zstd-ruby 1.3.4.0 → 1.3.5.0

data/ext/zstdruby/libzstd/compress/zstd_compress_internal.h

@@ -27,6 +27,7 @@
 extern "C" {
 #endif
 
+
 /*-*************************************
 *  Constants
 ***************************************/
@@ -37,7 +38,8 @@ extern "C" {
                                        It's not a big deal though : candidate will just be sorted again.
                                        Additionnally, candidate position 1 will be lost.
                                        But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss.
-                                       The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be misdhandled after table re-use with a different strategy */
+                                       The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be misdhandled after table re-use with a different strategy
+                                       Constant required by ZSTD_compressBlock_btlazy2() and ZSTD_reduceTable_internal() */
 
 
 /*-*************************************
@@ -46,6 +48,12 @@ extern "C" {
 typedef enum { ZSTDcs_created=0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
 typedef enum { zcss_init=0, zcss_load, zcss_flush } ZSTD_cStreamStage;
 
+typedef enum {
+    ZSTD_dictDefaultAttach = 0,
+    ZSTD_dictForceAttach = 1,
+    ZSTD_dictForceCopy = -1,
+} ZSTD_dictAttachPref_e;
+
 typedef struct ZSTD_prefixDict_s {
     const void* dict;
     size_t dictSize;
@@ -53,14 +61,22 @@ typedef struct ZSTD_prefixDict_s {
 } ZSTD_prefixDict;
 
 typedef struct {
-    U32 hufCTable[HUF_CTABLE_SIZE_U32(255)];
+    U32 CTable[HUF_CTABLE_SIZE_U32(255)];
+    HUF_repeat repeatMode;
+} ZSTD_hufCTables_t;
+
+typedef struct {
     FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)];
     FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)];
     FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
-    HUF_repeat hufCTable_repeatMode;
     FSE_repeat offcode_repeatMode;
     FSE_repeat matchlength_repeatMode;
     FSE_repeat litlength_repeatMode;
+} ZSTD_fseCTables_t;
+
+typedef struct {
+    ZSTD_hufCTables_t huf;
+    ZSTD_fseCTables_t fse;
 } ZSTD_entropyCTables_t;
 
 typedef struct {
@@ -76,26 +92,27 @@ typedef struct {
     U32 rep[ZSTD_REP_NUM];
 } ZSTD_optimal_t;
 
+typedef enum { zop_dynamic=0, zop_predef } ZSTD_OptPrice_e;
+
 typedef struct {
     /* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */
-    U32* litFreq;               /* table of literals statistics, of size 256 */
-    U32* litLengthFreq;         /* table of litLength statistics, of size (MaxLL+1) */
-    U32* matchLengthFreq;       /* table of matchLength statistics, of size (MaxML+1) */
-    U32* offCodeFreq;           /* table of offCode statistics, of size (MaxOff+1) */
-    ZSTD_match_t* matchTable;   /* list of found matches, of size ZSTD_OPT_NUM+1 */
-    ZSTD_optimal_t* priceTable; /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */
+    U32* litFreq;                /* table of literals statistics, of size 256 */
+    U32* litLengthFreq;          /* table of litLength statistics, of size (MaxLL+1) */
+    U32* matchLengthFreq;        /* table of matchLength statistics, of size (MaxML+1) */
+    U32* offCodeFreq;            /* table of offCode statistics, of size (MaxOff+1) */
+    ZSTD_match_t* matchTable;    /* list of found matches, of size ZSTD_OPT_NUM+1 */
+    ZSTD_optimal_t* priceTable;  /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */
 
     U32  litSum;                 /* nb of literals */
     U32  litLengthSum;           /* nb of litLength codes */
     U32  matchLengthSum;         /* nb of matchLength codes */
     U32  offCodeSum;             /* nb of offset codes */
-    /* begin updated by ZSTD_setLog2Prices */
-    U32  log2litSum;             /* pow2 to compare log2(litfreq) to */
-    U32  log2litLengthSum;       /* pow2 to compare log2(llfreq) to */
-    U32  log2matchLengthSum;     /* pow2 to compare log2(mlfreq) to */
-    U32  log2offCodeSum;         /* pow2 to compare log2(offreq) to */
-    /* end : updated by ZSTD_setLog2Prices */
-    U32  staticPrices;           /* prices follow a pre-defined cost structure, statistics are irrelevant */
+    U32  litSumBasePrice;        /* to compare to log2(litfreq) */
+    U32  litLengthSumBasePrice;  /* to compare to log2(llfreq)  */
+    U32  matchLengthSumBasePrice;/* to compare to log2(mlfreq)  */
+    U32  offCodeSumBasePrice;    /* to compare to log2(offreq)  */
+    ZSTD_OptPrice_e priceType;   /* prices can be determined dynamically, or follow a pre-defined cost structure */
+    const ZSTD_entropyCTables_t* symbolCosts;  /* pre-calculated dictionary statistics */
 } optState_t;
 
 typedef struct {
@@ -111,17 +128,19 @@ typedef struct {
     U32 lowLimit;           /* below that point, no more data */
 } ZSTD_window_t;
 
-typedef struct {
-    ZSTD_window_t window;      /* State for window round buffer management */
-    U32 loadedDictEnd;         /* index of end of dictionary */
-    U32 nextToUpdate;          /* index from which to continue table update */
-    U32 nextToUpdate3;         /* index from which to continue table update */
-    U32 hashLog3;              /* dispatch table : larger == faster, more memory */
+typedef struct ZSTD_matchState_t ZSTD_matchState_t;
+struct ZSTD_matchState_t {
+    ZSTD_window_t window;   /* State for window round buffer management */
+    U32 loadedDictEnd;      /* index of end of dictionary */
+    U32 nextToUpdate;       /* index from which to continue table update */
+    U32 nextToUpdate3;      /* index from which to continue table update */
+    U32 hashLog3;           /* dispatch table : larger == faster, more memory */
     U32* hashTable;
     U32* hashTable3;
     U32* chainTable;
     optState_t opt;         /* optimal parser state */
-} ZSTD_matchState_t;
+    const ZSTD_matchState_t *dictMatchState;
+};
 
 typedef struct {
     ZSTD_compressedBlockState_t* prevCBlock;
@@ -161,7 +180,7 @@ typedef struct {
   rawSeq* seq;     /* The start of the sequences */
   size_t pos;      /* The position where reading stopped. <= size. */
   size_t size;     /* The number of sequences. <= capacity. */
-  size_t capacity; /* The capacity of the `seq` pointer */
+  size_t capacity; /* The capacity starting from `seq` pointer */
 } rawSeqStore_t;
 
 struct ZSTD_CCtx_params_s {
@@ -170,10 +189,11 @@ struct ZSTD_CCtx_params_s {
     ZSTD_frameParameters fParams;
 
     int compressionLevel;
-    int disableLiteralCompression;
     int forceWindow;           /* force back-references to respect limit of
                                 * 1<<wLog, even for dictionary */
 
+    ZSTD_dictAttachPref_e attachDictPref;
+
     /* Multithreading: used to pass parameters to mtctx */
     unsigned nbWorkers;
     unsigned jobSize;
@@ -193,6 +213,8 @@ struct ZSTD_CCtx_s {
     ZSTD_CCtx_params requestedParams;
     ZSTD_CCtx_params appliedParams;
     U32   dictID;
+
+    int workSpaceOversizedDuration;
     void* workSpace;
     size_t workSpaceSize;
     size_t blockSize;
@@ -235,11 +257,15 @@ struct ZSTD_CCtx_s {
 #endif
 };
 
+typedef enum { ZSTD_dtlm_fast, ZSTD_dtlm_full } ZSTD_dictTableLoadMethod_e;
+
+typedef enum { ZSTD_noDict = 0, ZSTD_extDict = 1, ZSTD_dictMatchState = 2 } ZSTD_dictMode_e;
+
 
 typedef size_t (*ZSTD_blockCompressor) (
         ZSTD_matchState_t* bs, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         ZSTD_compressionParameters const* cParams, void const* src, size_t srcSize);
-ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, int extDict);
+ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode);
 
 
 MEM_STATIC U32 ZSTD_LLcode(U32 litLength)
@@ -280,11 +306,11 @@ MEM_STATIC U32 ZSTD_MLcode(U32 mlBase)
 */
 MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const void* literals, U32 offsetCode, size_t mlBase)
 {
-#if defined(ZSTD_DEBUG) && (ZSTD_DEBUG >= 6)
+#if defined(DEBUGLEVEL) && (DEBUGLEVEL >= 6)
     static const BYTE* g_start = NULL;
     if (g_start==NULL) g_start = (const BYTE*)literals;  /* note : index only works for compression within a single segment */
     {   U32 const pos = (U32)((const BYTE*)literals - g_start);
-        DEBUGLOG(6, "Cpos%7u :%3u literals, match%3u bytes at dist.code%7u",
+        DEBUGLOG(6, "Cpos%7u :%3u literals, match%4u bytes at offCode%7u",
                pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offsetCode);
     }
 #endif
@@ -420,6 +446,11 @@ ZSTD_count_2segments(const BYTE* ip, const BYTE* match,
     const BYTE* const vEnd = MIN( ip + (mEnd - match), iEnd);
     size_t const matchLength = ZSTD_count(ip, match, vEnd);
     if (match + matchLength != mEnd) return matchLength;
+    DEBUGLOG(7, "ZSTD_count_2segments: found a 2-parts match (current length==%zu)", matchLength);
+    DEBUGLOG(7, "distance from match beginning to end dictionary = %zi", mEnd - match);
+    DEBUGLOG(7, "distance from current pos to end buffer = %zi", iEnd - ip);
+    DEBUGLOG(7, "next byte : ip==%02X, istart==%02X", ip[matchLength], *iStart);
+    DEBUGLOG(7, "final match length = %zu", matchLength + ZSTD_count(ip+matchLength, iStart, iEnd));
     return matchLength + ZSTD_count(ip+matchLength, iStart, iEnd);
 }
 
@@ -496,6 +527,20 @@ MEM_STATIC U32 ZSTD_window_hasExtDict(ZSTD_window_t const window)
     return window.lowLimit < window.dictLimit;
 }
 
+/**
+ * ZSTD_matchState_dictMode():
+ * Inspects the provided matchState and figures out what dictMode should be
+ * passed to the compressor.
+ */
+MEM_STATIC ZSTD_dictMode_e ZSTD_matchState_dictMode(const ZSTD_matchState_t *ms)
+{
+    return ZSTD_window_hasExtDict(ms->window) ?
+        ZSTD_extDict :
+        ms->dictMatchState != NULL ?
+            ZSTD_dictMatchState :
+            ZSTD_noDict;
+}
+
 /**
  * ZSTD_window_needOverflowCorrection():
  * Returns non-zero if the indices are getting too large and need overflow
@@ -563,31 +608,41 @@ MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
  * ZSTD_window_enforceMaxDist():
  * Updates lowLimit so that:
  *    (srcEnd - base) - lowLimit == maxDist + loadedDictEnd
+ *
  * This allows a simple check that index >= lowLimit to see if index is valid.
  * This must be called before a block compression call, with srcEnd as the block
  * source end.
+ *
  * If loadedDictEndPtr is not NULL, we set it to zero once we update lowLimit.
  * This is because dictionaries are allowed to be referenced as long as the last
  * byte of the dictionary is in the window, but once they are out of range,
  * they cannot be referenced. If loadedDictEndPtr is NULL, we use
  * loadedDictEnd == 0.
+ *
+ * In normal dict mode, the dict is between lowLimit and dictLimit. In
+ * dictMatchState mode, lowLimit and dictLimit are the same, and the dictionary
+ * is below them. forceWindow and dictMatchState are therefore incompatible.
  */
 MEM_STATIC void ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
                                            void const* srcEnd, U32 maxDist,
-                                           U32* loadedDictEndPtr)
+                                           U32* loadedDictEndPtr,
+                                           const ZSTD_matchState_t** dictMatchStatePtr)
 {
     U32 const current = (U32)((BYTE const*)srcEnd - window->base);
     U32 loadedDictEnd = loadedDictEndPtr != NULL ? *loadedDictEndPtr : 0;
+    DEBUGLOG(5, "ZSTD_window_enforceMaxDist: current=%u, maxDist=%u", current, maxDist);
     if (current > maxDist + loadedDictEnd) {
         U32 const newLowLimit = current - maxDist;
         if (window->lowLimit < newLowLimit) window->lowLimit = newLowLimit;
         if (window->dictLimit < window->lowLimit) {
-            DEBUGLOG(5, "Update dictLimit from %u to %u", window->dictLimit,
-                     window->lowLimit);
+            DEBUGLOG(5, "Update dictLimit to match lowLimit, from %u to %u",
+                        window->dictLimit, window->lowLimit);
             window->dictLimit = window->lowLimit;
         }
         if (loadedDictEndPtr)
             *loadedDictEndPtr = 0;
+        if (dictMatchStatePtr)
+            *dictMatchStatePtr = NULL;
     }
 }
 
@@ -603,12 +658,12 @@ MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
 {
     BYTE const* const ip = (BYTE const*)src;
     U32 contiguous = 1;
+    DEBUGLOG(5, "ZSTD_window_update");
     /* Check if blocks follow each other */
     if (src != window->nextSrc) {
         /* not contiguous */
         size_t const distanceFromBase = (size_t)(window->nextSrc - window->base);
-        DEBUGLOG(5, "Non contiguous blocks, new segment starts at %u",
-                 window->dictLimit);
+        DEBUGLOG(5, "Non contiguous blocks, new segment starts at %u", window->dictLimit);
         window->lowLimit = window->dictLimit;
         assert(distanceFromBase == (size_t)(U32)distanceFromBase);  /* should never overflow */
         window->dictLimit = (U32)distanceFromBase;
@@ -625,10 +680,38 @@ MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
         ptrdiff_t const highInputIdx = (ip + srcSize) - window->dictBase;
         U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)window->dictLimit) ? window->dictLimit : (U32)highInputIdx;
         window->lowLimit = lowLimitMax;
+        DEBUGLOG(5, "Overlapping extDict and input : new lowLimit = %u", window->lowLimit);
     }
     return contiguous;
 }
 
+
+/* debug functions */
+
+MEM_STATIC double ZSTD_fWeight(U32 rawStat)
+{
+    U32 const fp_accuracy = 8;
+    U32 const fp_multiplier = (1 << fp_accuracy);
+    U32 const stat = rawStat + 1;
+    U32 const hb = ZSTD_highbit32(stat);
+    U32 const BWeight = hb * fp_multiplier;
+    U32 const FWeight = (stat << fp_accuracy) >> hb;
+    U32 const weight = BWeight + FWeight;
+    assert(hb + fp_accuracy < 31);
+    return (double)weight / fp_multiplier;
+}
+
+MEM_STATIC void ZSTD_debugTable(const U32* table, U32 max)
+{
+    unsigned u, sum;
+    for (u=0, sum=0; u<=max; u++) sum += table[u];
+    DEBUGLOG(2, "total nb elts: %u", sum);
+    for (u=0; u<=max; u++) {
+        DEBUGLOG(2, "%2u: %5u  (%.2f)",
+                u, table[u], ZSTD_fWeight(sum) - ZSTD_fWeight(table[u]) );
+    }
+}
+
 #if defined (__cplusplus)
 }
 #endif
@@ -640,7 +723,7 @@ MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
  * ============================================================== */
 
 /* ZSTD_getCParamsFromCCtxParams() :
- * cParams are built depending on compressionLevel, src size hints, 
+ * cParams are built depending on compressionLevel, src size hints,
  * LDM and manually set compression parameters.
  */
 ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
@@ -656,6 +739,8 @@ size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
                      const ZSTD_CDict* cdict,
                      ZSTD_CCtx_params  params, unsigned long long pledgedSrcSize);
 
+void ZSTD_resetSeqStore(seqStore_t* ssPtr);
+
 /*! ZSTD_compressStream_generic() :
  *  Private use only. To be called from zstdmt_compress.c in single-thread mode. */
 size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
@@ -672,6 +757,7 @@ ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict);
 size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
                                     const void* dict, size_t dictSize,
                                     ZSTD_dictContentType_e dictContentType,
+                                    ZSTD_dictTableLoadMethod_e dtlm,
                                     const ZSTD_CDict* cdict,
                                     ZSTD_CCtx_params params,
                                     unsigned long long pledgedSrcSize);

data/ext/zstdruby/libzstd/compress/zstd_double_fast.c

@@ -14,7 +14,7 @@
 
 void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
                               ZSTD_compressionParameters const* cParams,
-                              void const* end)
+                              void const* end, ZSTD_dictTableLoadMethod_e dtlm)
 {
     U32* const hashLarge = ms->hashTable;
     U32  const hBitsL = cParams->hashLog;
@@ -40,6 +40,9 @@ void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
                 hashSmall[smHash] = current + i;
             if (i == 0 || hashLarge[lgHash] == 0)
                 hashLarge[lgHash] = current + i;
+            /* Only load extra positions for ZSTD_dtlm_full */
+            if (dtlm == ZSTD_dtlm_fast)
+                break;
         }
     }
 }
@@ -49,7 +52,7 @@ FORCE_INLINE_TEMPLATE
 size_t ZSTD_compressBlock_doubleFast_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         ZSTD_compressionParameters const* cParams, void const* src, size_t srcSize,
-        U32 const mls /* template */)
+        U32 const mls /* template */, ZSTD_dictMode_e const dictMode)
 {
     U32* const hashLong = ms->hashTable;
     const U32 hBitsL = cParams->hashLog;
@@ -59,70 +62,178 @@ size_t ZSTD_compressBlock_doubleFast_generic(
     const BYTE* const istart = (const BYTE*)src;
     const BYTE* ip = istart;
     const BYTE* anchor = istart;
-    const U32 lowestIndex = ms->window.dictLimit;
-    const BYTE* const lowest = base + lowestIndex;
+    const U32 prefixLowestIndex = ms->window.dictLimit;
+    const BYTE* const prefixLowest = base + prefixLowestIndex;
     const BYTE* const iend = istart + srcSize;
     const BYTE* const ilimit = iend - HASH_READ_SIZE;
     U32 offset_1=rep[0], offset_2=rep[1];
     U32 offsetSaved = 0;
 
+    const ZSTD_matchState_t* const dms = ms->dictMatchState;
+    const U32* const dictHashLong  = dictMode == ZSTD_dictMatchState ?
+                                     dms->hashTable : NULL;
+    const U32* const dictHashSmall = dictMode == ZSTD_dictMatchState ?
+                                     dms->chainTable : NULL;
+    const U32 dictStartIndex       = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.dictLimit : 0;
+    const BYTE* const dictBase     = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.base : NULL;
+    const BYTE* const dictStart    = dictMode == ZSTD_dictMatchState ?
+                                     dictBase + dictStartIndex : NULL;
+    const BYTE* const dictEnd      = dictMode == ZSTD_dictMatchState ?
+                                     dms->window.nextSrc : NULL;
+    const U32 dictIndexDelta       = dictMode == ZSTD_dictMatchState ?
+                                     prefixLowestIndex - (U32)(dictEnd - dictBase) :
+                                     0;
+    const U32 dictAndPrefixLength  = (U32)(ip - prefixLowest + dictEnd - dictStart);
+
+    assert(dictMode == ZSTD_noDict || dictMode == ZSTD_dictMatchState);
+
     /* init */
-    ip += (ip==lowest);
-    {   U32 const maxRep = (U32)(ip-lowest);
+    ip += (dictAndPrefixLength == 0);
+    if (dictMode == ZSTD_noDict) {
+        U32 const maxRep = (U32)(ip - prefixLowest);
         if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
         if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
     }
+    if (dictMode == ZSTD_dictMatchState) {
+        /* dictMatchState repCode checks don't currently handle repCode == 0
+         * disabling. */
+        assert(offset_1 <= dictAndPrefixLength);
+        assert(offset_2 <= dictAndPrefixLength);
+    }
 
     /* Main Search Loop */
     while (ip < ilimit) {   /* < instead of <=, because repcode check at (ip+1) */
         size_t mLength;
+        U32 offset;
         size_t const h2 = ZSTD_hashPtr(ip, hBitsL, 8);
         size_t const h = ZSTD_hashPtr(ip, hBitsS, mls);
         U32 const current = (U32)(ip-base);
         U32 const matchIndexL = hashLong[h2];
-        U32 const matchIndexS = hashSmall[h];
+        U32 matchIndexS = hashSmall[h];
         const BYTE* matchLong = base + matchIndexL;
         const BYTE* match = base + matchIndexS;
+        const U32 repIndex = current + 1 - offset_1;
+        const BYTE* repMatch = (dictMode == ZSTD_dictMatchState
+                            && repIndex < prefixLowestIndex) ?
+                               dictBase + (repIndex - dictIndexDelta) :
+                               base + repIndex;
         hashLong[h2] = hashSmall[h] = current;   /* update hash tables */
 
-        assert(offset_1 <= current);   /* supposed guaranteed by construction */
-        if ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1))) {
-            /* favor repcode */
+        /* check dictMatchState repcode */
+        if (dictMode == ZSTD_dictMatchState
+            && ((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
+            && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
+            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
+            ip++;
+            ZSTD_storeSeq(seqStore, ip-anchor, anchor, 0, mLength-MINMATCH);
+            goto _match_stored;
+        }
+
+        /* check noDict repcode */
+        if ( dictMode == ZSTD_noDict
+          && ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
             mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
             ip++;
             ZSTD_storeSeq(seqStore, ip-anchor, anchor, 0, mLength-MINMATCH);
-        } else {
-            U32 offset;
-            if ( (matchIndexL > lowestIndex) && (MEM_read64(matchLong) == MEM_read64(ip)) ) {
-                mLength = ZSTD_count(ip+8, matchLong+8, iend) + 8;
-                offset = (U32)(ip-matchLong);
-                while (((ip>anchor) & (matchLong>lowest)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
-            } else if ( (matchIndexS > lowestIndex) && (MEM_read32(match) == MEM_read32(ip)) ) {
-                size_t const hl3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
-                U32 const matchIndexL3 = hashLong[hl3];
-                const BYTE* matchL3 = base + matchIndexL3;
-                hashLong[hl3] = current + 1;
-                if ( (matchIndexL3 > lowestIndex) && (MEM_read64(matchL3) == MEM_read64(ip+1)) ) {
-                    mLength = ZSTD_count(ip+9, matchL3+8, iend) + 8;
+            goto _match_stored;
+        }
+
+        /* check prefix long match */
+        if ( (matchIndexL > prefixLowestIndex) && (MEM_read64(matchLong) == MEM_read64(ip)) ) {
+            mLength = ZSTD_count(ip+8, matchLong+8, iend) + 8;
+            offset = (U32)(ip-matchLong);
+            while (((ip>anchor) & (matchLong>prefixLowest)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
+            goto _match_found;
+        }
+
+        /* check dictMatchState long match */
+        if (dictMode == ZSTD_dictMatchState) {
+            U32 const dictMatchIndexL = dictHashLong[h2];
+            const BYTE* dictMatchL = dictBase + dictMatchIndexL;
+            assert(dictMatchL < dictEnd);
+
+            if (dictMatchL > dictStart && MEM_read64(dictMatchL) == MEM_read64(ip)) {
+                mLength = ZSTD_count_2segments(ip+8, dictMatchL+8, iend, dictEnd, prefixLowest) + 8;
+                offset = (U32)(current - dictMatchIndexL - dictIndexDelta);
+                while (((ip>anchor) & (dictMatchL>dictStart)) && (ip[-1] == dictMatchL[-1])) { ip--; dictMatchL--; mLength++; } /* catch up */
+                goto _match_found;
+            }
+        }
+
+        /* check prefix short match */
+        if ( (matchIndexS > prefixLowestIndex) && (MEM_read32(match) == MEM_read32(ip)) ) {
+            goto _search_next_long;
+        }
+
+        /* check dictMatchState short match */
+        if (dictMode == ZSTD_dictMatchState) {
+            U32 const dictMatchIndexS = dictHashSmall[h];
+            match = dictBase + dictMatchIndexS;
+            matchIndexS = dictMatchIndexS + dictIndexDelta;
+
+            if (match > dictStart && MEM_read32(match) == MEM_read32(ip)) {
+                goto _search_next_long;
+            }
+        }
+
+        ip += ((ip-anchor) >> kSearchStrength) + 1;
+        continue;
+
+_search_next_long:
+
+        {
+            size_t const hl3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
+            U32 const matchIndexL3 = hashLong[hl3];
+            const BYTE* matchL3 = base + matchIndexL3;
+            hashLong[hl3] = current + 1;
+
+            /* check prefix long +1 match */
+            if ( (matchIndexL3 > prefixLowestIndex) && (MEM_read64(matchL3) == MEM_read64(ip+1)) ) {
+                mLength = ZSTD_count(ip+9, matchL3+8, iend) + 8;
+                ip++;
+                offset = (U32)(ip-matchL3);
+                while (((ip>anchor) & (matchL3>prefixLowest)) && (ip[-1] == matchL3[-1])) { ip--; matchL3--; mLength++; } /* catch up */
+                goto _match_found;
+            }
+
+            /* check dict long +1 match */
+            if (dictMode == ZSTD_dictMatchState) {
+                U32 const dictMatchIndexL3 = dictHashLong[hl3];
+                const BYTE* dictMatchL3 = dictBase + dictMatchIndexL3;
+                assert(dictMatchL3 < dictEnd);
+                if (dictMatchL3 > dictStart && MEM_read64(dictMatchL3) == MEM_read64(ip+1)) {
+                    mLength = ZSTD_count_2segments(ip+1+8, dictMatchL3+8, iend, dictEnd, prefixLowest) + 8;
                     ip++;
-                    offset = (U32)(ip-matchL3);
-                    while (((ip>anchor) & (matchL3>lowest)) && (ip[-1] == matchL3[-1])) { ip--; matchL3--; mLength++; } /* catch up */
-                } else {
-                    mLength = ZSTD_count(ip+4, match+4, iend) + 4;
-                    offset = (U32)(ip-match);
-                    while (((ip>anchor) & (match>lowest)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
+                    offset = (U32)(current + 1 - dictMatchIndexL3 - dictIndexDelta);
+                    while (((ip>anchor) & (dictMatchL3>dictStart)) && (ip[-1] == dictMatchL3[-1])) { ip--; dictMatchL3--; mLength++; } /* catch up */
+                    goto _match_found;
                 }
-            } else {
-                ip += ((ip-anchor) >> kSearchStrength) + 1;
-                continue;
             }
+        }
 
-            offset_2 = offset_1;
-            offset_1 = offset;
-
-            ZSTD_storeSeq(seqStore, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+        /* if no long +1 match, explore the short match we found */
+        if (dictMode == ZSTD_dictMatchState && matchIndexS < prefixLowestIndex) {
+            mLength = ZSTD_count_2segments(ip+4, match+4, iend, dictEnd, prefixLowest) + 4;
+            offset = (U32)(current - matchIndexS);
+            while (((ip>anchor) & (match>dictStart)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
+        } else {
+            mLength = ZSTD_count(ip+4, match+4, iend) + 4;
+            offset = (U32)(ip - match);
+            while (((ip>anchor) & (match>prefixLowest)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
         }
 
+        /* fall-through */
+
+_match_found:
+        offset_2 = offset_1;
+        offset_1 = offset;
+
+        ZSTD_storeSeq(seqStore, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
+
+_match_stored:
         /* match found */
         ip += mLength;
         anchor = ip;
@@ -135,19 +246,44 @@ size_t ZSTD_compressBlock_doubleFast_generic(
                 hashSmall[ZSTD_hashPtr(ip-2, hBitsS, mls)] = (U32)(ip-2-base);
 
             /* check immediate repcode */
-            while ( (ip <= ilimit)
-                 && ( (offset_2>0)
-                 & (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
-                /* store sequence */
-                size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
-                { U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; } /* swap offset_2 <=> offset_1 */
-                hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base);
-                hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base);
-                ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
-                ip += rLength;
-                anchor = ip;
-                continue;   /* faster when present ... (?) */
-    }   }   }
+            if (dictMode == ZSTD_dictMatchState) {
+                while (ip <= ilimit) {
+                    U32 const current2 = (U32)(ip-base);
+                    U32 const repIndex2 = current2 - offset_2;
+                    const BYTE* repMatch2 = dictMode == ZSTD_dictMatchState
+                        && repIndex2 < prefixLowestIndex ?
+                            dictBase - dictIndexDelta + repIndex2 :
+                            base + repIndex2;
+                    if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
+                       && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                        const BYTE* const repEnd2 = repIndex2 < prefixLowestIndex ? dictEnd : iend;
+                        size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixLowest) + 4;
+                        U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
+                        ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
+                        hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
+                        hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
+                        ip += repLength2;
+                        anchor = ip;
+                        continue;
+                    }
+                    break;
+                }
+            }
+
+            if (dictMode == ZSTD_noDict) {
+                while ( (ip <= ilimit)
+                     && ( (offset_2>0)
+                        & (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
+                    /* store sequence */
+                    size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
+                    U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff;  /* swap offset_2 <=> offset_1 */
+                    hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base);
+                    hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base);
+                    ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
+                    ip += rLength;
+                    anchor = ip;
+                    continue;   /* faster when present ... (?) */
+    }   }   }   }
 
     /* save reps for next block */
     rep[0] = offset_1 ? offset_1 : offsetSaved;
@@ -167,13 +303,33 @@ size_t ZSTD_compressBlock_doubleFast(
     {
     default: /* includes case 3 */
     case 4 :
-        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 4);
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 4, ZSTD_noDict);
     case 5 :
-        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 5);
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 5, ZSTD_noDict);
     case 6 :
-        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 6);
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 6, ZSTD_noDict);
     case 7 :
-        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 7);
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 7, ZSTD_noDict);
+    }
+}
+
+
+size_t ZSTD_compressBlock_doubleFast_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        ZSTD_compressionParameters const* cParams, void const* src, size_t srcSize)
+{
+    const U32 mls = cParams->searchLength;
+    switch(mls)
+    {
+    default: /* includes case 3 */
+    case 4 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 4, ZSTD_dictMatchState);
+    case 5 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 5, ZSTD_dictMatchState);
+    case 6 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 6, ZSTD_dictMatchState);
+    case 7 :
+        return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, cParams, src, srcSize, 7, ZSTD_dictMatchState);
     }
 }
 
@@ -187,75 +343,78 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
     U32  const hBitsL = cParams->hashLog;
     U32* const hashSmall = ms->chainTable;
     U32  const hBitsS = cParams->chainLog;
-    const BYTE* const base = ms->window.base;
-    const BYTE* const dictBase = ms->window.dictBase;
     const BYTE* const istart = (const BYTE*)src;
     const BYTE* ip = istart;
     const BYTE* anchor = istart;
-    const U32   lowestIndex = ms->window.lowLimit;
-    const BYTE* const dictStart = dictBase + lowestIndex;
-    const U32   dictLimit = ms->window.dictLimit;
-    const BYTE* const lowPrefixPtr = base + dictLimit;
-    const BYTE* const dictEnd = dictBase + dictLimit;
     const BYTE* const iend = istart + srcSize;
     const BYTE* const ilimit = iend - 8;
+    const U32   prefixStartIndex = ms->window.dictLimit;
+    const BYTE* const base = ms->window.base;
+    const BYTE* const prefixStart = base + prefixStartIndex;
+    const U32   dictStartIndex = ms->window.lowLimit;
+    const BYTE* const dictBase = ms->window.dictBase;
+    const BYTE* const dictStart = dictBase + dictStartIndex;
+    const BYTE* const dictEnd = dictBase + prefixStartIndex;
     U32 offset_1=rep[0], offset_2=rep[1];
 
+    DEBUGLOG(5, "ZSTD_compressBlock_doubleFast_extDict_generic (srcSize=%zu)", srcSize);
+
     /* Search Loop */
     while (ip < ilimit) {  /* < instead of <=, because (ip+1) */
         const size_t hSmall = ZSTD_hashPtr(ip, hBitsS, mls);
         const U32 matchIndex = hashSmall[hSmall];
-        const BYTE* matchBase = matchIndex < dictLimit ? dictBase : base;
+        const BYTE* const matchBase = matchIndex < prefixStartIndex ? dictBase : base;
         const BYTE* match = matchBase + matchIndex;
 
         const size_t hLong = ZSTD_hashPtr(ip, hBitsL, 8);
         const U32 matchLongIndex = hashLong[hLong];
-        const BYTE* matchLongBase = matchLongIndex < dictLimit ? dictBase : base;
+        const BYTE* const matchLongBase = matchLongIndex < prefixStartIndex ? dictBase : base;
         const BYTE* matchLong = matchLongBase + matchLongIndex;
 
         const U32 current = (U32)(ip-base);
         const U32 repIndex = current + 1 - offset_1;   /* offset_1 expected <= current +1 */
-        const BYTE* repBase = repIndex < dictLimit ? dictBase : base;
-        const BYTE* repMatch = repBase + repIndex;
+        const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
+        const BYTE* const repMatch = repBase + repIndex;
         size_t mLength;
         hashSmall[hSmall] = hashLong[hLong] = current;   /* update hash table */
 
-        if ( (((U32)((dictLimit-1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex))
-           && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
-            const BYTE* repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
-            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, lowPrefixPtr) + 4;
+        if ((((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex doesn't overlap dict + prefix */
+            & (repIndex > dictStartIndex))
+          && (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
+            const BYTE* repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
+            mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
             ip++;
             ZSTD_storeSeq(seqStore, ip-anchor, anchor, 0, mLength-MINMATCH);
         } else {
-            if ((matchLongIndex > lowestIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) {
-                const BYTE* matchEnd = matchLongIndex < dictLimit ? dictEnd : iend;
-                const BYTE* lowMatchPtr = matchLongIndex < dictLimit ? dictStart : lowPrefixPtr;
+            if ((matchLongIndex > dictStartIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) {
+                const BYTE* const matchEnd = matchLongIndex < prefixStartIndex ? dictEnd : iend;
+                const BYTE* const lowMatchPtr = matchLongIndex < prefixStartIndex ? dictStart : prefixStart;
                 U32 offset;
-                mLength = ZSTD_count_2segments(ip+8, matchLong+8, iend, matchEnd, lowPrefixPtr) + 8;
+                mLength = ZSTD_count_2segments(ip+8, matchLong+8, iend, matchEnd, prefixStart) + 8;
                 offset = current - matchLongIndex;
                 while (((ip>anchor) & (matchLong>lowMatchPtr)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; }   /* catch up */
                 offset_2 = offset_1;
                 offset_1 = offset;
                 ZSTD_storeSeq(seqStore, ip-anchor, anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
 
-            } else if ((matchIndex > lowestIndex) && (MEM_read32(match) == MEM_read32(ip))) {
+            } else if ((matchIndex > dictStartIndex) && (MEM_read32(match) == MEM_read32(ip))) {
                 size_t const h3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
                 U32 const matchIndex3 = hashLong[h3];
-                const BYTE* const match3Base = matchIndex3 < dictLimit ? dictBase : base;
+                const BYTE* const match3Base = matchIndex3 < prefixStartIndex ? dictBase : base;
                 const BYTE* match3 = match3Base + matchIndex3;
                 U32 offset;
                 hashLong[h3] = current + 1;
-                if ( (matchIndex3 > lowestIndex) && (MEM_read64(match3) == MEM_read64(ip+1)) ) {
-                    const BYTE* matchEnd = matchIndex3 < dictLimit ? dictEnd : iend;
-                    const BYTE* lowMatchPtr = matchIndex3 < dictLimit ? dictStart : lowPrefixPtr;
-                    mLength = ZSTD_count_2segments(ip+9, match3+8, iend, matchEnd, lowPrefixPtr) + 8;
+                if ( (matchIndex3 > dictStartIndex) && (MEM_read64(match3) == MEM_read64(ip+1)) ) {
+                    const BYTE* const matchEnd = matchIndex3 < prefixStartIndex ? dictEnd : iend;
+                    const BYTE* const lowMatchPtr = matchIndex3 < prefixStartIndex ? dictStart : prefixStart;
+                    mLength = ZSTD_count_2segments(ip+9, match3+8, iend, matchEnd, prefixStart) + 8;
                     ip++;
                     offset = current+1 - matchIndex3;
                     while (((ip>anchor) & (match3>lowMatchPtr)) && (ip[-1] == match3[-1])) { ip--; match3--; mLength++; } /* catch up */
                 } else {
-                    const BYTE* matchEnd = matchIndex < dictLimit ? dictEnd : iend;
-                    const BYTE* lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
-                    mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, lowPrefixPtr) + 4;
+                    const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
+                    const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
+                    mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
                     offset = current - matchIndex;
                     while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; }   /* catch up */
                 }
@@ -282,12 +441,13 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
             while (ip <= ilimit) {
                 U32 const current2 = (U32)(ip-base);
                 U32 const repIndex2 = current2 - offset_2;
-                const BYTE* repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
-                if ( (((U32)((dictLimit-1) - repIndex2) >= 3) & (repIndex2 > lowestIndex))  /* intentional overflow */
-                   && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
-                    const BYTE* const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
-                    size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, lowPrefixPtr) + 4;
-                    U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
+                const BYTE* repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
+                if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3)   /* intentional overflow : ensure repIndex2 doesn't overlap dict + prefix */
+                    & (repIndex2 > dictStartIndex))
+                  && (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
+                    const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
+                    size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
+                    U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset;   /* swap offset_2 <=> offset_1 */
                     ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
                     hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
                     hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;