xref: /freebsd/sys/contrib/zstd/lib/dictBuilder/zdict.c (revision 63a938566d524836885917d95bd491aa4400b181)
1 /*
2  * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
3  * All rights reserved.
4  *
5  * This source code is licensed under both the BSD-style license (found in the
6  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7  * in the COPYING file in the root directory of this source tree).
8  * You may select, at your option, one of the above-listed licenses.
9  */
10 
11 
12 /*-**************************************
13 *  Tuning parameters
14 ****************************************/
15 #define MINRATIO 4   /* minimum nb of apparition to be selected in dictionary */
16 #define ZDICT_MAX_SAMPLES_SIZE (2000U << 20)
17 #define ZDICT_MIN_SAMPLES_SIZE (ZDICT_CONTENTSIZE_MIN * MINRATIO)
18 
19 
20 /*-**************************************
21 *  Compiler Options
22 ****************************************/
23 /* Unix Large Files support (>4GB) */
24 #define _FILE_OFFSET_BITS 64
25 #if (defined(__sun__) && (!defined(__LP64__)))   /* Sun Solaris 32-bits requires specific definitions */
26 #  define _LARGEFILE_SOURCE
27 #elif ! defined(__LP64__)                        /* No point defining Large file for 64 bit */
28 #  define _LARGEFILE64_SOURCE
29 #endif
30 
31 
32 /*-*************************************
33 *  Dependencies
34 ***************************************/
35 #include <stdlib.h>        /* malloc, free */
36 #include <string.h>        /* memset */
37 #include <stdio.h>         /* fprintf, fopen, ftello64 */
38 #include <time.h>          /* clock */
39 
40 #include "mem.h"           /* read */
41 #include "fse.h"           /* FSE_normalizeCount, FSE_writeNCount */
42 #define HUF_STATIC_LINKING_ONLY
43 #include "huf.h"           /* HUF_buildCTable, HUF_writeCTable */
44 #include "zstd_internal.h" /* includes zstd.h */
45 #include "xxhash.h"        /* XXH64 */
46 #include "divsufsort.h"
47 #ifndef ZDICT_STATIC_LINKING_ONLY
48 #  define ZDICT_STATIC_LINKING_ONLY
49 #endif
50 #include "zdict.h"
51 
52 
53 /*-*************************************
54 *  Constants
55 ***************************************/
56 #define KB *(1 <<10)
57 #define MB *(1 <<20)
58 #define GB *(1U<<30)
59 
60 #define DICTLISTSIZE_DEFAULT 10000
61 
62 #define NOISELENGTH 32
63 
64 static const int g_compressionLevel_default = 3;
65 static const U32 g_selectivity_default = 9;
66 
67 
68 /*-*************************************
69 *  Console display
70 ***************************************/
71 #define DISPLAY(...)         { fprintf(stderr, __VA_ARGS__); fflush( stderr ); }
72 #define DISPLAYLEVEL(l, ...) if (notificationLevel>=l) { DISPLAY(__VA_ARGS__); }    /* 0 : no display;   1: errors;   2: default;  3: details;  4: debug */
73 
74 static clock_t ZDICT_clockSpan(clock_t nPrevious) { return clock() - nPrevious; }
75 
76 static void ZDICT_printHex(const void* ptr, size_t length)
77 {
78     const BYTE* const b = (const BYTE*)ptr;
79     size_t u;
80     for (u=0; u<length; u++) {
81         BYTE c = b[u];
82         if (c<32 || c>126) c = '.';   /* non-printable char */
83         DISPLAY("%c", c);
84     }
85 }
86 
87 
88 /*-********************************************************
89 *  Helper functions
90 **********************************************************/
91 unsigned ZDICT_isError(size_t errorCode) { return ERR_isError(errorCode); }
92 
93 const char* ZDICT_getErrorName(size_t errorCode) { return ERR_getErrorName(errorCode); }
94 
95 unsigned ZDICT_getDictID(const void* dictBuffer, size_t dictSize)
96 {
97     if (dictSize < 8) return 0;
98     if (MEM_readLE32(dictBuffer) != ZSTD_MAGIC_DICTIONARY) return 0;
99     return MEM_readLE32((const char*)dictBuffer + 4);
100 }
101 
102 
103 /*-********************************************************
104 *  Dictionary training functions
105 **********************************************************/
106 static unsigned ZDICT_NbCommonBytes (size_t val)
107 {
108     if (MEM_isLittleEndian()) {
109         if (MEM_64bits()) {
110 #       if defined(_MSC_VER) && defined(_WIN64)
111             unsigned long r = 0;
112             _BitScanForward64( &r, (U64)val );
113             return (unsigned)(r>>3);
114 #       elif defined(__GNUC__) && (__GNUC__ >= 3)
115             return (__builtin_ctzll((U64)val) >> 3);
116 #       else
117             static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2, 0, 3, 1, 3, 1, 4, 2, 7, 0, 2, 3, 6, 1, 5, 3, 5, 1, 3, 4, 4, 2, 5, 6, 7, 7, 0, 1, 2, 3, 3, 4, 6, 2, 6, 5, 5, 3, 4, 5, 6, 7, 1, 2, 4, 6, 4, 4, 5, 7, 2, 6, 5, 7, 6, 7, 7 };
118             return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
119 #       endif
120         } else { /* 32 bits */
121 #       if defined(_MSC_VER)
122             unsigned long r=0;
123             _BitScanForward( &r, (U32)val );
124             return (unsigned)(r>>3);
125 #       elif defined(__GNUC__) && (__GNUC__ >= 3)
126             return (__builtin_ctz((U32)val) >> 3);
127 #       else
128             static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0, 3, 2, 2, 1, 3, 2, 0, 1, 3, 3, 1, 2, 2, 2, 2, 0, 3, 1, 2, 0, 1, 0, 1, 1 };
129             return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27];
130 #       endif
131         }
132     } else {  /* Big Endian CPU */
133         if (MEM_64bits()) {
134 #       if defined(_MSC_VER) && defined(_WIN64)
135             unsigned long r = 0;
136             _BitScanReverse64( &r, val );
137             return (unsigned)(r>>3);
138 #       elif defined(__GNUC__) && (__GNUC__ >= 3)
139             return (__builtin_clzll(val) >> 3);
140 #       else
141             unsigned r;
142             const unsigned n32 = sizeof(size_t)*4;   /* calculate this way due to compiler complaining in 32-bits mode */
143             if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; }
144             if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
145             r += (!val);
146             return r;
147 #       endif
148         } else { /* 32 bits */
149 #       if defined(_MSC_VER)
150             unsigned long r = 0;
151             _BitScanReverse( &r, (unsigned long)val );
152             return (unsigned)(r>>3);
153 #       elif defined(__GNUC__) && (__GNUC__ >= 3)
154             return (__builtin_clz((U32)val) >> 3);
155 #       else
156             unsigned r;
157             if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
158             r += (!val);
159             return r;
160 #       endif
161     }   }
162 }
163 
164 
165 /*! ZDICT_count() :
166     Count the nb of common bytes between 2 pointers.
167     Note : this function presumes end of buffer followed by noisy guard band.
168 */
169 static size_t ZDICT_count(const void* pIn, const void* pMatch)
170 {
171     const char* const pStart = (const char*)pIn;
172     for (;;) {
173         size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
174         if (!diff) {
175             pIn = (const char*)pIn+sizeof(size_t);
176             pMatch = (const char*)pMatch+sizeof(size_t);
177             continue;
178         }
179         pIn = (const char*)pIn+ZDICT_NbCommonBytes(diff);
180         return (size_t)((const char*)pIn - pStart);
181     }
182 }
183 
184 
185 typedef struct {
186     U32 pos;
187     U32 length;
188     U32 savings;
189 } dictItem;
190 
191 static void ZDICT_initDictItem(dictItem* d)
192 {
193     d->pos = 1;
194     d->length = 0;
195     d->savings = (U32)(-1);
196 }
197 
198 
199 #define LLIMIT 64          /* heuristic determined experimentally */
200 #define MINMATCHLENGTH 7   /* heuristic determined experimentally */
201 static dictItem ZDICT_analyzePos(
202                        BYTE* doneMarks,
203                        const int* suffix, U32 start,
204                        const void* buffer, U32 minRatio, U32 notificationLevel)
205 {
206     U32 lengthList[LLIMIT] = {0};
207     U32 cumulLength[LLIMIT] = {0};
208     U32 savings[LLIMIT] = {0};
209     const BYTE* b = (const BYTE*)buffer;
210     size_t length;
211     size_t maxLength = LLIMIT;
212     size_t pos = suffix[start];
213     U32 end = start;
214     dictItem solution;
215 
216     /* init */
217     memset(&solution, 0, sizeof(solution));
218     doneMarks[pos] = 1;
219 
220     /* trivial repetition cases */
221     if ( (MEM_read16(b+pos+0) == MEM_read16(b+pos+2))
222        ||(MEM_read16(b+pos+1) == MEM_read16(b+pos+3))
223        ||(MEM_read16(b+pos+2) == MEM_read16(b+pos+4)) ) {
224         /* skip and mark segment */
225         U16 u16 = MEM_read16(b+pos+4);
226         U32 u, e = 6;
227         while (MEM_read16(b+pos+e) == u16) e+=2 ;
228         if (b[pos+e] == b[pos+e-1]) e++;
229         for (u=1; u<e; u++)
230             doneMarks[pos+u] = 1;
231         return solution;
232     }
233 
234     /* look forward */
235     do {
236         end++;
237         length = ZDICT_count(b + pos, b + suffix[end]);
238     } while (length >=MINMATCHLENGTH);
239 
240     /* look backward */
241     do {
242         length = ZDICT_count(b + pos, b + *(suffix+start-1));
243         if (length >=MINMATCHLENGTH) start--;
244     } while(length >= MINMATCHLENGTH);
245 
246     /* exit if not found a minimum nb of repetitions */
247     if (end-start < minRatio) {
248         U32 idx;
249         for(idx=start; idx<end; idx++)
250             doneMarks[suffix[idx]] = 1;
251         return solution;
252     }
253 
254     {   int i;
255         U32 searchLength;
256         U32 refinedStart = start;
257         U32 refinedEnd = end;
258 
259         DISPLAYLEVEL(4, "\n");
260         DISPLAYLEVEL(4, "found %3u matches of length >= %i at pos %7u  ", (U32)(end-start), MINMATCHLENGTH, (U32)pos);
261         DISPLAYLEVEL(4, "\n");
262 
263         for (searchLength = MINMATCHLENGTH ; ; searchLength++) {
264             BYTE currentChar = 0;
265             U32 currentCount = 0;
266             U32 currentID = refinedStart;
267             U32 id;
268             U32 selectedCount = 0;
269             U32 selectedID = currentID;
270             for (id =refinedStart; id < refinedEnd; id++) {
271                 if (b[ suffix[id] + searchLength] != currentChar) {
272                     if (currentCount > selectedCount) {
273                         selectedCount = currentCount;
274                         selectedID = currentID;
275                     }
276                     currentID = id;
277                     currentChar = b[ suffix[id] + searchLength];
278                     currentCount = 0;
279                 }
280                 currentCount ++;
281             }
282             if (currentCount > selectedCount) {  /* for last */
283                 selectedCount = currentCount;
284                 selectedID = currentID;
285             }
286 
287             if (selectedCount < minRatio)
288                 break;
289             refinedStart = selectedID;
290             refinedEnd = refinedStart + selectedCount;
291         }
292 
293         /* evaluate gain based on new ref */
294         start = refinedStart;
295         pos = suffix[refinedStart];
296         end = start;
297         memset(lengthList, 0, sizeof(lengthList));
298 
299         /* look forward */
300         do {
301             end++;
302             length = ZDICT_count(b + pos, b + suffix[end]);
303             if (length >= LLIMIT) length = LLIMIT-1;
304             lengthList[length]++;
305         } while (length >=MINMATCHLENGTH);
306 
307         /* look backward */
308         length = MINMATCHLENGTH;
309         while ((length >= MINMATCHLENGTH) & (start > 0)) {
310             length = ZDICT_count(b + pos, b + suffix[start - 1]);
311             if (length >= LLIMIT) length = LLIMIT - 1;
312             lengthList[length]++;
313             if (length >= MINMATCHLENGTH) start--;
314         }
315 
316         /* largest useful length */
317         memset(cumulLength, 0, sizeof(cumulLength));
318         cumulLength[maxLength-1] = lengthList[maxLength-1];
319         for (i=(int)(maxLength-2); i>=0; i--)
320             cumulLength[i] = cumulLength[i+1] + lengthList[i];
321 
322         for (i=LLIMIT-1; i>=MINMATCHLENGTH; i--) if (cumulLength[i]>=minRatio) break;
323         maxLength = i;
324 
325         /* reduce maxLength in case of final into repetitive data */
326         {   U32 l = (U32)maxLength;
327             BYTE const c = b[pos + maxLength-1];
328             while (b[pos+l-2]==c) l--;
329             maxLength = l;
330         }
331         if (maxLength < MINMATCHLENGTH) return solution;   /* skip : no long-enough solution */
332 
333         /* calculate savings */
334         savings[5] = 0;
335         for (i=MINMATCHLENGTH; i<=(int)maxLength; i++)
336             savings[i] = savings[i-1] + (lengthList[i] * (i-3));
337 
338         DISPLAYLEVEL(4, "Selected ref at position %u, of length %u : saves %u (ratio: %.2f)  \n",
339                      (U32)pos, (U32)maxLength, savings[maxLength], (double)savings[maxLength] / maxLength);
340 
341         solution.pos = (U32)pos;
342         solution.length = (U32)maxLength;
343         solution.savings = savings[maxLength];
344 
345         /* mark positions done */
346         {   U32 id;
347             for (id=start; id<end; id++) {
348                 U32 p, pEnd;
349                 U32 const testedPos = suffix[id];
350                 if (testedPos == pos)
351                     length = solution.length;
352                 else {
353                     length = ZDICT_count(b+pos, b+testedPos);
354                     if (length > solution.length) length = solution.length;
355                 }
356                 pEnd = (U32)(testedPos + length);
357                 for (p=testedPos; p<pEnd; p++)
358                     doneMarks[p] = 1;
359     }   }   }
360 
361     return solution;
362 }
363 
364 
365 static int isIncluded(const void* in, const void* container, size_t length)
366 {
367     const char* const ip = (const char*) in;
368     const char* const into = (const char*) container;
369     size_t u;
370 
371     for (u=0; u<length; u++) {  /* works because end of buffer is a noisy guard band */
372         if (ip[u] != into[u]) break;
373     }
374 
375     return u==length;
376 }
377 
378 /*! ZDICT_tryMerge() :
379     check if dictItem can be merged, do it if possible
380     @return : id of destination elt, 0 if not merged
381 */
382 static U32 ZDICT_tryMerge(dictItem* table, dictItem elt, U32 eltNbToSkip, const void* buffer)
383 {
384     const U32 tableSize = table->pos;
385     const U32 eltEnd = elt.pos + elt.length;
386     const char* const buf = (const char*) buffer;
387 
388     /* tail overlap */
389     U32 u; for (u=1; u<tableSize; u++) {
390         if (u==eltNbToSkip) continue;
391         if ((table[u].pos > elt.pos) && (table[u].pos <= eltEnd)) {  /* overlap, existing > new */
392             /* append */
393             U32 const addedLength = table[u].pos - elt.pos;
394             table[u].length += addedLength;
395             table[u].pos = elt.pos;
396             table[u].savings += elt.savings * addedLength / elt.length;   /* rough approx */
397             table[u].savings += elt.length / 8;    /* rough approx bonus */
398             elt = table[u];
399             /* sort : improve rank */
400             while ((u>1) && (table[u-1].savings < elt.savings))
401             table[u] = table[u-1], u--;
402             table[u] = elt;
403             return u;
404     }   }
405 
406     /* front overlap */
407     for (u=1; u<tableSize; u++) {
408         if (u==eltNbToSkip) continue;
409 
410         if ((table[u].pos + table[u].length >= elt.pos) && (table[u].pos < elt.pos)) {  /* overlap, existing < new */
411             /* append */
412             int const addedLength = (int)eltEnd - (table[u].pos + table[u].length);
413             table[u].savings += elt.length / 8;    /* rough approx bonus */
414             if (addedLength > 0) {   /* otherwise, elt fully included into existing */
415                 table[u].length += addedLength;
416                 table[u].savings += elt.savings * addedLength / elt.length;   /* rough approx */
417             }
418             /* sort : improve rank */
419             elt = table[u];
420             while ((u>1) && (table[u-1].savings < elt.savings))
421                 table[u] = table[u-1], u--;
422             table[u] = elt;
423             return u;
424         }
425 
426         if (MEM_read64(buf + table[u].pos) == MEM_read64(buf + elt.pos + 1)) {
427             if (isIncluded(buf + table[u].pos, buf + elt.pos + 1, table[u].length)) {
428                 size_t const addedLength = MAX( (int)elt.length - (int)table[u].length , 1 );
429                 table[u].pos = elt.pos;
430                 table[u].savings += (U32)(elt.savings * addedLength / elt.length);
431                 table[u].length = MIN(elt.length, table[u].length + 1);
432                 return u;
433             }
434         }
435     }
436 
437     return 0;
438 }
439 
440 
441 static void ZDICT_removeDictItem(dictItem* table, U32 id)
442 {
443     /* convention : table[0].pos stores nb of elts */
444     U32 const max = table[0].pos;
445     U32 u;
446     if (!id) return;   /* protection, should never happen */
447     for (u=id; u<max-1; u++)
448         table[u] = table[u+1];
449     table->pos--;
450 }
451 
452 
453 static void ZDICT_insertDictItem(dictItem* table, U32 maxSize, dictItem elt, const void* buffer)
454 {
455     /* merge if possible */
456     U32 mergeId = ZDICT_tryMerge(table, elt, 0, buffer);
457     if (mergeId) {
458         U32 newMerge = 1;
459         while (newMerge) {
460             newMerge = ZDICT_tryMerge(table, table[mergeId], mergeId, buffer);
461             if (newMerge) ZDICT_removeDictItem(table, mergeId);
462             mergeId = newMerge;
463         }
464         return;
465     }
466 
467     /* insert */
468     {   U32 current;
469         U32 nextElt = table->pos;
470         if (nextElt >= maxSize) nextElt = maxSize-1;
471         current = nextElt-1;
472         while (table[current].savings < elt.savings) {
473             table[current+1] = table[current];
474             current--;
475         }
476         table[current+1] = elt;
477         table->pos = nextElt+1;
478     }
479 }
480 
481 
482 static U32 ZDICT_dictSize(const dictItem* dictList)
483 {
484     U32 u, dictSize = 0;
485     for (u=1; u<dictList[0].pos; u++)
486         dictSize += dictList[u].length;
487     return dictSize;
488 }
489 
490 
491 static size_t ZDICT_trainBuffer_legacy(dictItem* dictList, U32 dictListSize,
492                             const void* const buffer, size_t bufferSize,   /* buffer must end with noisy guard band */
493                             const size_t* fileSizes, unsigned nbFiles,
494                             U32 minRatio, U32 notificationLevel)
495 {
496     int* const suffix0 = (int*)malloc((bufferSize+2)*sizeof(*suffix0));
497     int* const suffix = suffix0+1;
498     U32* reverseSuffix = (U32*)malloc((bufferSize)*sizeof(*reverseSuffix));
499     BYTE* doneMarks = (BYTE*)malloc((bufferSize+16)*sizeof(*doneMarks));   /* +16 for overflow security */
500     U32* filePos = (U32*)malloc(nbFiles * sizeof(*filePos));
501     size_t result = 0;
502     clock_t displayClock = 0;
503     clock_t const refreshRate = CLOCKS_PER_SEC * 3 / 10;
504 
505 #   define DISPLAYUPDATE(l, ...) if (notificationLevel>=l) { \
506             if (ZDICT_clockSpan(displayClock) > refreshRate)  \
507             { displayClock = clock(); DISPLAY(__VA_ARGS__); \
508             if (notificationLevel>=4) fflush(stderr); } }
509 
510     /* init */
511     DISPLAYLEVEL(2, "\r%70s\r", "");   /* clean display line */
512     if (!suffix0 || !reverseSuffix || !doneMarks || !filePos) {
513         result = ERROR(memory_allocation);
514         goto _cleanup;
515     }
516     if (minRatio < MINRATIO) minRatio = MINRATIO;
517     memset(doneMarks, 0, bufferSize+16);
518 
519     /* limit sample set size (divsufsort limitation)*/
520     if (bufferSize > ZDICT_MAX_SAMPLES_SIZE) DISPLAYLEVEL(3, "sample set too large : reduced to %u MB ...\n", (U32)(ZDICT_MAX_SAMPLES_SIZE>>20));
521     while (bufferSize > ZDICT_MAX_SAMPLES_SIZE) bufferSize -= fileSizes[--nbFiles];
522 
523     /* sort */
524     DISPLAYLEVEL(2, "sorting %u files of total size %u MB ...\n", nbFiles, (U32)(bufferSize>>20));
525     {   int const divSuftSortResult = divsufsort((const unsigned char*)buffer, suffix, (int)bufferSize, 0);
526         if (divSuftSortResult != 0) { result = ERROR(GENERIC); goto _cleanup; }
527     }
528     suffix[bufferSize] = (int)bufferSize;   /* leads into noise */
529     suffix0[0] = (int)bufferSize;           /* leads into noise */
530     /* build reverse suffix sort */
531     {   size_t pos;
532         for (pos=0; pos < bufferSize; pos++)
533             reverseSuffix[suffix[pos]] = (U32)pos;
534         /* note filePos tracks borders between samples.
535            It's not used at this stage, but planned to become useful in a later update */
536         filePos[0] = 0;
537         for (pos=1; pos<nbFiles; pos++)
538             filePos[pos] = (U32)(filePos[pos-1] + fileSizes[pos-1]);
539     }
540 
541     DISPLAYLEVEL(2, "finding patterns ... \n");
542     DISPLAYLEVEL(3, "minimum ratio : %u \n", minRatio);
543 
544     {   U32 cursor; for (cursor=0; cursor < bufferSize; ) {
545             dictItem solution;
546             if (doneMarks[cursor]) { cursor++; continue; }
547             solution = ZDICT_analyzePos(doneMarks, suffix, reverseSuffix[cursor], buffer, minRatio, notificationLevel);
548             if (solution.length==0) { cursor++; continue; }
549             ZDICT_insertDictItem(dictList, dictListSize, solution, buffer);
550             cursor += solution.length;
551             DISPLAYUPDATE(2, "\r%4.2f %% \r", (double)cursor / bufferSize * 100);
552     }   }
553 
554 _cleanup:
555     free(suffix0);
556     free(reverseSuffix);
557     free(doneMarks);
558     free(filePos);
559     return result;
560 }
561 
562 
563 static void ZDICT_fillNoise(void* buffer, size_t length)
564 {
565     unsigned const prime1 = 2654435761U;
566     unsigned const prime2 = 2246822519U;
567     unsigned acc = prime1;
568     size_t p=0;;
569     for (p=0; p<length; p++) {
570         acc *= prime2;
571         ((unsigned char*)buffer)[p] = (unsigned char)(acc >> 21);
572     }
573 }
574 
575 
576 typedef struct
577 {
578     ZSTD_CCtx* ref;
579     ZSTD_CCtx* zc;
580     void* workPlace;   /* must be ZSTD_BLOCKSIZE_MAX allocated */
581 } EStats_ress_t;
582 
583 #define MAXREPOFFSET 1024
584 
585 static void ZDICT_countEStats(EStats_ress_t esr, ZSTD_parameters params,
586                             U32* countLit, U32* offsetcodeCount, U32* matchlengthCount, U32* litlengthCount, U32* repOffsets,
587                             const void* src, size_t srcSize, U32 notificationLevel)
588 {
589     size_t const blockSizeMax = MIN (ZSTD_BLOCKSIZE_MAX, 1 << params.cParams.windowLog);
590     size_t cSize;
591 
592     if (srcSize > blockSizeMax) srcSize = blockSizeMax;   /* protection vs large samples */
593     {  size_t const errorCode = ZSTD_copyCCtx(esr.zc, esr.ref, 0);
594             if (ZSTD_isError(errorCode)) { DISPLAYLEVEL(1, "warning : ZSTD_copyCCtx failed \n"); return; }
595     }
596     cSize = ZSTD_compressBlock(esr.zc, esr.workPlace, ZSTD_BLOCKSIZE_MAX, src, srcSize);
597     if (ZSTD_isError(cSize)) { DISPLAYLEVEL(3, "warning : could not compress sample size %u \n", (U32)srcSize); return; }
598 
599     if (cSize) {  /* if == 0; block is not compressible */
600         const seqStore_t* seqStorePtr = ZSTD_getSeqStore(esr.zc);
601 
602         /* literals stats */
603         {   const BYTE* bytePtr;
604             for(bytePtr = seqStorePtr->litStart; bytePtr < seqStorePtr->lit; bytePtr++)
605                 countLit[*bytePtr]++;
606         }
607 
608         /* seqStats */
609         {   U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
610             ZSTD_seqToCodes(seqStorePtr);
611 
612             {   const BYTE* codePtr = seqStorePtr->ofCode;
613                 U32 u;
614                 for (u=0; u<nbSeq; u++) offsetcodeCount[codePtr[u]]++;
615             }
616 
617             {   const BYTE* codePtr = seqStorePtr->mlCode;
618                 U32 u;
619                 for (u=0; u<nbSeq; u++) matchlengthCount[codePtr[u]]++;
620             }
621 
622             {   const BYTE* codePtr = seqStorePtr->llCode;
623                 U32 u;
624                 for (u=0; u<nbSeq; u++) litlengthCount[codePtr[u]]++;
625             }
626 
627             if (nbSeq >= 2) { /* rep offsets */
628                 const seqDef* const seq = seqStorePtr->sequencesStart;
629                 U32 offset1 = seq[0].offset - 3;
630                 U32 offset2 = seq[1].offset - 3;
631                 if (offset1 >= MAXREPOFFSET) offset1 = 0;
632                 if (offset2 >= MAXREPOFFSET) offset2 = 0;
633                 repOffsets[offset1] += 3;
634                 repOffsets[offset2] += 1;
635     }   }   }
636 }
637 
638 static size_t ZDICT_totalSampleSize(const size_t* fileSizes, unsigned nbFiles)
639 {
640     size_t total=0;
641     unsigned u;
642     for (u=0; u<nbFiles; u++) total += fileSizes[u];
643     return total;
644 }
645 
646 typedef struct { U32 offset; U32 count; } offsetCount_t;
647 
648 static void ZDICT_insertSortCount(offsetCount_t table[ZSTD_REP_NUM+1], U32 val, U32 count)
649 {
650     U32 u;
651     table[ZSTD_REP_NUM].offset = val;
652     table[ZSTD_REP_NUM].count = count;
653     for (u=ZSTD_REP_NUM; u>0; u--) {
654         offsetCount_t tmp;
655         if (table[u-1].count >= table[u].count) break;
656         tmp = table[u-1];
657         table[u-1] = table[u];
658         table[u] = tmp;
659     }
660 }
661 
662 
663 #define OFFCODE_MAX 30  /* only applicable to first block */
664 static size_t ZDICT_analyzeEntropy(void*  dstBuffer, size_t maxDstSize,
665                                    unsigned compressionLevel,
666                              const void*  srcBuffer, const size_t* fileSizes, unsigned nbFiles,
667                              const void* dictBuffer, size_t  dictBufferSize,
668                                    unsigned notificationLevel)
669 {
670     U32 countLit[256];
671     HUF_CREATE_STATIC_CTABLE(hufTable, 255);
672     U32 offcodeCount[OFFCODE_MAX+1];
673     short offcodeNCount[OFFCODE_MAX+1];
674     U32 offcodeMax = ZSTD_highbit32((U32)(dictBufferSize + 128 KB));
675     U32 matchLengthCount[MaxML+1];
676     short matchLengthNCount[MaxML+1];
677     U32 litLengthCount[MaxLL+1];
678     short litLengthNCount[MaxLL+1];
679     U32 repOffset[MAXREPOFFSET];
680     offsetCount_t bestRepOffset[ZSTD_REP_NUM+1];
681     EStats_ress_t esr;
682     ZSTD_parameters params;
683     U32 u, huffLog = 11, Offlog = OffFSELog, mlLog = MLFSELog, llLog = LLFSELog, total;
684     size_t pos = 0, errorCode;
685     size_t eSize = 0;
686     size_t const totalSrcSize = ZDICT_totalSampleSize(fileSizes, nbFiles);
687     size_t const averageSampleSize = totalSrcSize / (nbFiles + !nbFiles);
688     BYTE* dstPtr = (BYTE*)dstBuffer;
689 
690     /* init */
691     esr.ref = ZSTD_createCCtx();
692     esr.zc = ZSTD_createCCtx();
693     esr.workPlace = malloc(ZSTD_BLOCKSIZE_MAX);
694     if (!esr.ref || !esr.zc || !esr.workPlace) {
695         eSize = ERROR(memory_allocation);
696         DISPLAYLEVEL(1, "Not enough memory \n");
697         goto _cleanup;
698     }
699     if (offcodeMax>OFFCODE_MAX) { eSize = ERROR(dictionaryCreation_failed); goto _cleanup; }   /* too large dictionary */
700     for (u=0; u<256; u++) countLit[u] = 1;   /* any character must be described */
701     for (u=0; u<=offcodeMax; u++) offcodeCount[u] = 1;
702     for (u=0; u<=MaxML; u++) matchLengthCount[u] = 1;
703     for (u=0; u<=MaxLL; u++) litLengthCount[u] = 1;
704     memset(repOffset, 0, sizeof(repOffset));
705     repOffset[1] = repOffset[4] = repOffset[8] = 1;
706     memset(bestRepOffset, 0, sizeof(bestRepOffset));
707     if (compressionLevel<=0) compressionLevel = g_compressionLevel_default;
708     params = ZSTD_getParams(compressionLevel, averageSampleSize, dictBufferSize);
709     {   size_t const beginResult = ZSTD_compressBegin_advanced(esr.ref, dictBuffer, dictBufferSize, params, 0);
710         if (ZSTD_isError(beginResult)) {
711             DISPLAYLEVEL(1, "error : ZSTD_compressBegin_advanced() failed : %s \n", ZSTD_getErrorName(beginResult));
712             eSize = ERROR(GENERIC);
713             goto _cleanup;
714     }   }
715 
716     /* collect stats on all files */
717     for (u=0; u<nbFiles; u++) {
718         ZDICT_countEStats(esr, params,
719                           countLit, offcodeCount, matchLengthCount, litLengthCount, repOffset,
720                          (const char*)srcBuffer + pos, fileSizes[u],
721                           notificationLevel);
722         pos += fileSizes[u];
723     }
724 
725     /* analyze */
726     errorCode = HUF_buildCTable (hufTable, countLit, 255, huffLog);
727     if (HUF_isError(errorCode)) {
728         eSize = ERROR(GENERIC);
729         DISPLAYLEVEL(1, "HUF_buildCTable error \n");
730         goto _cleanup;
731     }
732     huffLog = (U32)errorCode;
733 
734     /* looking for most common first offsets */
735     {   U32 offset;
736         for (offset=1; offset<MAXREPOFFSET; offset++)
737             ZDICT_insertSortCount(bestRepOffset, offset, repOffset[offset]);
738     }
739     /* note : the result of this phase should be used to better appreciate the impact on statistics */
740 
741     total=0; for (u=0; u<=offcodeMax; u++) total+=offcodeCount[u];
742     errorCode = FSE_normalizeCount(offcodeNCount, Offlog, offcodeCount, total, offcodeMax);
743     if (FSE_isError(errorCode)) {
744         eSize = ERROR(GENERIC);
745         DISPLAYLEVEL(1, "FSE_normalizeCount error with offcodeCount \n");
746         goto _cleanup;
747     }
748     Offlog = (U32)errorCode;
749 
750     total=0; for (u=0; u<=MaxML; u++) total+=matchLengthCount[u];
751     errorCode = FSE_normalizeCount(matchLengthNCount, mlLog, matchLengthCount, total, MaxML);
752     if (FSE_isError(errorCode)) {
753         eSize = ERROR(GENERIC);
754         DISPLAYLEVEL(1, "FSE_normalizeCount error with matchLengthCount \n");
755         goto _cleanup;
756     }
757     mlLog = (U32)errorCode;
758 
759     total=0; for (u=0; u<=MaxLL; u++) total+=litLengthCount[u];
760     errorCode = FSE_normalizeCount(litLengthNCount, llLog, litLengthCount, total, MaxLL);
761     if (FSE_isError(errorCode)) {
762         eSize = ERROR(GENERIC);
763         DISPLAYLEVEL(1, "FSE_normalizeCount error with litLengthCount \n");
764         goto _cleanup;
765     }
766     llLog = (U32)errorCode;
767 
768     /* write result to buffer */
769     {   size_t const hhSize = HUF_writeCTable(dstPtr, maxDstSize, hufTable, 255, huffLog);
770         if (HUF_isError(hhSize)) {
771             eSize = ERROR(GENERIC);
772             DISPLAYLEVEL(1, "HUF_writeCTable error \n");
773             goto _cleanup;
774         }
775         dstPtr += hhSize;
776         maxDstSize -= hhSize;
777         eSize += hhSize;
778     }
779 
780     {   size_t const ohSize = FSE_writeNCount(dstPtr, maxDstSize, offcodeNCount, OFFCODE_MAX, Offlog);
781         if (FSE_isError(ohSize)) {
782             eSize = ERROR(GENERIC);
783             DISPLAYLEVEL(1, "FSE_writeNCount error with offcodeNCount \n");
784             goto _cleanup;
785         }
786         dstPtr += ohSize;
787         maxDstSize -= ohSize;
788         eSize += ohSize;
789     }
790 
791     {   size_t const mhSize = FSE_writeNCount(dstPtr, maxDstSize, matchLengthNCount, MaxML, mlLog);
792         if (FSE_isError(mhSize)) {
793             eSize = ERROR(GENERIC);
794             DISPLAYLEVEL(1, "FSE_writeNCount error with matchLengthNCount \n");
795             goto _cleanup;
796         }
797         dstPtr += mhSize;
798         maxDstSize -= mhSize;
799         eSize += mhSize;
800     }
801 
802     {   size_t const lhSize = FSE_writeNCount(dstPtr, maxDstSize, litLengthNCount, MaxLL, llLog);
803         if (FSE_isError(lhSize)) {
804             eSize = ERROR(GENERIC);
805             DISPLAYLEVEL(1, "FSE_writeNCount error with litlengthNCount \n");
806             goto _cleanup;
807         }
808         dstPtr += lhSize;
809         maxDstSize -= lhSize;
810         eSize += lhSize;
811     }
812 
813     if (maxDstSize<12) {
814         eSize = ERROR(GENERIC);
815         DISPLAYLEVEL(1, "not enough space to write RepOffsets \n");
816         goto _cleanup;
817     }
818 # if 0
819     MEM_writeLE32(dstPtr+0, bestRepOffset[0].offset);
820     MEM_writeLE32(dstPtr+4, bestRepOffset[1].offset);
821     MEM_writeLE32(dstPtr+8, bestRepOffset[2].offset);
822 #else
823     /* at this stage, we don't use the result of "most common first offset",
824        as the impact of statistics is not properly evaluated */
825     MEM_writeLE32(dstPtr+0, repStartValue[0]);
826     MEM_writeLE32(dstPtr+4, repStartValue[1]);
827     MEM_writeLE32(dstPtr+8, repStartValue[2]);
828 #endif
829     eSize += 12;
830 
831 _cleanup:
832     ZSTD_freeCCtx(esr.ref);
833     ZSTD_freeCCtx(esr.zc);
834     free(esr.workPlace);
835 
836     return eSize;
837 }
838 
839 
840 
841 size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBufferCapacity,
842                           const void* customDictContent, size_t dictContentSize,
843                           const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
844                           ZDICT_params_t params)
845 {
846     size_t hSize;
847 #define HBUFFSIZE 256   /* should prove large enough for all entropy headers */
848     BYTE header[HBUFFSIZE];
849     int const compressionLevel = (params.compressionLevel <= 0) ? g_compressionLevel_default : params.compressionLevel;
850     U32 const notificationLevel = params.notificationLevel;
851 
852     /* check conditions */
853     if (dictBufferCapacity < dictContentSize) return ERROR(dstSize_tooSmall);
854     if (dictContentSize < ZDICT_CONTENTSIZE_MIN) return ERROR(srcSize_wrong);
855     if (dictBufferCapacity < ZDICT_DICTSIZE_MIN) return ERROR(dstSize_tooSmall);
856 
857     /* dictionary header */
858     MEM_writeLE32(header, ZSTD_MAGIC_DICTIONARY);
859     {   U64 const randomID = XXH64(customDictContent, dictContentSize, 0);
860         U32 const compliantID = (randomID % ((1U<<31)-32768)) + 32768;
861         U32 const dictID = params.dictID ? params.dictID : compliantID;
862         MEM_writeLE32(header+4, dictID);
863     }
864     hSize = 8;
865 
866     /* entropy tables */
867     DISPLAYLEVEL(2, "\r%70s\r", "");   /* clean display line */
868     DISPLAYLEVEL(2, "statistics ... \n");
869     {   size_t const eSize = ZDICT_analyzeEntropy(header+hSize, HBUFFSIZE-hSize,
870                                   compressionLevel,
871                                   samplesBuffer, samplesSizes, nbSamples,
872                                   customDictContent, dictContentSize,
873                                   notificationLevel);
874         if (ZDICT_isError(eSize)) return eSize;
875         hSize += eSize;
876     }
877 
878     /* copy elements in final buffer ; note : src and dst buffer can overlap */
879     if (hSize + dictContentSize > dictBufferCapacity) dictContentSize = dictBufferCapacity - hSize;
880     {   size_t const dictSize = hSize + dictContentSize;
881         char* dictEnd = (char*)dictBuffer + dictSize;
882         memmove(dictEnd - dictContentSize, customDictContent, dictContentSize);
883         memcpy(dictBuffer, header, hSize);
884         return dictSize;
885     }
886 }
887 
888 
889 size_t ZDICT_addEntropyTablesFromBuffer_advanced(void* dictBuffer, size_t dictContentSize, size_t dictBufferCapacity,
890                                                  const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
891                                                  ZDICT_params_t params)
892 {
893     int const compressionLevel = (params.compressionLevel <= 0) ? g_compressionLevel_default : params.compressionLevel;
894     U32 const notificationLevel = params.notificationLevel;
895     size_t hSize = 8;
896 
897     /* calculate entropy tables */
898     DISPLAYLEVEL(2, "\r%70s\r", "");   /* clean display line */
899     DISPLAYLEVEL(2, "statistics ... \n");
900     {   size_t const eSize = ZDICT_analyzeEntropy((char*)dictBuffer+hSize, dictBufferCapacity-hSize,
901                                   compressionLevel,
902                                   samplesBuffer, samplesSizes, nbSamples,
903                                   (char*)dictBuffer + dictBufferCapacity - dictContentSize, dictContentSize,
904                                   notificationLevel);
905         if (ZDICT_isError(eSize)) return eSize;
906         hSize += eSize;
907     }
908 
909     /* add dictionary header (after entropy tables) */
910     MEM_writeLE32(dictBuffer, ZSTD_MAGIC_DICTIONARY);
911     {   U64 const randomID = XXH64((char*)dictBuffer + dictBufferCapacity - dictContentSize, dictContentSize, 0);
912         U32 const compliantID = (randomID % ((1U<<31)-32768)) + 32768;
913         U32 const dictID = params.dictID ? params.dictID : compliantID;
914         MEM_writeLE32((char*)dictBuffer+4, dictID);
915     }
916 
917     if (hSize + dictContentSize < dictBufferCapacity)
918         memmove((char*)dictBuffer + hSize, (char*)dictBuffer + dictBufferCapacity - dictContentSize, dictContentSize);
919     return MIN(dictBufferCapacity, hSize+dictContentSize);
920 }
921 
922 
923 /*! ZDICT_trainFromBuffer_unsafe_legacy() :
924 *   Warning : `samplesBuffer` must be followed by noisy guard band.
925 *   @return : size of dictionary, or an error code which can be tested with ZDICT_isError()
926 */
927 size_t ZDICT_trainFromBuffer_unsafe_legacy(
928                             void* dictBuffer, size_t maxDictSize,
929                             const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
930                             ZDICT_legacy_params_t params)
931 {
932     U32 const dictListSize = MAX(MAX(DICTLISTSIZE_DEFAULT, nbSamples), (U32)(maxDictSize/16));
933     dictItem* const dictList = (dictItem*)malloc(dictListSize * sizeof(*dictList));
934     unsigned const selectivity = params.selectivityLevel == 0 ? g_selectivity_default : params.selectivityLevel;
935     unsigned const minRep = (selectivity > 30) ? MINRATIO : nbSamples >> selectivity;
936     size_t const targetDictSize = maxDictSize;
937     size_t const samplesBuffSize = ZDICT_totalSampleSize(samplesSizes, nbSamples);
938     size_t dictSize = 0;
939     U32 const notificationLevel = params.zParams.notificationLevel;
940 
941     /* checks */
942     if (!dictList) return ERROR(memory_allocation);
943     if (maxDictSize < ZDICT_DICTSIZE_MIN) { free(dictList); return ERROR(dstSize_tooSmall); }   /* requested dictionary size is too small */
944     if (samplesBuffSize < ZDICT_MIN_SAMPLES_SIZE) { free(dictList); return ERROR(dictionaryCreation_failed); }   /* not enough source to create dictionary */
945 
946     /* init */
947     ZDICT_initDictItem(dictList);
948 
949     /* build dictionary */
950     ZDICT_trainBuffer_legacy(dictList, dictListSize,
951                        samplesBuffer, samplesBuffSize,
952                        samplesSizes, nbSamples,
953                        minRep, notificationLevel);
954 
955     /* display best matches */
956     if (params.zParams.notificationLevel>= 3) {
957         U32 const nb = MIN(25, dictList[0].pos);
958         U32 const dictContentSize = ZDICT_dictSize(dictList);
959         U32 u;
960         DISPLAYLEVEL(3, "\n %u segments found, of total size %u \n", dictList[0].pos-1, dictContentSize);
961         DISPLAYLEVEL(3, "list %u best segments \n", nb-1);
962         for (u=1; u<nb; u++) {
963             U32 const pos = dictList[u].pos;
964             U32 const length = dictList[u].length;
965             U32 const printedLength = MIN(40, length);
966             if ((pos > samplesBuffSize) || ((pos + length) > samplesBuffSize))
967                 return ERROR(GENERIC);   /* should never happen */
968             DISPLAYLEVEL(3, "%3u:%3u bytes at pos %8u, savings %7u bytes |",
969                          u, length, pos, dictList[u].savings);
970             ZDICT_printHex((const char*)samplesBuffer+pos, printedLength);
971             DISPLAYLEVEL(3, "| \n");
972     }   }
973 
974 
975     /* create dictionary */
976     {   U32 dictContentSize = ZDICT_dictSize(dictList);
977         if (dictContentSize < ZDICT_CONTENTSIZE_MIN) { free(dictList); return ERROR(dictionaryCreation_failed); }   /* dictionary content too small */
978         if (dictContentSize < targetDictSize/4) {
979             DISPLAYLEVEL(2, "!  warning : selected content significantly smaller than requested (%u < %u) \n", dictContentSize, (U32)maxDictSize);
980             if (samplesBuffSize < 10 * targetDictSize)
981                 DISPLAYLEVEL(2, "!  consider increasing the number of samples (total size : %u MB)\n", (U32)(samplesBuffSize>>20));
982             if (minRep > MINRATIO) {
983                 DISPLAYLEVEL(2, "!  consider increasing selectivity to produce larger dictionary (-s%u) \n", selectivity+1);
984                 DISPLAYLEVEL(2, "!  note : larger dictionaries are not necessarily better, test its efficiency on samples \n");
985             }
986         }
987 
988         if ((dictContentSize > targetDictSize*3) && (nbSamples > 2*MINRATIO) && (selectivity>1)) {
989             U32 proposedSelectivity = selectivity-1;
990             while ((nbSamples >> proposedSelectivity) <= MINRATIO) { proposedSelectivity--; }
991             DISPLAYLEVEL(2, "!  note : calculated dictionary significantly larger than requested (%u > %u) \n", dictContentSize, (U32)maxDictSize);
992             DISPLAYLEVEL(2, "!  consider increasing dictionary size, or produce denser dictionary (-s%u) \n", proposedSelectivity);
993             DISPLAYLEVEL(2, "!  always test dictionary efficiency on real samples \n");
994         }
995 
996         /* limit dictionary size */
997         {   U32 const max = dictList->pos;   /* convention : nb of useful elts within dictList */
998             U32 currentSize = 0;
999             U32 n; for (n=1; n<max; n++) {
1000                 currentSize += dictList[n].length;
1001                 if (currentSize > targetDictSize) { currentSize -= dictList[n].length; break; }
1002             }
1003             dictList->pos = n;
1004             dictContentSize = currentSize;
1005         }
1006 
1007         /* build dict content */
1008         {   U32 u;
1009             BYTE* ptr = (BYTE*)dictBuffer + maxDictSize;
1010             for (u=1; u<dictList->pos; u++) {
1011                 U32 l = dictList[u].length;
1012                 ptr -= l;
1013                 if (ptr<(BYTE*)dictBuffer) { free(dictList); return ERROR(GENERIC); }   /* should not happen */
1014                 memcpy(ptr, (const char*)samplesBuffer+dictList[u].pos, l);
1015         }   }
1016 
1017         dictSize = ZDICT_addEntropyTablesFromBuffer_advanced(dictBuffer, dictContentSize, maxDictSize,
1018                                                              samplesBuffer, samplesSizes, nbSamples,
1019                                                              params.zParams);
1020     }
1021 
1022     /* clean up */
1023     free(dictList);
1024     return dictSize;
1025 }
1026 
1027 
1028 /* issue : samplesBuffer need to be followed by a noisy guard band.
1029 *  work around : duplicate the buffer, and add the noise */
1030 size_t ZDICT_trainFromBuffer_legacy(void* dictBuffer, size_t dictBufferCapacity,
1031                               const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
1032                               ZDICT_legacy_params_t params)
1033 {
1034     size_t result;
1035     void* newBuff;
1036     size_t const sBuffSize = ZDICT_totalSampleSize(samplesSizes, nbSamples);
1037     if (sBuffSize < ZDICT_MIN_SAMPLES_SIZE) return 0;   /* not enough content => no dictionary */
1038 
1039     newBuff = malloc(sBuffSize + NOISELENGTH);
1040     if (!newBuff) return ERROR(memory_allocation);
1041 
1042     memcpy(newBuff, samplesBuffer, sBuffSize);
1043     ZDICT_fillNoise((char*)newBuff + sBuffSize, NOISELENGTH);   /* guard band, for end of buffer condition */
1044 
1045     result =
1046         ZDICT_trainFromBuffer_unsafe_legacy(dictBuffer, dictBufferCapacity, newBuff,
1047                                             samplesSizes, nbSamples, params);
1048     free(newBuff);
1049     return result;
1050 }
1051 
1052 
1053 size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
1054                              const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples)
1055 {
1056     ZDICT_cover_params_t params;
1057     memset(&params, 0, sizeof(params));
1058     params.d = 8;
1059     params.steps = 4;
1060     /* Default to level 6 since no compression level information is avaialble */
1061     params.zParams.compressionLevel = 6;
1062     return ZDICT_optimizeTrainFromBuffer_cover(dictBuffer, dictBufferCapacity,
1063                                                samplesBuffer, samplesSizes,
1064                                                nbSamples, &params);
1065 }
1066 
1067 size_t ZDICT_addEntropyTablesFromBuffer(void* dictBuffer, size_t dictContentSize, size_t dictBufferCapacity,
1068                                         const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples)
1069 {
1070     ZDICT_params_t params;
1071     memset(&params, 0, sizeof(params));
1072     return ZDICT_addEntropyTablesFromBuffer_advanced(dictBuffer, dictContentSize, dictBufferCapacity,
1073                                                      samplesBuffer, samplesSizes, nbSamples,
1074                                                      params);
1075 }
1076