Adding targetCBlockSize param

diff --git a/doc/zstd_manual.html b/doc/zstd_manual.html
index 024c10e630f4e17e179acd09d6dafb0037715d85..c38c1f60cc3e70d2655026a365616ae7d00669d8 100644 (file)
--- a/doc/zstd_manual.html
+++ b/doc/zstd_manual.html
@@ -318,6 +318,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);
      * ZSTD_c_forceMaxWindow
      * ZSTD_c_forceAttachDict
      * ZSTD_c_literalCompressionMode
+     * ZSTD_c_targetCBlockSize
      * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
      * note : never ever use experimentalParam? names directly;
      *        also, the enums values themselves are unstable and can still change.
@@ -327,6 +328,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);
      ZSTD_c_experimentalParam3=1000,
      ZSTD_c_experimentalParam4=1001,
      ZSTD_c_experimentalParam5=1002,
+     ZSTD_c_experimentalParam6=1003,
 } ZSTD_cParameter;
 </b></pre><BR>
 <pre><b>typedef struct {

diff --git a/lib/compress/zstd_compress.c b/lib/compress/zstd_compress.c
index b4feaef83f75a5d6df74f7465cd11a5172089c90..bf173db5dd0d5e3f19b15231dbc8ea0d66b881c8 100644 (file)
--- a/lib/compress/zstd_compress.c
+++ b/lib/compress/zstd_compress.c
@@ -385,6 +385,11 @@ ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
         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;
@@ -452,6 +457,7 @@ static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
     case ZSTD_c_ldmHashRateLog:
     case ZSTD_c_forceAttachDict:
     case ZSTD_c_literalCompressionMode:
+    case ZSTD_c_targetCBlockSize:
     default:
         return 0;
     }
@@ -497,6 +503,7 @@ size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
     case ZSTD_c_ldmHashLog:
     case ZSTD_c_ldmMinMatch:
     case ZSTD_c_ldmBucketSizeLog:
+    case ZSTD_c_targetCBlockSize:
         break;
 
     default: RETURN_ERROR(parameter_unsupported);
@@ -671,6 +678,12 @@ size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
         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");
     }
 }
@@ -773,6 +786,9 @@ size_t ZSTD_CCtxParams_getParameter(
     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;

diff --git a/lib/compress/zstd_compress_internal.h b/lib/compress/zstd_compress_internal.h
index ee9d9e31a23fbf95967f1b1f03b21573dece6824..8a9001c413b36379a14700c32c8e47f6941fffe3 100644 (file)
--- a/lib/compress/zstd_compress_internal.h
+++ b/lib/compress/zstd_compress_internal.h
@@ -194,6 +194,9 @@ struct ZSTD_CCtx_params_s {
     int compressionLevel;
     int forceWindow;           /* force back-references to respect limit of
                                 * 1<<wLog, even for dictionary */
+    size_t targetCBlockSize;   /* Tries to fit compressed block size to be around targetCBlockSize.
+                                * No target when targetCBlockSize == 0.
+                                * There is no guarantee on compressed block size */
 
     ZSTD_dictAttachPref_e attachDictPref;
     ZSTD_literalCompressionMode_e literalCompressionMode;

diff --git a/lib/zstd.h b/lib/zstd.h
index 0cc72b4e4c1d6e1a5775c980aed76220f09c1728..922a2dfe7e88bfff8365596c9bf332333fb72163 100644 (file)
--- a/lib/zstd.h
+++ b/lib/zstd.h
@@ -380,6 +380,7 @@ typedef enum {
      * ZSTD_c_forceMaxWindow
      * ZSTD_c_forceAttachDict
      * ZSTD_c_literalCompressionMode
+     * ZSTD_c_targetCBlockSize
      * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
      * note : never ever use experimentalParam? names directly;
      *        also, the enums values themselves are unstable and can still change.
@@ -389,6 +390,7 @@ typedef enum {
      ZSTD_c_experimentalParam3=1000,
      ZSTD_c_experimentalParam4=1001,
      ZSTD_c_experimentalParam5=1002,
+     ZSTD_c_experimentalParam6=1003,
 } ZSTD_cParameter;
 
 typedef struct {
@@ -1053,6 +1055,10 @@ ZSTDLIB_API size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
 #define ZSTD_LDM_HASHRATELOG_MIN     0
 #define ZSTD_LDM_HASHRATELOG_MAX (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)
 
+/* Advanced parameter bounds */
+#define ZSTD_TARGETCBLOCKSIZE_MIN   64
+#define ZSTD_TARGETCBLOCKSIZE_MAX   ZSTD_BLOCKSIZE_MAX
+
 /* internal */
 #define ZSTD_HASHLOG3_MAX           17
 
@@ -1425,6 +1431,11 @@ ZSTDLIB_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const void* pre
  */
 #define ZSTD_c_literalCompressionMode ZSTD_c_experimentalParam5
 
+/* Tries to fit compressed block size to be around targetCBlockSize.
+ * No target when targetCBlockSize == 0.
+ * There is no guarantee on compressed block size (default:0) */
+#define ZSTD_c_targetCBlockSize ZSTD_c_experimentalParam6
+
 /*! ZSTD_CCtx_getParameter() :
  *  Get the requested compression parameter value, selected by enum ZSTD_cParameter,
  *  and store it into int* value.

diff --git a/programs/fileio.c b/programs/fileio.c
index 3c45a986434edcb82b0ce2a560f1ec7de2b48b2e..e55cabe09b188bc605a2ee5912140af431534a6a 100644 (file)
--- a/programs/fileio.c
+++ b/programs/fileio.c
@@ -304,6 +304,7 @@ struct FIO_prefs_s {
     int ldmMinMatch;
     int ldmBucketSizeLog;
     int ldmHashRateLog;
+    size_t targetCBlockSize;
     ZSTD_literalCompressionMode_e literalCompressionMode;
 
     /* IO preferences */
@@ -348,6 +349,7 @@ FIO_prefs_t* FIO_createPreferences(void)
     ret->ldmMinMatch = 0;
     ret->ldmBucketSizeLog = FIO_LDM_PARAM_NOTSET;
     ret->ldmHashRateLog = FIO_LDM_PARAM_NOTSET;
+    ret->targetCBlockSize = 0;
     ret->literalCompressionMode = ZSTD_lcm_auto;
     return ret;
 }
@@ -416,6 +418,10 @@ void FIO_setRsyncable(FIO_prefs_t* const prefs, int rsyncable) {
     prefs->rsyncable = rsyncable;
 }
 
+void FIO_setTargetCBlockSize(FIO_prefs_t* const prefs, size_t targetCBlockSize) {
+    prefs->targetCBlockSize = targetCBlockSize;
+}
+
 void FIO_setLiteralCompressionMode(
         FIO_prefs_t* const prefs,
         ZSTD_literalCompressionMode_e mode) {
@@ -656,6 +662,8 @@ static cRess_t FIO_createCResources(FIO_prefs_t* const prefs,
         CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_checksumFlag, prefs->checksumFlag) );
         /* compression level */
         CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_compressionLevel, cLevel) );
+        /* max compressed block size */
+        CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_targetCBlockSize, (int)prefs->targetCBlockSize) );
         /* long distance matching */
         CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_enableLongDistanceMatching, prefs->ldmFlag) );
         CHECK( ZSTD_CCtx_setParameter(ress.cctx, ZSTD_c_ldmHashLog, prefs->ldmHashLog) );

diff --git a/programs/fileio.h b/programs/fileio.h
index e46633752f4afef3a725a92c8e275ffa24c072da..311f8c0e1f087af132964f1c6a7177626e8ee37d 100644 (file)
--- a/programs/fileio.h
+++ b/programs/fileio.h
@@ -71,6 +71,7 @@ void FIO_setOverlapLog(FIO_prefs_t* const prefs, int overlapLog);
 void FIO_setRemoveSrcFile(FIO_prefs_t* const prefs, unsigned flag);
 void FIO_setSparseWrite(FIO_prefs_t* const prefs, unsigned sparse);  /**< 0: no sparse; 1: disable on stdout; 2: always enabled */
 void FIO_setRsyncable(FIO_prefs_t* const prefs, int rsyncable);
+void FIO_setTargetCBlockSize(FIO_prefs_t* const prefs, size_t targetCBlockSize);
 void FIO_setLiteralCompressionMode(
         FIO_prefs_t* const prefs,
         ZSTD_literalCompressionMode_e mode);

diff --git a/programs/zstdcli.c b/programs/zstdcli.c
index c0dd925cefa921108c238090a08eeb2b6ebe5fab..1d2e1ec73897f7386b991c98522aa0736674777c 100644 (file)
--- a/programs/zstdcli.c
+++ b/programs/zstdcli.c
@@ -141,6 +141,7 @@ static int usage_advanced(const char* programName)
     DISPLAY( "--long[=#]: enable long distance matching with given window log (default: %u)\n", g_defaultMaxWindowLog);
     DISPLAY( "--fast[=#]: switch to ultra fast compression level (default: %u)\n", 1);
     DISPLAY( "--adapt : dynamically adapt compression level to I/O conditions \n");
+    DISPLAY( "--target-compressed-block-size=# : make compressed block near targeted size \n");
 #ifdef ZSTD_MULTITHREAD
     DISPLAY( " -T#    : spawns # compression threads (default: 1, 0==# cores) \n");
     DISPLAY( " -B#    : select size of each job (default: 0==automatic) \n");
@@ -555,6 +556,7 @@ int main(int argCount, const char* argv[])
     const char* suffix = ZSTD_EXTENSION;
     unsigned maxDictSize = g_defaultMaxDictSize;
     unsigned dictID = 0;
+    size_t targetCBlockSize = 0;
     int dictCLevel = g_defaultDictCLevel;
     unsigned dictSelect = g_defaultSelectivityLevel;
 #ifdef UTIL_HAS_CREATEFILELIST
@@ -711,6 +713,7 @@ int main(int argCount, const char* argv[])
                     if (longCommandWArg(&argument, "--maxdict=")) { maxDictSize = readU32FromChar(&argument); continue; }
                     if (longCommandWArg(&argument, "--dictID=")) { dictID = readU32FromChar(&argument); continue; }
                     if (longCommandWArg(&argument, "--zstd=")) { if (!parseCompressionParameters(argument, &compressionParams)) CLEAN_RETURN(badusage(programName)); continue; }
+                    if (longCommandWArg(&argument, "--target-compressed-block-size=")) { targetCBlockSize = readU32FromChar(&argument); continue; }
                     if (longCommandWArg(&argument, "--long")) {
                         unsigned ldmWindowLog = 0;
                         ldmFlag = 1;
@@ -1115,6 +1118,7 @@ int main(int argCount, const char* argv[])
         FIO_setAdaptMin(prefs, adaptMin);
         FIO_setAdaptMax(prefs, adaptMax);
         FIO_setRsyncable(prefs, rsyncable);
+        FIO_setTargetCBlockSize(prefs, targetCBlockSize);
         FIO_setLiteralCompressionMode(prefs, literalCompressionMode);
         if (adaptMin > cLevel) cLevel = adaptMin;
         if (adaptMax < cLevel) cLevel = adaptMax;
@@ -1124,7 +1128,7 @@ int main(int argCount, const char* argv[])
         else
           operationResult = FIO_compressMultipleFilenames(prefs, filenameTable, filenameIdx, outFileName, suffix, dictFileName, cLevel, compressionParams);
 #else
-        (void)suffix; (void)adapt; (void)rsyncable; (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; /* not used when ZSTD_NOCOMPRESS set */
+        (void)suffix; (void)adapt; (void)rsyncable; (void)ultra; (void)cLevel; (void)ldmFlag; (void)literalCompressionMode; (void)targetCBlockSize; /* not used when ZSTD_NOCOMPRESS set */
         DISPLAY("Compression not supported \n");
 #endif
     } else {  /* decompression or test */