From 2e656cea7670c6b023f021c6eabfe9926f46d3dc Mon Sep 17 00:00:00 2001 From: Tino Reichardt Date: Mon, 6 Feb 2017 20:15:11 +0100 Subject: [PATCH] Update to LZ4 v1.7.5 --- C/lz4/lz4.c | 1 + C/lz4/lz4.h | 2 +- C/lz4/lz4frame.c | 29 ++-- C/lz4/lz4frame.h | 23 +-- C/lz4/lz4hc.c | 54 +++++-- C/lz4/lz4hc.h | 17 ++- C/lz4/lz4opt.h | 363 +++++++++++++++++++++++++++++++++++++++++++++++ 7 files changed, 450 insertions(+), 39 deletions(-) create mode 100644 C/lz4/lz4opt.h diff --git a/C/lz4/lz4.c b/C/lz4/lz4.c index fc19493d..143c36e1 100644 --- a/C/lz4/lz4.c +++ b/C/lz4/lz4.c @@ -391,6 +391,7 @@ typedef enum { full = 0, partial = 1 } earlyEnd_directive; * Local Utils **************************************/ int LZ4_versionNumber (void) { return LZ4_VERSION_NUMBER; } +const char* LZ4_versionString(void) { return LZ4_VERSION_STRING; } int LZ4_compressBound(int isize) { return LZ4_COMPRESSBOUND(isize); } int LZ4_sizeofState() { return LZ4_STREAMSIZE; } diff --git a/C/lz4/lz4.h b/C/lz4/lz4.h index 7420ad8d..0aae19c9 100644 --- a/C/lz4/lz4.h +++ b/C/lz4/lz4.h @@ -85,7 +85,7 @@ extern "C" { /*========== Version =========== */ #define LZ4_VERSION_MAJOR 1 /* for breaking interface changes */ #define LZ4_VERSION_MINOR 7 /* for new (non-breaking) interface capabilities */ -#define LZ4_VERSION_RELEASE 4 /* for tweaks, bug-fixes, or development */ +#define LZ4_VERSION_RELEASE 5 /* for tweaks, bug-fixes, or development */ #define LZ4_VERSION_NUMBER (LZ4_VERSION_MAJOR *100*100 + LZ4_VERSION_MINOR *100 + LZ4_VERSION_RELEASE) diff --git a/C/lz4/lz4frame.c b/C/lz4/lz4frame.c index 3c2b788d..a0a625b0 100644 --- a/C/lz4/lz4frame.c +++ b/C/lz4/lz4frame.c @@ -158,7 +158,7 @@ static void LZ4F_writeLE64 (void* dst, U64 value64) #define LZ4F_BLOCKSIZEID_DEFAULT LZ4F_max64KB static const size_t minFHSize = 7; -static const size_t maxFHSize = 15; /* max Frame Header Size */ +static const size_t maxFHSize = LZ4F_HEADER_SIZE_MAX; /* 15 */ static const size_t BHSize = 4; @@ -254,20 +254,27 @@ static LZ4F_blockSizeID_t LZ4F_optimalBSID(const LZ4F_blockSizeID_t requestedBSI return requestedBSID; } +/* LZ4F_compressBound() : + * Provides dstCapacity given a srcSize to guarantee operation success in worst case situations. + * prefsPtr is optional : you can provide NULL as argument, preferences will be set to cover worst case scenario. + * Result is always the same for a srcSize and prefsPtr, so it can be trusted to size reusable buffers. + * When srcSize==0, LZ4F_compressBound() provides an upper bound for LZ4F_flush() and LZ4F_compressEnd() operations. + */ static size_t LZ4F_compressBound_internal(size_t srcSize, const LZ4F_preferences_t* preferencesPtr, size_t alreadyBuffered) { LZ4F_preferences_t prefsNull; memset(&prefsNull, 0, sizeof(prefsNull)); prefsNull.frameInfo.contentChecksumFlag = LZ4F_contentChecksumEnabled; /* worst case */ { const LZ4F_preferences_t* const prefsPtr = (preferencesPtr==NULL) ? &prefsNull : preferencesPtr; + U32 const flush = prefsPtr->autoFlush | (srcSize==0); LZ4F_blockSizeID_t const bid = prefsPtr->frameInfo.blockSizeID; size_t const blockSize = LZ4F_getBlockSize(bid); size_t const maxBuffered = blockSize - 1; size_t const bufferedSize = MIN(alreadyBuffered, maxBuffered); size_t const maxSrcSize = srcSize + bufferedSize; unsigned const nbFullBlocks = (unsigned)(maxSrcSize / blockSize); - size_t const partialBlockSize = srcSize & (blockSize-1); - size_t const lastBlockSize = prefsPtr->autoFlush ? partialBlockSize : 0; + size_t const partialBlockSize = (srcSize - (srcSize==0)) & (blockSize-1); /* 0 => -1 == MAX => blockSize-1 */ + size_t const lastBlockSize = flush ? partialBlockSize : 0; unsigned const nbBlocks = nbFullBlocks + (lastBlockSize>0); size_t const blockHeaderSize = 4; /* default, without block CRC option (which cannot be generated with current API) */ @@ -322,7 +329,7 @@ size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity, const void* srcBu if (prefs.frameInfo.contentSize != 0) prefs.frameInfo.contentSize = (U64)srcSize; /* auto-correct content size if selected (!=0) */ - if (prefs.compressionLevel < LZ4HC_MIN_CLEVEL) { + if (prefs.compressionLevel < LZ4HC_CLEVEL_MIN) { cctxI.lz4CtxPtr = &lz4ctx; cctxI.lz4CtxLevel = 1; } @@ -349,7 +356,7 @@ size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity, const void* srcBu if (LZ4F_isError(tailSize)) return tailSize; dstPtr += tailSize; } - if (prefs.compressionLevel >= LZ4HC_MIN_CLEVEL) /* no allocation done with lz4 fast */ + if (prefs.compressionLevel >= LZ4HC_CLEVEL_MIN) /* no allocation done with lz4 fast */ FREEMEM(cctxI.lz4CtxPtr); return (dstPtr - dstStart); @@ -398,7 +405,7 @@ LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_compressionContext_t LZ4F_comp /*! LZ4F_compressBegin() : * will write the frame header into dstBuffer. - * dstBuffer must be large enough to accommodate a header (dstCapacity). Maximum header size is LZ4F_MAXHEADERFRAME_SIZE bytes. + * dstBuffer must be large enough to accommodate a header (dstCapacity). Maximum header size is LZ4F_HEADER_SIZE_MAX bytes. * @return : number of bytes written into dstBuffer for the header * or an error code (can be tested using LZ4F_isError()) */ @@ -417,10 +424,10 @@ size_t LZ4F_compressBegin(LZ4F_cctx* cctxPtr, void* dstBuffer, size_t dstCapacit cctxPtr->prefs = *preferencesPtr; /* ctx Management */ - { U32 const tableID = (cctxPtr->prefs.compressionLevel < LZ4HC_MIN_CLEVEL) ? 1 : 2; /* 0:nothing ; 1:LZ4 table ; 2:HC tables */ + { U32 const tableID = (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) ? 1 : 2; /* 0:nothing ; 1:LZ4 table ; 2:HC tables */ if (cctxPtr->lz4CtxLevel < tableID) { FREEMEM(cctxPtr->lz4CtxPtr); - if (cctxPtr->prefs.compressionLevel < LZ4HC_MIN_CLEVEL) + if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) cctxPtr->lz4CtxPtr = (void*)LZ4_createStream(); else cctxPtr->lz4CtxPtr = (void*)LZ4_createStreamHC(); @@ -445,7 +452,7 @@ size_t LZ4F_compressBegin(LZ4F_cctx* cctxPtr, void* dstBuffer, size_t dstCapacit cctxPtr->tmpIn = cctxPtr->tmpBuff; cctxPtr->tmpInSize = 0; XXH32_reset(&(cctxPtr->xxh), 0); - if (cctxPtr->prefs.compressionLevel < LZ4HC_MIN_CLEVEL) + if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) LZ4_resetStream((LZ4_stream_t*)(cctxPtr->lz4CtxPtr)); else LZ4_resetStreamHC((LZ4_streamHC_t*)(cctxPtr->lz4CtxPtr), cctxPtr->prefs.compressionLevel); @@ -526,7 +533,7 @@ static int LZ4F_localLZ4_compressHC_limitedOutput_continue(void* ctx, const char static compressFunc_t LZ4F_selectCompression(LZ4F_blockMode_t blockMode, int level) { - if (level < LZ4HC_MIN_CLEVEL) { + if (level < LZ4HC_CLEVEL_MIN) { if (blockMode == LZ4F_blockIndependent) return LZ4F_localLZ4_compress_limitedOutput_withState; return LZ4F_localLZ4_compress_limitedOutput_continue; } @@ -536,7 +543,7 @@ static compressFunc_t LZ4F_selectCompression(LZ4F_blockMode_t blockMode, int lev static int LZ4F_localSaveDict(LZ4F_cctx_t* cctxPtr) { - if (cctxPtr->prefs.compressionLevel < LZ4HC_MIN_CLEVEL) + if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) return LZ4_saveDict ((LZ4_stream_t*)(cctxPtr->lz4CtxPtr), (char*)(cctxPtr->tmpBuff), 64 KB); return LZ4_saveDictHC ((LZ4_streamHC_t*)(cctxPtr->lz4CtxPtr), (char*)(cctxPtr->tmpBuff), 64 KB); } diff --git a/C/lz4/lz4frame.h b/C/lz4/lz4frame.h index 3104d2ef..a4a4ce65 100644 --- a/C/lz4/lz4frame.h +++ b/C/lz4/lz4frame.h @@ -206,10 +206,11 @@ LZ4FLIB_API LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_cctx* cctx); /* Compression */ +#define LZ4F_HEADER_SIZE_MAX 15 LZ4FLIB_API size_t LZ4F_compressBegin(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapacity, const LZ4F_preferences_t* prefsPtr); /* LZ4F_compressBegin() : * will write the frame header into dstBuffer. - * dstBuffer must be large enough to accommodate a header. Maximum header size is 15 bytes. + * dstCapacity must be large enough to store the header. Maximum header size is LZ4F_HEADER_SIZE_MAX bytes. * `prefsPtr` is optional : you can provide NULL as argument, all preferences will then be set to default. * @return : number of bytes written into dstBuffer for the header * or an error code (which can be tested using LZ4F_isError()) @@ -217,20 +218,20 @@ LZ4FLIB_API size_t LZ4F_compressBegin(LZ4F_cctx* cctx, void* dstBuffer, size_t d LZ4FLIB_API size_t LZ4F_compressBound(size_t srcSize, const LZ4F_preferences_t* prefsPtr); /* LZ4F_compressBound() : - * Provides the minimum size of Dst buffer given srcSize to handle worst case situations. - * Different preferences can produce different results. - * prefsPtr is optional : you can provide NULL as argument, all preferences will then be set to cover worst case. - * This function includes frame termination cost (4 bytes, or 8 if frame checksum is enabled) + * Provides dstCapacity given a srcSize to guarantee operation success in worst case situations. + * prefsPtr is optional : you can provide NULL as argument, preferences will be set to cover worst case scenario. + * Result is always the same for a srcSize and prefsPtr, so it can be trusted to size reusable buffers. + * When srcSize==0, LZ4F_compressBound() provides an upper bound for LZ4F_flush() and LZ4F_compressEnd() operations. */ LZ4FLIB_API size_t LZ4F_compressUpdate(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapacity, const void* srcBuffer, size_t srcSize, const LZ4F_compressOptions_t* cOptPtr); /* LZ4F_compressUpdate() : * LZ4F_compressUpdate() can be called repetitively to compress as much data as necessary. - * An important rule is that dstBuffer MUST be large enough (dstCapacity) to ensure compression completion even in worst case. - * This value is provided by using LZ4F_compressBound(). + * An important rule is that dstCapacity MUST be large enough to ensure operation success even in worst case situations. + * This value is provided by LZ4F_compressBound(). * If this condition is not respected, LZ4F_compress() will fail (result is an errorCode). * LZ4F_compressUpdate() doesn't guarantee error recovery. When an error occurs, compression context must be freed or resized. - * `cOptPtr` is optional : it's possible to provide NULL, all options will be set to default. + * `cOptPtr` is optional : NULL can be provided, in which case all options are set to default. * @return : number of bytes written into `dstBuffer` (it can be zero, meaning input data was just buffered). * or an error code if it fails (which can be tested using LZ4F_isError()) */ @@ -245,12 +246,12 @@ LZ4FLIB_API size_t LZ4F_flush(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapaci * or an error code if it fails (which can be tested using LZ4F_isError()) */ -LZ4FLIB_API size_t LZ4F_compressEnd(LZ4F_cctx* cctx, void* dstBuffer, size_t dstMaxSize, const LZ4F_compressOptions_t* cOptPtr); +LZ4FLIB_API size_t LZ4F_compressEnd(LZ4F_cctx* cctx, void* dstBuffer, size_t dstCapacity, const LZ4F_compressOptions_t* cOptPtr); /* LZ4F_compressEnd() : - * To properly finish the compressed frame, invoke LZ4F_compressEnd(). + * To properly finish an LZ4 frame, invoke LZ4F_compressEnd(). * It will flush whatever data remained within `cctx` (like LZ4_flush()) * and properly finalize the frame, with an endMark and a checksum. - * `cOptPtr` is optional : it's possible to provide NULL, all options will be set to default. + * `cOptPtr` is optional : NULL can be provided, in which case all options will be set to default. * @return : number of bytes written into dstBuffer (necessarily >= 4 (endMark), or 8 if optional frame checksum is enabled) * or an error code if it fails (which can be tested using LZ4F_isError()) * A successful call to LZ4F_compressEnd() makes `cctx` available again for another compression task. diff --git a/C/lz4/lz4hc.c b/C/lz4/lz4hc.c index 298550c0..5d4ea3e6 100644 --- a/C/lz4/lz4hc.c +++ b/C/lz4/lz4hc.c @@ -84,7 +84,7 @@ * Local Macros **************************************/ #define HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8)-LZ4HC_HASH_LOG)) -/* #define DELTANEXTU16(p) chainTable[(p) & LZ4HC_MAXD_MASK] */ /* flexible, LZ4HC_MAXD dependent */ +#define DELTANEXTMAXD(p) chainTable[(p) & LZ4HC_MAXD_MASK] /* flexible, LZ4HC_MAXD dependent */ #define DELTANEXTU16(p) chainTable[(U16)(p)] /* faster */ static U32 LZ4HC_hashPtr(const void* ptr) { return HASH_FUNCTION(LZ4_read32(ptr)); } @@ -305,14 +305,15 @@ FORCE_INLINE int LZ4HC_encodeSequence ( return 0; } +#include "lz4opt.h" -static int LZ4HC_compress_generic ( +static int LZ4HC_compress_hashChain ( LZ4HC_CCtx_internal* const ctx, const char* const source, char* const dest, int const inputSize, int const maxOutputSize, - int compressionLevel, + unsigned maxNbAttempts, limitedOutput_directive limit ) { @@ -325,7 +326,6 @@ static int LZ4HC_compress_generic ( BYTE* op = (BYTE*) dest; BYTE* const oend = op + maxOutputSize; - unsigned maxNbAttempts; int ml, ml2, ml3, ml0; const BYTE* ref = NULL; const BYTE* start2 = NULL; @@ -336,9 +336,6 @@ static int LZ4HC_compress_generic ( const BYTE* ref0; /* init */ - if (compressionLevel > LZ4HC_MAX_CLEVEL) compressionLevel = LZ4HC_MAX_CLEVEL; - if (compressionLevel < 1) compressionLevel = LZ4HC_DEFAULT_CLEVEL; - maxNbAttempts = 1 << (compressionLevel-1); ctx->end += inputSize; ip++; @@ -489,6 +486,37 @@ _Search3: return (int) (((char*)op)-dest); } +static int LZ4HC_getSearchNum(int compressionLevel) +{ + switch (compressionLevel) { + default: return 0; /* unused */ + case 11: return 128; + case 12: return 1<<10; + } +} + +static int LZ4HC_compress_generic ( + LZ4HC_CCtx_internal* const ctx, + const char* const source, + char* const dest, + int const inputSize, + int const maxOutputSize, + int compressionLevel, + limitedOutput_directive limit + ) +{ + if (compressionLevel < 1) compressionLevel = LZ4HC_CLEVEL_DEFAULT; + if (compressionLevel > 9) { + switch (compressionLevel) { + case 10: return LZ4HC_compress_hashChain(ctx, source, dest, inputSize, maxOutputSize, 1 << (16-1), limit); + case 11: ctx->searchNum = LZ4HC_getSearchNum(compressionLevel); return LZ4HC_compress_optimal(ctx, source, dest, inputSize, maxOutputSize, limit, 128, 0); + default: + case 12: ctx->searchNum = LZ4HC_getSearchNum(compressionLevel); return LZ4HC_compress_optimal(ctx, source, dest, inputSize, maxOutputSize, limit, LZ4_OPT_NUM, 1); + } + } + return LZ4HC_compress_hashChain(ctx, source, dest, inputSize, maxOutputSize, 1 << (compressionLevel-1), limit); +} + int LZ4_sizeofStateHC(void) { return sizeof(LZ4_streamHC_t); } @@ -534,6 +562,7 @@ void LZ4_resetStreamHC (LZ4_streamHC_t* LZ4_streamHCPtr, int compressionLevel) LZ4_STATIC_ASSERT(sizeof(LZ4HC_CCtx_internal) <= sizeof(size_t) * LZ4_STREAMHCSIZE_SIZET); /* if compilation fails here, LZ4_STREAMHCSIZE must be increased */ LZ4_streamHCPtr->internal_donotuse.base = NULL; LZ4_streamHCPtr->internal_donotuse.compressionLevel = (unsigned)compressionLevel; + LZ4_streamHCPtr->internal_donotuse.searchNum = LZ4HC_getSearchNum(compressionLevel); } int LZ4_loadDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, const char* dictionary, int dictSize) @@ -544,8 +573,11 @@ int LZ4_loadDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, const char* dictionary, int dictSize = 64 KB; } LZ4HC_init (ctxPtr, (const BYTE*)dictionary); - if (dictSize >= 4) LZ4HC_Insert (ctxPtr, (const BYTE*)dictionary +(dictSize-3)); ctxPtr->end = (const BYTE*)dictionary + dictSize; + if (ctxPtr->compressionLevel >= LZ4HC_CLEVEL_OPT_MIN) + LZ4HC_updateBinTree(ctxPtr, ctxPtr->end - MFLIMIT, ctxPtr->end - LASTLITERALS); + else + if (dictSize >= 4) LZ4HC_Insert (ctxPtr, ctxPtr->end-3); return dictSize; } @@ -554,7 +586,11 @@ int LZ4_loadDictHC (LZ4_streamHC_t* LZ4_streamHCPtr, const char* dictionary, int static void LZ4HC_setExternalDict(LZ4HC_CCtx_internal* ctxPtr, const BYTE* newBlock) { - if (ctxPtr->end >= ctxPtr->base + 4) LZ4HC_Insert (ctxPtr, ctxPtr->end-3); /* Referencing remaining dictionary content */ + if (ctxPtr->compressionLevel >= LZ4HC_CLEVEL_OPT_MIN) + LZ4HC_updateBinTree(ctxPtr, ctxPtr->end - MFLIMIT, ctxPtr->end - LASTLITERALS); + else + if (ctxPtr->end >= ctxPtr->base + 4) LZ4HC_Insert (ctxPtr, ctxPtr->end-3); /* Referencing remaining dictionary content */ + /* Only one memory segment for extDict, so any previous extDict is lost at this stage */ ctxPtr->lowLimit = ctxPtr->dictLimit; ctxPtr->dictLimit = (U32)(ctxPtr->end - ctxPtr->base); diff --git a/C/lz4/lz4hc.h b/C/lz4/lz4hc.h index d8ac3c0e..1036fd0b 100644 --- a/C/lz4/lz4hc.h +++ b/C/lz4/lz4hc.h @@ -44,9 +44,10 @@ extern "C" { /* --- Useful constants --- */ -#define LZ4HC_MIN_CLEVEL 3 -#define LZ4HC_DEFAULT_CLEVEL 9 -#define LZ4HC_MAX_CLEVEL 16 +#define LZ4HC_CLEVEL_MIN 3 +#define LZ4HC_CLEVEL_DEFAULT 9 +#define LZ4HC_CLEVEL_OPT_MIN 11 +#define LZ4HC_CLEVEL_MAX 12 /*-************************************ @@ -58,7 +59,7 @@ extern "C" { * Compression is guaranteed to succeed if `dstCapacity >= LZ4_compressBound(srcSize)` (see "lz4.h") * Max supported `srcSize` value is LZ4_MAX_INPUT_SIZE (see "lz4.h") * `compressionLevel` : Recommended values are between 4 and 9, although any value between 1 and LZ4HC_MAX_CLEVEL will work. - * Values >LZ4HC_MAX_CLEVEL behave the same as 16. + * Values >LZ4HC_MAX_CLEVEL behave the same as LZ4HC_MAX_CLEVEL. * @return : the number of bytes written into 'dst' * or 0 if compression fails. */ @@ -130,11 +131,11 @@ LZ4LIB_API int LZ4_saveDictHC (LZ4_streamHC_t* streamHCPtr, char* safeBuffer, in * They are exposed to allow static allocation of `LZ4_streamHC_t`. * Using these definitions makes the code vulnerable to potential API break when upgrading LZ4 **************************************/ -#define LZ4HC_DICTIONARY_LOGSIZE 16 +#define LZ4HC_DICTIONARY_LOGSIZE 17 #define LZ4HC_MAXD (1< + Note : this file is intended to be included within lz4hc.c + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - LZ4 source repository : https://github.com/lz4/lz4 + - LZ4 public forum : https://groups.google.com/forum/#!forum/lz4c +*/ + +#define LZ4_OPT_NUM (1<<12) + + +typedef struct +{ + int off; + int len; +} LZ4HC_match_t; + +typedef struct +{ + int price; + int off; + int mlen; + int litlen; +} LZ4HC_optimal_t; + + +/* price in bits */ +FORCE_INLINE size_t LZ4HC_literalsPrice(size_t litlen) +{ + size_t price = 8*litlen; + if (litlen >= (size_t)RUN_MASK) price+=8*(1+(litlen-RUN_MASK)/255); + return price; +} + + +/* requires mlen >= MINMATCH */ +FORCE_INLINE size_t LZ4HC_sequencePrice(size_t litlen, size_t mlen) +{ + size_t price = 16 + 8; /* 16-bit offset + token */ + + price += LZ4HC_literalsPrice(litlen); + + mlen -= MINMATCH; + if (mlen >= (size_t)ML_MASK) price+=8*(1+(mlen-ML_MASK)/255); + + return price; +} + + +/*-************************************* +* Binary Tree search +***************************************/ +FORCE_INLINE int LZ4HC_BinTree_InsertAndGetAllMatches ( + LZ4HC_CCtx_internal* ctx, + const BYTE* const ip, + const BYTE* const iHighLimit, + size_t best_mlen, + LZ4HC_match_t* matches, + int* matchNum) +{ + U16* const chainTable = ctx->chainTable; + U32* const HashTable = ctx->hashTable; + const BYTE* const base = ctx->base; + const U32 dictLimit = ctx->dictLimit; + const U32 current = (U32)(ip - base); + const U32 lowLimit = (ctx->lowLimit + MAX_DISTANCE > current) ? ctx->lowLimit : current - (MAX_DISTANCE - 1); + const BYTE* const dictBase = ctx->dictBase; + const BYTE* match; + int nbAttempts = ctx->searchNum; + int mnum = 0; + U16 *ptr0, *ptr1, delta0, delta1; + U32 matchIndex; + size_t matchLength = 0; + U32* HashPos; + + if (ip + MINMATCH > iHighLimit) return 1; + + /* HC4 match finder */ + HashPos = &HashTable[LZ4HC_hashPtr(ip)]; + matchIndex = *HashPos; + *HashPos = current; + + ptr0 = &DELTANEXTMAXD(current*2+1); + ptr1 = &DELTANEXTMAXD(current*2); + delta0 = delta1 = (U16)(current - matchIndex); + + while ((matchIndex < current) && (matchIndex>=lowLimit) && (nbAttempts)) { + nbAttempts--; + if (matchIndex >= dictLimit) { + match = base + matchIndex; + matchLength = LZ4_count(ip, match, iHighLimit); + } else { + const BYTE* vLimit = ip + (dictLimit - matchIndex); + match = dictBase + matchIndex; + if (vLimit > iHighLimit) vLimit = iHighLimit; + matchLength = LZ4_count(ip, match, vLimit); + if ((ip+matchLength == vLimit) && (vLimit < iHighLimit)) + matchLength += LZ4_count(ip+matchLength, base+dictLimit, iHighLimit); + } + + if (matchLength > best_mlen) { + best_mlen = matchLength; + if (matches) { + if (matchIndex >= dictLimit) + matches[mnum].off = (int)(ip - match); + else + matches[mnum].off = (int)(ip - (base + matchIndex)); /* virtual matchpos */ + matches[mnum].len = (int)matchLength; + mnum++; + } + if (best_mlen > LZ4_OPT_NUM) break; + } + + if (ip+matchLength >= iHighLimit) /* equal : no way to know if inf or sup */ + break; /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt the tree */ + + if (*(ip+matchLength) < *(match+matchLength)) { + *ptr0 = delta0; + ptr0 = &DELTANEXTMAXD(matchIndex*2); + if (*ptr0 == (U16)-1) break; + delta0 = *ptr0; + delta1 += delta0; + matchIndex -= delta0; + } else { + *ptr1 = delta1; + ptr1 = &DELTANEXTMAXD(matchIndex*2+1); + if (*ptr1 == (U16)-1) break; + delta1 = *ptr1; + delta0 += delta1; + matchIndex -= delta1; + } + } + + *ptr0 = (U16)-1; + *ptr1 = (U16)-1; + if (matchNum) *matchNum = mnum; + /* if (best_mlen > 8) return best_mlen-8; */ + if (!matchNum) return 1; + return 1; +} + + +FORCE_INLINE void LZ4HC_updateBinTree(LZ4HC_CCtx_internal* ctx, const BYTE* const ip, const BYTE* const iHighLimit) +{ + const BYTE* const base = ctx->base; + const U32 target = (U32)(ip - base); + U32 idx = ctx->nextToUpdate; + while(idx < target) + idx += LZ4HC_BinTree_InsertAndGetAllMatches(ctx, base+idx, iHighLimit, 8, NULL, NULL); +} + + +/** Tree updater, providing best match */ +FORCE_INLINE int LZ4HC_BinTree_GetAllMatches ( + LZ4HC_CCtx_internal* ctx, + const BYTE* const ip, const BYTE* const iHighLimit, + size_t best_mlen, LZ4HC_match_t* matches, const int fullUpdate) +{ + int mnum = 0; + if (ip < ctx->base + ctx->nextToUpdate) return 0; /* skipped area */ + if (fullUpdate) LZ4HC_updateBinTree(ctx, ip, iHighLimit); + best_mlen = LZ4HC_BinTree_InsertAndGetAllMatches(ctx, ip, iHighLimit, best_mlen, matches, &mnum); + ctx->nextToUpdate = (U32)(ip - ctx->base + best_mlen); + return mnum; +} + + +#define SET_PRICE(pos, mlen, offset, litlen, price) \ +{ \ + while (last_pos < pos) { opt[last_pos+1].price = 1<<30; last_pos++; } \ + opt[pos].mlen = (int)mlen; \ + opt[pos].off = (int)offset; \ + opt[pos].litlen = (int)litlen; \ + opt[pos].price = (int)price; \ +} + + +static int LZ4HC_compress_optimal ( + LZ4HC_CCtx_internal* ctx, + const char* const source, + char* dest, + int inputSize, + int maxOutputSize, + limitedOutput_directive limit, + const size_t sufficient_len, + const int fullUpdate + ) +{ + LZ4HC_optimal_t opt[LZ4_OPT_NUM + 1]; + LZ4HC_match_t matches[LZ4_OPT_NUM + 1]; + const BYTE *inr = NULL; + size_t res, cur, cur2; + size_t i, llen, litlen, mlen, best_mlen, price, offset, best_off, match_num, last_pos; + + const BYTE* ip = (const BYTE*) source; + const BYTE* anchor = ip; + const BYTE* const iend = ip + inputSize; + const BYTE* const mflimit = iend - MFLIMIT; + const BYTE* const matchlimit = (iend - LASTLITERALS); + BYTE* op = (BYTE*) dest; + BYTE* const oend = op + maxOutputSize; + + /* init */ + ctx->end += inputSize; + ip++; + + /* Main Loop */ + while (ip < mflimit) { + memset(opt, 0, sizeof(LZ4HC_optimal_t)); + last_pos = 0; + llen = ip - anchor; + match_num = LZ4HC_BinTree_GetAllMatches(ctx, ip, matchlimit, MINMATCH-1, matches, fullUpdate); + if (!match_num) { ip++; continue; } + + if ((size_t)matches[match_num-1].len > sufficient_len) { + best_mlen = matches[match_num-1].len; + best_off = matches[match_num-1].off; + cur = 0; + last_pos = 1; + goto encode; + } + + /* set prices using matches at position = 0 */ + for (i = 0; i < match_num; i++) { + mlen = (i>0) ? (size_t)matches[i-1].len+1 : MINMATCH; + best_mlen = (matches[i].len < LZ4_OPT_NUM) ? matches[i].len : LZ4_OPT_NUM; + while (mlen <= best_mlen) { + litlen = 0; + price = LZ4HC_sequencePrice(llen + litlen, mlen) - LZ4HC_literalsPrice(llen); + SET_PRICE(mlen, mlen, matches[i].off, litlen, price); + mlen++; + } + } + + if (last_pos < MINMATCH) { ip++; continue; } + + /* check further positions */ + opt[0].mlen = opt[1].mlen = 1; + for (cur = 1; cur <= last_pos; cur++) { + inr = ip + cur; + + if (opt[cur-1].mlen == 1) { + litlen = opt[cur-1].litlen + 1; + if (cur != litlen) { + price = opt[cur - litlen].price + LZ4HC_literalsPrice(litlen); + } else { + price = LZ4HC_literalsPrice(llen + litlen) - LZ4HC_literalsPrice(llen); + } + } else { + litlen = 1; + price = opt[cur - 1].price + LZ4HC_literalsPrice(litlen); + } + + mlen = 1; + best_mlen = 0; + if (cur > last_pos || price < (size_t)opt[cur].price) + SET_PRICE(cur, mlen, best_mlen, litlen, price); + + if (cur == last_pos || inr >= mflimit) break; + + match_num = LZ4HC_BinTree_GetAllMatches(ctx, inr, matchlimit, MINMATCH-1, matches, fullUpdate); + if (match_num > 0 && (size_t)matches[match_num-1].len > sufficient_len) { + best_mlen = matches[match_num-1].len; + best_off = matches[match_num-1].off; + last_pos = cur + 1; + goto encode; + } + + /* set prices using matches at position = cur */ + for (i = 0; i < match_num; i++) { + mlen = (i>0) ? (size_t)matches[i-1].len+1 : MINMATCH; + cur2 = cur; + best_mlen = (cur2 + matches[i].len < LZ4_OPT_NUM) ? (size_t)matches[i].len : LZ4_OPT_NUM - cur2; + + while (mlen <= best_mlen) { + if (opt[cur2].mlen == 1) { + litlen = opt[cur2].litlen; + + if (cur2 != litlen) + price = opt[cur2 - litlen].price + LZ4HC_sequencePrice(litlen, mlen); + else + price = LZ4HC_sequencePrice(llen + litlen, mlen) - LZ4HC_literalsPrice(llen); + } else { + litlen = 0; + price = opt[cur2].price + LZ4HC_sequencePrice(litlen, mlen); + } + + if (cur2 + mlen > last_pos || price < (size_t)opt[cur2 + mlen].price) { // || (((int)price == opt[cur2 + mlen].price) && (opt[cur2 + mlen-1].mlen == 1))) { + SET_PRICE(cur2 + mlen, mlen, matches[i].off, litlen, price); + } + mlen++; + } + } + } /* for (cur = 1; cur <= last_pos; cur++) */ + + best_mlen = opt[last_pos].mlen; + best_off = opt[last_pos].off; + cur = last_pos - best_mlen; + +encode: /* cur, last_pos, best_mlen, best_off have to be set */ + opt[0].mlen = 1; + while (1) { + mlen = opt[cur].mlen; + offset = opt[cur].off; + opt[cur].mlen = (int)best_mlen; + opt[cur].off = (int)best_off; + best_mlen = mlen; + best_off = offset; + if (mlen > cur) break; + cur -= mlen; + } + + cur = 0; + while (cur < last_pos) { + mlen = opt[cur].mlen; + if (mlen == 1) { ip++; cur++; continue; } + offset = opt[cur].off; + cur += mlen; + + res = LZ4HC_encodeSequence(&ip, &op, &anchor, (int)mlen, ip - offset, limit, oend); + if (res) return 0; + } + } + + /* Encode Last Literals */ + { int lastRun = (int)(iend - anchor); + if ((limit) && (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize)) return 0; /* Check output limit */ + if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK< 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; } + else *op++ = (BYTE)(lastRun<