vendor/github.com/DataDog/zstd/zstd_compress.c

/*
 * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
 * All rights reserved.
 *
 * This source code is licensed under both the BSD-style license (found in the
 * LICENSE file in the root directory of this source tree) and the GPLv2 (found
 * in the COPYING file in the root directory of this source tree).
 * You may select, at your option, one of the above-listed licenses.
 */

/*-*************************************
*  Dependencies
***************************************/
#include <limits.h>         /* INT_MAX */
#include <string.h>         /* memset */
#include "cpu.h"
#include "mem.h"
#include "hist.h"           /* HIST_countFast_wksp */
#define FSE_STATIC_LINKING_ONLY   /* FSE_encodeSymbol */
#include "fse.h"
#define HUF_STATIC_LINKING_ONLY
#include "huf.h"
#include "zstd_compress_internal.h"
#include "zstd_fast.h"
#include "zstd_double_fast.h"
#include "zstd_lazy.h"
#include "zstd_opt.h"
#include "zstd_ldm.h"


/*-*************************************
*  Helper functions
***************************************/
size_t ZSTD_compressBound(size_t srcSize) {
    return ZSTD_COMPRESSBOUND(srcSize);
}


/*-*************************************
*  Context memory management
***************************************/
struct ZSTD_CDict_s {
    void* dictBuffer;
    const void* dictContent;
    size_t dictContentSize;
    void* workspace;
    size_t workspaceSize;
    ZSTD_matchState_t matchState;
    ZSTD_compressedBlockState_t cBlockState;
    ZSTD_customMem customMem;
    U32 dictID;
};  /* typedef'd to ZSTD_CDict within "zstd.h" */

ZSTD_CCtx* ZSTD_createCCtx(void)
{
    return ZSTD_createCCtx_advanced(ZSTD_defaultCMem);
}

static void ZSTD_initCCtx(ZSTD_CCtx* cctx, ZSTD_customMem memManager)
{
    assert(cctx != NULL);
    memset(cctx, 0, sizeof(*cctx));
    cctx->customMem = memManager;
    cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
    {   size_t const err = ZSTD_CCtx_reset(cctx, ZSTD_reset_parameters);
        assert(!ZSTD_isError(err));
        (void)err;
    }
}

ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
{
    ZSTD_STATIC_ASSERT(zcss_init==0);
    ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN==(0ULL - 1));
    if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
    {   ZSTD_CCtx* const cctx = (ZSTD_CCtx*)ZSTD_malloc(sizeof(ZSTD_CCtx), customMem);
        if (!cctx) return NULL;
        ZSTD_initCCtx(cctx, customMem);
        return cctx;
    }
}

ZSTD_CCtx* ZSTD_initStaticCCtx(void *workspace, size_t workspaceSize)
{
    ZSTD_CCtx* const cctx = (ZSTD_CCtx*) workspace;
    if (workspaceSize <= sizeof(ZSTD_CCtx)) return NULL;  /* minimum size */
    if ((size_t)workspace & 7) return NULL;  /* must be 8-aligned */
    memset(workspace, 0, workspaceSize);   /* may be a bit generous, could memset be smaller ? */
    cctx->staticSize = workspaceSize;
    cctx->workSpace = (void*)(cctx+1);
    cctx->workSpaceSize = workspaceSize - sizeof(ZSTD_CCtx);

    /* statically sized space. entropyWorkspace never moves (but prev/next block swap places) */
    if (cctx->workSpaceSize < HUF_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t)) return NULL;
    assert(((size_t)cctx->workSpace & (sizeof(void*)-1)) == 0);   /* ensure correct alignment */
    cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)cctx->workSpace;
    cctx->blockState.nextCBlock = cctx->blockState.prevCBlock + 1;
    {
        void* const ptr = cctx->blockState.nextCBlock + 1;
        cctx->entropyWorkspace = (U32*)ptr;
    }
    cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
    return cctx;
}

/**
 * Clears and frees all of the dictionaries in the CCtx.
 */
static void ZSTD_clearAllDicts(ZSTD_CCtx* cctx)
{
    ZSTD_free(cctx->localDict.dictBuffer, cctx->customMem);
    ZSTD_freeCDict(cctx->localDict.cdict);
    memset(&cctx->localDict, 0, sizeof(cctx->localDict));
    memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));
    cctx->cdict = NULL;
}

static size_t ZSTD_sizeof_localDict(ZSTD_localDict dict)
{
    size_t const bufferSize = dict.dictBuffer != NULL ? dict.dictSize : 0;
    size_t const cdictSize = ZSTD_sizeof_CDict(dict.cdict);
    return bufferSize + cdictSize;
}

static void ZSTD_freeCCtxContent(ZSTD_CCtx* cctx)
{
    assert(cctx != NULL);
    assert(cctx->staticSize == 0);
    ZSTD_free(cctx->workSpace, cctx->customMem); cctx->workSpace = NULL;
    ZSTD_clearAllDicts(cctx);
#ifdef ZSTD_MULTITHREAD
    ZSTDMT_freeCCtx(cctx->mtctx); cctx->mtctx = NULL;
#endif
}

size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx)
{
    if (cctx==NULL) return 0;   /* support free on NULL */
    RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
                    "not compatible with static CCtx");
    ZSTD_freeCCtxContent(cctx);
    ZSTD_free(cctx, cctx->customMem);
    return 0;
}


static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
    return ZSTDMT_sizeof_CCtx(cctx->mtctx);
#else
    (void)cctx;
    return 0;
#endif
}


size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx)
{
    if (cctx==NULL) return 0;   /* support sizeof on NULL */
    return sizeof(*cctx) + cctx->workSpaceSize
           + ZSTD_sizeof_localDict(cctx->localDict)
           + ZSTD_sizeof_mtctx(cctx);
}

size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs)
{
    return ZSTD_sizeof_CCtx(zcs);  /* same object */
}

/* private API call, for dictBuilder only */
const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx) { return &(ctx->seqStore); }

static ZSTD_CCtx_params ZSTD_makeCCtxParamsFromCParams(
        ZSTD_compressionParameters cParams)
{
    ZSTD_CCtx_params cctxParams;
    memset(&cctxParams, 0, sizeof(cctxParams));
    cctxParams.cParams = cParams;
    cctxParams.compressionLevel = ZSTD_CLEVEL_DEFAULT;  /* should not matter, as all cParams are presumed properly defined */
    assert(!ZSTD_checkCParams(cParams));
    cctxParams.fParams.contentSizeFlag = 1;
    return cctxParams;
}

static ZSTD_CCtx_params* ZSTD_createCCtxParams_advanced(
        ZSTD_customMem customMem)
{
    ZSTD_CCtx_params* params;
    if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
    params = (ZSTD_CCtx_params*)ZSTD_calloc(
            sizeof(ZSTD_CCtx_params), customMem);
    if (!params) { return NULL; }
    params->customMem = customMem;
    params->compressionLevel = ZSTD_CLEVEL_DEFAULT;
    params->fParams.contentSizeFlag = 1;
    return params;
}

ZSTD_CCtx_params* ZSTD_createCCtxParams(void)
{
    return ZSTD_createCCtxParams_advanced(ZSTD_defaultCMem);
}

size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params)
{
    if (params == NULL) { return 0; }
    ZSTD_free(params, params->customMem);
    return 0;
}

size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params)
{
    return ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
}

size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) {
    RETURN_ERROR_IF(!cctxParams, GENERIC);
    memset(cctxParams, 0, sizeof(*cctxParams));
    cctxParams->compressionLevel = compressionLevel;
    cctxParams->fParams.contentSizeFlag = 1;
    return 0;
}

size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params)
{
    RETURN_ERROR_IF(!cctxParams, GENERIC);
    FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) );
    memset(cctxParams, 0, sizeof(*cctxParams));
    cctxParams->cParams = params.cParams;
    cctxParams->fParams = params.fParams;
    cctxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT;   /* should not matter, as all cParams are presumed properly defined */
    assert(!ZSTD_checkCParams(params.cParams));
    return 0;
}

/* ZSTD_assignParamsToCCtxParams() :
 * params is presumed valid at this stage */
static ZSTD_CCtx_params ZSTD_assignParamsToCCtxParams(
        ZSTD_CCtx_params cctxParams, ZSTD_parameters params)
{
    ZSTD_CCtx_params ret = cctxParams;
    ret.cParams = params.cParams;
    ret.fParams = params.fParams;
    ret.compressionLevel = ZSTD_CLEVEL_DEFAULT;   /* should not matter, as all cParams are presumed properly defined */
    assert(!ZSTD_checkCParams(params.cParams));
    return ret;
}

ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
{
    ZSTD_bounds bounds = { 0, 0, 0 };

    switch(param)
    {
    case ZSTD_c_compressionLevel:
        bounds.lowerBound = ZSTD_minCLevel();
        bounds.upperBound = ZSTD_maxCLevel();
        return bounds;

    case ZSTD_c_windowLog:
        bounds.lowerBound = ZSTD_WINDOWLOG_MIN;
        bounds.upperBound = ZSTD_WINDOWLOG_MAX;
        return bounds;

    case ZSTD_c_hashLog:
        bounds.lowerBound = ZSTD_HASHLOG_MIN;
        bounds.upperBound = ZSTD_HASHLOG_MAX;
        return bounds;

    case ZSTD_c_chainLog:
        bounds.lowerBound = ZSTD_CHAINLOG_MIN;
        bounds.upperBound = ZSTD_CHAINLOG_MAX;
        return bounds;

    case ZSTD_c_searchLog:
        bounds.lowerBound = ZSTD_SEARCHLOG_MIN;
        bounds.upperBound = ZSTD_SEARCHLOG_MAX;
        return bounds;

    case ZSTD_c_minMatch:
        bounds.lowerBound = ZSTD_MINMATCH_MIN;
        bounds.upperBound = ZSTD_MINMATCH_MAX;
        return bounds;

    case ZSTD_c_targetLength:
        bounds.lowerBound = ZSTD_TARGETLENGTH_MIN;
        bounds.upperBound = ZSTD_TARGETLENGTH_MAX;
        return bounds;

    case ZSTD_c_strategy:
        bounds.lowerBound = ZSTD_STRATEGY_MIN;
        bounds.upperBound = ZSTD_STRATEGY_MAX;
        return bounds;

    case ZSTD_c_contentSizeFlag:
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_checksumFlag:
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_dictIDFlag:
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_nbWorkers:
        bounds.lowerBound = 0;
#ifdef ZSTD_MULTITHREAD
        bounds.upperBound = ZSTDMT_NBWORKERS_MAX;
#else
        bounds.upperBound = 0;
#endif
        return bounds;

    case ZSTD_c_jobSize:
        bounds.lowerBound = 0;
#ifdef ZSTD_MULTITHREAD
        bounds.upperBound = ZSTDMT_JOBSIZE_MAX;
#else
        bounds.upperBound = 0;
#endif
        return bounds;

    case ZSTD_c_overlapLog:
        bounds.lowerBound = ZSTD_OVERLAPLOG_MIN;
        bounds.upperBound = ZSTD_OVERLAPLOG_MAX;
        return bounds;

    case ZSTD_c_enableLongDistanceMatching:
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_ldmHashLog:
        bounds.lowerBound = ZSTD_LDM_HASHLOG_MIN;
        bounds.upperBound = ZSTD_LDM_HASHLOG_MAX;
        return bounds;

    case ZSTD_c_ldmMinMatch:
        bounds.lowerBound = ZSTD_LDM_MINMATCH_MIN;
        bounds.upperBound = ZSTD_LDM_MINMATCH_MAX;
        return bounds;

    case ZSTD_c_ldmBucketSizeLog:
        bounds.lowerBound = ZSTD_LDM_BUCKETSIZELOG_MIN;
        bounds.upperBound = ZSTD_LDM_BUCKETSIZELOG_MAX;
        return bounds;

    case ZSTD_c_ldmHashRateLog:
        bounds.lowerBound = ZSTD_LDM_HASHRATELOG_MIN;
        bounds.upperBound = ZSTD_LDM_HASHRATELOG_MAX;
        return bounds;

    /* experimental parameters */
    case ZSTD_c_rsyncable:
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_forceMaxWindow :
        bounds.lowerBound = 0;
        bounds.upperBound = 1;
        return bounds;

    case ZSTD_c_format:
        ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
        bounds.lowerBound = ZSTD_f_zstd1;
        bounds.upperBound = ZSTD_f_zstd1_magicless;   /* note : how to ensure at compile time that this is the highest value enum ? */
        return bounds;

    case ZSTD_c_forceAttachDict:
        ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceCopy);
        bounds.lowerBound = ZSTD_dictDefaultAttach;
        bounds.upperBound = ZSTD_dictForceCopy;       /* note : how to ensure at compile time that this is the highest value enum ? */
        return bounds;

    case ZSTD_c_literalCompressionMode:
        ZSTD_STATIC_ASSERT(ZSTD_lcm_auto < ZSTD_lcm_huffman && ZSTD_lcm_huffman < ZSTD_lcm_uncompressed);
        bounds.lowerBound = ZSTD_lcm_auto;
        bounds.upperBound = ZSTD_lcm_uncompressed;
        return bounds;

    case ZSTD_c_targetCBlockSize:
        bounds.lowerBound = ZSTD_TARGETCBLOCKSIZE_MIN;
        bounds.upperBound = ZSTD_TARGETCBLOCKSIZE_MAX;
        return bounds;

    default:
        {   ZSTD_bounds const boundError = { ERROR(parameter_unsupported), 0, 0 };
            return boundError;
        }
    }
}

/* ZSTD_cParam_withinBounds:
 * @return 1 if value is within cParam bounds,
 * 0 otherwise */
static int ZSTD_cParam_withinBounds(ZSTD_cParameter cParam, int value)
{
    ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
    if (ZSTD_isError(bounds.error)) return 0;
    if (value < bounds.lowerBound) return 0;
    if (value > bounds.upperBound) return 0;
    return 1;
}

/* ZSTD_cParam_clampBounds:
 * Clamps the value into the bounded range.
 */
static size_t ZSTD_cParam_clampBounds(ZSTD_cParameter cParam, int* value)
{
    ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
    if (ZSTD_isError(bounds.error)) return bounds.error;
    if (*value < bounds.lowerBound) *value = bounds.lowerBound;
    if (*value > bounds.upperBound) *value = bounds.upperBound;
    return 0;
}

#define BOUNDCHECK(cParam, val) { \
    RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \
                    parameter_outOfBound); \
}


static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
{
    switch(param)
    {
    case ZSTD_c_compressionLevel:
    case ZSTD_c_hashLog:
    case ZSTD_c_chainLog:
    case ZSTD_c_searchLog:
    case ZSTD_c_minMatch:
    case ZSTD_c_targetLength:
    case ZSTD_c_strategy:
        return 1;

    case ZSTD_c_format:
    case ZSTD_c_windowLog:
    case ZSTD_c_contentSizeFlag:
    case ZSTD_c_checksumFlag:
    case ZSTD_c_dictIDFlag:
    case ZSTD_c_forceMaxWindow :
    case ZSTD_c_nbWorkers:
    case ZSTD_c_jobSize:
    case ZSTD_c_overlapLog:
    case ZSTD_c_rsyncable:
    case ZSTD_c_enableLongDistanceMatching:
    case ZSTD_c_ldmHashLog:
    case ZSTD_c_ldmMinMatch:
    case ZSTD_c_ldmBucketSizeLog:
    case ZSTD_c_ldmHashRateLog:
    case ZSTD_c_forceAttachDict:
    case ZSTD_c_literalCompressionMode:
    case ZSTD_c_targetCBlockSize:
    default:
        return 0;
    }
}

size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
{
    DEBUGLOG(4, "ZSTD_CCtx_setParameter (%i, %i)", (int)param, value);
    if (cctx->streamStage != zcss_init) {
        if (ZSTD_isUpdateAuthorized(param)) {
            cctx->cParamsChanged = 1;
        } else {
            RETURN_ERROR(stage_wrong);
    }   }

    switch(param)
    {
    case ZSTD_c_nbWorkers:
        RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported,
                        "MT not compatible with static alloc");
        break;

    case ZSTD_c_compressionLevel:
    case ZSTD_c_windowLog:
    case ZSTD_c_hashLog:
    case ZSTD_c_chainLog:
    case ZSTD_c_searchLog:
    case ZSTD_c_minMatch:
    case ZSTD_c_targetLength:
    case ZSTD_c_strategy:
    case ZSTD_c_ldmHashRateLog:
    case ZSTD_c_format:
    case ZSTD_c_contentSizeFlag:
    case ZSTD_c_checksumFlag:
    case ZSTD_c_dictIDFlag:
    case ZSTD_c_forceMaxWindow:
    case ZSTD_c_forceAttachDict:
    case ZSTD_c_literalCompressionMode:
    case ZSTD_c_jobSize:
    case ZSTD_c_overlapLog:
    case ZSTD_c_rsyncable:
    case ZSTD_c_enableLongDistanceMatching:
    case ZSTD_c_ldmHashLog:
    case ZSTD_c_ldmMinMatch:
    case ZSTD_c_ldmBucketSizeLog:
    case ZSTD_c_targetCBlockSize:
        break;

    default: RETURN_ERROR(parameter_unsupported);
    }
    return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value);
}

size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
                                    ZSTD_cParameter param, int value)
{
    DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value);
    switch(param)
    {
    case ZSTD_c_format :
        BOUNDCHECK(ZSTD_c_format, value);
        CCtxParams->format = (ZSTD_format_e)value;
        return (size_t)CCtxParams->format;

    case ZSTD_c_compressionLevel : {
        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value));
        if (value) {  /* 0 : does not change current level */
            CCtxParams->compressionLevel = value;
        }
        if (CCtxParams->compressionLevel >= 0) return CCtxParams->compressionLevel;
        return 0;  /* return type (size_t) cannot represent negative values */
    }

    case ZSTD_c_windowLog :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_windowLog, value);
        CCtxParams->cParams.windowLog = value;
        return CCtxParams->cParams.windowLog;

    case ZSTD_c_hashLog :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_hashLog, value);
        CCtxParams->cParams.hashLog = value;
        return CCtxParams->cParams.hashLog;

    case ZSTD_c_chainLog :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_chainLog, value);
        CCtxParams->cParams.chainLog = value;
        return CCtxParams->cParams.chainLog;

    case ZSTD_c_searchLog :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_searchLog, value);
        CCtxParams->cParams.searchLog = value;
        return value;

    case ZSTD_c_minMatch :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_minMatch, value);
        CCtxParams->cParams.minMatch = value;
        return CCtxParams->cParams.minMatch;

    case ZSTD_c_targetLength :
        BOUNDCHECK(ZSTD_c_targetLength, value);
        CCtxParams->cParams.targetLength = value;
        return CCtxParams->cParams.targetLength;

    case ZSTD_c_strategy :
        if (value!=0)   /* 0 => use default */
            BOUNDCHECK(ZSTD_c_strategy, value);
        CCtxParams->cParams.strategy = (ZSTD_strategy)value;
        return (size_t)CCtxParams->cParams.strategy;

    case ZSTD_c_contentSizeFlag :
        /* Content size written in frame header _when known_ (default:1) */
        DEBUGLOG(4, "set content size flag = %u", (value!=0));
        CCtxParams->fParams.contentSizeFlag = value != 0;
        return CCtxParams->fParams.contentSizeFlag;

    case ZSTD_c_checksumFlag :
        /* A 32-bits content checksum will be calculated and written at end of frame (default:0) */
        CCtxParams->fParams.checksumFlag = value != 0;
        return CCtxParams->fParams.checksumFlag;

    case ZSTD_c_dictIDFlag : /* When applicable, dictionary's dictID is provided in frame header (default:1) */
        DEBUGLOG(4, "set dictIDFlag = %u", (value!=0));
        CCtxParams->fParams.noDictIDFlag = !value;
        return !CCtxParams->fParams.noDictIDFlag;

    case ZSTD_c_forceMaxWindow :
        CCtxParams->forceWindow = (value != 0);
        return CCtxParams->forceWindow;

    case ZSTD_c_forceAttachDict : {
        const ZSTD_dictAttachPref_e pref = (ZSTD_dictAttachPref_e)value;
        BOUNDCHECK(ZSTD_c_forceAttachDict, pref);
        CCtxParams->attachDictPref = pref;
        return CCtxParams->attachDictPref;
    }

    case ZSTD_c_literalCompressionMode : {
        const ZSTD_literalCompressionMode_e lcm = (ZSTD_literalCompressionMode_e)value;
        BOUNDCHECK(ZSTD_c_literalCompressionMode, lcm);
        CCtxParams->literalCompressionMode = lcm;
        return CCtxParams->literalCompressionMode;
    }

    case ZSTD_c_nbWorkers :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
        return 0;
#else
        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value));
        CCtxParams->nbWorkers = value;
        return CCtxParams->nbWorkers;
#endif

    case ZSTD_c_jobSize :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
        return 0;
#else
        /* Adjust to the minimum non-default value. */
        if (value != 0 && value < ZSTDMT_JOBSIZE_MIN)
            value = ZSTDMT_JOBSIZE_MIN;
        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value));
        assert(value >= 0);
        CCtxParams->jobSize = value;
        return CCtxParams->jobSize;
#endif

    case ZSTD_c_overlapLog :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
        return 0;
#else
        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value));
        CCtxParams->overlapLog = value;
        return CCtxParams->overlapLog;
#endif

    case ZSTD_c_rsyncable :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
        return 0;
#else
        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value));
        CCtxParams->rsyncable = value;
        return CCtxParams->rsyncable;
#endif

    case ZSTD_c_enableLongDistanceMatching :
        CCtxParams->ldmParams.enableLdm = (value!=0);
        return CCtxParams->ldmParams.enableLdm;

    case ZSTD_c_ldmHashLog :
        if (value!=0)   /* 0 ==> auto */
            BOUNDCHECK(ZSTD_c_ldmHashLog, value);
        CCtxParams->ldmParams.hashLog = value;
        return CCtxParams->ldmParams.hashLog;

    case ZSTD_c_ldmMinMatch :
        if (value!=0)   /* 0 ==> default */
            BOUNDCHECK(ZSTD_c_ldmMinMatch, value);
        CCtxParams->ldmParams.minMatchLength = value;
        return CCtxParams->ldmParams.minMatchLength;

    case ZSTD_c_ldmBucketSizeLog :
        if (value!=0)   /* 0 ==> default */
            BOUNDCHECK(ZSTD_c_ldmBucketSizeLog, value);
        CCtxParams->ldmParams.bucketSizeLog = value;
        return CCtxParams->ldmParams.bucketSizeLog;

    case ZSTD_c_ldmHashRateLog :
        RETURN_ERROR_IF(value > ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN,
                        parameter_outOfBound);
        CCtxParams->ldmParams.hashRateLog = value;
        return CCtxParams->ldmParams.hashRateLog;

    case ZSTD_c_targetCBlockSize :
        if (value!=0)   /* 0 ==> default */
            BOUNDCHECK(ZSTD_c_targetCBlockSize, value);
        CCtxParams->targetCBlockSize = value;
        return CCtxParams->targetCBlockSize;

    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
    }
}

size_t ZSTD_CCtx_getParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value)
{
    return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value);
}

size_t ZSTD_CCtxParams_getParameter(
        ZSTD_CCtx_params* CCtxParams, ZSTD_cParameter param, int* value)
{
    switch(param)
    {
    case ZSTD_c_format :
        *value = CCtxParams->format;
        break;
    case ZSTD_c_compressionLevel :
        *value = CCtxParams->compressionLevel;
        break;
    case ZSTD_c_windowLog :
        *value = (int)CCtxParams->cParams.windowLog;
        break;
    case ZSTD_c_hashLog :
        *value = (int)CCtxParams->cParams.hashLog;
        break;
    case ZSTD_c_chainLog :
        *value = (int)CCtxParams->cParams.chainLog;
        break;
    case ZSTD_c_searchLog :
        *value = CCtxParams->cParams.searchLog;
        break;
    case ZSTD_c_minMatch :
        *value = CCtxParams->cParams.minMatch;
        break;
    case ZSTD_c_targetLength :
        *value = CCtxParams->cParams.targetLength;
        break;
    case ZSTD_c_strategy :
        *value = (unsigned)CCtxParams->cParams.strategy;
        break;
    case ZSTD_c_contentSizeFlag :
        *value = CCtxParams->fParams.contentSizeFlag;
        break;
    case ZSTD_c_checksumFlag :
        *value = CCtxParams->fParams.checksumFlag;
        break;
    case ZSTD_c_dictIDFlag :
        *value = !CCtxParams->fParams.noDictIDFlag;
        break;
    case ZSTD_c_forceMaxWindow :
        *value = CCtxParams->forceWindow;
        break;
    case ZSTD_c_forceAttachDict :
        *value = CCtxParams->attachDictPref;
        break;
    case ZSTD_c_literalCompressionMode :
        *value = CCtxParams->literalCompressionMode;
        break;
    case ZSTD_c_nbWorkers :
#ifndef ZSTD_MULTITHREAD
        assert(CCtxParams->nbWorkers == 0);
#endif
        *value = CCtxParams->nbWorkers;
        break;
    case ZSTD_c_jobSize :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
#else
        assert(CCtxParams->jobSize <= INT_MAX);
        *value = (int)CCtxParams->jobSize;
        break;
#endif
    case ZSTD_c_overlapLog :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
#else
        *value = CCtxParams->overlapLog;
        break;
#endif
    case ZSTD_c_rsyncable :
#ifndef ZSTD_MULTITHREAD
        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
#else
        *value = CCtxParams->rsyncable;
        break;
#endif
    case ZSTD_c_enableLongDistanceMatching :
        *value = CCtxParams->ldmParams.enableLdm;
        break;
    case ZSTD_c_ldmHashLog :
        *value = CCtxParams->ldmParams.hashLog;
        break;
    case ZSTD_c_ldmMinMatch :
        *value = CCtxParams->ldmParams.minMatchLength;
        break;
    case ZSTD_c_ldmBucketSizeLog :
        *value = CCtxParams->ldmParams.bucketSizeLog;
        break;
    case ZSTD_c_ldmHashRateLog :
        *value = CCtxParams->ldmParams.hashRateLog;
        break;
    case ZSTD_c_targetCBlockSize :
        *value = (int)CCtxParams->targetCBlockSize;
        break;
    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
    }
    return 0;
}

/** ZSTD_CCtx_setParametersUsingCCtxParams() :
 *  just applies `params` into `cctx`
 *  no action is performed, parameters are merely stored.
 *  If ZSTDMT is enabled, parameters are pushed to cctx->mtctx.
 *    This is possible even if a compression is ongoing.
 *    In which case, new parameters will be applied on the fly, starting with next compression job.
 */
size_t ZSTD_CCtx_setParametersUsingCCtxParams(
        ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params)
{
    DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams");
    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
    RETURN_ERROR_IF(cctx->cdict, stage_wrong);

    cctx->requestedParams = *params;
    return 0;
}

ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize)
{
    DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %u bytes", (U32)pledgedSrcSize);
    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
    cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
    return 0;
}

/**
 * Initializes the local dict using the requested parameters.
 * NOTE: This does not use the pledged src size, because it may be used for more
 * than one compression.
 */
static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx)
{
    ZSTD_localDict* const dl = &cctx->localDict;
    ZSTD_compressionParameters const cParams = ZSTD_getCParamsFromCCtxParams(
            &cctx->requestedParams, 0, dl->dictSize);
    if (dl->dict == NULL) {
        /* No local dictionary. */
        assert(dl->dictBuffer == NULL);
        assert(dl->cdict == NULL);
        assert(dl->dictSize == 0);
        return 0;
    }
    if (dl->cdict != NULL) {
        assert(cctx->cdict == dl->cdict);
        /* Local dictionary already initialized. */
        return 0;
    }
    assert(dl->dictSize > 0);
    assert(cctx->cdict == NULL);
    assert(cctx->prefixDict.dict == NULL);

    dl->cdict = ZSTD_createCDict_advanced(
            dl->dict,
            dl->dictSize,
            ZSTD_dlm_byRef,
            dl->dictContentType,
            cParams,
            cctx->customMem);
    RETURN_ERROR_IF(!dl->cdict, memory_allocation);
    cctx->cdict = dl->cdict;
    return 0;
}

size_t ZSTD_CCtx_loadDictionary_advanced(
        ZSTD_CCtx* cctx, const void* dict, size_t dictSize,
        ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType)
{
    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
    RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
                    "no malloc for static CCtx");
    DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize);
    ZSTD_clearAllDicts(cctx);  /* in case one already exists */
    if (dict == NULL || dictSize == 0)  /* no dictionary mode */
        return 0;
    if (dictLoadMethod == ZSTD_dlm_byRef) {
        cctx->localDict.dict = dict;
    } else {
        void* dictBuffer = ZSTD_malloc(dictSize, cctx->customMem);
        RETURN_ERROR_IF(!dictBuffer, memory_allocation);
        memcpy(dictBuffer, dict, dictSize);
        cctx->localDict.dictBuffer = dictBuffer;
        cctx->localDict.dict = dictBuffer;
    }
    cctx->localDict.dictSize = dictSize;
    cctx->localDict.dictContentType = dictContentType;
    return 0;
}

ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(
      ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
{
    return ZSTD_CCtx_loadDictionary_advanced(
            cctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
}

ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
{
    return ZSTD_CCtx_loadDictionary_advanced(
            cctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
}


size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
{
    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
    /* Free the existing local cdict (if any) to save memory. */
    ZSTD_clearAllDicts(cctx);
    cctx->cdict = cdict;
    return 0;
}

size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize)
{
    return ZSTD_CCtx_refPrefix_advanced(cctx, prefix, prefixSize, ZSTD_dct_rawContent);
}

size_t ZSTD_CCtx_refPrefix_advanced(
        ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
{
    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
    ZSTD_clearAllDicts(cctx);
    cctx->prefixDict.dict = prefix;
    cctx->prefixDict.dictSize = prefixSize;
    cctx->prefixDict.dictContentType = dictContentType;
    return 0;
}

/*! ZSTD_CCtx_reset() :
 *  Also dumps dictionary */
size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
{
    if ( (reset == ZSTD_reset_session_only)
      || (reset == ZSTD_reset_session_and_parameters) ) {
        cctx->streamStage = zcss_init;
        cctx->pledgedSrcSizePlusOne = 0;
    }
    if ( (reset == ZSTD_reset_parameters)
      || (reset == ZSTD_reset_session_and_parameters) ) {
        RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong);
        ZSTD_clearAllDicts(cctx);
        return ZSTD_CCtxParams_reset(&cctx->requestedParams);
    }
    return 0;
}


/** ZSTD_checkCParams() :
    control CParam values remain within authorized range.
    @return : 0, or an error code if one value is beyond authorized range */
size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
{
    BOUNDCHECK(ZSTD_c_windowLog, (int)cParams.windowLog);
    BOUNDCHECK(ZSTD_c_chainLog,  (int)cParams.chainLog);
    BOUNDCHECK(ZSTD_c_hashLog,   (int)cParams.hashLog);
    BOUNDCHECK(ZSTD_c_searchLog, (int)cParams.searchLog);
    BOUNDCHECK(ZSTD_c_minMatch,  (int)cParams.minMatch);
    BOUNDCHECK(ZSTD_c_targetLength,(int)cParams.targetLength);
    BOUNDCHECK(ZSTD_c_strategy,  cParams.strategy);
    return 0;
}

/** ZSTD_clampCParams() :
 *  make CParam values within valid range.
 *  @return : valid CParams */
static ZSTD_compressionParameters
ZSTD_clampCParams(ZSTD_compressionParameters cParams)
{
#   define CLAMP_TYPE(cParam, val, type) {                                \
        ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);         \
        if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound;      \
        else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \
    }
#   define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, unsigned)
    CLAMP(ZSTD_c_windowLog, cParams.windowLog);
    CLAMP(ZSTD_c_chainLog,  cParams.chainLog);
    CLAMP(ZSTD_c_hashLog,   cParams.hashLog);
    CLAMP(ZSTD_c_searchLog, cParams.searchLog);
    CLAMP(ZSTD_c_minMatch,  cParams.minMatch);
    CLAMP(ZSTD_c_targetLength,cParams.targetLength);
    CLAMP_TYPE(ZSTD_c_strategy,cParams.strategy, ZSTD_strategy);
    return cParams;
}

/** ZSTD_cycleLog() :
 *  condition for correct operation : hashLog > 1 */
static U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
{
    U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
    return hashLog - btScale;
}

/** ZSTD_adjustCParams_internal() :
 *  optimize `cPar` for a specified input (`srcSize` and `dictSize`).
 *  mostly downsize to reduce memory consumption and initialization latency.
 * `srcSize` can be ZSTD_CONTENTSIZE_UNKNOWN when not known.
 *  note : for the time being, `srcSize==0` means "unknown" too, for compatibility with older convention.
 *  condition : cPar is presumed validated (can be checked using ZSTD_checkCParams()). */
static ZSTD_compressionParameters
ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
                            unsigned long long srcSize,
                            size_t dictSize)
{
    static const U64 minSrcSize = 513; /* (1<<9) + 1 */
    static const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
    assert(ZSTD_checkCParams(cPar)==0);

    if (dictSize && (srcSize+1<2) /* ZSTD_CONTENTSIZE_UNKNOWN and 0 mean "unknown" */ )
        srcSize = minSrcSize;  /* presumed small when there is a dictionary */
    else if (srcSize == 0)
        srcSize = ZSTD_CONTENTSIZE_UNKNOWN;  /* 0 == unknown : presumed large */

    /* resize windowLog if input is small enough, to use less memory */
    if ( (srcSize < maxWindowResize)
      && (dictSize < maxWindowResize) )  {
        U32 const tSize = (U32)(srcSize + dictSize);
        static U32 const hashSizeMin = 1 << ZSTD_HASHLOG_MIN;
        U32 const srcLog = (tSize < hashSizeMin) ? ZSTD_HASHLOG_MIN :
                            ZSTD_highbit32(tSize-1) + 1;
        if (cPar.windowLog > srcLog) cPar.windowLog = srcLog;
    }
    if (cPar.hashLog > cPar.windowLog+1) cPar.hashLog = cPar.windowLog+1;
    {   U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
        if (cycleLog > cPar.windowLog)
            cPar.chainLog -= (cycleLog - cPar.windowLog);
    }

    if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
        cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN;  /* minimum wlog required for valid frame header */

    return cPar;
}

ZSTD_compressionParameters
ZSTD_adjustCParams(ZSTD_compressionParameters cPar,
                   unsigned long long srcSize,
                   size_t dictSize)
{
    cPar = ZSTD_clampCParams(cPar);   /* resulting cPar is necessarily valid (all parameters within range) */
    return ZSTD_adjustCParams_internal(cPar, srcSize, dictSize);
}

ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
        const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize)
{
    ZSTD_compressionParameters cParams = ZSTD_getCParams(CCtxParams->compressionLevel, srcSizeHint, dictSize);
    if (CCtxParams->ldmParams.enableLdm) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG;
    if (CCtxParams->cParams.windowLog) cParams.windowLog = CCtxParams->cParams.windowLog;
    if (CCtxParams->cParams.hashLog) cParams.hashLog = CCtxParams->cParams.hashLog;
    if (CCtxParams->cParams.chainLog) cParams.chainLog = CCtxParams->cParams.chainLog;
    if (CCtxParams->cParams.searchLog) cParams.searchLog = CCtxParams->cParams.searchLog;
    if (CCtxParams->cParams.minMatch) cParams.minMatch = CCtxParams->cParams.minMatch;
    if (CCtxParams->cParams.targetLength) cParams.targetLength = CCtxParams->cParams.targetLength;
    if (CCtxParams->cParams.strategy) cParams.strategy = CCtxParams->cParams.strategy;
    assert(!ZSTD_checkCParams(cParams));
    return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize);
}

static size_t
ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
                       const U32 forCCtx)
{
    size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
    size_t const hSize = ((size_t)1) << cParams->hashLog;
    U32    const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
    size_t const h3Size = ((size_t)1) << hashLog3;
    size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
    size_t const optPotentialSpace = ((MaxML+1) + (MaxLL+1) + (MaxOff+1) + (1<<Litbits)) * sizeof(U32)
                          + (ZSTD_OPT_NUM+1) * (sizeof(ZSTD_match_t)+sizeof(ZSTD_optimal_t));
    size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
                                ? optPotentialSpace
                                : 0;
    DEBUGLOG(4, "chainSize: %u - hSize: %u - h3Size: %u",
                (U32)chainSize, (U32)hSize, (U32)h3Size);
    return tableSpace + optSpace;
}

size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params)
{
    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
    {   ZSTD_compressionParameters const cParams =
                ZSTD_getCParamsFromCCtxParams(params, 0, 0);
        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
        U32    const divider = (cParams.minMatch==3) ? 3 : 4;
        size_t const maxNbSeq = blockSize / divider;
        size_t const tokenSpace = WILDCOPY_OVERLENGTH + blockSize + 11*maxNbSeq;
        size_t const entropySpace = HUF_WORKSPACE_SIZE;
        size_t const blockStateSpace = 2 * sizeof(ZSTD_compressedBlockState_t);
        size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 1);

        size_t const ldmSpace = ZSTD_ldm_getTableSize(params->ldmParams);
        size_t const ldmSeqSpace = ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize) * sizeof(rawSeq);

        size_t const neededSpace = entropySpace + blockStateSpace + tokenSpace +
                                   matchStateSize + ldmSpace + ldmSeqSpace;

        DEBUGLOG(5, "sizeof(ZSTD_CCtx) : %u", (U32)sizeof(ZSTD_CCtx));
        DEBUGLOG(5, "estimate workSpace : %u", (U32)neededSpace);
        return sizeof(ZSTD_CCtx) + neededSpace;
    }
}

size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams)
{
    ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
    return ZSTD_estimateCCtxSize_usingCCtxParams(&params);
}

static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel)
{
    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, 0, 0);
    return ZSTD_estimateCCtxSize_usingCParams(cParams);
}

size_t ZSTD_estimateCCtxSize(int compressionLevel)
{
    int level;
    size_t memBudget = 0;
    for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
        size_t const newMB = ZSTD_estimateCCtxSize_internal(level);
        if (newMB > memBudget) memBudget = newMB;
    }
    return memBudget;
}

size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params)
{
    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
    {   ZSTD_compressionParameters const cParams =
                ZSTD_getCParamsFromCCtxParams(params, 0, 0);
        size_t const CCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(params);
        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
        size_t const inBuffSize = ((size_t)1 << cParams.windowLog) + blockSize;
        size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1;
        size_t const streamingSize = inBuffSize + outBuffSize;

        return CCtxSize + streamingSize;
    }
}

size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams)
{
    ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
    return ZSTD_estimateCStreamSize_usingCCtxParams(&params);
}

static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel)
{
    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, 0, 0);
    return ZSTD_estimateCStreamSize_usingCParams(cParams);
}

size_t ZSTD_estimateCStreamSize(int compressionLevel)
{
    int level;
    size_t memBudget = 0;
    for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
        size_t const newMB = ZSTD_estimateCStreamSize_internal(level);
        if (newMB > memBudget) memBudget = newMB;
    }
    return memBudget;
}

/* ZSTD_getFrameProgression():
 * tells how much data has been consumed (input) and produced (output) for current frame.
 * able to count progression inside worker threads (non-blocking mode).
 */
ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
    if (cctx->appliedParams.nbWorkers > 0) {
        return ZSTDMT_getFrameProgression(cctx->mtctx);
    }
#endif
    {   ZSTD_frameProgression fp;
        size_t const buffered = (cctx->inBuff == NULL) ? 0 :
                                cctx->inBuffPos - cctx->inToCompress;
        if (buffered) assert(cctx->inBuffPos >= cctx->inToCompress);
        assert(buffered <= ZSTD_BLOCKSIZE_MAX);
        fp.ingested = cctx->consumedSrcSize + buffered;
        fp.consumed = cctx->consumedSrcSize;
        fp.produced = cctx->producedCSize;
        fp.flushed  = cctx->producedCSize;   /* simplified; some data might still be left within streaming output buffer */
        fp.currentJobID = 0;
        fp.nbActiveWorkers = 0;
        return fp;
}   }

/*! ZSTD_toFlushNow()
 *  Only useful for multithreading scenarios currently (nbWorkers >= 1).
 */
size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
    if (cctx->appliedParams.nbWorkers > 0) {
        return ZSTDMT_toFlushNow(cctx->mtctx);
    }
#endif
    (void)cctx;
    return 0;   /* over-simplification; could also check if context is currently running in streaming mode, and in which case, report how many bytes are left to be flushed within output buffer */
}



static U32 ZSTD_equivalentCParams(ZSTD_compressionParameters cParams1,
                                  ZSTD_compressionParameters cParams2)
{
    return (cParams1.hashLog  == cParams2.hashLog)
         & (cParams1.chainLog == cParams2.chainLog)
         & (cParams1.strategy == cParams2.strategy)   /* opt parser space */
         & ((cParams1.minMatch==3) == (cParams2.minMatch==3));  /* hashlog3 space */
}

static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
                                    ZSTD_compressionParameters cParams2)
{
    (void)cParams1;
    (void)cParams2;
    assert(cParams1.windowLog    == cParams2.windowLog);
    assert(cParams1.chainLog     == cParams2.chainLog);
    assert(cParams1.hashLog      == cParams2.hashLog);
    assert(cParams1.searchLog    == cParams2.searchLog);
    assert(cParams1.minMatch     == cParams2.minMatch);
    assert(cParams1.targetLength == cParams2.targetLength);
    assert(cParams1.strategy     == cParams2.strategy);
}

/** The parameters are equivalent if ldm is not enabled in both sets or
 *  all the parameters are equivalent. */
static U32 ZSTD_equivalentLdmParams(ldmParams_t ldmParams1,
                                    ldmParams_t ldmParams2)
{
    return (!ldmParams1.enableLdm && !ldmParams2.enableLdm) ||
           (ldmParams1.enableLdm == ldmParams2.enableLdm &&
            ldmParams1.hashLog == ldmParams2.hashLog &&
            ldmParams1.bucketSizeLog == ldmParams2.bucketSizeLog &&
            ldmParams1.minMatchLength == ldmParams2.minMatchLength &&
            ldmParams1.hashRateLog == ldmParams2.hashRateLog);
}

typedef enum { ZSTDb_not_buffered, ZSTDb_buffered } ZSTD_buffered_policy_e;

/* ZSTD_sufficientBuff() :
 * check internal buffers exist for streaming if buffPol == ZSTDb_buffered .
 * Note : they are assumed to be correctly sized if ZSTD_equivalentCParams()==1 */
static U32 ZSTD_sufficientBuff(size_t bufferSize1, size_t maxNbSeq1,
                            size_t maxNbLit1,
                            ZSTD_buffered_policy_e buffPol2,
                            ZSTD_compressionParameters cParams2,
                            U64 pledgedSrcSize)
{
    size_t const windowSize2 = MAX(1, (size_t)MIN(((U64)1 << cParams2.windowLog), pledgedSrcSize));
    size_t const blockSize2 = MIN(ZSTD_BLOCKSIZE_MAX, windowSize2);
    size_t const maxNbSeq2 = blockSize2 / ((cParams2.minMatch == 3) ? 3 : 4);
    size_t const maxNbLit2 = blockSize2;
    size_t const neededBufferSize2 = (buffPol2==ZSTDb_buffered) ? windowSize2 + blockSize2 : 0;
    DEBUGLOG(4, "ZSTD_sufficientBuff: is neededBufferSize2=%u <= bufferSize1=%u",
                (U32)neededBufferSize2, (U32)bufferSize1);
    DEBUGLOG(4, "ZSTD_sufficientBuff: is maxNbSeq2=%u <= maxNbSeq1=%u",
                (U32)maxNbSeq2, (U32)maxNbSeq1);
    DEBUGLOG(4, "ZSTD_sufficientBuff: is maxNbLit2=%u <= maxNbLit1=%u",
                (U32)maxNbLit2, (U32)maxNbLit1);
    return (maxNbLit2 <= maxNbLit1)
         & (maxNbSeq2 <= maxNbSeq1)
         & (neededBufferSize2 <= bufferSize1);
}

/** Equivalence for resetCCtx purposes */
static U32 ZSTD_equivalentParams(ZSTD_CCtx_params params1,
                                 ZSTD_CCtx_params params2,
                                 size_t buffSize1,
                                 size_t maxNbSeq1, size_t maxNbLit1,
                                 ZSTD_buffered_policy_e buffPol2,
                                 U64 pledgedSrcSize)
{
    DEBUGLOG(4, "ZSTD_equivalentParams: pledgedSrcSize=%u", (U32)pledgedSrcSize);
    if (!ZSTD_equivalentCParams(params1.cParams, params2.cParams)) {
      DEBUGLOG(4, "ZSTD_equivalentCParams() == 0");
      return 0;
    }
    if (!ZSTD_equivalentLdmParams(params1.ldmParams, params2.ldmParams)) {
      DEBUGLOG(4, "ZSTD_equivalentLdmParams() == 0");
      return 0;
    }
    if (!ZSTD_sufficientBuff(buffSize1, maxNbSeq1, maxNbLit1, buffPol2,
                             params2.cParams, pledgedSrcSize)) {
      DEBUGLOG(4, "ZSTD_sufficientBuff() == 0");
      return 0;
    }
    return 1;
}

static void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
{
    int i;
    for (i = 0; i < ZSTD_REP_NUM; ++i)
        bs->rep[i] = repStartValue[i];
    bs->entropy.huf.repeatMode = HUF_repeat_none;
    bs->entropy.fse.offcode_repeatMode = FSE_repeat_none;
    bs->entropy.fse.matchlength_repeatMode = FSE_repeat_none;
    bs->entropy.fse.litlength_repeatMode = FSE_repeat_none;
}

/*! ZSTD_invalidateMatchState()
 *  Invalidate all the matches in the match finder tables.
 *  Requires nextSrc and base to be set (can be NULL).
 */
static void ZSTD_invalidateMatchState(ZSTD_matchState_t* ms)
{
    ZSTD_window_clear(&ms->window);

    ms->nextToUpdate = ms->window.dictLimit;
    ms->loadedDictEnd = 0;
    ms->opt.litLengthSum = 0;  /* force reset of btopt stats */
    ms->dictMatchState = NULL;
}

/*! ZSTD_continueCCtx() :
 *  reuse CCtx without reset (note : requires no dictionary) */
static size_t ZSTD_continueCCtx(ZSTD_CCtx* cctx, ZSTD_CCtx_params params, U64 pledgedSrcSize)
{
    size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize));
    size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
    DEBUGLOG(4, "ZSTD_continueCCtx: re-use context in place");

    cctx->blockSize = blockSize;   /* previous block size could be different even for same windowLog, due to pledgedSrcSize */
    cctx->appliedParams = params;
    cctx->blockState.matchState.cParams = params.cParams;
    cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
    cctx->consumedSrcSize = 0;
    cctx->producedCSize = 0;
    if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
        cctx->appliedParams.fParams.contentSizeFlag = 0;
    DEBUGLOG(4, "pledged content size : %u ; flag : %u",
        (U32)pledgedSrcSize, cctx->appliedParams.fParams.contentSizeFlag);
    cctx->stage = ZSTDcs_init;
    cctx->dictID = 0;
    if (params.ldmParams.enableLdm)
        ZSTD_window_clear(&cctx->ldmState.window);
    ZSTD_referenceExternalSequences(cctx, NULL, 0);
    ZSTD_invalidateMatchState(&cctx->blockState.matchState);
    ZSTD_reset_compressedBlockState(cctx->blockState.prevCBlock);
    XXH64_reset(&cctx->xxhState, 0);
    return 0;
}

typedef enum { ZSTDcrp_continue, ZSTDcrp_noMemset } ZSTD_compResetPolicy_e;

typedef enum { ZSTD_resetTarget_CDict, ZSTD_resetTarget_CCtx } ZSTD_resetTarget_e;

static void*
ZSTD_reset_matchState(ZSTD_matchState_t* ms,
                      void* ptr,
                const ZSTD_compressionParameters* cParams,
                      ZSTD_compResetPolicy_e const crp, ZSTD_resetTarget_e const forWho)
{
    size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
    size_t const hSize = ((size_t)1) << cParams->hashLog;
    U32    const hashLog3 = ((forWho == ZSTD_resetTarget_CCtx) && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
    size_t const h3Size = ((size_t)1) << hashLog3;
    size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);

    assert(((size_t)ptr & 3) == 0);

    ms->hashLog3 = hashLog3;
    memset(&ms->window, 0, sizeof(ms->window));
    ms->window.dictLimit = 1;    /* start from 1, so that 1st position is valid */
    ms->window.lowLimit = 1;     /* it ensures first and later CCtx usages compress the same */
    ms->window.nextSrc = ms->window.base + 1;   /* see issue #1241 */
    ZSTD_invalidateMatchState(ms);

    /* opt parser space */
    if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) {
        DEBUGLOG(4, "reserving optimal parser space");
        ms->opt.litFreq = (unsigned*)ptr;
        ms->opt.litLengthFreq = ms->opt.litFreq + (1<<Litbits);
        ms->opt.matchLengthFreq = ms->opt.litLengthFreq + (MaxLL+1);
        ms->opt.offCodeFreq = ms->opt.matchLengthFreq + (MaxML+1);
        ptr = ms->opt.offCodeFreq + (MaxOff+1);
        ms->opt.matchTable = (ZSTD_match_t*)ptr;
        ptr = ms->opt.matchTable + ZSTD_OPT_NUM+1;
        ms->opt.priceTable = (ZSTD_optimal_t*)ptr;
        ptr = ms->opt.priceTable + ZSTD_OPT_NUM+1;
    }

    /* table Space */
    DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_noMemset);
    assert(((size_t)ptr & 3) == 0);  /* ensure ptr is properly aligned */
    if (crp!=ZSTDcrp_noMemset) memset(ptr, 0, tableSpace);   /* reset tables only */
    ms->hashTable = (U32*)(ptr);
    ms->chainTable = ms->hashTable + hSize;
    ms->hashTable3 = ms->chainTable + chainSize;
    ptr = ms->hashTable3 + h3Size;

    ms->cParams = *cParams;

    assert(((size_t)ptr & 3) == 0);
    return ptr;
}

/* ZSTD_indexTooCloseToMax() :
 * minor optimization : prefer memset() rather than reduceIndex()
 * which is measurably slow in some circumstances (reported for Visual Studio).
 * Works when re-using a context for a lot of smallish inputs :
 * if all inputs are smaller than ZSTD_INDEXOVERFLOW_MARGIN,
 * memset() will be triggered before reduceIndex().
 */
#define ZSTD_INDEXOVERFLOW_MARGIN (16 MB)
static int ZSTD_indexTooCloseToMax(ZSTD_window_t w)
{
    return (size_t)(w.nextSrc - w.base) > (ZSTD_CURRENT_MAX - ZSTD_INDEXOVERFLOW_MARGIN);
}

#define ZSTD_WORKSPACETOOLARGE_FACTOR 3 /* define "workspace is too large" as this number of times larger than needed */
#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128  /* when workspace is continuously too large
                                         * during at least this number of times,
                                         * context's memory usage is considered wasteful,
                                         * because it's sized to handle a worst case scenario which rarely happens.
                                         * In which case, resize it down to free some memory */

/*! ZSTD_resetCCtx_internal() :
    note : `params` are assumed fully validated at this stage */
static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
                                      ZSTD_CCtx_params params,
                                      U64 const pledgedSrcSize,
                                      ZSTD_compResetPolicy_e const crp,
                                      ZSTD_buffered_policy_e const zbuff)
{
    DEBUGLOG(4, "ZSTD_resetCCtx_internal: pledgedSrcSize=%u, wlog=%u",
                (U32)pledgedSrcSize, params.cParams.windowLog);
    assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));

    if (crp == ZSTDcrp_continue) {
        if (ZSTD_equivalentParams(zc->appliedParams, params,
                                  zc->inBuffSize,
                                  zc->seqStore.maxNbSeq, zc->seqStore.maxNbLit,
                                  zbuff, pledgedSrcSize) ) {
            DEBUGLOG(4, "ZSTD_equivalentParams()==1 -> consider continue mode");
            zc->workSpaceOversizedDuration += (zc->workSpaceOversizedDuration > 0);   /* if it was too large, it still is */
            if (zc->workSpaceOversizedDuration <= ZSTD_WORKSPACETOOLARGE_MAXDURATION) {
                DEBUGLOG(4, "continue mode confirmed (wLog1=%u, blockSize1=%zu)",
                            zc->appliedParams.cParams.windowLog, zc->blockSize);
                if (ZSTD_indexTooCloseToMax(zc->blockState.matchState.window)) {
                    /* prefer a reset, faster than a rescale */
                    ZSTD_reset_matchState(&zc->blockState.matchState,
                                           zc->entropyWorkspace + HUF_WORKSPACE_SIZE_U32,
                                          &params.cParams,
                                           crp, ZSTD_resetTarget_CCtx);
                }
                return ZSTD_continueCCtx(zc, params, pledgedSrcSize);
    }   }   }
    DEBUGLOG(4, "ZSTD_equivalentParams()==0 -> reset CCtx");

    if (params.ldmParams.enableLdm) {
        /* Adjust long distance matching parameters */
        ZSTD_ldm_adjustParameters(&params.ldmParams, &params.cParams);
        assert(params.ldmParams.hashLog >= params.ldmParams.bucketSizeLog);
        assert(params.ldmParams.hashRateLog < 32);
        zc->ldmState.hashPower = ZSTD_rollingHash_primePower(params.ldmParams.minMatchLength);
    }

    {   size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize));
        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
        U32    const divider = (params.cParams.minMatch==3) ? 3 : 4;
        size_t const maxNbSeq = blockSize / divider;
        size_t const tokenSpace = WILDCOPY_OVERLENGTH + blockSize + 11*maxNbSeq;
        size_t const buffOutSize = (zbuff==ZSTDb_buffered) ? ZSTD_compressBound(blockSize)+1 : 0;
        size_t const buffInSize = (zbuff==ZSTDb_buffered) ? windowSize + blockSize : 0;
        size_t const matchStateSize = ZSTD_sizeof_matchState(&params.cParams, /* forCCtx */ 1);
        size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params.ldmParams, blockSize);
        void* ptr;   /* used to partition workSpace */

        /* Check if workSpace is large enough, alloc a new one if needed */
        {   size_t const entropySpace = HUF_WORKSPACE_SIZE;
            size_t const blockStateSpace = 2 * sizeof(ZSTD_compressedBlockState_t);
            size_t const bufferSpace = buffInSize + buffOutSize;
            size_t const ldmSpace = ZSTD_ldm_getTableSize(params.ldmParams);
            size_t const ldmSeqSpace = maxNbLdmSeq * sizeof(rawSeq);

            size_t const neededSpace = entropySpace + blockStateSpace + ldmSpace +
                                       ldmSeqSpace + matchStateSize + tokenSpace +
                                       bufferSpace;

            int const workSpaceTooSmall = zc->workSpaceSize < neededSpace;
            int const workSpaceTooLarge = zc->workSpaceSize > ZSTD_WORKSPACETOOLARGE_FACTOR * neededSpace;
            int const workSpaceWasteful = workSpaceTooLarge && (zc->workSpaceOversizedDuration > ZSTD_WORKSPACETOOLARGE_MAXDURATION);
            zc->workSpaceOversizedDuration = workSpaceTooLarge ? zc->workSpaceOversizedDuration+1 : 0;

            DEBUGLOG(4, "Need %zuKB workspace, including %zuKB for match state, and %zuKB for buffers",
                        neededSpace>>10, matchStateSize>>10, bufferSpace>>10);
            DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize);

            if (workSpaceTooSmall || workSpaceWasteful) {
                DEBUGLOG(4, "Resize workSpaceSize from %zuKB to %zuKB",
                            zc->workSpaceSize >> 10,
                            neededSpace >> 10);

                RETURN_ERROR_IF(zc->staticSize, memory_allocation, "static cctx : no resize");

                zc->workSpaceSize = 0;
                ZSTD_free(zc->workSpace, zc->customMem);
                zc->workSpace = ZSTD_malloc(neededSpace, zc->customMem);
                RETURN_ERROR_IF(zc->workSpace == NULL, memory_allocation);
                zc->workSpaceSize = neededSpace;
                zc->workSpaceOversizedDuration = 0;

                /* Statically sized space.
                 * entropyWorkspace never moves,
                 * though prev/next block swap places */
                assert(((size_t)zc->workSpace & 3) == 0);   /* ensure correct alignment */
                assert(zc->workSpaceSize >= 2 * sizeof(ZSTD_compressedBlockState_t));
                zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)zc->workSpace;
                zc->blockState.nextCBlock = zc->blockState.prevCBlock + 1;
                ptr = zc->blockState.nextCBlock + 1;
                zc->entropyWorkspace = (U32*)ptr;
        }   }

        /* init params */
        zc->appliedParams = params;
        zc->blockState.matchState.cParams = params.cParams;
        zc->pledgedSrcSizePlusOne = pledgedSrcSize+1;
        zc->consumedSrcSize = 0;
        zc->producedCSize = 0;
        if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
            zc->appliedParams.fParams.contentSizeFlag = 0;
        DEBUGLOG(4, "pledged content size : %u ; flag : %u",
            (unsigned)pledgedSrcSize, zc->appliedParams.fParams.contentSizeFlag);
        zc->blockSize = blockSize;

        XXH64_reset(&zc->xxhState, 0);
        zc->stage = ZSTDcs_init;
        zc->dictID = 0;

        ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock);

        ptr = ZSTD_reset_matchState(&zc->blockState.matchState,
                                     zc->entropyWorkspace + HUF_WORKSPACE_SIZE_U32,
                                    &params.cParams,
                                     crp, ZSTD_resetTarget_CCtx);

        /* ldm hash table */
        /* initialize bucketOffsets table later for pointer alignment */
        if (params.ldmParams.enableLdm) {
            size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog;
            memset(ptr, 0, ldmHSize * sizeof(ldmEntry_t));
            assert(((size_t)ptr & 3) == 0); /* ensure ptr is properly aligned */
            zc->ldmState.hashTable = (ldmEntry_t*)ptr;
            ptr = zc->ldmState.hashTable + ldmHSize;
            zc->ldmSequences = (rawSeq*)ptr;
            ptr = zc->ldmSequences + maxNbLdmSeq;
            zc->maxNbLdmSequences = maxNbLdmSeq;

            memset(&zc->ldmState.window, 0, sizeof(zc->ldmState.window));
        }
        assert(((size_t)ptr & 3) == 0); /* ensure ptr is properly aligned */

        /* sequences storage */
        zc->seqStore.maxNbSeq = maxNbSeq;
        zc->seqStore.sequencesStart = (seqDef*)ptr;
        ptr = zc->seqStore.sequencesStart + maxNbSeq;
        zc->seqStore.llCode = (BYTE*) ptr;
        zc->seqStore.mlCode = zc->seqStore.llCode + maxNbSeq;
        zc->seqStore.ofCode = zc->seqStore.mlCode + maxNbSeq;
        zc->seqStore.litStart = zc->seqStore.ofCode + maxNbSeq;
        /* ZSTD_wildcopy() is used to copy into the literals buffer,
         * so we have to oversize the buffer by WILDCOPY_OVERLENGTH bytes.
         */
        zc->seqStore.maxNbLit = blockSize;
        ptr = zc->seqStore.litStart + blockSize + WILDCOPY_OVERLENGTH;

        /* ldm bucketOffsets table */
        if (params.ldmParams.enableLdm) {
            size_t const ldmBucketSize =
                  ((size_t)1) << (params.ldmParams.hashLog -
                                  params.ldmParams.bucketSizeLog);
            memset(ptr, 0, ldmBucketSize);
            zc->ldmState.bucketOffsets = (BYTE*)ptr;
            ptr = zc->ldmState.bucketOffsets + ldmBucketSize;
            ZSTD_window_clear(&zc->ldmState.window);
        }
        ZSTD_referenceExternalSequences(zc, NULL, 0);

        /* buffers */
        zc->inBuffSize = buffInSize;
        zc->inBuff = (char*)ptr;
        zc->outBuffSize = buffOutSize;
        zc->outBuff = zc->inBuff + buffInSize;

        return 0;
    }
}

/* ZSTD_invalidateRepCodes() :
 * ensures next compression will not use repcodes from previous block.
 * Note : only works with regular variant;
 *        do not use with extDict variant ! */
void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx) {
    int i;
    for (i=0; i<ZSTD_REP_NUM; i++) cctx->blockState.prevCBlock->rep[i] = 0;
    assert(!ZSTD_window_hasExtDict(cctx->blockState.matchState.window));
}

/* These are the approximate sizes for each strategy past which copying the
 * dictionary tables into the working context is faster than using them
 * in-place.
 */
static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = {
    8 KB,  /* unused */
    8 KB,  /* ZSTD_fast */
    16 KB, /* ZSTD_dfast */
    32 KB, /* ZSTD_greedy */
    32 KB, /* ZSTD_lazy */
    32 KB, /* ZSTD_lazy2 */
    32 KB, /* ZSTD_btlazy2 */
    32 KB, /* ZSTD_btopt */
    8 KB,  /* ZSTD_btultra */
    8 KB   /* ZSTD_btultra2 */
};

static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict,
                                 ZSTD_CCtx_params params,
                                 U64 pledgedSrcSize)
{
    size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy];
    return ( pledgedSrcSize <= cutoff
          || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
          || params.attachDictPref == ZSTD_dictForceAttach )
        && params.attachDictPref != ZSTD_dictForceCopy
        && !params.forceWindow; /* dictMatchState isn't correctly
                                 * handled in _enforceMaxDist */
}

static size_t
ZSTD_resetCCtx_byAttachingCDict(ZSTD_CCtx* cctx,
                        const ZSTD_CDict* cdict,
                        ZSTD_CCtx_params params,
                        U64 pledgedSrcSize,
                        ZSTD_buffered_policy_e zbuff)
{
    {   const ZSTD_compressionParameters* const cdict_cParams = &cdict->matchState.cParams;
        unsigned const windowLog = params.cParams.windowLog;
        assert(windowLog != 0);
        /* Resize working context table params for input only, since the dict
         * has its own tables. */
        params.cParams = ZSTD_adjustCParams_internal(*cdict_cParams, pledgedSrcSize, 0);
        params.cParams.windowLog = windowLog;
        ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
                                ZSTDcrp_continue, zbuff);
        assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
    }

    {   const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
                                  - cdict->matchState.window.base);
        const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit;
        if (cdictLen == 0) {
            /* don't even attach dictionaries with no contents */
            DEBUGLOG(4, "skipping attaching empty dictionary");
        } else {
            DEBUGLOG(4, "attaching dictionary into context");
            cctx->blockState.matchState.dictMatchState = &cdict->matchState;

            /* prep working match state so dict matches never have negative indices
             * when they are translated to the working context's index space. */
            if (cctx->blockState.matchState.window.dictLimit < cdictEnd) {
                cctx->blockState.matchState.window.nextSrc =
                    cctx->blockState.matchState.window.base + cdictEnd;
                ZSTD_window_clear(&cctx->blockState.matchState.window);
            }
            /* loadedDictEnd is expressed within the referential of the active context */
            cctx->blockState.matchState.loadedDictEnd = cctx->blockState.matchState.window.dictLimit;
    }   }

    cctx->dictID = cdict->dictID;

    /* copy block state */
    memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));

    return 0;
}

static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
                            const ZSTD_CDict* cdict,
                            ZSTD_CCtx_params params,
                            U64 pledgedSrcSize,
                            ZSTD_buffered_policy_e zbuff)
{
    const ZSTD_compressionParameters *cdict_cParams = &cdict->matchState.cParams;

    DEBUGLOG(4, "copying dictionary into context");

    {   unsigned const windowLog = params.cParams.windowLog;
        assert(windowLog != 0);
        /* Copy only compression parameters related to tables. */
        params.cParams = *cdict_cParams;
        params.cParams.windowLog = windowLog;
        ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
                                ZSTDcrp_noMemset, zbuff);
        assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
        assert(cctx->appliedParams.cParams.hashLog == cdict_cParams->hashLog);
        assert(cctx->appliedParams.cParams.chainLog == cdict_cParams->chainLog);
    }

    /* copy tables */
    {   size_t const chainSize = (cdict_cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cdict_cParams->chainLog);
        size_t const hSize =  (size_t)1 << cdict_cParams->hashLog;
        size_t const tableSpace = (chainSize + hSize) * sizeof(U32);
        assert((U32*)cctx->blockState.matchState.chainTable == (U32*)cctx->blockState.matchState.hashTable + hSize);  /* chainTable must follow hashTable */
        assert((U32*)cctx->blockState.matchState.hashTable3 == (U32*)cctx->blockState.matchState.chainTable + chainSize);
        assert((U32*)cdict->matchState.chainTable == (U32*)cdict->matchState.hashTable + hSize);  /* chainTable must follow hashTable */
        assert((U32*)cdict->matchState.hashTable3 == (U32*)cdict->matchState.chainTable + chainSize);
        memcpy(cctx->blockState.matchState.hashTable, cdict->matchState.hashTable, tableSpace);   /* presumes all tables follow each other */
    }

    /* Zero the hashTable3, since the cdict never fills it */
    {   size_t const h3Size = (size_t)1 << cctx->blockState.matchState.hashLog3;
        assert(cdict->matchState.hashLog3 == 0);
        memset(cctx->blockState.matchState.hashTable3, 0, h3Size * sizeof(U32));
    }

    /* copy dictionary offsets */
    {   ZSTD_matchState_t const* srcMatchState = &cdict->matchState;
        ZSTD_matchState_t* dstMatchState = &cctx->blockState.matchState;
        dstMatchState->window       = srcMatchState->window;
        dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
        dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
    }

    cctx->dictID = cdict->dictID;

    /* copy block state */
    memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));

    return 0;
}

/* We have a choice between copying the dictionary context into the working
 * context, or referencing the dictionary context from the working context
 * in-place. We decide here which strategy to use. */
static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
                            const ZSTD_CDict* cdict,
                            ZSTD_CCtx_params params,
                            U64 pledgedSrcSize,
                            ZSTD_buffered_policy_e zbuff)
{

    DEBUGLOG(4, "ZSTD_resetCCtx_usingCDict (pledgedSrcSize=%u)",
                (unsigned)pledgedSrcSize);

    if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) {
        return ZSTD_resetCCtx_byAttachingCDict(
            cctx, cdict, params, pledgedSrcSize, zbuff);
    } else {
        return ZSTD_resetCCtx_byCopyingCDict(
            cctx, cdict, params, pledgedSrcSize, zbuff);
    }
}

/*! ZSTD_copyCCtx_internal() :
 *  Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
 *  Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
 *  The "context", in this case, refers to the hash and chain tables,
 *  entropy tables, and dictionary references.
 * `windowLog` value is enforced if != 0, otherwise value is copied from srcCCtx.
 * @return : 0, or an error code */
static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
                            const ZSTD_CCtx* srcCCtx,
                            ZSTD_frameParameters fParams,
                            U64 pledgedSrcSize,
                            ZSTD_buffered_policy_e zbuff)
{
    DEBUGLOG(5, "ZSTD_copyCCtx_internal");
    RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong);

    memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
    {   ZSTD_CCtx_params params = dstCCtx->requestedParams;
        /* Copy only compression parameters related to tables. */
        params.cParams = srcCCtx->appliedParams.cParams;
        params.fParams = fParams;
        ZSTD_resetCCtx_internal(dstCCtx, params, pledgedSrcSize,
                                ZSTDcrp_noMemset, zbuff);
        assert(dstCCtx->appliedParams.cParams.windowLog == srcCCtx->appliedParams.cParams.windowLog);
        assert(dstCCtx->appliedParams.cParams.strategy == srcCCtx->appliedParams.cParams.strategy);
        assert(dstCCtx->appliedParams.cParams.hashLog == srcCCtx->appliedParams.cParams.hashLog);
        assert(dstCCtx->appliedParams.cParams.chainLog == srcCCtx->appliedParams.cParams.chainLog);
        assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3);
    }

    /* copy tables */
    {   size_t const chainSize = (srcCCtx->appliedParams.cParams.strategy == ZSTD_fast) ? 0 : ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog);
        size_t const hSize =  (size_t)1 << srcCCtx->appliedParams.cParams.hashLog;
        size_t const h3Size = (size_t)1 << srcCCtx->blockState.matchState.hashLog3;
        size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
        assert((U32*)dstCCtx->blockState.matchState.chainTable == (U32*)dstCCtx->blockState.matchState.hashTable + hSize);  /* chainTable must follow hashTable */
        assert((U32*)dstCCtx->blockState.matchState.hashTable3 == (U32*)dstCCtx->blockState.matchState.chainTable + chainSize);
        memcpy(dstCCtx->blockState.matchState.hashTable, srcCCtx->blockState.matchState.hashTable, tableSpace);   /* presumes all tables follow each other */
    }

    /* copy dictionary offsets */
    {
        const ZSTD_matchState_t* srcMatchState = &srcCCtx->blockState.matchState;
        ZSTD_matchState_t* dstMatchState = &dstCCtx->blockState.matchState;
        dstMatchState->window       = srcMatchState->window;
        dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
        dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
    }
    dstCCtx->dictID = srcCCtx->dictID;

    /* copy block state */
    memcpy(dstCCtx->blockState.prevCBlock, srcCCtx->blockState.prevCBlock, sizeof(*srcCCtx->blockState.prevCBlock));

    return 0;
}

/*! ZSTD_copyCCtx() :
 *  Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
 *  Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
 *  pledgedSrcSize==0 means "unknown".
*   @return : 0, or an error code */
size_t ZSTD_copyCCtx(ZSTD_CCtx* dstCCtx, const ZSTD_CCtx* srcCCtx, unsigned long long pledgedSrcSize)
{
    ZSTD_frameParameters fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ };
    ZSTD_buffered_policy_e const zbuff = (ZSTD_buffered_policy_e)(srcCCtx->inBuffSize>0);
    ZSTD_STATIC_ASSERT((U32)ZSTDb_buffered==1);
    if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;
    fParams.contentSizeFlag = (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN);

    return ZSTD_copyCCtx_internal(dstCCtx, srcCCtx,
                                fParams, pledgedSrcSize,
                                zbuff);
}


#define ZSTD_ROWSIZE 16
/*! ZSTD_reduceTable() :
 *  reduce table indexes by `reducerValue`, or squash to zero.
 *  PreserveMark preserves "unsorted mark" for btlazy2 strategy.
 *  It must be set to a clear 0/1 value, to remove branch during inlining.
 *  Presume table size is a multiple of ZSTD_ROWSIZE
 *  to help auto-vectorization */
FORCE_INLINE_TEMPLATE void
ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerValue, int const preserveMark)
{
    int const nbRows = (int)size / ZSTD_ROWSIZE;
    int cellNb = 0;
    int rowNb;
    assert((size & (ZSTD_ROWSIZE-1)) == 0);  /* multiple of ZSTD_ROWSIZE */
    assert(size < (1U<<31));   /* can be casted to int */
    for (rowNb=0 ; rowNb < nbRows ; rowNb++) {
        int column;
        for (column=0; column<ZSTD_ROWSIZE; column++) {
            if (preserveMark) {
                U32 const adder = (table[cellNb] == ZSTD_DUBT_UNSORTED_MARK) ? reducerValue : 0;
                table[cellNb] += adder;
            }
            if (table[cellNb] < reducerValue) table[cellNb] = 0;
            else table[cellNb] -= reducerValue;
            cellNb++;
    }   }
}

static void ZSTD_reduceTable(U32* const table, U32 const size, U32 const reducerValue)
{
    ZSTD_reduceTable_internal(table, size, reducerValue, 0);
}

static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const reducerValue)
{
    ZSTD_reduceTable_internal(table, size, reducerValue, 1);
}

/*! ZSTD_reduceIndex() :
*   rescale all indexes to avoid future overflow (indexes are U32) */
static void ZSTD_reduceIndex (ZSTD_matchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue)
{
    {   U32 const hSize = (U32)1 << params->cParams.hashLog;
        ZSTD_reduceTable(ms->hashTable, hSize, reducerValue);
    }

    if (params->cParams.strategy != ZSTD_fast) {
        U32 const chainSize = (U32)1 << params->cParams.chainLog;
        if (params->cParams.strategy == ZSTD_btlazy2)
            ZSTD_reduceTable_btlazy2(ms->chainTable, chainSize, reducerValue);
        else
            ZSTD_reduceTable(ms->chainTable, chainSize, reducerValue);
    }

    if (ms->hashLog3) {
        U32 const h3Size = (U32)1 << ms->hashLog3;
        ZSTD_reduceTable(ms->hashTable3, h3Size, reducerValue);
    }
}


/*-*******************************************************
*  Block entropic compression
*********************************************************/

/* See doc/zstd_compression_format.md for detailed format description */

static size_t ZSTD_noCompressBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize, U32 lastBlock)
{
    U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw)<<1) + (U32)(srcSize << 3);
    RETURN_ERROR_IF(srcSize + ZSTD_blockHeaderSize > dstCapacity,
                    dstSize_tooSmall);
    MEM_writeLE24(dst, cBlockHeader24);
    memcpy((BYTE*)dst + ZSTD_blockHeaderSize, src, srcSize);
    return ZSTD_blockHeaderSize + srcSize;
}

static size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
{
    BYTE* const ostart = (BYTE* const)dst;
    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);

    RETURN_ERROR_IF(srcSize + flSize > dstCapacity, dstSize_tooSmall);

    switch(flSize)
    {
        case 1: /* 2 - 1 - 5 */
            ostart[0] = (BYTE)((U32)set_basic + (srcSize<<3));
            break;
        case 2: /* 2 - 2 - 12 */
            MEM_writeLE16(ostart, (U16)((U32)set_basic + (1<<2) + (srcSize<<4)));
            break;
        case 3: /* 2 - 2 - 20 */
            MEM_writeLE32(ostart, (U32)((U32)set_basic + (3<<2) + (srcSize<<4)));
            break;
        default:   /* not necessary : flSize is {1,2,3} */
            assert(0);
    }

    memcpy(ostart + flSize, src, srcSize);
    return srcSize + flSize;
}

static size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
{
    BYTE* const ostart = (BYTE* const)dst;
    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);

    (void)dstCapacity;  /* dstCapacity already guaranteed to be >=4, hence large enough */

    switch(flSize)
    {
        case 1: /* 2 - 1 - 5 */
            ostart[0] = (BYTE)((U32)set_rle + (srcSize<<3));
            break;
        case 2: /* 2 - 2 - 12 */
            MEM_writeLE16(ostart, (U16)((U32)set_rle + (1<<2) + (srcSize<<4)));
            break;
        case 3: /* 2 - 2 - 20 */
            MEM_writeLE32(ostart, (U32)((U32)set_rle + (3<<2) + (srcSize<<4)));
            break;
        default:   /* not necessary : flSize is {1,2,3} */
            assert(0);
    }

    ostart[flSize] = *(const BYTE*)src;
    return flSize+1;
}


/* ZSTD_minGain() :
 * minimum compression required
 * to generate a compress block or a compressed literals section.
 * note : use same formula for both situations */
static size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat)
{
    U32 const minlog = (strat>=ZSTD_btultra) ? (U32)(strat) - 1 : 6;
    ZSTD_STATIC_ASSERT(ZSTD_btultra == 8);
    assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
    return (srcSize >> minlog) + 2;
}

static size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
                                     ZSTD_hufCTables_t* nextHuf,
                                     ZSTD_strategy strategy, int disableLiteralCompression,
                                     void* dst, size_t dstCapacity,
                               const void* src, size_t srcSize,
                                     void* workspace, size_t wkspSize,
                               const int bmi2)
{
    size_t const minGain = ZSTD_minGain(srcSize, strategy);
    size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
    BYTE*  const ostart = (BYTE*)dst;
    U32 singleStream = srcSize < 256;
    symbolEncodingType_e hType = set_compressed;
    size_t cLitSize;

    DEBUGLOG(5,"ZSTD_compressLiterals (disableLiteralCompression=%i)",
                disableLiteralCompression);

    /* Prepare nextEntropy assuming reusing the existing table */
    memcpy(nextHuf, prevHuf, sizeof(*prevHuf));

    if (disableLiteralCompression)
        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);

    /* small ? don't even attempt compression (speed opt) */
#   define COMPRESS_LITERALS_SIZE_MIN 63
    {   size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
        if (srcSize <= minLitSize) return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
    }

    RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
    {   HUF_repeat repeat = prevHuf->repeatMode;
        int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
        if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
        cLitSize = singleStream ? HUF_compress1X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
                                      workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2)
                                : HUF_compress4X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
                                      workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
        if (repeat != HUF_repeat_none) {
            /* reused the existing table */
            hType = set_repeat;
        }
    }

    if ((cLitSize==0) | (cLitSize >= srcSize - minGain) | ERR_isError(cLitSize)) {
        memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
    }
    if (cLitSize==1) {
        memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
        return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
    }

    if (hType == set_compressed) {
        /* using a newly constructed table */
        nextHuf->repeatMode = HUF_repeat_check;
    }

    /* Build header */
    switch(lhSize)
    {
    case 3: /* 2 - 2 - 10 - 10 */
        {   U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<14);
            MEM_writeLE24(ostart, lhc);
            break;
        }
    case 4: /* 2 - 2 - 14 - 14 */
        {   U32 const lhc = hType + (2 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<18);
            MEM_writeLE32(ostart, lhc);
            break;
        }
    case 5: /* 2 - 2 - 18 - 18 */
        {   U32 const lhc = hType + (3 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<22);
            MEM_writeLE32(ostart, lhc);
            ostart[4] = (BYTE)(cLitSize >> 10);
            break;
        }
    default:  /* not possible : lhSize is {3,4,5} */
        assert(0);
    }
    return lhSize+cLitSize;
}


void ZSTD_seqToCodes(const seqStore_t* seqStorePtr)
{
    const seqDef* const sequences = seqStorePtr->sequencesStart;
    BYTE* const llCodeTable = seqStorePtr->llCode;
    BYTE* const ofCodeTable = seqStorePtr->ofCode;
    BYTE* const mlCodeTable = seqStorePtr->mlCode;
    U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
    U32 u;
    assert(nbSeq <= seqStorePtr->maxNbSeq);
    for (u=0; u<nbSeq; u++) {
        U32 const llv = sequences[u].litLength;
        U32 const mlv = sequences[u].matchLength;
        llCodeTable[u] = (BYTE)ZSTD_LLcode(llv);
        ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset);
        mlCodeTable[u] = (BYTE)ZSTD_MLcode(mlv);
    }
    if (seqStorePtr->longLengthID==1)
        llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
    if (seqStorePtr->longLengthID==2)
        mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
}


/**
 * -log2(x / 256) lookup table for x in [0, 256).
 * If x == 0: Return 0
 * Else: Return floor(-log2(x / 256) * 256)
 */
static unsigned const kInverseProbabilityLog256[256] = {
    0,    2048, 1792, 1642, 1536, 1453, 1386, 1329, 1280, 1236, 1197, 1162,
    1130, 1100, 1073, 1047, 1024, 1001, 980,  960,  941,  923,  906,  889,
    874,  859,  844,  830,  817,  804,  791,  779,  768,  756,  745,  734,
    724,  714,  704,  694,  685,  676,  667,  658,  650,  642,  633,  626,
    618,  610,  603,  595,  588,  581,  574,  567,  561,  554,  548,  542,
    535,  529,  523,  517,  512,  506,  500,  495,  489,  484,  478,  473,
    468,  463,  458,  453,  448,  443,  438,  434,  429,  424,  420,  415,
    411,  407,  402,  398,  394,  390,  386,  382,  377,  373,  370,  366,
    362,  358,  354,  350,  347,  343,  339,  336,  332,  329,  325,  322,
    318,  315,  311,  308,  305,  302,  298,  295,  292,  289,  286,  282,
    279,  276,  273,  270,  267,  264,  261,  258,  256,  253,  250,  247,
    244,  241,  239,  236,  233,  230,  228,  225,  222,  220,  217,  215,
    212,  209,  207,  204,  202,  199,  197,  194,  192,  190,  187,  185,
    182,  180,  178,  175,  173,  171,  168,  166,  164,  162,  159,  157,
    155,  153,  151,  149,  146,  144,  142,  140,  138,  136,  134,  132,
    130,  128,  126,  123,  121,  119,  117,  115,  114,  112,  110,  108,
    106,  104,  102,  100,  98,   96,   94,   93,   91,   89,   87,   85,
    83,   82,   80,   78,   76,   74,   73,   71,   69,   67,   66,   64,
    62,   61,   59,   57,   55,   54,   52,   50,   49,   47,   46,   44,
    42,   41,   39,   37,   36,   34,   33,   31,   30,   28,   26,   25,
    23,   22,   20,   19,   17,   16,   14,   13,   11,   10,   8,    7,
    5,    4,    2,    1,
};


/**
 * Returns the cost in bits of encoding the distribution described by count
 * using the entropy bound.
 */
static size_t ZSTD_entropyCost(unsigned const* count, unsigned const max, size_t const total)
{
    unsigned cost = 0;
    unsigned s;
    for (s = 0; s <= max; ++s) {
        unsigned norm = (unsigned)((256 * count[s]) / total);
        if (count[s] != 0 && norm == 0)
            norm = 1;
        assert(count[s] < total);
        cost += count[s] * kInverseProbabilityLog256[norm];
    }
    return cost >> 8;
}


/**
 * Returns the cost in bits of encoding the distribution in count using the
 * table described by norm. The max symbol support by norm is assumed >= max.
 * norm must be valid for every symbol with non-zero probability in count.
 */
static size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
                                    unsigned const* count, unsigned const max)
{
    unsigned const shift = 8 - accuracyLog;
    size_t cost = 0;
    unsigned s;
    assert(accuracyLog <= 8);
    for (s = 0; s <= max; ++s) {
        unsigned const normAcc = norm[s] != -1 ? norm[s] : 1;
        unsigned const norm256 = normAcc << shift;
        assert(norm256 > 0);
        assert(norm256 < 256);
        cost += count[s] * kInverseProbabilityLog256[norm256];
    }
    return cost >> 8;
}


static unsigned ZSTD_getFSEMaxSymbolValue(FSE_CTable const* ctable) {
  void const* ptr = ctable;
  U16 const* u16ptr = (U16 const*)ptr;
  U32 const maxSymbolValue = MEM_read16(u16ptr + 1);
  return maxSymbolValue;
}


/**
 * Returns the cost in bits of encoding the distribution in count using ctable.
 * Returns an error if ctable cannot represent all the symbols in count.
 */
static size_t ZSTD_fseBitCost(
    FSE_CTable const* ctable,
    unsigned const* count,
    unsigned const max)
{
    unsigned const kAccuracyLog = 8;
    size_t cost = 0;
    unsigned s;
    FSE_CState_t cstate;
    FSE_initCState(&cstate, ctable);
    RETURN_ERROR_IF(ZSTD_getFSEMaxSymbolValue(ctable) < max, GENERIC,
                    "Repeat FSE_CTable has maxSymbolValue %u < %u",
                    ZSTD_getFSEMaxSymbolValue(ctable), max);
    for (s = 0; s <= max; ++s) {
        unsigned const tableLog = cstate.stateLog;
        unsigned const badCost = (tableLog + 1) << kAccuracyLog;
        unsigned const bitCost = FSE_bitCost(cstate.symbolTT, tableLog, s, kAccuracyLog);
        if (count[s] == 0)
            continue;
        RETURN_ERROR_IF(bitCost >= badCost, GENERIC,
                        "Repeat FSE_CTable has Prob[%u] == 0", s);
        cost += count[s] * bitCost;
    }
    return cost >> kAccuracyLog;
}

/**
 * Returns the cost in bytes of encoding the normalized count header.
 * Returns an error if any of the helper functions return an error.
 */
static size_t ZSTD_NCountCost(unsigned const* count, unsigned const max,
                              size_t const nbSeq, unsigned const FSELog)
{
    BYTE wksp[FSE_NCOUNTBOUND];
    S16 norm[MaxSeq + 1];
    const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
    FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq, max));
    return FSE_writeNCount(wksp, sizeof(wksp), norm, max, tableLog);
}


typedef enum {
    ZSTD_defaultDisallowed = 0,
    ZSTD_defaultAllowed = 1
} ZSTD_defaultPolicy_e;

MEM_STATIC symbolEncodingType_e
ZSTD_selectEncodingType(
        FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
        size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
        FSE_CTable const* prevCTable,
        short const* defaultNorm, U32 defaultNormLog,
        ZSTD_defaultPolicy_e const isDefaultAllowed,
        ZSTD_strategy const strategy)
{
    ZSTD_STATIC_ASSERT(ZSTD_defaultDisallowed == 0 && ZSTD_defaultAllowed != 0);
    if (mostFrequent == nbSeq) {
        *repeatMode = FSE_repeat_none;
        if (isDefaultAllowed && nbSeq <= 2) {
            /* Prefer set_basic over set_rle when there are 2 or less symbols,
             * since RLE uses 1 byte, but set_basic uses 5-6 bits per symbol.
             * If basic encoding isn't possible, always choose RLE.
             */
            DEBUGLOG(5, "Selected set_basic");
            return set_basic;
        }
        DEBUGLOG(5, "Selected set_rle");
        return set_rle;
    }
    if (strategy < ZSTD_lazy) {
        if (isDefaultAllowed) {
            size_t const staticFse_nbSeq_max = 1000;
            size_t const mult = 10 - strategy;
            size_t const baseLog = 3;
            size_t const dynamicFse_nbSeq_min = (((size_t)1 << defaultNormLog) * mult) >> baseLog;  /* 28-36 for offset, 56-72 for lengths */
            assert(defaultNormLog >= 5 && defaultNormLog <= 6);  /* xx_DEFAULTNORMLOG */
            assert(mult <= 9 && mult >= 7);
            if ( (*repeatMode == FSE_repeat_valid)
              && (nbSeq < staticFse_nbSeq_max) ) {
                DEBUGLOG(5, "Selected set_repeat");
                return set_repeat;
            }
            if ( (nbSeq < dynamicFse_nbSeq_min)
              || (mostFrequent < (nbSeq >> (defaultNormLog-1))) ) {
                DEBUGLOG(5, "Selected set_basic");
                /* The format allows default tables to be repeated, but it isn't useful.
                 * When using simple heuristics to select encoding type, we don't want
                 * to confuse these tables with dictionaries. When running more careful
                 * analysis, we don't need to waste time checking both repeating tables
                 * and default tables.
                 */
                *repeatMode = FSE_repeat_none;
                return set_basic;
            }
        }
    } else {
        size_t const basicCost = isDefaultAllowed ? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, count, max) : ERROR(GENERIC);
        size_t const repeatCost = *repeatMode != FSE_repeat_none ? ZSTD_fseBitCost(prevCTable, count, max) : ERROR(GENERIC);
        size_t const NCountCost = ZSTD_NCountCost(count, max, nbSeq, FSELog);
        size_t const compressedCost = (NCountCost << 3) + ZSTD_entropyCost(count, max, nbSeq);

        if (isDefaultAllowed) {
            assert(!ZSTD_isError(basicCost));
            assert(!(*repeatMode == FSE_repeat_valid && ZSTD_isError(repeatCost)));
        }
        assert(!ZSTD_isError(NCountCost));
        assert(compressedCost < ERROR(maxCode));
        DEBUGLOG(5, "Estimated bit costs: basic=%u\trepeat=%u\tcompressed=%u",
                    (unsigned)basicCost, (unsigned)repeatCost, (unsigned)compressedCost);
        if (basicCost <= repeatCost && basicCost <= compressedCost) {
            DEBUGLOG(5, "Selected set_basic");
            assert(isDefaultAllowed);
            *repeatMode = FSE_repeat_none;
            return set_basic;
        }
        if (repeatCost <= compressedCost) {
            DEBUGLOG(5, "Selected set_repeat");
            assert(!ZSTD_isError(repeatCost));
            return set_repeat;
        }
        assert(compressedCost < basicCost && compressedCost < repeatCost);
    }
    DEBUGLOG(5, "Selected set_compressed");
    *repeatMode = FSE_repeat_check;
    return set_compressed;
}

MEM_STATIC size_t
ZSTD_buildCTable(void* dst, size_t dstCapacity,
                FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
                unsigned* count, U32 max,
                const BYTE* codeTable, size_t nbSeq,
                const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
                const FSE_CTable* prevCTable, size_t prevCTableSize,
                void* workspace, size_t workspaceSize)
{
    BYTE* op = (BYTE*)dst;
    const BYTE* const oend = op + dstCapacity;
    DEBUGLOG(6, "ZSTD_buildCTable (dstCapacity=%u)", (unsigned)dstCapacity);

    switch (type) {
    case set_rle:
        FORWARD_IF_ERROR(FSE_buildCTable_rle(nextCTable, (BYTE)max));
        RETURN_ERROR_IF(dstCapacity==0, dstSize_tooSmall);
        *op = codeTable[0];
        return 1;
    case set_repeat:
        memcpy(nextCTable, prevCTable, prevCTableSize);
        return 0;
    case set_basic:
        FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, workspace, workspaceSize));  /* note : could be pre-calculated */
        return 0;
    case set_compressed: {
        S16 norm[MaxSeq + 1];
        size_t nbSeq_1 = nbSeq;
        const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
        if (count[codeTable[nbSeq-1]] > 1) {
            count[codeTable[nbSeq-1]]--;
            nbSeq_1--;
        }
        assert(nbSeq_1 > 1);
        FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max));
        {   size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog);   /* overflow protected */
            FORWARD_IF_ERROR(NCountSize);
            FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, norm, max, tableLog, workspace, workspaceSize));
            return NCountSize;
        }
    }
    default: assert(0); RETURN_ERROR(GENERIC);
    }
}

FORCE_INLINE_TEMPLATE size_t
ZSTD_encodeSequences_body(
            void* dst, size_t dstCapacity,
            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
            seqDef const* sequences, size_t nbSeq, int longOffsets)
{
    BIT_CStream_t blockStream;
    FSE_CState_t  stateMatchLength;
    FSE_CState_t  stateOffsetBits;
    FSE_CState_t  stateLitLength;

    RETURN_ERROR_IF(
        ERR_isError(BIT_initCStream(&blockStream, dst, dstCapacity)),
        dstSize_tooSmall, "not enough space remaining");
    DEBUGLOG(6, "available space for bitstream : %i  (dstCapacity=%u)",
                (int)(blockStream.endPtr - blockStream.startPtr),
                (unsigned)dstCapacity);

    /* first symbols */
    FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq-1]);
    FSE_initCState2(&stateOffsetBits,  CTable_OffsetBits,  ofCodeTable[nbSeq-1]);
    FSE_initCState2(&stateLitLength,   CTable_LitLength,   llCodeTable[nbSeq-1]);
    BIT_addBits(&blockStream, sequences[nbSeq-1].litLength, LL_bits[llCodeTable[nbSeq-1]]);
    if (MEM_32bits()) BIT_flushBits(&blockStream);
    BIT_addBits(&blockStream, sequences[nbSeq-1].matchLength, ML_bits[mlCodeTable[nbSeq-1]]);
    if (MEM_32bits()) BIT_flushBits(&blockStream);
    if (longOffsets) {
        U32 const ofBits = ofCodeTable[nbSeq-1];
        int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
        if (extraBits) {
            BIT_addBits(&blockStream, sequences[nbSeq-1].offset, extraBits);
            BIT_flushBits(&blockStream);
        }
        BIT_addBits(&blockStream, sequences[nbSeq-1].offset >> extraBits,
                    ofBits - extraBits);
    } else {
        BIT_addBits(&blockStream, sequences[nbSeq-1].offset, ofCodeTable[nbSeq-1]);
    }
    BIT_flushBits(&blockStream);

    {   size_t n;
        for (n=nbSeq-2 ; n<nbSeq ; n--) {      /* intentional underflow */
            BYTE const llCode = llCodeTable[n];
            BYTE const ofCode = ofCodeTable[n];
            BYTE const mlCode = mlCodeTable[n];
            U32  const llBits = LL_bits[llCode];
            U32  const ofBits = ofCode;
            U32  const mlBits = ML_bits[mlCode];
            DEBUGLOG(6, "encoding: litlen:%2u - matchlen:%2u - offCode:%7u",
                        (unsigned)sequences[n].litLength,
                        (unsigned)sequences[n].matchLength + MINMATCH,
                        (unsigned)sequences[n].offset);
                                                                            /* 32b*/  /* 64b*/
                                                                            /* (7)*/  /* (7)*/
            FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode);       /* 15 */  /* 15 */
            FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode);      /* 24 */  /* 24 */
            if (MEM_32bits()) BIT_flushBits(&blockStream);                  /* (7)*/
            FSE_encodeSymbol(&blockStream, &stateLitLength, llCode);        /* 16 */  /* 33 */
            if (MEM_32bits() || (ofBits+mlBits+llBits >= 64-7-(LLFSELog+MLFSELog+OffFSELog)))
                BIT_flushBits(&blockStream);                                /* (7)*/
            BIT_addBits(&blockStream, sequences[n].litLength, llBits);
            if (MEM_32bits() && ((llBits+mlBits)>24)) BIT_flushBits(&blockStream);
            BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
            if (MEM_32bits() || (ofBits+mlBits+llBits > 56)) BIT_flushBits(&blockStream);
            if (longOffsets) {
                int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
                if (extraBits) {
                    BIT_addBits(&blockStream, sequences[n].offset, extraBits);