | |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1655 | int FIO_decompressFilename(const char* dstFileName, const char* srcFileName, |
| 1656 | const char* dictFileName) |
| 1657 | { |
| 1658 | int missingFiles = 0; |
| 1659 | dRess_t ress = FIO_createDResources(dictFileName); |
| 1660 | |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 1661 | missingFiles += FIO_decompressDstFile(ress, dstFileName, srcFileName); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1662 | |
| 1663 | FIO_freeDResources(ress); |
| 1664 | return missingFiles; |
| 1665 | } |
| 1666 | |
| 1667 | |
| 1668 | #define MAXSUFFIXSIZE 8 |
| 1669 | int FIO_decompressMultipleFilenames(const char** srcNamesTable, unsigned nbFiles, |
| 1670 | const char* suffix, |
| 1671 | const char* dictFileName) |
| 1672 | { |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1673 | int skippedFiles = 0; |
| 1674 | int missingFiles = 0; |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1675 | dRess_t ress = FIO_createDResources(dictFileName); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1676 | |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1677 | if (suffix==NULL) |
| 1678 | EXM_THROW(70, "zstd: decompression: unknown dst"); /* should never happen */ |
Yann Collet | 44f684d | 2016-07-13 19:30:40 +0200 | [diff] [blame] | 1679 | |
Yann Collet | 743b33f | 2016-12-02 15:18:57 -0800 | [diff] [blame] | 1680 | if (!strcmp(suffix, stdoutmark) || !strcmp(suffix, nulmark)) { /* special cases : -c or -t */ |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1681 | unsigned u; |
Yann Collet | accfd80 | 2016-02-15 19:33:16 +0100 | [diff] [blame] | 1682 | ress.dstFile = FIO_openDstFile(suffix); |
| 1683 | if (ress.dstFile == 0) EXM_THROW(71, "cannot open %s", suffix); |
| 1684 | for (u=0; u<nbFiles; u++) |
Yann Collet | 743b33f | 2016-12-02 15:18:57 -0800 | [diff] [blame] | 1685 | missingFiles += FIO_decompressSrcFile(ress, suffix, srcNamesTable[u]); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1686 | if (fclose(ress.dstFile)) |
| 1687 | EXM_THROW(72, "Write error : cannot properly close stdout"); |
Yann Collet | accfd80 | 2016-02-15 19:33:16 +0100 | [diff] [blame] | 1688 | } else { |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1689 | size_t suffixSize; |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1690 | size_t dfnSize = FNSPACE; |
Yann Collet | f849462 | 2016-05-07 22:43:40 +0200 | [diff] [blame] | 1691 | unsigned u; |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1692 | char* dstFileName = (char*)malloc(FNSPACE); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1693 | if (dstFileName==NULL) |
| 1694 | EXM_THROW(73, "not enough memory for dstFileName"); |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1695 | for (u=0; u<nbFiles; u++) { /* create dstFileName */ |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1696 | const char* const srcFileName = srcNamesTable[u]; |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1697 | const char* const suffixPtr = strrchr(srcFileName, '.'); |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1698 | size_t const sfnSize = strlen(srcFileName); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1699 | if (!suffixPtr) { |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1700 | DISPLAYLEVEL(1, "zstd: %s: unknown suffix -- ignored \n", |
| 1701 | srcFileName); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1702 | skippedFiles++; |
| 1703 | continue; |
| 1704 | } |
| 1705 | suffixSize = strlen(suffixPtr); |
Yann Collet | accfd80 | 2016-02-15 19:33:16 +0100 | [diff] [blame] | 1706 | if (dfnSize+suffixSize <= sfnSize+1) { |
| 1707 | free(dstFileName); |
| 1708 | dfnSize = sfnSize + 20; |
| 1709 | dstFileName = (char*)malloc(dfnSize); |
Yann Collet | e12ae02 | 2017-05-16 17:32:33 -0700 | [diff] [blame] | 1710 | if (dstFileName==NULL) |
| 1711 | EXM_THROW(74, "not enough memory for dstFileName"); |
Yann Collet | accfd80 | 2016-02-15 19:33:16 +0100 | [diff] [blame] | 1712 | } |
Yann Collet | 01a1abf | 2017-05-05 19:15:24 -0700 | [diff] [blame] | 1713 | if (sfnSize <= suffixSize |
| 1714 | || (strcmp(suffixPtr, GZ_EXTENSION) |
| 1715 | && strcmp(suffixPtr, XZ_EXTENSION) |
| 1716 | && strcmp(suffixPtr, ZSTD_EXTENSION) |
| 1717 | && strcmp(suffixPtr, LZMA_EXTENSION) |
| 1718 | && strcmp(suffixPtr, LZ4_EXTENSION)) ) { |
| 1719 | DISPLAYLEVEL(1, "zstd: %s: unknown suffix (%s/%s/%s/%s/%s expected) -- ignored \n", |
| 1720 | srcFileName, GZ_EXTENSION, XZ_EXTENSION, ZSTD_EXTENSION, LZMA_EXTENSION, LZ4_EXTENSION); |
Nick Terrell | aa8bcf3 | 2017-03-13 18:11:07 -0700 | [diff] [blame] | 1721 | skippedFiles++; |
| 1722 | continue; |
Przemyslaw Skibinski | 0e14675 | 2016-11-30 13:34:21 +0100 | [diff] [blame] | 1723 | } else { |
| 1724 | memcpy(dstFileName, srcFileName, sfnSize - suffixSize); |
| 1725 | dstFileName[sfnSize-suffixSize] = '\0'; |
Yann Collet | 1f1f239 | 2016-02-12 18:33:26 +0100 | [diff] [blame] | 1726 | } |
Yann Collet | b09b12c | 2016-06-09 22:59:51 +0200 | [diff] [blame] | 1727 | missingFiles += FIO_decompressDstFile(ress, dstFileName, srcFileName); |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1728 | } |
| 1729 | free(dstFileName); |
| 1730 | } |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1731 | |
| 1732 | FIO_freeDResources(ress); |
Yann Collet | deb078b | 2015-12-17 20:30:14 +0100 | [diff] [blame] | 1733 | return missingFiles + skippedFiles; |
| 1734 | } |
Yann Collet | accfd80 | 2016-02-15 19:33:16 +0100 | [diff] [blame] | 1735 | |
Yann Collet | 8b23eea | 2016-05-10 05:37:43 +0200 | [diff] [blame] | 1736 | #endif /* #ifndef ZSTD_NODECOMPRESS */ |