| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1 | /* |
| 2 | zstd - standard compression library |
| Yann Collet | ae7aa06 | 2016-02-03 02:46:46 +0100 | [diff] [blame] | 3 | Copyright (C) 2014-2016, Yann Collet. |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 4 | |
| 5 | BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) |
| 6 | |
| 7 | Redistribution and use in source and binary forms, with or without |
| 8 | modification, are permitted provided that the following conditions are |
| 9 | met: |
| 10 | * Redistributions of source code must retain the above copyright |
| 11 | notice, this list of conditions and the following disclaimer. |
| 12 | * Redistributions in binary form must reproduce the above |
| 13 | copyright notice, this list of conditions and the following disclaimer |
| 14 | in the documentation and/or other materials provided with the |
| 15 | distribution. |
| 16 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 17 | "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 18 | LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| 19 | A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| 20 | OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| 21 | SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| 22 | LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| 23 | DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| 24 | THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 25 | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 26 | OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 27 | |
| 28 | You can contact the author at : |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 29 | - zstd homepage : http://www.zstd.net |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 30 | */ |
| 31 | |
| 32 | /* *************************************************************** |
| 33 | * Tuning parameters |
| 34 | *****************************************************************/ |
| 35 | /*! |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 36 | * HEAPMODE : |
| Yann Collet | 3a3b72f | 2016-01-11 12:56:11 +0100 | [diff] [blame] | 37 | * Select how default decompression function ZSTD_decompress() will allocate memory, |
| 38 | * in memory stack (0), or in memory heap (1, requires malloc()) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 39 | */ |
| 40 | #ifndef ZSTD_HEAPMODE |
| 41 | # define ZSTD_HEAPMODE 1 |
| Yann Collet | 3a3b72f | 2016-01-11 12:56:11 +0100 | [diff] [blame] | 42 | #endif |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 43 | |
| 44 | /*! |
| 45 | * LEGACY_SUPPORT : |
| Yann Collet | ae7aa06 | 2016-02-03 02:46:46 +0100 | [diff] [blame] | 46 | * if set to 1, ZSTD_decompress() can decode older formats (v0.1+) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 47 | */ |
| 48 | #ifndef ZSTD_LEGACY_SUPPORT |
| Yann Collet | fba6aed | 2016-01-18 12:03:27 +0100 | [diff] [blame] | 49 | # define ZSTD_LEGACY_SUPPORT 0 |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 50 | #endif |
| 51 | |
| 52 | |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 53 | /*-******************************************************* |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 54 | * Dependencies |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 55 | *********************************************************/ |
| 56 | #include <stdlib.h> /* calloc */ |
| 57 | #include <string.h> /* memcpy, memmove */ |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 58 | #include <stdio.h> /* debug only : printf */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 59 | #include "mem.h" /* low level memory routines */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 60 | #include "zstd_internal.h" |
| 61 | #include "fse_static.h" |
| Yann Collet | afe0709 | 2016-01-25 04:10:46 +0100 | [diff] [blame] | 62 | #include "huff0_static.h" |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 63 | |
| 64 | #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT==1) |
| 65 | # include "zstd_legacy.h" |
| 66 | #endif |
| 67 | |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 68 | |
| 69 | /*-******************************************************* |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 70 | * Compiler specifics |
| 71 | *********************************************************/ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 72 | #ifdef _MSC_VER /* Visual Studio */ |
| 73 | # define FORCE_INLINE static __forceinline |
| 74 | # include <intrin.h> /* For Visual 2005 */ |
| 75 | # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ |
| 76 | # pragma warning(disable : 4324) /* disable: C4324: padded structure */ |
| 77 | #else |
| 78 | # define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) |
| 79 | # ifdef __GNUC__ |
| 80 | # define FORCE_INLINE static inline __attribute__((always_inline)) |
| 81 | # else |
| 82 | # define FORCE_INLINE static inline |
| 83 | # endif |
| 84 | #endif |
| 85 | |
| 86 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 87 | /*_******************************************************* |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 88 | * Memory operations |
| 89 | **********************************************************/ |
| 90 | static void ZSTD_copy4(void* dst, const void* src) { memcpy(dst, src, 4); } |
| 91 | |
| 92 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 93 | /*-************************************* |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 94 | * Error Management |
| 95 | ***************************************/ |
| Yann Collet | 14983e7 | 2015-11-11 21:38:21 +0100 | [diff] [blame] | 96 | unsigned ZSTD_versionNumber (void) { return ZSTD_VERSION_NUMBER; } |
| 97 | |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 98 | /*! ZSTD_isError() : |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 99 | * tells if a return value is an error code */ |
| 100 | unsigned ZSTD_isError(size_t code) { return ERR_isError(code); } |
| 101 | |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 102 | /*! ZSTD_getError() : |
| Yann Collet | 72bff50 | 2016-02-03 12:06:24 +0100 | [diff] [blame] | 103 | * convert a `size_t` function result into a proper ZSTD_errorCode enum */ |
| Yann Collet | 982ffc7 | 2016-02-05 02:33:10 +0100 | [diff] [blame] | 104 | ZSTD_ErrorCode ZSTD_getError(size_t code) { return ERR_getError(code); } |
| Yann Collet | 72bff50 | 2016-02-03 12:06:24 +0100 | [diff] [blame] | 105 | |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 106 | /*! ZSTD_getErrorName() : |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 107 | * provides error code string (useful for debugging) */ |
| 108 | const char* ZSTD_getErrorName(size_t code) { return ERR_getErrorName(code); } |
| 109 | |
| 110 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 111 | /*-************************************************************* |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 112 | * Context management |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 113 | ***************************************************************/ |
| Yann Collet | e4fdad5 | 2015-11-25 21:09:17 +0100 | [diff] [blame] | 114 | typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader, |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 115 | ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock } ZSTD_dStage; |
| 116 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 117 | struct ZSTD_DCtx_s |
| 118 | { |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 119 | FSE_DTable LLTable[FSE_DTABLE_SIZE_U32(LLFSELog)]; |
| 120 | FSE_DTable OffTable[FSE_DTABLE_SIZE_U32(OffFSELog)]; |
| 121 | FSE_DTable MLTable[FSE_DTABLE_SIZE_U32(MLFSELog)]; |
| 122 | unsigned hufTableX4[HUF_DTABLE_SIZE(HufLog)]; |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 123 | const void* previousDstEnd; |
| 124 | const void* base; |
| 125 | const void* vBase; |
| 126 | const void* dictEnd; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 127 | size_t expected; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 128 | size_t headerSize; |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 129 | ZSTD_frameParams fParams; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 130 | blockType_t bType; /* used in ZSTD_decompressContinue(), to transfer blockType between header decoding and block decoding stages */ |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 131 | ZSTD_dStage stage; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 132 | U32 flagStaticTables; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 133 | const BYTE* litPtr; |
| 134 | size_t litBufSize; |
| 135 | size_t litSize; |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 136 | BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH]; |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 137 | BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 138 | }; /* typedef'd to ZSTD_DCtx within "zstd_static.h" */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 139 | |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 140 | size_t sizeofDCtx (void) { return sizeof(ZSTD_DCtx); } |
| 141 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 142 | size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx) |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 143 | { |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 144 | dctx->expected = ZSTD_frameHeaderSize_min; |
| 145 | dctx->stage = ZSTDds_getFrameHeaderSize; |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 146 | dctx->previousDstEnd = NULL; |
| 147 | dctx->base = NULL; |
| 148 | dctx->vBase = NULL; |
| 149 | dctx->dictEnd = NULL; |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 150 | dctx->hufTableX4[0] = HufLog; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 151 | dctx->flagStaticTables = 0; |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 152 | dctx->fParams.mml = MINMATCH; /* overwritten by frame but forces ZSTD_btopt to MINMATCH in block mode */ |
| inikep | a4dde25 | 2016-03-01 14:14:35 +0100 | [diff] [blame] | 153 | ZSTD_LOG_BLOCK("%p: ZSTD_decompressBegin searchLength=%d\n", dctx->base, dctx->params.searchLength); |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 154 | return 0; |
| 155 | } |
| 156 | |
| 157 | ZSTD_DCtx* ZSTD_createDCtx(void) |
| 158 | { |
| 159 | ZSTD_DCtx* dctx = (ZSTD_DCtx*)malloc(sizeof(ZSTD_DCtx)); |
| 160 | if (dctx==NULL) return NULL; |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 161 | ZSTD_decompressBegin(dctx); |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 162 | return dctx; |
| 163 | } |
| 164 | |
| 165 | size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx) |
| 166 | { |
| 167 | free(dctx); |
| Yann Collet | 982ffc7 | 2016-02-05 02:33:10 +0100 | [diff] [blame] | 168 | return 0; /* reserved as a potential error code in the future */ |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 169 | } |
| 170 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 171 | void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx) |
| 172 | { |
| 173 | memcpy(dstDCtx, srcDCtx, |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 174 | sizeof(ZSTD_DCtx) - (ZSTD_BLOCKSIZE_MAX+WILDCOPY_OVERLENGTH + ZSTD_frameHeaderSize_max)); /* no need to copy workspace */ |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 175 | } |
| 176 | |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 177 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 178 | /*-************************************************************* |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 179 | * Decompression section |
| 180 | ***************************************************************/ |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 181 | |
| 182 | /* Frame format description |
| 183 | Frame Header - [ Block Header - Block ] - Frame End |
| 184 | 1) Frame Header |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 185 | - 4 bytes - Magic Number : ZSTD_MAGICNUMBER (defined within zstd_static.h) |
| 186 | - 1 byte - Frame Descriptor |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 187 | 2) Block Header |
| 188 | - 3 bytes, starting with a 2-bits descriptor |
| 189 | Uncompressed, Compressed, Frame End, unused |
| 190 | 3) Block |
| 191 | See Block Format Description |
| 192 | 4) Frame End |
| 193 | - 3 bytes, compatible with Block Header |
| 194 | */ |
| 195 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 196 | |
| 197 | /* Frame descriptor |
| 198 | |
| 199 | 1 byte, using : |
| 200 | bit 0-3 : windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN (see zstd_internal.h) |
| 201 | bit 4 : minmatch 4(0) or 3(1) |
| 202 | bit 5 : reserved (must be zero) |
| 203 | bit 6-7 : Frame content size : unknown, 1 byte, 2 bytes, 8 bytes |
| Yann Collet | fb79735 | 2016-03-13 11:08:40 +0100 | [diff] [blame] | 204 | |
| 205 | Optional : content size (0, 1, 2 or 8 bytes) |
| 206 | 0 : unknown |
| 207 | 1 : 0-255 bytes |
| 208 | 2 : 256 - 65535+256 |
| 209 | 8 : up to 16 exa |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 210 | */ |
| 211 | |
| 212 | |
| 213 | /* Compressed Block, format description |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 214 | |
| 215 | Block = Literal Section - Sequences Section |
| 216 | Prerequisite : size of (compressed) block, maximum size of regenerated data |
| 217 | |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 218 | 1) Literal Section |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 219 | |
| 220 | 1.1) Header : 1-5 bytes |
| 221 | flags: 2 bits |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 222 | 00 compressed by Huff0 |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 223 | 01 unused |
| 224 | 10 is Raw (uncompressed) |
| 225 | 11 is Rle |
| 226 | Note : using 01 => Huff0 with precomputed table ? |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 227 | Note : delta map ? => compressed ? |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 228 | |
| 229 | 1.1.1) Huff0-compressed literal block : 3-5 bytes |
| 230 | srcSize < 1 KB => 3 bytes (2-2-10-10) => single stream |
| 231 | srcSize < 1 KB => 3 bytes (2-2-10-10) |
| 232 | srcSize < 16KB => 4 bytes (2-2-14-14) |
| 233 | else => 5 bytes (2-2-18-18) |
| 234 | big endian convention |
| 235 | |
| 236 | 1.1.2) Raw (uncompressed) literal block header : 1-3 bytes |
| 237 | size : 5 bits: (IS_RAW<<6) + (0<<4) + size |
| 238 | 12 bits: (IS_RAW<<6) + (2<<4) + (size>>8) |
| 239 | size&255 |
| 240 | 20 bits: (IS_RAW<<6) + (3<<4) + (size>>16) |
| 241 | size>>8&255 |
| 242 | size&255 |
| 243 | |
| 244 | 1.1.3) Rle (repeated single byte) literal block header : 1-3 bytes |
| 245 | size : 5 bits: (IS_RLE<<6) + (0<<4) + size |
| 246 | 12 bits: (IS_RLE<<6) + (2<<4) + (size>>8) |
| 247 | size&255 |
| 248 | 20 bits: (IS_RLE<<6) + (3<<4) + (size>>16) |
| 249 | size>>8&255 |
| 250 | size&255 |
| 251 | |
| 252 | 1.1.4) Huff0-compressed literal block, using precomputed CTables : 3-5 bytes |
| 253 | srcSize < 1 KB => 3 bytes (2-2-10-10) => single stream |
| 254 | srcSize < 1 KB => 3 bytes (2-2-10-10) |
| 255 | srcSize < 16KB => 4 bytes (2-2-14-14) |
| 256 | else => 5 bytes (2-2-18-18) |
| 257 | big endian convention |
| 258 | |
| 259 | 1- CTable available (stored into workspace ?) |
| 260 | 2- Small input (fast heuristic ? Full comparison ? depend on clevel ?) |
| 261 | |
| 262 | |
| 263 | 1.2) Literal block content |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 264 | |
| 265 | 1.2.1) Huff0 block, using sizes from header |
| 266 | See Huff0 format |
| 267 | |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 268 | 1.2.2) Huff0 block, using prepared table |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 269 | |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 270 | 1.2.3) Raw content |
| 271 | |
| 272 | 1.2.4) single byte |
| 273 | |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 274 | |
| 275 | 2) Sequences section |
| 276 | TO DO |
| 277 | */ |
| 278 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 279 | static const size_t ZSTD_fcs_fieldSize[4] = { 0, 1, 2, 8 }; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 280 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 281 | /** ZSTD_frameHeaderSize() : |
| 282 | * srcSize must be >= ZSTD_frameHeaderSize_min. |
| 283 | * @return : size of the Frame Header */ |
| 284 | static size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize) |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 285 | { |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 286 | U32 fcsId; |
| 287 | if (srcSize < ZSTD_frameHeaderSize_min) return ERROR(srcSize_wrong); |
| 288 | fcsId = (((const BYTE*)src)[4]) >> 6; |
| 289 | return ZSTD_frameHeaderSize_min + ZSTD_fcs_fieldSize[fcsId]; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 290 | } |
| 291 | |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 292 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 293 | /** ZSTD_getFrameParams() : |
| 294 | * decode Frame Header, or provide expected `srcSize`. |
| 295 | * @return : 0, `fparamsPtr` is correctly filled, |
| Yann Collet | 1c2c2bc | 2016-03-15 01:33:36 +0100 | [diff] [blame] | 296 | * >0, `srcSize` is too small, result is expected `srcSize`, |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 297 | * or an error code, which can be tested using ZSTD_isError() */ |
| Yann Collet | 09b21ee | 2016-03-15 12:56:03 +0100 | [diff] [blame] | 298 | size_t ZSTD_getFrameParams(ZSTD_frameParams* fparamsPtr, const void* src, size_t srcSize) |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 299 | { |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 300 | const BYTE* ip = (const BYTE*)src; |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 301 | |
| 302 | if (srcSize < ZSTD_frameHeaderSize_min) return ZSTD_frameHeaderSize_min; |
| Yann Collet | d1b2684 | 2016-03-15 01:24:33 +0100 | [diff] [blame] | 303 | if (MEM_readLE32(src) != ZSTD_MAGICNUMBER) return ERROR(prefix_unknown); |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 304 | |
| Yann Collet | 1c2c2bc | 2016-03-15 01:33:36 +0100 | [diff] [blame] | 305 | /* ensure there is enough `srcSize` to fully read/decode frame header */ |
| Yann Collet | d1b2684 | 2016-03-15 01:24:33 +0100 | [diff] [blame] | 306 | { size_t const fhsize = ZSTD_frameHeaderSize(src, srcSize); |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 307 | if (srcSize < fhsize) return fhsize; } |
| 308 | |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 309 | memset(fparamsPtr, 0, sizeof(*fparamsPtr)); |
| Yann Collet | 1c2c2bc | 2016-03-15 01:33:36 +0100 | [diff] [blame] | 310 | { BYTE const frameDesc = ip[4]; |
| 311 | fparamsPtr->windowLog = (frameDesc & 0xF) + ZSTD_WINDOWLOG_ABSOLUTEMIN; |
| 312 | fparamsPtr->mml = (frameDesc & 0x10) ? MINMATCH-1 : MINMATCH; |
| 313 | if ((frameDesc & 0x20) != 0) return ERROR(frameParameter_unsupported); /* reserved 1 bit */ |
| 314 | switch(frameDesc >> 6) /* fcsId */ |
| 315 | { |
| 316 | default: /* impossible */ |
| 317 | case 0 : fparamsPtr->frameContentSize = 0; break; |
| 318 | case 1 : fparamsPtr->frameContentSize = ip[5]; break; |
| 319 | case 2 : fparamsPtr->frameContentSize = MEM_readLE16(ip+5)+256; break; |
| 320 | case 3 : fparamsPtr->frameContentSize = MEM_readLE64(ip+5); break; |
| 321 | } } |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 322 | return 0; |
| 323 | } |
| 324 | |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 325 | |
| 326 | /** ZSTD_decodeFrameHeader() : |
| Yann Collet | 1c2c2bc | 2016-03-15 01:33:36 +0100 | [diff] [blame] | 327 | * `srcSize` must be the size provided by ZSTD_frameHeaderSize(). |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 328 | * @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */ |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 329 | static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* zc, const void* src, size_t srcSize) |
| Yann Collet | 26415d3 | 2015-11-26 12:43:28 +0100 | [diff] [blame] | 330 | { |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 331 | size_t result = ZSTD_getFrameParams(&(zc->fParams), src, srcSize); |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 332 | if ((MEM_32bits()) && (zc->fParams.windowLog > 25)) return ERROR(frameParameter_unsupportedBy32bits); |
| Yann Collet | 00fd7a2 | 2015-11-28 16:03:22 +0100 | [diff] [blame] | 333 | return result; |
| Yann Collet | 26415d3 | 2015-11-26 12:43:28 +0100 | [diff] [blame] | 334 | } |
| 335 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 336 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 337 | typedef struct |
| 338 | { |
| 339 | blockType_t blockType; |
| 340 | U32 origSize; |
| 341 | } blockProperties_t; |
| 342 | |
| 343 | /*! ZSTD_getcBlockSize() : |
| 344 | * Provides the size of compressed block from block header `src` */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 345 | size_t ZSTD_getcBlockSize(const void* src, size_t srcSize, blockProperties_t* bpPtr) |
| 346 | { |
| 347 | const BYTE* const in = (const BYTE* const)src; |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 348 | size_t cSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 349 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 350 | if (srcSize < ZSTD_blockHeaderSize) return ERROR(srcSize_wrong); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 351 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 352 | bpPtr->blockType = (blockType_t)((*in) >> 6); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 353 | cSize = in[2] + (in[1]<<8) + ((in[0] & 7)<<16); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 354 | bpPtr->origSize = (bpPtr->blockType == bt_rle) ? cSize : 0; |
| 355 | |
| 356 | if (bpPtr->blockType == bt_end) return 0; |
| 357 | if (bpPtr->blockType == bt_rle) return 1; |
| 358 | return cSize; |
| 359 | } |
| 360 | |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 361 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 362 | static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity, const void* src, size_t srcSize) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 363 | { |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 364 | if (srcSize > dstCapacity) return ERROR(dstSize_tooSmall); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 365 | memcpy(dst, src, srcSize); |
| 366 | return srcSize; |
| 367 | } |
| 368 | |
| 369 | |
| Yann Collet | 953ce72 | 2016-02-04 15:28:14 +0100 | [diff] [blame] | 370 | /*! ZSTD_decodeLiteralsBlock() : |
| Yann Collet | 14983e7 | 2015-11-11 21:38:21 +0100 | [diff] [blame] | 371 | @return : nb of bytes read from src (< srcSize ) */ |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 372 | size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx, |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 373 | const void* src, size_t srcSize) /* note : srcSize < BLOCKSIZE */ |
| 374 | { |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 375 | const BYTE* const istart = (const BYTE*) src; |
| 376 | |
| 377 | /* any compressed block with literals segment must be at least this size */ |
| 378 | if (srcSize < MIN_CBLOCK_SIZE) return ERROR(corruption_detected); |
| 379 | |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 380 | switch(istart[0]>> 6) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 381 | { |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 382 | case IS_HUF: |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 383 | { |
| Yann Collet | afe0709 | 2016-01-25 04:10:46 +0100 | [diff] [blame] | 384 | size_t litSize, litCSize, singleStream=0; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 385 | U32 lhSize = ((istart[0]) >> 4) & 3; |
| 386 | switch(lhSize) |
| 387 | { |
| 388 | case 0: case 1: default: /* note : default is impossible, since lhSize into [0..3] */ |
| 389 | /* 2 - 2 - 10 - 10 */ |
| 390 | lhSize=3; |
| Yann Collet | afe0709 | 2016-01-25 04:10:46 +0100 | [diff] [blame] | 391 | singleStream = istart[0] & 16; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 392 | litSize = ((istart[0] & 15) << 6) + (istart[1] >> 2); |
| 393 | litCSize = ((istart[1] & 3) << 8) + istart[2]; |
| 394 | break; |
| 395 | case 2: |
| 396 | /* 2 - 2 - 14 - 14 */ |
| 397 | lhSize=4; |
| 398 | litSize = ((istart[0] & 15) << 10) + (istart[1] << 2) + (istart[2] >> 6); |
| 399 | litCSize = ((istart[2] & 63) << 8) + istart[3]; |
| 400 | break; |
| 401 | case 3: |
| 402 | /* 2 - 2 - 18 - 18 */ |
| 403 | lhSize=5; |
| 404 | litSize = ((istart[0] & 15) << 14) + (istart[1] << 6) + (istart[2] >> 2); |
| 405 | litCSize = ((istart[2] & 3) << 16) + (istart[3] << 8) + istart[4]; |
| 406 | break; |
| 407 | } |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 408 | if (litSize > ZSTD_BLOCKSIZE_MAX) return ERROR(corruption_detected); |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 409 | |
| Yann Collet | afe0709 | 2016-01-25 04:10:46 +0100 | [diff] [blame] | 410 | if (HUF_isError(singleStream ? |
| 411 | HUF_decompress1X2(dctx->litBuffer, litSize, istart+lhSize, litCSize) : |
| 412 | HUF_decompress (dctx->litBuffer, litSize, istart+lhSize, litCSize) )) |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 413 | return ERROR(corruption_detected); |
| 414 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 415 | dctx->litPtr = dctx->litBuffer; |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 416 | dctx->litBufSize = ZSTD_BLOCKSIZE_MAX+8; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 417 | dctx->litSize = litSize; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 418 | return litCSize + lhSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 419 | } |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 420 | case IS_PCH: |
| 421 | { |
| 422 | size_t errorCode; |
| 423 | size_t litSize, litCSize; |
| 424 | U32 lhSize = ((istart[0]) >> 4) & 3; |
| 425 | if (lhSize != 1) /* only case supported for now : small litSize, single stream */ |
| 426 | return ERROR(corruption_detected); |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 427 | if (!dctx->flagStaticTables) |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 428 | return ERROR(dictionary_corrupted); |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 429 | |
| 430 | /* 2 - 2 - 10 - 10 */ |
| 431 | lhSize=3; |
| 432 | litSize = ((istart[0] & 15) << 6) + (istart[1] >> 2); |
| 433 | litCSize = ((istart[1] & 3) << 8) + istart[2]; |
| 434 | |
| 435 | errorCode = HUF_decompress1X4_usingDTable(dctx->litBuffer, litSize, istart+lhSize, litCSize, dctx->hufTableX4); |
| 436 | if (HUF_isError(errorCode)) return ERROR(corruption_detected); |
| 437 | |
| 438 | dctx->litPtr = dctx->litBuffer; |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 439 | dctx->litBufSize = ZSTD_BLOCKSIZE_MAX+WILDCOPY_OVERLENGTH; |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 440 | dctx->litSize = litSize; |
| 441 | return litCSize + lhSize; |
| 442 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 443 | case IS_RAW: |
| 444 | { |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 445 | size_t litSize; |
| 446 | U32 lhSize = ((istart[0]) >> 4) & 3; |
| 447 | switch(lhSize) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 448 | { |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 449 | case 0: case 1: default: /* note : default is impossible, since lhSize into [0..3] */ |
| 450 | lhSize=1; |
| 451 | litSize = istart[0] & 31; |
| 452 | break; |
| 453 | case 2: |
| 454 | litSize = ((istart[0] & 15) << 8) + istart[1]; |
| 455 | break; |
| 456 | case 3: |
| 457 | litSize = ((istart[0] & 15) << 16) + (istart[1] << 8) + istart[2]; |
| 458 | break; |
| 459 | } |
| 460 | |
| Yann Collet | 61e16ce | 2016-01-31 02:04:15 +0100 | [diff] [blame] | 461 | if (lhSize+litSize+WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */ |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 462 | if (litSize+lhSize > srcSize) return ERROR(corruption_detected); |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 463 | memcpy(dctx->litBuffer, istart+lhSize, litSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 464 | dctx->litPtr = dctx->litBuffer; |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 465 | dctx->litBufSize = ZSTD_BLOCKSIZE_MAX+8; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 466 | dctx->litSize = litSize; |
| Yann Collet | bc4c8aa | 2016-01-25 17:26:01 +0100 | [diff] [blame] | 467 | return lhSize+litSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 468 | } |
| 469 | /* direct reference into compressed stream */ |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 470 | dctx->litPtr = istart+lhSize; |
| 471 | dctx->litBufSize = srcSize-lhSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 472 | dctx->litSize = litSize; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 473 | return lhSize+litSize; |
| 474 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 475 | case IS_RLE: |
| 476 | { |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 477 | size_t litSize; |
| 478 | U32 lhSize = ((istart[0]) >> 4) & 3; |
| 479 | switch(lhSize) |
| 480 | { |
| 481 | case 0: case 1: default: /* note : default is impossible, since lhSize into [0..3] */ |
| 482 | lhSize = 1; |
| 483 | litSize = istart[0] & 31; |
| 484 | break; |
| 485 | case 2: |
| 486 | litSize = ((istart[0] & 15) << 8) + istart[1]; |
| 487 | break; |
| 488 | case 3: |
| 489 | litSize = ((istart[0] & 15) << 16) + (istart[1] << 8) + istart[2]; |
| 490 | break; |
| 491 | } |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 492 | if (litSize > ZSTD_BLOCKSIZE_MAX) return ERROR(corruption_detected); |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 493 | memset(dctx->litBuffer, istart[lhSize], litSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 494 | dctx->litPtr = dctx->litBuffer; |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 495 | dctx->litBufSize = ZSTD_BLOCKSIZE_MAX+WILDCOPY_OVERLENGTH; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 496 | dctx->litSize = litSize; |
| Yann Collet | 59d1f79 | 2016-01-23 19:28:41 +0100 | [diff] [blame] | 497 | return lhSize+1; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 498 | } |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 499 | default: |
| 500 | return ERROR(corruption_detected); /* impossible */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 501 | } |
| 502 | } |
| 503 | |
| 504 | |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 505 | /*! ZSTD_buildSeqTable() : |
| 506 | @return : nb bytes read from src, |
| 507 | or an error code if it fails, testable with ZSTD_isError() |
| 508 | */ |
| 509 | static size_t ZSTD_buildSeqTable(FSE_DTable* DTable, U32 type, U32 rawBits, U32 maxLog, |
| 510 | const void* src, size_t srcSize) |
| 511 | { |
| Yann Collet | d1d210f | 2016-03-19 12:12:07 +0100 | [diff] [blame] | 512 | U32 max = (1<<rawBits)-1; |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 513 | switch(type) |
| 514 | { |
| 515 | case FSE_ENCODING_RLE : |
| 516 | if (!srcSize) return ERROR(srcSize_wrong); |
| Yann Collet | d1d210f | 2016-03-19 12:12:07 +0100 | [diff] [blame] | 517 | FSE_buildDTable_rle(DTable, (*(const BYTE*)src) & max); /* if *src > max, data is corrupted */ |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 518 | return 1; |
| 519 | case FSE_ENCODING_RAW : |
| 520 | FSE_buildDTable_raw(DTable, rawBits); |
| 521 | return 0; |
| 522 | case FSE_ENCODING_STATIC: |
| 523 | return 0; |
| 524 | default : /* impossible */ |
| 525 | case FSE_ENCODING_DYNAMIC : |
| Yann Collet | d1d210f | 2016-03-19 12:12:07 +0100 | [diff] [blame] | 526 | { U32 tableLog; |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 527 | S16 norm[MaxSeq+1]; |
| 528 | size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize); |
| 529 | if (FSE_isError(headerSize)) return ERROR(GENERIC); |
| 530 | if (tableLog > maxLog) return ERROR(corruption_detected); |
| 531 | FSE_buildDTable(DTable, norm, max, tableLog); |
| 532 | return headerSize; |
| 533 | } } |
| 534 | } |
| 535 | |
| 536 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 537 | size_t ZSTD_decodeSeqHeaders(int* nbSeq, const BYTE** dumpsPtr, size_t* dumpsLengthPtr, |
| 538 | FSE_DTable* DTableLL, FSE_DTable* DTableML, FSE_DTable* DTableOffb, |
| 539 | const void* src, size_t srcSize) |
| 540 | { |
| 541 | const BYTE* const istart = (const BYTE* const)src; |
| 542 | const BYTE* ip = istart; |
| 543 | const BYTE* const iend = istart + srcSize; |
| 544 | U32 LLtype, Offtype, MLtype; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 545 | |
| 546 | /* check */ |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 547 | if (srcSize < MIN_SEQUENCES_SIZE) return ERROR(srcSize_wrong); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 548 | |
| 549 | /* SeqHead */ |
| Yann Collet | 61e16ce | 2016-01-31 02:04:15 +0100 | [diff] [blame] | 550 | *nbSeq = *ip++; |
| 551 | if (*nbSeq==0) return 1; |
| Yann Collet | d409db6 | 2016-03-04 14:45:31 +0100 | [diff] [blame] | 552 | if (*nbSeq >= 0x7F) { |
| 553 | if (*nbSeq == 0xFF) |
| 554 | *nbSeq = MEM_readLE16(ip) + LONGNBSEQ, ip+=2; |
| 555 | else |
| 556 | *nbSeq = ((nbSeq[0]-0x80)<<8) + *ip++; |
| 557 | } |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 558 | |
| Yann Collet | d409db6 | 2016-03-04 14:45:31 +0100 | [diff] [blame] | 559 | /* FSE table descriptors */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 560 | LLtype = *ip >> 6; |
| 561 | Offtype = (*ip >> 4) & 3; |
| 562 | MLtype = (*ip >> 2) & 3; |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 563 | { size_t dumpsLength; |
| 564 | if (*ip & 2) { |
| 565 | dumpsLength = ip[2]; |
| 566 | dumpsLength += ip[1] << 8; |
| 567 | ip += 3; |
| 568 | } else { |
| 569 | dumpsLength = ip[1]; |
| 570 | dumpsLength += (ip[0] & 1) << 8; |
| 571 | ip += 2; |
| 572 | } |
| 573 | *dumpsPtr = ip; |
| 574 | ip += dumpsLength; |
| 575 | *dumpsLengthPtr = dumpsLength; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 576 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 577 | |
| 578 | /* check */ |
| 579 | if (ip > iend-3) return ERROR(srcSize_wrong); /* min : all 3 are "raw", hence no header, but at least xxLog bits per type */ |
| 580 | |
| Yann Collet | 39c596c | 2016-03-18 21:40:56 +0100 | [diff] [blame] | 581 | /* Build DTables */ |
| Yann Collet | 4db09ef | 2016-03-18 22:23:49 +0100 | [diff] [blame] | 582 | { size_t const bhSize = ZSTD_buildSeqTable(DTableLL, LLtype, LLbits, LLFSELog, ip, iend-ip); |
| 583 | if (ZSTD_isError(bhSize)) return ERROR(corruption_detected); |
| 584 | ip += bhSize; |
| 585 | } |
| 586 | { size_t const bhSize = ZSTD_buildSeqTable(DTableOffb, Offtype, Offbits, OffFSELog, ip, iend-ip); |
| 587 | if (ZSTD_isError(bhSize)) return ERROR(corruption_detected); |
| 588 | ip += bhSize; |
| 589 | } |
| 590 | { size_t const bhSize = ZSTD_buildSeqTable(DTableML, MLtype, MLbits, MLFSELog, ip, iend-ip); |
| 591 | if (ZSTD_isError(bhSize)) return ERROR(corruption_detected); |
| 592 | ip += bhSize; |
| 593 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 594 | |
| 595 | return ip-istart; |
| 596 | } |
| 597 | |
| 598 | |
| 599 | typedef struct { |
| 600 | size_t litLength; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 601 | size_t matchLength; |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 602 | size_t offset; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 603 | } seq_t; |
| 604 | |
| 605 | typedef struct { |
| 606 | BIT_DStream_t DStream; |
| 607 | FSE_DState_t stateLL; |
| 608 | FSE_DState_t stateOffb; |
| 609 | FSE_DState_t stateML; |
| 610 | size_t prevOffset; |
| 611 | const BYTE* dumps; |
| 612 | const BYTE* dumpsEnd; |
| 613 | } seqState_t; |
| 614 | |
| inikep | e9f30ea | 2016-02-03 12:53:07 +0100 | [diff] [blame] | 615 | |
| inikep | 6b3739c | 2016-02-22 15:53:42 +0100 | [diff] [blame] | 616 | static void ZSTD_decodeSequence(seq_t* seq, seqState_t* seqState, const U32 mls) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 617 | { |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 618 | const BYTE* dumps = seqState->dumps; |
| 619 | const BYTE* const de = seqState->dumpsEnd; |
| Yann Collet | 3fd164e | 2016-03-17 12:56:46 +0100 | [diff] [blame] | 620 | size_t litLength, offset; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 621 | |
| 622 | /* Literal length */ |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 623 | litLength = FSE_peakSymbol(&(seqState->stateLL)); |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 624 | if (litLength == MaxLL) { |
| Yann Collet | 3fd164e | 2016-03-17 12:56:46 +0100 | [diff] [blame] | 625 | const U32 add = *dumps++; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 626 | if (add < 255) litLength += add; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 627 | else { |
| Yann Collet | 7d8e6bd | 2016-02-02 17:30:37 +0100 | [diff] [blame] | 628 | litLength = MEM_readLE32(dumps) & 0xFFFFFF; /* no risk : dumps is always followed by seq tables > 1 byte */ |
| 629 | if (litLength&1) litLength>>=1, dumps += 3; |
| 630 | else litLength = (U16)(litLength)>>1, dumps += 2; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 631 | } |
| 632 | if (dumps >= de) dumps = de-1; /* late correction, to avoid read overflow (data is now corrupted anyway) */ |
| 633 | } |
| 634 | |
| 635 | /* Offset */ |
| Yann Collet | d1d210f | 2016-03-19 12:12:07 +0100 | [diff] [blame] | 636 | { static const U32 offsetPrefix[MaxOff+1] = { |
| 637 | 1 /*fake*/, 1, 2, 4, 8, 0x10, 0x20, 0x40, |
| 638 | 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, |
| 639 | 0x8000, 0x10000, 0x20000, 0x40000, 0x80000, 0x100000, 0x200000, 0x400000, |
| 640 | 0x800000, 0x1000000, 0x2000000, 0x4000000, /*fake*/ 1, 1, 1, 1 }; |
| 641 | U32 const offsetCode = FSE_peakSymbol(&(seqState->stateOffb)); /* <= maxOff, by table construction */ |
| 642 | U32 const nbBits = offsetCode ? offsetCode-1 : 0; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 643 | offset = offsetPrefix[offsetCode] + BIT_readBits(&(seqState->DStream), nbBits); |
| 644 | if (MEM_32bits()) BIT_reloadDStream(&(seqState->DStream)); |
| Yann Collet | b40287a | 2016-03-17 12:11:16 +0100 | [diff] [blame] | 645 | if (offsetCode==0) offset = litLength ? seq->offset : seqState->prevOffset; |
| Yann Collet | 55aa7f9 | 2015-11-20 12:04:52 +0100 | [diff] [blame] | 646 | if (offsetCode | !litLength) seqState->prevOffset = seq->offset; /* cmove */ |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 647 | FSE_decodeSymbol(&(seqState->stateOffb), &(seqState->DStream)); /* update */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 648 | } |
| 649 | |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 650 | /* Literal length update */ |
| 651 | FSE_decodeSymbol(&(seqState->stateLL), &(seqState->DStream)); /* update */ |
| 652 | if (MEM_32bits()) BIT_reloadDStream(&(seqState->DStream)); |
| 653 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 654 | /* MatchLength */ |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 655 | { size_t matchLength = FSE_decodeSymbol(&(seqState->stateML), &(seqState->DStream)); |
| Yann Collet | 3fd164e | 2016-03-17 12:56:46 +0100 | [diff] [blame] | 656 | if (matchLength == MaxML) { |
| 657 | const U32 add = *dumps++; |
| 658 | if (add < 255) matchLength += add; |
| 659 | else { |
| 660 | matchLength = MEM_readLE32(dumps) & 0xFFFFFF; /* no pb : dumps is always followed by seq tables > 1 byte */ |
| 661 | if (matchLength&1) matchLength>>=1, dumps += 3; |
| 662 | else matchLength = (U16)(matchLength)>>1, dumps += 2; |
| 663 | } |
| 664 | if (dumps >= de) dumps = de-1; /* late correction, to avoid read overflow (data is now corrupted anyway) */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 665 | } |
| Yann Collet | 3fd164e | 2016-03-17 12:56:46 +0100 | [diff] [blame] | 666 | matchLength += mls; |
| 667 | seq->matchLength = matchLength; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 668 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 669 | |
| 670 | /* save result */ |
| 671 | seq->litLength = litLength; |
| 672 | seq->offset = offset; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 673 | seqState->dumps = dumps; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 674 | |
| Yann Collet | 7d8e6bd | 2016-02-02 17:30:37 +0100 | [diff] [blame] | 675 | #if 0 /* debug */ |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 676 | { |
| 677 | static U64 totalDecoded = 0; |
| 678 | printf("pos %6u : %3u literals & match %3u bytes at distance %6u \n", |
| 679 | (U32)(totalDecoded), (U32)litLength, (U32)matchLength, (U32)offset); |
| 680 | totalDecoded += litLength + matchLength; |
| 681 | } |
| 682 | #endif |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 683 | } |
| 684 | |
| 685 | |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 686 | FORCE_INLINE size_t ZSTD_execSequence(BYTE* op, |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 687 | BYTE* const oend, seq_t sequence, |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 688 | const BYTE** litPtr, const BYTE* const litLimit_8, |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 689 | const BYTE* const base, const BYTE* const vBase, const BYTE* const dictEnd) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 690 | { |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 691 | static const int dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 }; /* added */ |
| 692 | static const int dec64table[] = { 8, 8, 8, 7, 8, 9,10,11 }; /* substracted */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 693 | BYTE* const oLitEnd = op + sequence.litLength; |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 694 | const size_t sequenceLength = sequence.litLength + sequence.matchLength; |
| 695 | BYTE* const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 696 | BYTE* const oend_8 = oend-8; |
| 697 | const BYTE* const litEnd = *litPtr + sequence.litLength; |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 698 | const BYTE* match = oLitEnd - sequence.offset; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 699 | |
| 700 | /* check */ |
| 701 | if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of 8 from oend */ |
| 702 | if (oMatchEnd > oend) return ERROR(dstSize_tooSmall); /* overwrite beyond dst buffer */ |
| 703 | if (litEnd > litLimit_8) return ERROR(corruption_detected); /* risk read beyond lit buffer */ |
| 704 | |
| 705 | /* copy Literals */ |
| 706 | ZSTD_wildcopy(op, *litPtr, sequence.litLength); /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */ |
| 707 | op = oLitEnd; |
| 708 | *litPtr = litEnd; /* update for next sequence */ |
| 709 | |
| 710 | /* copy Match */ |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 711 | if (sequence.offset > (size_t)(oLitEnd - base)) { |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 712 | /* offset beyond prefix */ |
| Yann Collet | 9f5ab1a | 2015-12-11 00:27:41 +0100 | [diff] [blame] | 713 | if (sequence.offset > (size_t)(oLitEnd - vBase)) |
| 714 | return ERROR(corruption_detected); |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 715 | match = dictEnd - (base-match); |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 716 | if (match + sequence.matchLength <= dictEnd) { |
| Yann Collet | 4bfe415 | 2015-12-06 13:18:37 +0100 | [diff] [blame] | 717 | memmove(oLitEnd, match, sequence.matchLength); |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 718 | return sequenceLength; |
| 719 | } |
| 720 | /* span extDict & currentPrefixSegment */ |
| Yann Collet | d1d210f | 2016-03-19 12:12:07 +0100 | [diff] [blame] | 721 | { size_t const length1 = dictEnd - match; |
| Yann Collet | 4bfe415 | 2015-12-06 13:18:37 +0100 | [diff] [blame] | 722 | memmove(oLitEnd, match, length1); |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 723 | op = oLitEnd + length1; |
| 724 | sequence.matchLength -= length1; |
| 725 | match = base; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 726 | } } |
| Yann Collet | 0f366c6 | 2015-11-12 16:19:30 +0100 | [diff] [blame] | 727 | |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 728 | /* match within prefix */ |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 729 | if (sequence.offset < 8) { |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 730 | /* close range match, overlap */ |
| 731 | const int sub2 = dec64table[sequence.offset]; |
| 732 | op[0] = match[0]; |
| 733 | op[1] = match[1]; |
| 734 | op[2] = match[2]; |
| 735 | op[3] = match[3]; |
| 736 | match += dec32table[sequence.offset]; |
| 737 | ZSTD_copy4(op+4, match); |
| 738 | match -= sub2; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 739 | } else { |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 740 | ZSTD_copy8(op, match); |
| 741 | } |
| 742 | op += 8; match += 8; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 743 | |
| inikep | a4dde25 | 2016-03-01 14:14:35 +0100 | [diff] [blame] | 744 | if (oMatchEnd > oend-(16-3)) { // 3 = MINMATCH |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 745 | if (op < oend_8) { |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 746 | ZSTD_wildcopy(op, match, oend_8 - op); |
| 747 | match += oend_8 - op; |
| 748 | op = oend_8; |
| 749 | } |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 750 | while (op < oMatchEnd) *op++ = *match++; |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 751 | } else { |
| Yann Collet | 44287a3 | 2015-11-30 23:13:56 +0100 | [diff] [blame] | 752 | ZSTD_wildcopy(op, match, sequence.matchLength-8); /* works even if matchLength < 8 */ |
| 753 | } |
| 754 | return sequenceLength; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 755 | } |
| 756 | |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 757 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 758 | static size_t ZSTD_decompressSequences( |
| Yann Collet | 5b78d2f | 2015-11-12 15:36:05 +0100 | [diff] [blame] | 759 | ZSTD_DCtx* dctx, |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 760 | void* dst, size_t maxDstSize, |
| 761 | const void* seqStart, size_t seqSize) |
| 762 | { |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 763 | const BYTE* ip = (const BYTE*)seqStart; |
| 764 | const BYTE* const iend = ip + seqSize; |
| 765 | BYTE* const ostart = (BYTE* const)dst; |
| 766 | BYTE* op = ostart; |
| 767 | BYTE* const oend = ostart + maxDstSize; |
| 768 | size_t errorCode, dumpsLength; |
| 769 | const BYTE* litPtr = dctx->litPtr; |
| 770 | const BYTE* const litLimit_8 = litPtr + dctx->litBufSize - 8; |
| 771 | const BYTE* const litEnd = litPtr + dctx->litSize; |
| 772 | int nbSeq; |
| 773 | const BYTE* dumps; |
| 774 | U32* DTableLL = dctx->LLTable; |
| 775 | U32* DTableML = dctx->MLTable; |
| 776 | U32* DTableOffb = dctx->OffTable; |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 777 | const BYTE* const base = (const BYTE*) (dctx->base); |
| 778 | const BYTE* const vBase = (const BYTE*) (dctx->vBase); |
| 779 | const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd); |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 780 | const U32 mls = dctx->fParams.mml; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 781 | |
| 782 | /* Build Decoding Tables */ |
| 783 | errorCode = ZSTD_decodeSeqHeaders(&nbSeq, &dumps, &dumpsLength, |
| 784 | DTableLL, DTableML, DTableOffb, |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 785 | ip, seqSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 786 | if (ZSTD_isError(errorCode)) return errorCode; |
| 787 | ip += errorCode; |
| 788 | |
| 789 | /* Regen sequences */ |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 790 | if (nbSeq) { |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 791 | seq_t sequence; |
| 792 | seqState_t seqState; |
| 793 | |
| 794 | memset(&sequence, 0, sizeof(sequence)); |
| Yann Collet | 61e16ce | 2016-01-31 02:04:15 +0100 | [diff] [blame] | 795 | sequence.offset = REPCODE_STARTVALUE; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 796 | seqState.dumps = dumps; |
| 797 | seqState.dumpsEnd = dumps + dumpsLength; |
| Yann Collet | 61e16ce | 2016-01-31 02:04:15 +0100 | [diff] [blame] | 798 | seqState.prevOffset = REPCODE_STARTVALUE; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 799 | errorCode = BIT_initDStream(&(seqState.DStream), ip, iend-ip); |
| 800 | if (ERR_isError(errorCode)) return ERROR(corruption_detected); |
| 801 | FSE_initDState(&(seqState.stateLL), &(seqState.DStream), DTableLL); |
| 802 | FSE_initDState(&(seqState.stateOffb), &(seqState.DStream), DTableOffb); |
| 803 | FSE_initDState(&(seqState.stateML), &(seqState.DStream), DTableML); |
| 804 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 805 | for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && nbSeq ; ) { |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 806 | size_t oneSeqSize; |
| 807 | nbSeq--; |
| inikep | 6b3739c | 2016-02-22 15:53:42 +0100 | [diff] [blame] | 808 | ZSTD_decodeSequence(&sequence, &seqState, mls); |
| Yann Collet | b3a2af9 | 2015-11-19 17:13:19 +0100 | [diff] [blame] | 809 | oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litLimit_8, base, vBase, dictEnd); |
| Yann Collet | 370b08e | 2016-03-08 00:03:59 +0100 | [diff] [blame] | 810 | if (ZSTD_isError(oneSeqSize)) |
| 811 | return oneSeqSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 812 | op += oneSeqSize; |
| 813 | } |
| 814 | |
| 815 | /* check if reached exact end */ |
| Yann Collet | 35f7de5 | 2016-01-31 02:51:03 +0100 | [diff] [blame] | 816 | if (nbSeq) return ERROR(corruption_detected); |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 817 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 818 | |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 819 | /* last literal segment */ |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 820 | { size_t const lastLLSize = litEnd - litPtr; |
| Yann Collet | e93d6ce | 2016-01-31 00:58:06 +0100 | [diff] [blame] | 821 | if (litPtr > litEnd) return ERROR(corruption_detected); /* too many literals already used */ |
| 822 | if (op+lastLLSize > oend) return ERROR(dstSize_tooSmall); |
| 823 | memcpy(op, litPtr, lastLLSize); |
| 824 | op += lastLLSize; |
| 825 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 826 | |
| 827 | return op-ostart; |
| 828 | } |
| 829 | |
| 830 | |
| Yann Collet | b012510 | 2016-01-09 02:00:10 +0100 | [diff] [blame] | 831 | static void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst) |
| 832 | { |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 833 | if (dst != dctx->previousDstEnd) { /* not contiguous */ |
| Yann Collet | b012510 | 2016-01-09 02:00:10 +0100 | [diff] [blame] | 834 | dctx->dictEnd = dctx->previousDstEnd; |
| 835 | dctx->vBase = (const char*)dst - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->base)); |
| 836 | dctx->base = dst; |
| 837 | dctx->previousDstEnd = dst; |
| 838 | } |
| 839 | } |
| 840 | |
| 841 | |
| 842 | static size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx, |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 843 | void* dst, size_t dstCapacity, |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 844 | const void* src, size_t srcSize) |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 845 | { /* blockType == blockCompressed */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 846 | const BYTE* ip = (const BYTE*)src; |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 847 | size_t litCSize; |
| 848 | |
| Yann Collet | 569b81a | 2016-03-16 15:26:51 +0100 | [diff] [blame] | 849 | if (srcSize >= ZSTD_BLOCKSIZE_MAX) return ERROR(srcSize_wrong); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 850 | |
| inikep | a4dde25 | 2016-03-01 14:14:35 +0100 | [diff] [blame] | 851 | ZSTD_LOG_BLOCK("%p: ZSTD_decompressBlock_internal searchLength=%d\n", dctx->base, dctx->params.searchLength); |
| 852 | |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 853 | /* Decode literals sub-block */ |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 854 | litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 855 | if (ZSTD_isError(litCSize)) return litCSize; |
| 856 | ip += litCSize; |
| 857 | srcSize -= litCSize; |
| 858 | |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 859 | return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 860 | } |
| 861 | |
| 862 | |
| Yann Collet | b012510 | 2016-01-09 02:00:10 +0100 | [diff] [blame] | 863 | size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 864 | void* dst, size_t dstCapacity, |
| Yann Collet | b012510 | 2016-01-09 02:00:10 +0100 | [diff] [blame] | 865 | const void* src, size_t srcSize) |
| 866 | { |
| 867 | ZSTD_checkContinuity(dctx, dst); |
| Yann Collet | b010b3b | 2016-02-03 12:39:34 +0100 | [diff] [blame] | 868 | return ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize); |
| Yann Collet | b012510 | 2016-01-09 02:00:10 +0100 | [diff] [blame] | 869 | } |
| 870 | |
| 871 | |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 872 | /*! ZSTD_decompress_continueDCtx() : |
| 873 | * `dctx` must have been properly initialized */ |
| 874 | static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx, |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 875 | void* dst, size_t dstCapacity, |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 876 | const void* src, size_t srcSize) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 877 | { |
| 878 | const BYTE* ip = (const BYTE*)src; |
| 879 | const BYTE* iend = ip + srcSize; |
| 880 | BYTE* const ostart = (BYTE* const)dst; |
| 881 | BYTE* op = ostart; |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 882 | BYTE* const oend = ostart + dstCapacity; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 883 | size_t remainingSize = srcSize; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 884 | blockProperties_t blockProperties; |
| 885 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 886 | /* check */ |
| 887 | if (srcSize < ZSTD_frameHeaderSize_min+ZSTD_blockHeaderSize) return ERROR(srcSize_wrong); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 888 | #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT==1) |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 889 | { const U32 magicNumber = MEM_readLE32(src); |
| 890 | if (ZSTD_isLegacy(magicNumber)) |
| 891 | return ZSTD_decompressLegacy(dst, dstCapacity, src, srcSize, magicNumber); |
| 892 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 893 | #endif |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 894 | |
| 895 | /* Frame Header */ |
| 896 | { size_t const frameHeaderSize = ZSTD_frameHeaderSize(src, ZSTD_frameHeaderSize_min); |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 897 | if (ZSTD_isError(frameHeaderSize)) return frameHeaderSize; |
| 898 | if (srcSize < frameHeaderSize+ZSTD_blockHeaderSize) return ERROR(srcSize_wrong); |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 899 | if (ZSTD_decodeFrameHeader(dctx, src, frameHeaderSize)) return ERROR(corruption_detected); |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 900 | ip += frameHeaderSize; remainingSize -= frameHeaderSize; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 901 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 902 | |
| 903 | /* Loop on each block */ |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 904 | while (1) { |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 905 | size_t decodedSize=0; |
| 906 | size_t cBlockSize = ZSTD_getcBlockSize(ip, iend-ip, &blockProperties); |
| 907 | if (ZSTD_isError(cBlockSize)) return cBlockSize; |
| 908 | |
| 909 | ip += ZSTD_blockHeaderSize; |
| 910 | remainingSize -= ZSTD_blockHeaderSize; |
| 911 | if (cBlockSize > remainingSize) return ERROR(srcSize_wrong); |
| 912 | |
| 913 | switch(blockProperties.blockType) |
| 914 | { |
| 915 | case bt_compressed: |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 916 | decodedSize = ZSTD_decompressBlock_internal(dctx, op, oend-op, ip, cBlockSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 917 | break; |
| 918 | case bt_raw : |
| Yann Collet | 0f366c6 | 2015-11-12 16:19:30 +0100 | [diff] [blame] | 919 | decodedSize = ZSTD_copyRawBlock(op, oend-op, ip, cBlockSize); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 920 | break; |
| 921 | case bt_rle : |
| 922 | return ERROR(GENERIC); /* not yet supported */ |
| 923 | break; |
| 924 | case bt_end : |
| 925 | /* end of frame */ |
| 926 | if (remainingSize) return ERROR(srcSize_wrong); |
| 927 | break; |
| 928 | default: |
| 929 | return ERROR(GENERIC); /* impossible */ |
| 930 | } |
| 931 | if (cBlockSize == 0) break; /* bt_end */ |
| 932 | |
| 933 | if (ZSTD_isError(decodedSize)) return decodedSize; |
| 934 | op += decodedSize; |
| 935 | ip += cBlockSize; |
| 936 | remainingSize -= cBlockSize; |
| 937 | } |
| 938 | |
| 939 | return op-ostart; |
| 940 | } |
| 941 | |
| Yann Collet | 31683c0 | 2015-12-18 01:26:48 +0100 | [diff] [blame] | 942 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 943 | size_t ZSTD_decompress_usingPreparedDCtx(ZSTD_DCtx* dctx, const ZSTD_DCtx* refDCtx, |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 944 | void* dst, size_t dstCapacity, |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 945 | const void* src, size_t srcSize) |
| 946 | { |
| 947 | ZSTD_copyDCtx(dctx, refDCtx); |
| 948 | ZSTD_checkContinuity(dctx, dst); |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 949 | return ZSTD_decompressFrame(dctx, dst, dstCapacity, src, srcSize); |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 950 | } |
| 951 | |
| 952 | |
| 953 | size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx, |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 954 | void* dst, size_t dstCapacity, |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 955 | const void* src, size_t srcSize, |
| 956 | const void* dict, size_t dictSize) |
| 957 | { |
| 958 | ZSTD_decompressBegin_usingDict(dctx, dict, dictSize); |
| inikep | a4dde25 | 2016-03-01 14:14:35 +0100 | [diff] [blame] | 959 | ZSTD_LOG_BLOCK("%p: ZSTD_decompressBegin_usingDict searchLength=%d\n", dctx->base, dctx->params.searchLength); |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 960 | ZSTD_checkContinuity(dctx, dst); |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 961 | return ZSTD_decompressFrame(dctx, dst, dstCapacity, src, srcSize); |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 962 | } |
| 963 | |
| 964 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 965 | size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize) |
| Yann Collet | 31683c0 | 2015-12-18 01:26:48 +0100 | [diff] [blame] | 966 | { |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 967 | return ZSTD_decompress_usingDict(dctx, dst, dstCapacity, src, srcSize, NULL, 0); |
| Yann Collet | 31683c0 | 2015-12-18 01:26:48 +0100 | [diff] [blame] | 968 | } |
| 969 | |
| Yann Collet | 0e491c0 | 2016-03-11 21:58:04 +0100 | [diff] [blame] | 970 | |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 971 | size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t srcSize) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 972 | { |
| Yann Collet | 3a3b72f | 2016-01-11 12:56:11 +0100 | [diff] [blame] | 973 | #if defined(ZSTD_HEAPMODE) && (ZSTD_HEAPMODE==1) |
| 974 | size_t regenSize; |
| 975 | ZSTD_DCtx* dctx = ZSTD_createDCtx(); |
| 976 | if (dctx==NULL) return ERROR(memory_allocation); |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 977 | regenSize = ZSTD_decompressDCtx(dctx, dst, dstCapacity, src, srcSize); |
| Yann Collet | 3a3b72f | 2016-01-11 12:56:11 +0100 | [diff] [blame] | 978 | ZSTD_freeDCtx(dctx); |
| 979 | return regenSize; |
| 980 | #else |
| Yann Collet | 31683c0 | 2015-12-18 01:26:48 +0100 | [diff] [blame] | 981 | ZSTD_DCtx dctx; |
| Yann Collet | 01e5b95 | 2016-03-19 14:14:31 +0100 | [diff] [blame^] | 982 | return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize); |
| Yann Collet | a768a30 | 2016-01-21 16:04:35 +0100 | [diff] [blame] | 983 | #endif |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 984 | } |
| 985 | |
| 986 | |
| Yann Collet | 346bffb | 2016-03-15 15:24:52 +0100 | [diff] [blame] | 987 | /*_****************************** |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 988 | * Streaming Decompression API |
| 989 | ********************************/ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 990 | size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx) |
| 991 | { |
| 992 | return dctx->expected; |
| 993 | } |
| 994 | |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 995 | size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 996 | { |
| 997 | /* Sanity check */ |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 998 | if (srcSize != dctx->expected) return ERROR(srcSize_wrong); |
| 999 | ZSTD_checkContinuity(dctx, dst); |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1000 | |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1001 | /* Decompress : frame header; part 1 */ |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1002 | switch (dctx->stage) |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1003 | { |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1004 | case ZSTDds_getFrameHeaderSize : |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1005 | { |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1006 | /* get frame header size */ |
| 1007 | if (srcSize != ZSTD_frameHeaderSize_min) return ERROR(srcSize_wrong); /* impossible */ |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 1008 | dctx->headerSize = ZSTD_frameHeaderSize(src, ZSTD_frameHeaderSize_min); |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1009 | if (ZSTD_isError(dctx->headerSize)) return dctx->headerSize; |
| 1010 | memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_min); |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1011 | if (dctx->headerSize > ZSTD_frameHeaderSize_min) { |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1012 | dctx->expected = dctx->headerSize - ZSTD_frameHeaderSize_min; |
| 1013 | dctx->stage = ZSTDds_decodeFrameHeader; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1014 | return 0; |
| 1015 | } |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1016 | dctx->expected = 0; /* not necessary to copy more */ |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1017 | } |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1018 | case ZSTDds_decodeFrameHeader: |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1019 | { |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1020 | /* get frame header */ |
| 1021 | size_t result; |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1022 | memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_min, src, dctx->expected); |
| Yann Collet | 03ea59b | 2016-03-12 01:25:40 +0100 | [diff] [blame] | 1023 | result = ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize); |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1024 | if (ZSTD_isError(result)) return result; |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1025 | dctx->expected = ZSTD_blockHeaderSize; |
| 1026 | dctx->stage = ZSTDds_decodeBlockHeader; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1027 | return 0; |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1028 | } |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1029 | case ZSTDds_decodeBlockHeader: |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1030 | { |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1031 | /* Decode block header */ |
| 1032 | blockProperties_t bp; |
| 1033 | size_t blockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp); |
| 1034 | if (ZSTD_isError(blockSize)) return blockSize; |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1035 | if (bp.blockType == bt_end) { |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1036 | dctx->expected = 0; |
| 1037 | dctx->stage = ZSTDds_getFrameHeaderSize; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1038 | } |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1039 | else { |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1040 | dctx->expected = blockSize; |
| 1041 | dctx->bType = bp.blockType; |
| 1042 | dctx->stage = ZSTDds_decompressBlock; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1043 | } |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1044 | return 0; |
| 1045 | } |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 1046 | case ZSTDds_decompressBlock: |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1047 | { |
| 1048 | /* Decompress : block content */ |
| 1049 | size_t rSize; |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1050 | switch(dctx->bType) |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1051 | { |
| 1052 | case bt_compressed: |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1053 | rSize = ZSTD_decompressBlock_internal(dctx, dst, maxDstSize, src, srcSize); |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1054 | break; |
| 1055 | case bt_raw : |
| 1056 | rSize = ZSTD_copyRawBlock(dst, maxDstSize, src, srcSize); |
| 1057 | break; |
| 1058 | case bt_rle : |
| 1059 | return ERROR(GENERIC); /* not yet handled */ |
| 1060 | break; |
| 1061 | case bt_end : /* should never happen (filtered at phase 1) */ |
| 1062 | rSize = 0; |
| 1063 | break; |
| 1064 | default: |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1065 | return ERROR(GENERIC); /* impossible */ |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1066 | } |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1067 | dctx->stage = ZSTDds_decodeBlockHeader; |
| 1068 | dctx->expected = ZSTD_blockHeaderSize; |
| 1069 | dctx->previousDstEnd = (char*)dst + rSize; |
| Yann Collet | 88fcd29 | 2015-11-25 14:42:45 +0100 | [diff] [blame] | 1070 | return rSize; |
| 1071 | } |
| 1072 | default: |
| 1073 | return ERROR(GENERIC); /* impossible */ |
| 1074 | } |
| Yann Collet | 5be2dd2 | 2015-11-11 13:43:58 +0100 | [diff] [blame] | 1075 | } |
| 1076 | |
| 1077 | |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1078 | static void ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize) |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 1079 | { |
| Yann Collet | 3742219 | 2016-01-25 16:54:05 +0100 | [diff] [blame] | 1080 | dctx->dictEnd = dctx->previousDstEnd; |
| 1081 | dctx->vBase = (const char*)dict - ((const char*)(dctx->previousDstEnd) - (const char*)(dctx->base)); |
| 1082 | dctx->base = dict; |
| 1083 | dctx->previousDstEnd = (const char*)dict + dictSize; |
| Yann Collet | 417890c | 2015-12-04 17:16:37 +0100 | [diff] [blame] | 1084 | } |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1085 | |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1086 | static size_t ZSTD_loadEntropy(ZSTD_DCtx* dctx, const void* dict, size_t dictSize) |
| 1087 | { |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 1088 | size_t hSize, offcodeHeaderSize, matchlengthHeaderSize, errorCode, litlengthHeaderSize; |
| 1089 | short offcodeNCount[MaxOff+1]; |
| 1090 | U32 offcodeMaxValue=MaxOff, offcodeLog=OffFSELog; |
| 1091 | short matchlengthNCount[MaxML+1]; |
| 1092 | unsigned matchlengthMaxValue = MaxML, matchlengthLog = MLFSELog; |
| 1093 | short litlengthNCount[MaxLL+1]; |
| 1094 | unsigned litlengthMaxValue = MaxLL, litlengthLog = LLFSELog; |
| 1095 | |
| 1096 | hSize = HUF_readDTableX4(dctx->hufTableX4, dict, dictSize); |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1097 | if (HUF_isError(hSize)) return ERROR(dictionary_corrupted); |
| Yann Collet | fb810d6 | 2016-01-28 00:18:06 +0100 | [diff] [blame] | 1098 | dict = (const char*)dict + hSize; |
| 1099 | dictSize -= hSize; |
| 1100 | |
| 1101 | offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dict, dictSize); |
| 1102 | if (FSE_isError(offcodeHeaderSize)) return ERROR(dictionary_corrupted); |
| 1103 | errorCode = FSE_buildDTable(dctx->OffTable, offcodeNCount, offcodeMaxValue, offcodeLog); |
| 1104 | if (FSE_isError(errorCode)) return ERROR(dictionary_corrupted); |
| 1105 | dict = (const char*)dict + offcodeHeaderSize; |
| 1106 | dictSize -= offcodeHeaderSize; |
| 1107 | |
| 1108 | matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dict, dictSize); |
| 1109 | if (FSE_isError(matchlengthHeaderSize)) return ERROR(dictionary_corrupted); |
| 1110 | errorCode = FSE_buildDTable(dctx->MLTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog); |
| 1111 | if (FSE_isError(errorCode)) return ERROR(dictionary_corrupted); |
| 1112 | dict = (const char*)dict + matchlengthHeaderSize; |
| 1113 | dictSize -= matchlengthHeaderSize; |
| 1114 | |
| 1115 | litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dict, dictSize); |
| 1116 | if (FSE_isError(litlengthHeaderSize)) return ERROR(dictionary_corrupted); |
| 1117 | errorCode = FSE_buildDTable(dctx->LLTable, litlengthNCount, litlengthMaxValue, litlengthLog); |
| 1118 | if (FSE_isError(errorCode)) return ERROR(dictionary_corrupted); |
| 1119 | |
| 1120 | dctx->flagStaticTables = 1; |
| 1121 | return hSize + offcodeHeaderSize + matchlengthHeaderSize + litlengthHeaderSize; |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1122 | } |
| 1123 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1124 | static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize) |
| Yann Collet | b923f65 | 2016-01-26 03:14:20 +0100 | [diff] [blame] | 1125 | { |
| 1126 | size_t eSize; |
| 1127 | U32 magic = MEM_readLE32(dict); |
| 1128 | if (magic != ZSTD_DICT_MAGIC) { |
| 1129 | /* pure content mode */ |
| 1130 | ZSTD_refDictContent(dctx, dict, dictSize); |
| 1131 | return 0; |
| 1132 | } |
| 1133 | /* load entropy tables */ |
| 1134 | dict = (const char*)dict + 4; |
| 1135 | dictSize -= 4; |
| 1136 | eSize = ZSTD_loadEntropy(dctx, dict, dictSize); |
| 1137 | if (ZSTD_isError(eSize)) return ERROR(dictionary_corrupted); |
| 1138 | |
| 1139 | /* reference dictionary content */ |
| 1140 | dict = (const char*)dict + eSize; |
| 1141 | dictSize -= eSize; |
| 1142 | ZSTD_refDictContent(dctx, dict, dictSize); |
| 1143 | |
| 1144 | return 0; |
| 1145 | } |
| 1146 | |
| Yann Collet | 7b51a29 | 2016-01-26 15:58:49 +0100 | [diff] [blame] | 1147 | |
| 1148 | size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize) |
| 1149 | { |
| 1150 | size_t errorCode; |
| 1151 | errorCode = ZSTD_decompressBegin(dctx); |
| 1152 | if (ZSTD_isError(errorCode)) return errorCode; |
| 1153 | |
| 1154 | if (dict && dictSize) { |
| 1155 | errorCode = ZSTD_decompress_insertDictionary(dctx, dict, dictSize); |
| 1156 | if (ZSTD_isError(errorCode)) return ERROR(dictionary_corrupted); |
| 1157 | } |
| 1158 | |
| 1159 | return 0; |
| 1160 | } |
| 1161 | |