zstd-ruby 1.4.1.0 → 1.5.0.0

data/ext/zstdruby/libzstd/legacy/zstd_v05.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -19,7 +19,7 @@ extern "C" {
 *  Dependencies
 ***************************************/
 #include <stddef.h>   /* size_t */
-#include "mem.h"      /* U64, U32 */
+#include "../common/mem.h"      /* U64, U32 */
 
 
 /* *************************************

data/ext/zstdruby/libzstd/legacy/zstd_v06.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -14,7 +14,7 @@
 #include <stddef.h>    /* size_t, ptrdiff_t */
 #include <string.h>    /* memcpy */
 #include <stdlib.h>    /* malloc, free, qsort */
-#include "error_private.h"
+#include "../common/error_private.h"
 
 
 
@@ -82,7 +82,11 @@ extern "C" {
 *  Basic Types
 *****************************************************************/
 #if  !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
-# include <stdint.h>
+# if defined(_AIX)
+#  include <inttypes.h>
+# else
+#  include <stdint.h> /* intptr_t */
+# endif
   typedef  uint8_t BYTE;
   typedef uint16_t U16;
   typedef  int16_t S16;
@@ -860,7 +864,7 @@ MEM_STATIC unsigned BITv06_highbit32 ( U32 val)
     _BitScanReverse ( &r, val );
     return (unsigned) r;
 #   elif defined(__GNUC__) && (__GNUC__ >= 3)   /* Use GCC Intrinsic */
-    return 31 - __builtin_clz (val);
+    return __builtin_clz (val) ^ 31;
 #   else   /* Software version */
     static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
     U32 v = val;
@@ -1862,7 +1866,7 @@ MEM_STATIC size_t HUFv06_readStats(BYTE* huffWeight, size_t hwSize, U32* rankSta
 
     if (!srcSize) return ERROR(srcSize_wrong);
     iSize = ip[0];
-    //memset(huffWeight, 0, hwSize);   /* is not necessary, even though some analyzer complain ... */
+    /* memset(huffWeight, 0, hwSize); */   /* is not necessary, even though some analyzer complain ... */
 
     if (iSize >= 128)  { /* special header */
         if (iSize >= (242)) {  /* RLE */
@@ -2014,7 +2018,7 @@ size_t HUFv06_readDTableX2 (U16* DTable, const void* src, size_t srcSize)
     HUFv06_DEltX2* const dt = (HUFv06_DEltX2*)dtPtr;
 
     HUFv06_STATIC_ASSERT(sizeof(HUFv06_DEltX2) == sizeof(U16));   /* if compilation fails here, assertion is false */
-    //memset(huffWeight, 0, sizeof(huffWeight));   /* is not necessary, even though some analyzer complain ... */
+    /* memset(huffWeight, 0, sizeof(huffWeight)); */   /* is not necessary, even though some analyzer complain ... */
 
     iSize = HUFv06_readStats(huffWeight, HUFv06_MAX_SYMBOL_VALUE + 1, rankVal, &nbSymbols, &tableLog, src, srcSize);
     if (HUFv06_isError(iSize)) return iSize;
@@ -2340,7 +2344,7 @@ size_t HUFv06_readDTableX4 (U32* DTable, const void* src, size_t srcSize)
 
     HUFv06_STATIC_ASSERT(sizeof(HUFv06_DEltX4) == sizeof(U32));   /* if compilation fails here, assertion is false */
     if (memLog > HUFv06_ABSOLUTEMAX_TABLELOG) return ERROR(tableLog_tooLarge);
-    //memset(weightList, 0, sizeof(weightList));   /* is not necessary, even though some analyzer complain ... */
+    /* memset(weightList, 0, sizeof(weightList)); */   /* is not necessary, even though some analyzer complain ... */
 
     iSize = HUFv06_readStats(weightList, HUFv06_MAX_SYMBOL_VALUE + 1, rankStats, &nbSymbols, &tableLog, src, srcSize);
     if (HUFv06_isError(iSize)) return iSize;
@@ -2664,13 +2668,13 @@ size_t HUFv06_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cS
 
     {   U32 algoNb = 0;
         if (Dtime[1] < Dtime[0]) algoNb = 1;
-        // if (Dtime[2] < Dtime[algoNb]) algoNb = 2;   /* current speed of HUFv06_decompress4X6 is not good */
+        /* if (Dtime[2] < Dtime[algoNb]) algoNb = 2; */   /* current speed of HUFv06_decompress4X6 is not good */
         return decompress[algoNb](dst, dstSize, cSrc, cSrcSize);
     }
 
-    //return HUFv06_decompress4X2(dst, dstSize, cSrc, cSrcSize);   /* multi-streams single-symbol decoding */
-    //return HUFv06_decompress4X4(dst, dstSize, cSrc, cSrcSize);   /* multi-streams double-symbols decoding */
-    //return HUFv06_decompress4X6(dst, dstSize, cSrc, cSrcSize);   /* multi-streams quad-symbols decoding */
+    /* return HUFv06_decompress4X2(dst, dstSize, cSrc, cSrcSize); */   /* multi-streams single-symbol decoding */
+    /* return HUFv06_decompress4X4(dst, dstSize, cSrc, cSrcSize); */   /* multi-streams double-symbols decoding */
+    /* return HUFv06_decompress4X6(dst, dstSize, cSrc, cSrcSize); */   /* multi-streams quad-symbols decoding */
 }
 /*
     Common functions of Zstd compression library
@@ -3025,7 +3029,7 @@ typedef struct
 *   Provides the size of compressed block from block header `src` */
 static size_t ZSTDv06_getcBlockSize(const void* src, size_t srcSize, blockProperties_t* bpPtr)
 {
-    const BYTE* const in = (const BYTE* const)src;
+    const BYTE* const in = (const BYTE*)src;
     U32 cSize;
 
     if (srcSize < ZSTDv06_blockHeaderSize) return ERROR(srcSize_wrong);
@@ -3219,7 +3223,7 @@ static size_t ZSTDv06_decodeSeqHeaders(int* nbSeqPtr,
                              FSEv06_DTable* DTableLL, FSEv06_DTable* DTableML, FSEv06_DTable* DTableOffb, U32 flagRepeatTable,
                              const void* src, size_t srcSize)
 {
-    const BYTE* const istart = (const BYTE* const)src;
+    const BYTE* const istart = (const BYTE*)src;
     const BYTE* const iend = istart + srcSize;
     const BYTE* ip = istart;
 
@@ -3441,7 +3445,7 @@ static size_t ZSTDv06_decompressSequences(
 {
     const BYTE* ip = (const BYTE*)seqStart;
     const BYTE* const iend = ip + seqSize;
-    BYTE* const ostart = (BYTE* const)dst;
+    BYTE* const ostart = (BYTE*)dst;
     BYTE* const oend = ostart + maxDstSize;
     BYTE* op = ostart;
     const BYTE* litPtr = dctx->litPtr;
@@ -3501,8 +3505,10 @@ static size_t ZSTDv06_decompressSequences(
     {   size_t const lastLLSize = litEnd - litPtr;
         if (litPtr > litEnd) return ERROR(corruption_detected);   /* too many literals already used */
         if (op+lastLLSize > oend) return ERROR(dstSize_tooSmall);
-        memcpy(op, litPtr, lastLLSize);
-        op += lastLLSize;
+        if (lastLLSize > 0) {
+            memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
     }
 
     return op-ostart;
@@ -3555,7 +3561,7 @@ static size_t ZSTDv06_decompressFrame(ZSTDv06_DCtx* dctx,
 {
     const BYTE* ip = (const BYTE*)src;
     const BYTE* const iend = ip + srcSize;
-    BYTE* const ostart = (BYTE* const)dst;
+    BYTE* const ostart = (BYTE*)dst;
     BYTE* op = ostart;
     BYTE* const oend = ostart + dstCapacity;
     size_t remainingSize = srcSize;
@@ -4000,7 +4006,9 @@ size_t ZBUFFv06_decompressInit(ZBUFFv06_DCtx* zbd)
 MEM_STATIC size_t ZBUFFv06_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
 {
     size_t length = MIN(dstCapacity, srcSize);
-    memcpy(dst, src, length);
+    if (length > 0) {
+        memcpy(dst, src, length);
+    }
     return length;
 }
 
@@ -4109,7 +4117,7 @@ size_t ZBUFFv06_decompressContinue(ZBUFFv06_DCtx* zbd,
                     if (!decodedSize) { zbd->stage = ZBUFFds_read; break; }   /* this was just a header */
                     zbd->outEnd = zbd->outStart +  decodedSize;
                     zbd->stage = ZBUFFds_flush;
-                    // break; /* ZBUFFds_flush follows */
+                    /* break; */ /* ZBUFFds_flush follows */
                 }
 	    }
 	    /* fall-through */

data/ext/zstdruby/libzstd/legacy/zstd_v06.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the

data/ext/zstdruby/libzstd/legacy/zstd_v07.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -17,14 +17,14 @@
 #ifndef XXH_STATIC_LINKING_ONLY
 #  define XXH_STATIC_LINKING_ONLY    /* XXH64_state_t */
 #endif
-#include "xxhash.h"                  /* XXH64_* */
+#include "../common/xxhash.h"                  /* XXH64_* */
 #include "zstd_v07.h"
 
 #define FSEv07_STATIC_LINKING_ONLY   /* FSEv07_MIN_TABLELOG */
 #define HUFv07_STATIC_LINKING_ONLY   /* HUFv07_TABLELOG_ABSOLUTEMAX */
 #define ZSTDv07_STATIC_LINKING_ONLY
 
-#include "error_private.h"
+#include "../common/error_private.h"
 
 
 #ifdef ZSTDv07_STATIC_LINKING_ONLY
@@ -242,7 +242,11 @@ extern "C" {
 *  Basic Types
 *****************************************************************/
 #if  !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
-# include <stdint.h>
+# if defined(_AIX)
+#  include <inttypes.h>
+# else
+#  include <stdint.h> /* intptr_t */
+# endif
   typedef  uint8_t BYTE;
   typedef uint16_t U16;
   typedef  int16_t S16;
@@ -530,7 +534,7 @@ MEM_STATIC unsigned BITv07_highbit32 (U32 val)
     _BitScanReverse ( &r, val );
     return (unsigned) r;
 #   elif defined(__GNUC__) && (__GNUC__ >= 3)   /* Use GCC Intrinsic */
-    return 31 - __builtin_clz (val);
+    return __builtin_clz (val) ^ 31;
 #   else   /* Software version */
     static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
     U32 v = val;
@@ -1314,7 +1318,7 @@ size_t HUFv07_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats,
 
     if (!srcSize) return ERROR(srcSize_wrong);
     iSize = ip[0];
-    //memset(huffWeight, 0, hwSize);   /* is not necessary, even though some analyzer complain ... */
+    /* memset(huffWeight, 0, hwSize); */   /* is not necessary, even though some analyzer complain ... */
 
     if (iSize >= 128)  { /* special header */
         if (iSize >= (242)) {  /* RLE */
@@ -1784,7 +1788,7 @@ size_t HUFv07_readDTableX2 (HUFv07_DTable* DTable, const void* src, size_t srcSi
     HUFv07_DEltX2* const dt = (HUFv07_DEltX2*)dtPtr;
 
     HUFv07_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUFv07_DTable));
-    //memset(huffWeight, 0, sizeof(huffWeight));   /* is not necessary, even though some analyzer complain ... */
+    /* memset(huffWeight, 0, sizeof(huffWeight)); */   /* is not necessary, even though some analyzer complain ... */
 
     iSize = HUFv07_readStats(huffWeight, HUFv07_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize);
     if (HUFv07_isError(iSize)) return iSize;
@@ -2148,7 +2152,7 @@ size_t HUFv07_readDTableX4 (HUFv07_DTable* DTable, const void* src, size_t srcSi
 
     HUFv07_STATIC_ASSERT(sizeof(HUFv07_DEltX4) == sizeof(HUFv07_DTable));   /* if compilation fails here, assertion is false */
     if (maxTableLog > HUFv07_TABLELOG_ABSOLUTEMAX) return ERROR(tableLog_tooLarge);
-    //memset(weightList, 0, sizeof(weightList));   /* is not necessary, even though some analyzer complain ... */
+    /* memset(weightList, 0, sizeof(weightList)); */   /* is not necessary, even though some analyzer complain ... */
 
     iSize = HUFv07_readStats(weightList, HUFv07_SYMBOLVALUE_MAX + 1, rankStats, &nbSymbols, &tableLog, src, srcSize);
     if (HUFv07_isError(iSize)) return iSize;
@@ -2530,8 +2534,8 @@ size_t HUFv07_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cS
         return decompress[algoNb](dst, dstSize, cSrc, cSrcSize);
     }
 
-    //return HUFv07_decompress4X2(dst, dstSize, cSrc, cSrcSize);   /* multi-streams single-symbol decoding */
-    //return HUFv07_decompress4X4(dst, dstSize, cSrc, cSrcSize);   /* multi-streams double-symbols decoding */
+    /* return HUFv07_decompress4X2(dst, dstSize, cSrc, cSrcSize); */   /* multi-streams single-symbol decoding */
+    /* return HUFv07_decompress4X4(dst, dstSize, cSrc, cSrcSize); */   /* multi-streams double-symbols decoding */
 }
 
 size_t HUFv07_decompress4X_DCtx (HUFv07_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
@@ -3254,7 +3258,7 @@ typedef struct
 *   Provides the size of compressed block from block header `src` */
 static size_t ZSTDv07_getcBlockSize(const void* src, size_t srcSize, blockProperties_t* bpPtr)
 {
-    const BYTE* const in = (const BYTE* const)src;
+    const BYTE* const in = (const BYTE*)src;
     U32 cSize;
 
     if (srcSize < ZSTDv07_blockHeaderSize) return ERROR(srcSize_wrong);
@@ -3272,7 +3276,9 @@ static size_t ZSTDv07_getcBlockSize(const void* src, size_t srcSize, blockProper
 static size_t ZSTDv07_copyRawBlock(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
 {
     if (srcSize > dstCapacity) return ERROR(dstSize_tooSmall);
-    memcpy(dst, src, srcSize);
+    if (srcSize > 0) {
+        memcpy(dst, src, srcSize);
+    }
     return srcSize;
 }
 
@@ -3447,7 +3453,7 @@ static size_t ZSTDv07_decodeSeqHeaders(int* nbSeqPtr,
                              FSEv07_DTable* DTableLL, FSEv07_DTable* DTableML, FSEv07_DTable* DTableOffb, U32 flagRepeatTable,
                              const void* src, size_t srcSize)
 {
-    const BYTE* const istart = (const BYTE* const)src;
+    const BYTE* const istart = (const BYTE*)src;
     const BYTE* const iend = istart + srcSize;
     const BYTE* ip = istart;
 
@@ -3666,7 +3672,7 @@ static size_t ZSTDv07_decompressSequences(
 {
     const BYTE* ip = (const BYTE*)seqStart;
     const BYTE* const iend = ip + seqSize;
-    BYTE* const ostart = (BYTE* const)dst;
+    BYTE* const ostart = (BYTE*)dst;
     BYTE* const oend = ostart + maxDstSize;
     BYTE* op = ostart;
     const BYTE* litPtr = dctx->litPtr;
@@ -3712,10 +3718,12 @@ static size_t ZSTDv07_decompressSequences(
 
     /* last literal segment */
     {   size_t const lastLLSize = litEnd - litPtr;
-        //if (litPtr > litEnd) return ERROR(corruption_detected);   /* too many literals already used */
+        /* if (litPtr > litEnd) return ERROR(corruption_detected); */   /* too many literals already used */
         if (lastLLSize > (size_t)(oend-op)) return ERROR(dstSize_tooSmall);
-        memcpy(op, litPtr, lastLLSize);
-        op += lastLLSize;
+        if (lastLLSize > 0) {
+            memcpy(op, litPtr, lastLLSize);
+            op += lastLLSize;
+        }
     }
 
     return op-ostart;
@@ -3776,7 +3784,9 @@ ZSTDLIBv07_API size_t ZSTDv07_insertBlock(ZSTDv07_DCtx* dctx, const void* blockS
 static size_t ZSTDv07_generateNxBytes(void* dst, size_t dstCapacity, BYTE byte, size_t length)
 {
     if (length > dstCapacity) return ERROR(dstSize_tooSmall);
-    memset(dst, byte, length);
+    if (length > 0) {
+        memset(dst, byte, length);
+    }
     return length;
 }
 
@@ -3789,7 +3799,7 @@ static size_t ZSTDv07_decompressFrame(ZSTDv07_DCtx* dctx,
 {
     const BYTE* ip = (const BYTE*)src;
     const BYTE* const iend = ip + srcSize;
-    BYTE* const ostart = (BYTE* const)dst;
+    BYTE* const ostart = (BYTE*)dst;
     BYTE* const oend = ostart + dstCapacity;
     BYTE* op = ostart;
     size_t remainingSize = srcSize;
@@ -4378,7 +4388,9 @@ size_t ZBUFFv07_decompressInit(ZBUFFv07_DCtx* zbd)
 MEM_STATIC size_t ZBUFFv07_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
 {
     size_t const length = MIN(dstCapacity, srcSize);
-    memcpy(dst, src, length);
+    if (length > 0) {
+        memcpy(dst, src, length);
+    }
     return length;
 }
 

data/ext/zstdruby/libzstd/legacy/zstd_v07.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the

data/ext/zstdruby/libzstd/libzstd.pc.in

@@ -3,8 +3,9 @@
 #   BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
 
 prefix=@PREFIX@
-libdir=@LIBDIR@
+exec_prefix=@EXEC_PREFIX@
 includedir=@INCLUDEDIR@
+libdir=@LIBDIR@
 
 Name: zstd
 Description: fast lossless compression algorithm library

data/ext/zstdruby/libzstd/{dictBuilder/zdict.h → zdict.h}

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -36,6 +36,145 @@ extern "C" {
 #  define ZDICTLIB_API ZDICTLIB_VISIBILITY
 #endif
 
+/*******************************************************************************
+ * Zstd dictionary builder
+ *
+ * FAQ
+ * ===
+ * Why should I use a dictionary?
+ * ------------------------------
+ *
+ * Zstd can use dictionaries to improve compression ratio of small data.
+ * Traditionally small files don't compress well because there is very little
+ * repetion in a single sample, since it is small. But, if you are compressing
+ * many similar files, like a bunch of JSON records that share the same
+ * structure, you can train a dictionary on ahead of time on some samples of
+ * these files. Then, zstd can use the dictionary to find repetitions that are
+ * present across samples. This can vastly improve compression ratio.
+ *
+ * When is a dictionary useful?
+ * ----------------------------
+ *
+ * Dictionaries are useful when compressing many small files that are similar.
+ * The larger a file is, the less benefit a dictionary will have. Generally,
+ * we don't expect dictionary compression to be effective past 100KB. And the
+ * smaller a file is, the more we would expect the dictionary to help.
+ *
+ * How do I use a dictionary?
+ * --------------------------
+ *
+ * Simply pass the dictionary to the zstd compressor with
+ * `ZSTD_CCtx_loadDictionary()`. The same dictionary must then be passed to
+ * the decompressor, using `ZSTD_DCtx_loadDictionary()`. There are other
+ * more advanced functions that allow selecting some options, see zstd.h for
+ * complete documentation.
+ *
+ * What is a zstd dictionary?
+ * --------------------------
+ *
+ * A zstd dictionary has two pieces: Its header, and its content. The header
+ * contains a magic number, the dictionary ID, and entropy tables. These
+ * entropy tables allow zstd to save on header costs in the compressed file,
+ * which really matters for small data. The content is just bytes, which are
+ * repeated content that is common across many samples.
+ *
+ * What is a raw content dictionary?
+ * ---------------------------------
+ *
+ * A raw content dictionary is just bytes. It doesn't have a zstd dictionary
+ * header, a dictionary ID, or entropy tables. Any buffer is a valid raw
+ * content dictionary.
+ *
+ * How do I train a dictionary?
+ * ----------------------------
+ *
+ * Gather samples from your use case. These samples should be similar to each
+ * other. If you have several use cases, you could try to train one dictionary
+ * per use case.
+ *
+ * Pass those samples to `ZDICT_trainFromBuffer()` and that will train your
+ * dictionary. There are a few advanced versions of this function, but this
+ * is a great starting point. If you want to further tune your dictionary
+ * you could try `ZDICT_optimizeTrainFromBuffer_cover()`. If that is too slow
+ * you can try `ZDICT_optimizeTrainFromBuffer_fastCover()`.
+ *
+ * If the dictionary training function fails, that is likely because you
+ * either passed too few samples, or a dictionary would not be effective
+ * for your data. Look at the messages that the dictionary trainer printed,
+ * if it doesn't say too few samples, then a dictionary would not be effective.
+ *
+ * How large should my dictionary be?
+ * ----------------------------------
+ *
+ * A reasonable dictionary size, the `dictBufferCapacity`, is about 100KB.
+ * The zstd CLI defaults to a 110KB dictionary. You likely don't need a
+ * dictionary larger than that. But, most use cases can get away with a
+ * smaller dictionary. The advanced dictionary builders can automatically
+ * shrink the dictionary for you, and select a the smallest size that
+ * doesn't hurt compression ratio too much. See the `shrinkDict` parameter.
+ * A smaller dictionary can save memory, and potentially speed up
+ * compression.
+ *
+ * How many samples should I provide to the dictionary builder?
+ * ------------------------------------------------------------
+ *
+ * We generally recommend passing ~100x the size of the dictionary
+ * in samples. A few thousand should suffice. Having too few samples
+ * can hurt the dictionaries effectiveness. Having more samples will
+ * only improve the dictionaries effectiveness. But having too many
+ * samples can slow down the dictionary builder.
+ *
+ * How do I determine if a dictionary will be effective?
+ * -----------------------------------------------------
+ *
+ * Simply train a dictionary and try it out. You can use zstd's built in
+ * benchmarking tool to test the dictionary effectiveness.
+ *
+ *   # Benchmark levels 1-3 without a dictionary
+ *   zstd -b1e3 -r /path/to/my/files
+ *   # Benchmark levels 1-3 with a dictioanry
+ *   zstd -b1e3 -r /path/to/my/files -D /path/to/my/dictionary
+ *
+ * When should I retrain a dictionary?
+ * -----------------------------------
+ *
+ * You should retrain a dictionary when its effectiveness drops. Dictionary
+ * effectiveness drops as the data you are compressing changes. Generally, we do
+ * expect dictionaries to "decay" over time, as your data changes, but the rate
+ * at which they decay depends on your use case. Internally, we regularly
+ * retrain dictionaries, and if the new dictionary performs significantly
+ * better than the old dictionary, we will ship the new dictionary.
+ *
+ * I have a raw content dictionary, how do I turn it into a zstd dictionary?
+ * -------------------------------------------------------------------------
+ *
+ * If you have a raw content dictionary, e.g. by manually constructing it, or
+ * using a third-party dictionary builder, you can turn it into a zstd
+ * dictionary by using `ZDICT_finalizeDictionary()`. You'll also have to
+ * provide some samples of the data. It will add the zstd header to the
+ * raw content, which contains a dictionary ID and entropy tables, which
+ * will improve compression ratio, and allow zstd to write the dictionary ID
+ * into the frame, if you so choose.
+ *
+ * Do I have to use zstd's dictionary builder?
+ * -------------------------------------------
+ *
+ * No! You can construct dictionary content however you please, it is just
+ * bytes. It will always be valid as a raw content dictionary. If you want
+ * a zstd dictionary, which can improve compression ratio, use
+ * `ZDICT_finalizeDictionary()`.
+ *
+ * What is the attack surface of a zstd dictionary?
+ * ------------------------------------------------
+ *
+ * Zstd is heavily fuzz tested, including loading fuzzed dictionaries, so
+ * zstd should never crash, or access out-of-bounds memory no matter what
+ * the dictionary is. However, if an attacker can control the dictionary
+ * during decompression, they can cause zstd to generate arbitrary bytes,
+ * just like if they controlled the compressed data.
+ *
+ ******************************************************************************/
+
 
 /*! ZDICT_trainFromBuffer():
  *  Train a dictionary from an array of samples.
@@ -61,9 +200,64 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCap
                                     const void* samplesBuffer,
                                     const size_t* samplesSizes, unsigned nbSamples);
 
+typedef struct {
+    int      compressionLevel;   /*< optimize for a specific zstd compression level; 0 means default */
+    unsigned notificationLevel;  /*< Write log to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
+    unsigned dictID;             /*< force dictID value; 0 means auto mode (32-bits random value)
+                                  *   NOTE: The zstd format reserves some dictionary IDs for future use.
+                                  *         You may use them in private settings, but be warned that they
+                                  *         may be used by zstd in a public dictionary registry in the future.
+                                  *         These dictionary IDs are:
+                                  *           - low range  : <= 32767
+                                  *           - high range : >= (2^31)
+                                  */
+} ZDICT_params_t;
+
+/*! ZDICT_finalizeDictionary():
+ * Given a custom content as a basis for dictionary, and a set of samples,
+ * finalize dictionary by adding headers and statistics according to the zstd
+ * dictionary format.
+ *
+ * Samples must be stored concatenated in a flat buffer `samplesBuffer`,
+ * supplied with an array of sizes `samplesSizes`, providing the size of each
+ * sample in order. The samples are used to construct the statistics, so they
+ * should be representative of what you will compress with this dictionary.
+ *
+ * The compression level can be set in `parameters`. You should pass the
+ * compression level you expect to use in production. The statistics for each
+ * compression level differ, so tuning the dictionary for the compression level
+ * can help quite a bit.
+ *
+ * You can set an explicit dictionary ID in `parameters`, or allow us to pick
+ * a random dictionary ID for you, but we can't guarantee no collisions.
+ *
+ * The dstDictBuffer and the dictContent may overlap, and the content will be
+ * appended to the end of the header. If the header + the content doesn't fit in
+ * maxDictSize the beginning of the content is truncated to make room, since it
+ * is presumed that the most profitable content is at the end of the dictionary,
+ * since that is the cheapest to reference.
+ *
+ * `dictContentSize` must be >= ZDICT_CONTENTSIZE_MIN bytes.
+ * `maxDictSize` must be >= max(dictContentSize, ZSTD_DICTSIZE_MIN).
+ *
+ * @return: size of dictionary stored into `dstDictBuffer` (<= `maxDictSize`),
+ *          or an error code, which can be tested by ZDICT_isError().
+ * Note: ZDICT_finalizeDictionary() will push notifications into stderr if
+ *       instructed to, using notificationLevel>0.
+ * NOTE: This function currently may fail in several edge cases including:
+ *         * Not enough samples
+ *         * Samples are uncompressible
+ *         * Samples are all exactly the same
+ */
+ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dstDictBuffer, size_t maxDictSize,
+                                const void* dictContent, size_t dictContentSize,
+                                const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
+                                ZDICT_params_t parameters);
+
 
 /*======   Helper functions   ======*/
 ZDICTLIB_API unsigned ZDICT_getDictID(const void* dictBuffer, size_t dictSize);  /**< extracts dictID; @return zero if error (not a valid dictionary) */
+ZDICTLIB_API size_t ZDICT_getDictHeaderSize(const void* dictBuffer, size_t dictSize);  /* returns dict header size; returns a ZSTD error code on failure */
 ZDICTLIB_API unsigned ZDICT_isError(size_t errorCode);
 ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode);
 
@@ -78,11 +272,8 @@ ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode);
  * Use them only in association with static linking.
  * ==================================================================================== */
 
-typedef struct {
-    int      compressionLevel;   /* optimize for a specific zstd compression level; 0 means default */
-    unsigned notificationLevel;  /* Write log to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
-    unsigned dictID;             /* force dictID value; 0 means auto mode (32-bits random value) */
-} ZDICT_params_t;
+#define ZDICT_CONTENTSIZE_MIN 128
+#define ZDICT_DICTSIZE_MIN    256
 
 /*! ZDICT_cover_params_t:
  *  k and d are the only required parameters.
@@ -198,28 +389,6 @@ ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_fastCover(void* dictBuffer,
                     const size_t* samplesSizes, unsigned nbSamples,
                     ZDICT_fastCover_params_t* parameters);
 
-/*! ZDICT_finalizeDictionary():
- * Given a custom content as a basis for dictionary, and a set of samples,
- * finalize dictionary by adding headers and statistics.
- *
- * Samples must be stored concatenated in a flat buffer `samplesBuffer`,
- * supplied with an array of sizes `samplesSizes`, providing the size of each sample in order.
- *
- * dictContentSize must be >= ZDICT_CONTENTSIZE_MIN bytes.
- * maxDictSize must be >= dictContentSize, and must be >= ZDICT_DICTSIZE_MIN bytes.
- *
- * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`),
- *          or an error code, which can be tested by ZDICT_isError().
- * Note: ZDICT_finalizeDictionary() will push notifications into stderr if instructed to, using notificationLevel>0.
- * Note 2: dictBuffer and dictContent can overlap
- */
-#define ZDICT_CONTENTSIZE_MIN 128
-#define ZDICT_DICTSIZE_MIN    256
-ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBufferCapacity,
-                                const void* dictContent, size_t dictContentSize,
-                                const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
-                                ZDICT_params_t parameters);
-
 typedef struct {
     unsigned selectivityLevel;   /* 0 means default; larger => select more => larger dictionary */
     ZDICT_params_t zParams;
@@ -241,10 +410,11 @@ typedef struct {
  *  Note: ZDICT_trainFromBuffer_legacy() will send notifications into stderr if instructed to, using notificationLevel>0.
  */
 ZDICTLIB_API size_t ZDICT_trainFromBuffer_legacy(
-    void *dictBuffer, size_t dictBufferCapacity,
-    const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples,
+    void* dictBuffer, size_t dictBufferCapacity,
+    const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
     ZDICT_legacy_params_t parameters);
 
+
 /* Deprecation warnings */
 /* It is generally possible to disable deprecation warnings from compiler,
    for example with -Wno-deprecated-declarations for gcc
@@ -256,7 +426,7 @@ ZDICTLIB_API size_t ZDICT_trainFromBuffer_legacy(
 #  define ZDICT_GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
 #  if defined (__cplusplus) && (__cplusplus >= 201402) /* C++14 or greater */
 #    define ZDICT_DEPRECATED(message) [[deprecated(message)]] ZDICTLIB_API
-#  elif (ZDICT_GCC_VERSION >= 405) || defined(__clang__)
+#  elif defined(__clang__) || (ZDICT_GCC_VERSION >= 405)
 #    define ZDICT_DEPRECATED(message) ZDICTLIB_API __attribute__((deprecated(message)))
 #  elif (ZDICT_GCC_VERSION >= 301)
 #    define ZDICT_DEPRECATED(message) ZDICTLIB_API __attribute__((deprecated))