[8ebc79b] | 1 | /* ****************************************************************** |
---|
| 2 | FSE : Finite State Entropy codec |
---|
| 3 | Public Prototypes declaration |
---|
| 4 | Copyright (C) 2013-2016, Yann Collet. |
---|
| 5 | |
---|
| 6 | BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) |
---|
| 7 | |
---|
| 8 | Redistribution and use in source and binary forms, with or without |
---|
| 9 | modification, are permitted provided that the following conditions are |
---|
| 10 | met: |
---|
| 11 | |
---|
| 12 | * Redistributions of source code must retain the above copyright |
---|
| 13 | notice, this list of conditions and the following disclaimer. |
---|
| 14 | * Redistributions in binary form must reproduce the above |
---|
| 15 | copyright notice, this list of conditions and the following disclaimer |
---|
| 16 | in the documentation and/or other materials provided with the |
---|
| 17 | distribution. |
---|
| 18 | |
---|
| 19 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
---|
| 20 | "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
---|
| 21 | LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
---|
| 22 | A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
---|
| 23 | OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
---|
| 24 | SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
---|
| 25 | LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
---|
| 26 | DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
---|
| 27 | THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
---|
| 28 | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
---|
| 29 | OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
---|
| 30 | |
---|
| 31 | You can contact the author at : |
---|
| 32 | - Source repository : https://github.com/Cyan4973/FiniteStateEntropy |
---|
| 33 | ****************************************************************** */ |
---|
| 34 | #ifndef FSE_H |
---|
| 35 | #define FSE_H |
---|
| 36 | |
---|
| 37 | #if defined (__cplusplus) |
---|
| 38 | extern "C" { |
---|
| 39 | #endif |
---|
| 40 | |
---|
| 41 | |
---|
| 42 | /*-***************************************** |
---|
| 43 | * Dependencies |
---|
| 44 | ******************************************/ |
---|
| 45 | #include <stddef.h> /* size_t, ptrdiff_t */ |
---|
| 46 | |
---|
| 47 | |
---|
| 48 | /*-**************************************** |
---|
| 49 | * FSE simple functions |
---|
| 50 | ******************************************/ |
---|
| 51 | /*! FSE_compress() : |
---|
| 52 | Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'. |
---|
| 53 | 'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize). |
---|
| 54 | @return : size of compressed data (<= dstCapacity). |
---|
| 55 | Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!! |
---|
| 56 | if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead. |
---|
| 57 | if FSE_isError(return), compression failed (more details using FSE_getErrorName()) |
---|
| 58 | */ |
---|
| 59 | size_t FSE_compress(void* dst, size_t dstCapacity, |
---|
| 60 | const void* src, size_t srcSize); |
---|
| 61 | |
---|
| 62 | /*! FSE_decompress(): |
---|
| 63 | Decompress FSE data from buffer 'cSrc', of size 'cSrcSize', |
---|
| 64 | into already allocated destination buffer 'dst', of size 'dstCapacity'. |
---|
| 65 | @return : size of regenerated data (<= maxDstSize), |
---|
| 66 | or an error code, which can be tested using FSE_isError() . |
---|
| 67 | |
---|
| 68 | ** Important ** : FSE_decompress() does not decompress non-compressible nor RLE data !!! |
---|
| 69 | Why ? : making this distinction requires a header. |
---|
| 70 | Header management is intentionally delegated to the user layer, which can better manage special cases. |
---|
| 71 | */ |
---|
| 72 | size_t FSE_decompress(void* dst, size_t dstCapacity, |
---|
| 73 | const void* cSrc, size_t cSrcSize); |
---|
| 74 | |
---|
| 75 | |
---|
| 76 | /*-***************************************** |
---|
| 77 | * Tool functions |
---|
| 78 | ******************************************/ |
---|
| 79 | size_t FSE_compressBound(size_t size); /* maximum compressed size */ |
---|
| 80 | |
---|
| 81 | /* Error Management */ |
---|
| 82 | unsigned FSE_isError(size_t code); /* tells if a return value is an error code */ |
---|
| 83 | const char* FSE_getErrorName(size_t code); /* provides error code string (useful for debugging) */ |
---|
| 84 | |
---|
| 85 | |
---|
| 86 | /*-***************************************** |
---|
| 87 | * FSE advanced functions |
---|
| 88 | ******************************************/ |
---|
| 89 | /*! FSE_compress2() : |
---|
| 90 | Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog' |
---|
| 91 | Both parameters can be defined as '0' to mean : use default value |
---|
| 92 | @return : size of compressed data |
---|
| 93 | Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!! |
---|
| 94 | if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression. |
---|
| 95 | if FSE_isError(return), it's an error code. |
---|
| 96 | */ |
---|
| 97 | size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog); |
---|
| 98 | |
---|
| 99 | |
---|
| 100 | /*-***************************************** |
---|
| 101 | * FSE detailed API |
---|
| 102 | ******************************************/ |
---|
| 103 | /*! |
---|
| 104 | FSE_compress() does the following: |
---|
| 105 | 1. count symbol occurrence from source[] into table count[] |
---|
| 106 | 2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog) |
---|
| 107 | 3. save normalized counters to memory buffer using writeNCount() |
---|
| 108 | 4. build encoding table 'CTable' from normalized counters |
---|
| 109 | 5. encode the data stream using encoding table 'CTable' |
---|
| 110 | |
---|
| 111 | FSE_decompress() does the following: |
---|
| 112 | 1. read normalized counters with readNCount() |
---|
| 113 | 2. build decoding table 'DTable' from normalized counters |
---|
| 114 | 3. decode the data stream using decoding table 'DTable' |
---|
| 115 | |
---|
| 116 | The following API allows targeting specific sub-functions for advanced tasks. |
---|
| 117 | For example, it's possible to compress several blocks using the same 'CTable', |
---|
| 118 | or to save and provide normalized distribution using external method. |
---|
| 119 | */ |
---|
| 120 | |
---|
| 121 | /* *** COMPRESSION *** */ |
---|
| 122 | |
---|
| 123 | /*! FSE_count(): |
---|
| 124 | Provides the precise count of each byte within a table 'count'. |
---|
| 125 | 'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1). |
---|
| 126 | *maxSymbolValuePtr will be updated if detected smaller than initial value. |
---|
| 127 | @return : the count of the most frequent symbol (which is not identified). |
---|
| 128 | if return == srcSize, there is only one symbol. |
---|
| 129 | Can also return an error code, which can be tested with FSE_isError(). */ |
---|
| 130 | size_t FSE_count(unsigned* count, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize); |
---|
| 131 | |
---|
| 132 | /*! FSE_optimalTableLog(): |
---|
| 133 | dynamically downsize 'tableLog' when conditions are met. |
---|
| 134 | It saves CPU time, by using smaller tables, while preserving or even improving compression ratio. |
---|
| 135 | @return : recommended tableLog (necessarily <= 'maxTableLog') */ |
---|
| 136 | unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue); |
---|
| 137 | |
---|
| 138 | /*! FSE_normalizeCount(): |
---|
| 139 | normalize counts so that sum(count[]) == Power_of_2 (2^tableLog) |
---|
| 140 | 'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1). |
---|
| 141 | @return : tableLog, |
---|
| 142 | or an errorCode, which can be tested using FSE_isError() */ |
---|
| 143 | size_t FSE_normalizeCount(short* normalizedCounter, unsigned tableLog, const unsigned* count, size_t srcSize, unsigned maxSymbolValue); |
---|
| 144 | |
---|
| 145 | /*! FSE_NCountWriteBound(): |
---|
| 146 | Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'. |
---|
| 147 | Typically useful for allocation purpose. */ |
---|
| 148 | size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog); |
---|
| 149 | |
---|
| 150 | /*! FSE_writeNCount(): |
---|
| 151 | Compactly save 'normalizedCounter' into 'buffer'. |
---|
| 152 | @return : size of the compressed table, |
---|
| 153 | or an errorCode, which can be tested using FSE_isError(). */ |
---|
| 154 | size_t FSE_writeNCount (void* buffer, size_t bufferSize, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog); |
---|
| 155 | |
---|
| 156 | |
---|
| 157 | /*! Constructor and Destructor of FSE_CTable. |
---|
| 158 | Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */ |
---|
| 159 | typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */ |
---|
| 160 | FSE_CTable* FSE_createCTable (unsigned tableLog, unsigned maxSymbolValue); |
---|
| 161 | void FSE_freeCTable (FSE_CTable* ct); |
---|
| 162 | |
---|
| 163 | /*! FSE_buildCTable(): |
---|
| 164 | Builds `ct`, which must be already allocated, using FSE_createCTable(). |
---|
| 165 | @return : 0, or an errorCode, which can be tested using FSE_isError() */ |
---|
| 166 | size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog); |
---|
| 167 | |
---|
| 168 | /*! FSE_compress_usingCTable(): |
---|
| 169 | Compress `src` using `ct` into `dst` which must be already allocated. |
---|
| 170 | @return : size of compressed data (<= `dstCapacity`), |
---|
| 171 | or 0 if compressed data could not fit into `dst`, |
---|
| 172 | or an errorCode, which can be tested using FSE_isError() */ |
---|
| 173 | size_t FSE_compress_usingCTable (void* dst, size_t dstCapacity, const void* src, size_t srcSize, const FSE_CTable* ct); |
---|
| 174 | |
---|
| 175 | /*! |
---|
| 176 | Tutorial : |
---|
| 177 | ---------- |
---|
| 178 | The first step is to count all symbols. FSE_count() does this job very fast. |
---|
| 179 | Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells. |
---|
| 180 | 'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0] |
---|
| 181 | maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value) |
---|
| 182 | FSE_count() will return the number of occurrence of the most frequent symbol. |
---|
| 183 | This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility. |
---|
| 184 | If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()). |
---|
| 185 | |
---|
| 186 | The next step is to normalize the frequencies. |
---|
| 187 | FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'. |
---|
| 188 | It also guarantees a minimum of 1 to any Symbol with frequency >= 1. |
---|
| 189 | You can use 'tableLog'==0 to mean "use default tableLog value". |
---|
| 190 | If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(), |
---|
| 191 | which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default"). |
---|
| 192 | |
---|
| 193 | The result of FSE_normalizeCount() will be saved into a table, |
---|
| 194 | called 'normalizedCounter', which is a table of signed short. |
---|
| 195 | 'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells. |
---|
| 196 | The return value is tableLog if everything proceeded as expected. |
---|
| 197 | It is 0 if there is a single symbol within distribution. |
---|
| 198 | If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()). |
---|
| 199 | |
---|
| 200 | 'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount(). |
---|
| 201 | 'buffer' must be already allocated. |
---|
| 202 | For guaranteed success, buffer size must be at least FSE_headerBound(). |
---|
| 203 | The result of the function is the number of bytes written into 'buffer'. |
---|
| 204 | If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small). |
---|
| 205 | |
---|
| 206 | 'normalizedCounter' can then be used to create the compression table 'CTable'. |
---|
| 207 | The space required by 'CTable' must be already allocated, using FSE_createCTable(). |
---|
| 208 | You can then use FSE_buildCTable() to fill 'CTable'. |
---|
| 209 | If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()). |
---|
| 210 | |
---|
| 211 | 'CTable' can then be used to compress 'src', with FSE_compress_usingCTable(). |
---|
| 212 | Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize' |
---|
| 213 | The function returns the size of compressed data (without header), necessarily <= `dstCapacity`. |
---|
| 214 | If it returns '0', compressed data could not fit into 'dst'. |
---|
| 215 | If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()). |
---|
| 216 | */ |
---|
| 217 | |
---|
| 218 | |
---|
| 219 | /* *** DECOMPRESSION *** */ |
---|
| 220 | |
---|
| 221 | /*! FSE_readNCount(): |
---|
| 222 | Read compactly saved 'normalizedCounter' from 'rBuffer'. |
---|
| 223 | @return : size read from 'rBuffer', |
---|
| 224 | or an errorCode, which can be tested using FSE_isError(). |
---|
| 225 | maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */ |
---|
| 226 | size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSymbolValuePtr, unsigned* tableLogPtr, const void* rBuffer, size_t rBuffSize); |
---|
| 227 | |
---|
| 228 | /*! Constructor and Destructor of FSE_DTable. |
---|
| 229 | Note that its size depends on 'tableLog' */ |
---|
| 230 | typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */ |
---|
| 231 | FSE_DTable* FSE_createDTable(unsigned tableLog); |
---|
| 232 | void FSE_freeDTable(FSE_DTable* dt); |
---|
| 233 | |
---|
| 234 | /*! FSE_buildDTable(): |
---|
| 235 | Builds 'dt', which must be already allocated, using FSE_createDTable(). |
---|
| 236 | return : 0, or an errorCode, which can be tested using FSE_isError() */ |
---|
| 237 | size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog); |
---|
| 238 | |
---|
| 239 | /*! FSE_decompress_usingDTable(): |
---|
| 240 | Decompress compressed source `cSrc` of size `cSrcSize` using `dt` |
---|
| 241 | into `dst` which must be already allocated. |
---|
| 242 | @return : size of regenerated data (necessarily <= `dstCapacity`), |
---|
| 243 | or an errorCode, which can be tested using FSE_isError() */ |
---|
| 244 | size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt); |
---|
| 245 | |
---|
| 246 | /*! |
---|
| 247 | Tutorial : |
---|
| 248 | ---------- |
---|
| 249 | (Note : these functions only decompress FSE-compressed blocks. |
---|
| 250 | If block is uncompressed, use memcpy() instead |
---|
| 251 | If block is a single repeated byte, use memset() instead ) |
---|
| 252 | |
---|
| 253 | The first step is to obtain the normalized frequencies of symbols. |
---|
| 254 | This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount(). |
---|
| 255 | 'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short. |
---|
| 256 | In practice, that means it's necessary to know 'maxSymbolValue' beforehand, |
---|
| 257 | or size the table to handle worst case situations (typically 256). |
---|
| 258 | FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'. |
---|
| 259 | The result of FSE_readNCount() is the number of bytes read from 'rBuffer'. |
---|
| 260 | Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that. |
---|
| 261 | If there is an error, the function will return an error code, which can be tested using FSE_isError(). |
---|
| 262 | |
---|
| 263 | The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'. |
---|
| 264 | This is performed by the function FSE_buildDTable(). |
---|
| 265 | The space required by 'FSE_DTable' must be already allocated using FSE_createDTable(). |
---|
| 266 | If there is an error, the function will return an error code, which can be tested using FSE_isError(). |
---|
| 267 | |
---|
| 268 | `FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable(). |
---|
| 269 | `cSrcSize` must be strictly correct, otherwise decompression will fail. |
---|
| 270 | FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`). |
---|
| 271 | If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small) |
---|
| 272 | */ |
---|
| 273 | |
---|
| 274 | |
---|
| 275 | #ifdef FSE_STATIC_LINKING_ONLY |
---|
| 276 | |
---|
| 277 | /* *** Dependency *** */ |
---|
| 278 | #include "bitstream.h" |
---|
| 279 | |
---|
| 280 | |
---|
| 281 | /* ***************************************** |
---|
| 282 | * Static allocation |
---|
| 283 | *******************************************/ |
---|
| 284 | /* FSE buffer bounds */ |
---|
| 285 | #define FSE_NCOUNTBOUND 512 |
---|
| 286 | #define FSE_BLOCKBOUND(size) (size + (size>>7)) |
---|
| 287 | #define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */ |
---|
| 288 | |
---|
| 289 | /* It is possible to statically allocate FSE CTable/DTable as a table of unsigned using below macros */ |
---|
| 290 | #define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1<<(maxTableLog-1)) + ((maxSymbolValue+1)*2)) |
---|
| 291 | #define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1<<maxTableLog)) |
---|
| 292 | |
---|
| 293 | |
---|
| 294 | /* ***************************************** |
---|
| 295 | * FSE advanced API |
---|
| 296 | *******************************************/ |
---|
| 297 | size_t FSE_countFast(unsigned* count, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize); |
---|
| 298 | /**< same as FSE_count(), but blindly trusts that all byte values within src are <= *maxSymbolValuePtr */ |
---|
| 299 | |
---|
| 300 | unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus); |
---|
| 301 | /**< same as FSE_optimalTableLog(), which used `minus==2` */ |
---|
| 302 | |
---|
| 303 | size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits); |
---|
| 304 | /**< build a fake FSE_CTable, designed to not compress an input, where each symbol uses nbBits */ |
---|
| 305 | |
---|
| 306 | size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue); |
---|
| 307 | /**< build a fake FSE_CTable, designed to compress always the same symbolValue */ |
---|
| 308 | |
---|
| 309 | size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits); |
---|
| 310 | /**< build a fake FSE_DTable, designed to read an uncompressed bitstream where each symbol uses nbBits */ |
---|
| 311 | |
---|
| 312 | size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue); |
---|
| 313 | /**< build a fake FSE_DTable, designed to always generate the same symbolValue */ |
---|
| 314 | |
---|
| 315 | |
---|
| 316 | /* ***************************************** |
---|
| 317 | * FSE symbol compression API |
---|
| 318 | *******************************************/ |
---|
| 319 | /*! |
---|
| 320 | This API consists of small unitary functions, which highly benefit from being inlined. |
---|
| 321 | You will want to enable link-time-optimization to ensure these functions are properly inlined in your binary. |
---|
| 322 | Visual seems to do it automatically. |
---|
| 323 | For gcc or clang, you'll need to add -flto flag at compilation and linking stages. |
---|
| 324 | If none of these solutions is applicable, include "fse.c" directly. |
---|
| 325 | */ |
---|
| 326 | typedef struct |
---|
| 327 | { |
---|
| 328 | ptrdiff_t value; |
---|
| 329 | const void* stateTable; |
---|
| 330 | const void* symbolTT; |
---|
| 331 | unsigned stateLog; |
---|
| 332 | } FSE_CState_t; |
---|
| 333 | |
---|
| 334 | static void FSE_initCState(FSE_CState_t* CStatePtr, const FSE_CTable* ct); |
---|
| 335 | |
---|
| 336 | static void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* CStatePtr, unsigned symbol); |
---|
| 337 | |
---|
| 338 | static void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* CStatePtr); |
---|
| 339 | |
---|
| 340 | /**< |
---|
| 341 | These functions are inner components of FSE_compress_usingCTable(). |
---|
| 342 | They allow the creation of custom streams, mixing multiple tables and bit sources. |
---|
| 343 | |
---|
| 344 | A key property to keep in mind is that encoding and decoding are done **in reverse direction**. |
---|
| 345 | So the first symbol you will encode is the last you will decode, like a LIFO stack. |
---|
| 346 | |
---|
| 347 | You will need a few variables to track your CStream. They are : |
---|
| 348 | |
---|
| 349 | FSE_CTable ct; // Provided by FSE_buildCTable() |
---|
| 350 | BIT_CStream_t bitStream; // bitStream tracking structure |
---|
| 351 | FSE_CState_t state; // State tracking structure (can have several) |
---|
| 352 | |
---|
| 353 | |
---|
| 354 | The first thing to do is to init bitStream and state. |
---|
| 355 | size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize); |
---|
| 356 | FSE_initCState(&state, ct); |
---|
| 357 | |
---|
| 358 | Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError(); |
---|
| 359 | You can then encode your input data, byte after byte. |
---|
| 360 | FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time. |
---|
| 361 | Remember decoding will be done in reverse direction. |
---|
| 362 | FSE_encodeByte(&bitStream, &state, symbol); |
---|
| 363 | |
---|
| 364 | At any time, you can also add any bit sequence. |
---|
| 365 | Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders |
---|
| 366 | BIT_addBits(&bitStream, bitField, nbBits); |
---|
| 367 | |
---|
| 368 | The above methods don't commit data to memory, they just store it into local register, for speed. |
---|
| 369 | Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t). |
---|
| 370 | Writing data to memory is a manual operation, performed by the flushBits function. |
---|
| 371 | BIT_flushBits(&bitStream); |
---|
| 372 | |
---|
| 373 | Your last FSE encoding operation shall be to flush your last state value(s). |
---|
| 374 | FSE_flushState(&bitStream, &state); |
---|
| 375 | |
---|
| 376 | Finally, you must close the bitStream. |
---|
| 377 | The function returns the size of CStream in bytes. |
---|
| 378 | If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible) |
---|
| 379 | If there is an error, it returns an errorCode (which can be tested using FSE_isError()). |
---|
| 380 | size_t size = BIT_closeCStream(&bitStream); |
---|
| 381 | */ |
---|
| 382 | |
---|
| 383 | |
---|
| 384 | /* ***************************************** |
---|
| 385 | * FSE symbol decompression API |
---|
| 386 | *******************************************/ |
---|
| 387 | typedef struct |
---|
| 388 | { |
---|
| 389 | size_t state; |
---|
| 390 | const void* table; /* precise table may vary, depending on U16 */ |
---|
| 391 | } FSE_DState_t; |
---|
| 392 | |
---|
| 393 | |
---|
| 394 | static void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt); |
---|
| 395 | |
---|
| 396 | static unsigned char FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD); |
---|
| 397 | |
---|
| 398 | static unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr); |
---|
| 399 | |
---|
| 400 | /**< |
---|
| 401 | Let's now decompose FSE_decompress_usingDTable() into its unitary components. |
---|
| 402 | You will decode FSE-encoded symbols from the bitStream, |
---|
| 403 | and also any other bitFields you put in, **in reverse order**. |
---|
| 404 | |
---|
| 405 | You will need a few variables to track your bitStream. They are : |
---|
| 406 | |
---|
| 407 | BIT_DStream_t DStream; // Stream context |
---|
| 408 | FSE_DState_t DState; // State context. Multiple ones are possible |
---|
| 409 | FSE_DTable* DTablePtr; // Decoding table, provided by FSE_buildDTable() |
---|
| 410 | |
---|
| 411 | The first thing to do is to init the bitStream. |
---|
| 412 | errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize); |
---|
| 413 | |
---|
| 414 | You should then retrieve your initial state(s) |
---|
| 415 | (in reverse flushing order if you have several ones) : |
---|
| 416 | errorCode = FSE_initDState(&DState, &DStream, DTablePtr); |
---|
| 417 | |
---|
| 418 | You can then decode your data, symbol after symbol. |
---|
| 419 | For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'. |
---|
| 420 | Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out). |
---|
| 421 | unsigned char symbol = FSE_decodeSymbol(&DState, &DStream); |
---|
| 422 | |
---|
| 423 | You can retrieve any bitfield you eventually stored into the bitStream (in reverse order) |
---|
| 424 | Note : maximum allowed nbBits is 25, for 32-bits compatibility |
---|
| 425 | size_t bitField = BIT_readBits(&DStream, nbBits); |
---|
| 426 | |
---|
| 427 | All above operations only read from local register (which size depends on size_t). |
---|
| 428 | Refueling the register from memory is manually performed by the reload method. |
---|
| 429 | endSignal = FSE_reloadDStream(&DStream); |
---|
| 430 | |
---|
| 431 | BIT_reloadDStream() result tells if there is still some more data to read from DStream. |
---|
| 432 | BIT_DStream_unfinished : there is still some data left into the DStream. |
---|
| 433 | BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled. |
---|
| 434 | BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed. |
---|
| 435 | BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted. |
---|
| 436 | |
---|
| 437 | When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop, |
---|
| 438 | to properly detect the exact end of stream. |
---|
| 439 | After each decoded symbol, check if DStream is fully consumed using this simple test : |
---|
| 440 | BIT_reloadDStream(&DStream) >= BIT_DStream_completed |
---|
| 441 | |
---|
| 442 | When it's done, verify decompression is fully completed, by checking both DStream and the relevant states. |
---|
| 443 | Checking if DStream has reached its end is performed by : |
---|
| 444 | BIT_endOfDStream(&DStream); |
---|
| 445 | Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible. |
---|
| 446 | FSE_endOfDState(&DState); |
---|
| 447 | */ |
---|
| 448 | |
---|
| 449 | |
---|
| 450 | /* ***************************************** |
---|
| 451 | * FSE unsafe API |
---|
| 452 | *******************************************/ |
---|
| 453 | static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD); |
---|
| 454 | /* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */ |
---|
| 455 | |
---|
| 456 | |
---|
| 457 | /* ***************************************** |
---|
| 458 | * Implementation of inlined functions |
---|
| 459 | *******************************************/ |
---|
| 460 | typedef struct { |
---|
| 461 | int deltaFindState; |
---|
| 462 | U32 deltaNbBits; |
---|
| 463 | } FSE_symbolCompressionTransform; /* total 8 bytes */ |
---|
| 464 | |
---|
| 465 | MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct) |
---|
| 466 | { |
---|
| 467 | const void* ptr = ct; |
---|
| 468 | const U16* u16ptr = (const U16*) ptr; |
---|
| 469 | const U32 tableLog = MEM_read16(ptr); |
---|
| 470 | statePtr->value = (ptrdiff_t)1<<tableLog; |
---|
| 471 | statePtr->stateTable = u16ptr+2; |
---|
| 472 | statePtr->symbolTT = ((const U32*)ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1)); |
---|
| 473 | statePtr->stateLog = tableLog; |
---|
| 474 | } |
---|
| 475 | |
---|
| 476 | |
---|
| 477 | /*! FSE_initCState2() : |
---|
| 478 | * Same as FSE_initCState(), but the first symbol to include (which will be the last to be read) |
---|
| 479 | * uses the smallest state value possible, saving the cost of this symbol */ |
---|
| 480 | MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U32 symbol) |
---|
| 481 | { |
---|
| 482 | FSE_initCState(statePtr, ct); |
---|
| 483 | { const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol]; |
---|
| 484 | const U16* stateTable = (const U16*)(statePtr->stateTable); |
---|
| 485 | U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1<<15)) >> 16); |
---|
| 486 | statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits; |
---|
| 487 | statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState]; |
---|
| 488 | } |
---|
| 489 | } |
---|
| 490 | |
---|
| 491 | MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, U32 symbol) |
---|
| 492 | { |
---|
| 493 | const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol]; |
---|
| 494 | const U16* const stateTable = (const U16*)(statePtr->stateTable); |
---|
| 495 | U32 nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16); |
---|
| 496 | BIT_addBits(bitC, statePtr->value, nbBitsOut); |
---|
| 497 | statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState]; |
---|
| 498 | } |
---|
| 499 | |
---|
| 500 | MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr) |
---|
| 501 | { |
---|
| 502 | BIT_addBits(bitC, statePtr->value, statePtr->stateLog); |
---|
| 503 | BIT_flushBits(bitC); |
---|
| 504 | } |
---|
| 505 | |
---|
| 506 | /*<===== Decompression =====>*/ |
---|
| 507 | |
---|
| 508 | typedef struct { |
---|
| 509 | U16 tableLog; |
---|
| 510 | U16 fastMode; |
---|
| 511 | } FSE_DTableHeader; /* sizeof U32 */ |
---|
| 512 | |
---|
| 513 | typedef struct |
---|
| 514 | { |
---|
| 515 | unsigned short newState; |
---|
| 516 | unsigned char symbol; |
---|
| 517 | unsigned char nbBits; |
---|
| 518 | } FSE_decode_t; /* size == U32 */ |
---|
| 519 | |
---|
| 520 | MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt) |
---|
| 521 | { |
---|
| 522 | const void* ptr = dt; |
---|
| 523 | const FSE_DTableHeader* const DTableH = (const FSE_DTableHeader*)ptr; |
---|
| 524 | DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog); |
---|
| 525 | BIT_reloadDStream(bitD); |
---|
| 526 | DStatePtr->table = dt + 1; |
---|
| 527 | } |
---|
| 528 | |
---|
| 529 | MEM_STATIC BYTE FSE_peekSymbol(const FSE_DState_t* DStatePtr) |
---|
| 530 | { |
---|
| 531 | FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; |
---|
| 532 | return DInfo.symbol; |
---|
| 533 | } |
---|
| 534 | |
---|
| 535 | MEM_STATIC void FSE_updateState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) |
---|
| 536 | { |
---|
| 537 | FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; |
---|
| 538 | U32 const nbBits = DInfo.nbBits; |
---|
| 539 | size_t const lowBits = BIT_readBits(bitD, nbBits); |
---|
| 540 | DStatePtr->state = DInfo.newState + lowBits; |
---|
| 541 | } |
---|
| 542 | |
---|
| 543 | MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) |
---|
| 544 | { |
---|
| 545 | FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; |
---|
| 546 | U32 const nbBits = DInfo.nbBits; |
---|
| 547 | BYTE const symbol = DInfo.symbol; |
---|
| 548 | size_t const lowBits = BIT_readBits(bitD, nbBits); |
---|
| 549 | |
---|
| 550 | DStatePtr->state = DInfo.newState + lowBits; |
---|
| 551 | return symbol; |
---|
| 552 | } |
---|
| 553 | |
---|
| 554 | /*! FSE_decodeSymbolFast() : |
---|
| 555 | unsafe, only works if no symbol has a probability > 50% */ |
---|
| 556 | MEM_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) |
---|
| 557 | { |
---|
| 558 | FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; |
---|
| 559 | U32 const nbBits = DInfo.nbBits; |
---|
| 560 | BYTE const symbol = DInfo.symbol; |
---|
| 561 | size_t const lowBits = BIT_readBitsFast(bitD, nbBits); |
---|
| 562 | |
---|
| 563 | DStatePtr->state = DInfo.newState + lowBits; |
---|
| 564 | return symbol; |
---|
| 565 | } |
---|
| 566 | |
---|
| 567 | MEM_STATIC unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr) |
---|
| 568 | { |
---|
| 569 | return DStatePtr->state == 0; |
---|
| 570 | } |
---|
| 571 | |
---|
| 572 | |
---|
| 573 | |
---|
| 574 | #ifndef FSE_COMMONDEFS_ONLY |
---|
| 575 | |
---|
| 576 | /* ************************************************************** |
---|
| 577 | * Tuning parameters |
---|
| 578 | ****************************************************************/ |
---|
| 579 | /*!MEMORY_USAGE : |
---|
| 580 | * Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.) |
---|
| 581 | * Increasing memory usage improves compression ratio |
---|
| 582 | * Reduced memory usage can improve speed, due to cache effect |
---|
| 583 | * Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */ |
---|
| 584 | #define FSE_MAX_MEMORY_USAGE 14 |
---|
| 585 | #define FSE_DEFAULT_MEMORY_USAGE 13 |
---|
| 586 | |
---|
| 587 | /*!FSE_MAX_SYMBOL_VALUE : |
---|
| 588 | * Maximum symbol value authorized. |
---|
| 589 | * Required for proper stack allocation */ |
---|
| 590 | #define FSE_MAX_SYMBOL_VALUE 255 |
---|
| 591 | |
---|
| 592 | |
---|
| 593 | /* ************************************************************** |
---|
| 594 | * template functions type & suffix |
---|
| 595 | ****************************************************************/ |
---|
| 596 | #define FSE_FUNCTION_TYPE BYTE |
---|
| 597 | #define FSE_FUNCTION_EXTENSION |
---|
| 598 | #define FSE_DECODE_TYPE FSE_decode_t |
---|
| 599 | |
---|
| 600 | |
---|
| 601 | #endif /* !FSE_COMMONDEFS_ONLY */ |
---|
| 602 | |
---|
| 603 | |
---|
| 604 | /* *************************************************************** |
---|
| 605 | * Constants |
---|
| 606 | *****************************************************************/ |
---|
| 607 | #define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE-2) |
---|
| 608 | #define FSE_MAX_TABLESIZE (1U<<FSE_MAX_TABLELOG) |
---|
| 609 | #define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE-1) |
---|
| 610 | #define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE-2) |
---|
| 611 | #define FSE_MIN_TABLELOG 5 |
---|
| 612 | |
---|
| 613 | #define FSE_TABLELOG_ABSOLUTE_MAX 15 |
---|
| 614 | #if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX |
---|
| 615 | # error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported" |
---|
| 616 | #endif |
---|
| 617 | |
---|
| 618 | #define FSE_TABLESTEP(tableSize) ((tableSize>>1) + (tableSize>>3) + 3) |
---|
| 619 | |
---|
| 620 | |
---|
| 621 | #endif /* FSE_STATIC_LINKING_ONLY */ |
---|
| 622 | |
---|
| 623 | |
---|
| 624 | #if defined (__cplusplus) |
---|
| 625 | } |
---|
| 626 | #endif |
---|
| 627 | |
---|
| 628 | #endif /* FSE_H */ |
---|