xref: /dflybsd-src/contrib/zstd/programs/datagen.c (revision a28cd43d19e8b720a6c852a4bbc5ae147a26165a)
1a28cd43dSSascha Wildner /*
2a28cd43dSSascha Wildner  * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
3a28cd43dSSascha Wildner  * All rights reserved.
4a28cd43dSSascha Wildner  *
5a28cd43dSSascha Wildner  * This source code is licensed under both the BSD-style license (found in the
6a28cd43dSSascha Wildner  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7a28cd43dSSascha Wildner  * in the COPYING file in the root directory of this source tree).
8a28cd43dSSascha Wildner  * You may select, at your option, one of the above-listed licenses.
9a28cd43dSSascha Wildner  */
10a28cd43dSSascha Wildner 
11a28cd43dSSascha Wildner 
12a28cd43dSSascha Wildner 
13a28cd43dSSascha Wildner /*-************************************
14a28cd43dSSascha Wildner *  Dependencies
15a28cd43dSSascha Wildner **************************************/
16a28cd43dSSascha Wildner #include "datagen.h"
17a28cd43dSSascha Wildner #include "platform.h"  /* SET_BINARY_MODE */
18a28cd43dSSascha Wildner #include <stdlib.h>    /* malloc, free */
19a28cd43dSSascha Wildner #include <stdio.h>     /* FILE, fwrite, fprintf */
20a28cd43dSSascha Wildner #include <string.h>    /* memcpy */
21a28cd43dSSascha Wildner #include "../lib/common/mem.h"  /* U32 */
22a28cd43dSSascha Wildner 
23a28cd43dSSascha Wildner 
24a28cd43dSSascha Wildner /*-************************************
25a28cd43dSSascha Wildner *  Macros
26a28cd43dSSascha Wildner **************************************/
27a28cd43dSSascha Wildner #define KB *(1 <<10)
28a28cd43dSSascha Wildner #define MIN(a,b)  ( (a) < (b) ? (a) : (b) )
29a28cd43dSSascha Wildner 
30a28cd43dSSascha Wildner #define RDG_DEBUG 0
31a28cd43dSSascha Wildner #define TRACE(...)   if (RDG_DEBUG) fprintf(stderr, __VA_ARGS__ )
32a28cd43dSSascha Wildner 
33a28cd43dSSascha Wildner 
34a28cd43dSSascha Wildner /*-************************************
35a28cd43dSSascha Wildner *  Local constants
36a28cd43dSSascha Wildner **************************************/
37a28cd43dSSascha Wildner #define LTLOG 13
38a28cd43dSSascha Wildner #define LTSIZE (1<<LTLOG)
39a28cd43dSSascha Wildner #define LTMASK (LTSIZE-1)
40a28cd43dSSascha Wildner 
41a28cd43dSSascha Wildner 
42a28cd43dSSascha Wildner /*-*******************************************************
43a28cd43dSSascha Wildner *  Local Functions
44a28cd43dSSascha Wildner *********************************************************/
45a28cd43dSSascha Wildner #define RDG_rotl32(x,r) ((x << r) | (x >> (32 - r)))
RDG_rand(U32 * src)46a28cd43dSSascha Wildner static U32 RDG_rand(U32* src)
47a28cd43dSSascha Wildner {
48a28cd43dSSascha Wildner     static const U32 prime1 = 2654435761U;
49a28cd43dSSascha Wildner     static const U32 prime2 = 2246822519U;
50a28cd43dSSascha Wildner     U32 rand32 = *src;
51a28cd43dSSascha Wildner     rand32 *= prime1;
52a28cd43dSSascha Wildner     rand32 ^= prime2;
53a28cd43dSSascha Wildner     rand32  = RDG_rotl32(rand32, 13);
54a28cd43dSSascha Wildner     *src = rand32;
55a28cd43dSSascha Wildner     return rand32 >> 5;
56a28cd43dSSascha Wildner }
57a28cd43dSSascha Wildner 
58a28cd43dSSascha Wildner typedef U32 fixedPoint_24_8;
59a28cd43dSSascha Wildner 
RDG_fillLiteralDistrib(BYTE * ldt,fixedPoint_24_8 ld)60a28cd43dSSascha Wildner static void RDG_fillLiteralDistrib(BYTE* ldt, fixedPoint_24_8 ld)
61a28cd43dSSascha Wildner {
62a28cd43dSSascha Wildner     BYTE const firstChar = (ld<=0.0) ?   0 : '(';
63a28cd43dSSascha Wildner     BYTE const lastChar  = (ld<=0.0) ? 255 : '}';
64a28cd43dSSascha Wildner     BYTE character = (ld<=0.0) ? 0 : '0';
65a28cd43dSSascha Wildner     U32 u;
66a28cd43dSSascha Wildner 
67a28cd43dSSascha Wildner     if (ld<=0) ld = 0;
68a28cd43dSSascha Wildner     for (u=0; u<LTSIZE; ) {
69a28cd43dSSascha Wildner         U32 const weight = (((LTSIZE - u) * ld) >> 8) + 1;
70a28cd43dSSascha Wildner         U32 const end = MIN ( u + weight , LTSIZE);
71a28cd43dSSascha Wildner         while (u < end) ldt[u++] = character;
72a28cd43dSSascha Wildner         character++;
73a28cd43dSSascha Wildner         if (character > lastChar) character = firstChar;
74a28cd43dSSascha Wildner     }
75a28cd43dSSascha Wildner }
76a28cd43dSSascha Wildner 
77a28cd43dSSascha Wildner 
RDG_genChar(U32 * seed,const BYTE * ldt)78a28cd43dSSascha Wildner static BYTE RDG_genChar(U32* seed, const BYTE* ldt)
79a28cd43dSSascha Wildner {
80a28cd43dSSascha Wildner     U32 const id = RDG_rand(seed) & LTMASK;
81a28cd43dSSascha Wildner     return ldt[id];  /* memory-sanitizer fails here, stating "uninitialized value" when table initialized with P==0.0. Checked : table is fully initialized */
82a28cd43dSSascha Wildner }
83a28cd43dSSascha Wildner 
84a28cd43dSSascha Wildner 
RDG_rand15Bits(U32 * seedPtr)85a28cd43dSSascha Wildner static U32 RDG_rand15Bits (U32* seedPtr)
86a28cd43dSSascha Wildner {
87a28cd43dSSascha Wildner     return RDG_rand(seedPtr) & 0x7FFF;
88a28cd43dSSascha Wildner }
89a28cd43dSSascha Wildner 
RDG_randLength(U32 * seedPtr)90a28cd43dSSascha Wildner static U32 RDG_randLength(U32* seedPtr)
91a28cd43dSSascha Wildner {
92a28cd43dSSascha Wildner     if (RDG_rand(seedPtr) & 7) return (RDG_rand(seedPtr) & 0xF);   /* small length */
93a28cd43dSSascha Wildner     return (RDG_rand(seedPtr) & 0x1FF) + 0xF;
94a28cd43dSSascha Wildner }
95a28cd43dSSascha Wildner 
RDG_genBlock(void * buffer,size_t buffSize,size_t prefixSize,double matchProba,const BYTE * ldt,U32 * seedPtr)96a28cd43dSSascha Wildner static void RDG_genBlock(void* buffer, size_t buffSize, size_t prefixSize,
97a28cd43dSSascha Wildner                          double matchProba, const BYTE* ldt, U32* seedPtr)
98a28cd43dSSascha Wildner {
99a28cd43dSSascha Wildner     BYTE* const buffPtr = (BYTE*)buffer;
100a28cd43dSSascha Wildner     U32 const matchProba32 = (U32)(32768 * matchProba);
101a28cd43dSSascha Wildner     size_t pos = prefixSize;
102a28cd43dSSascha Wildner     U32 prevOffset = 1;
103a28cd43dSSascha Wildner 
104a28cd43dSSascha Wildner     /* special case : sparse content */
105a28cd43dSSascha Wildner     while (matchProba >= 1.0) {
106a28cd43dSSascha Wildner         size_t size0 = RDG_rand(seedPtr) & 3;
107a28cd43dSSascha Wildner         size0  = (size_t)1 << (16 + size0 * 2);
108a28cd43dSSascha Wildner         size0 += RDG_rand(seedPtr) & (size0-1);   /* because size0 is power of 2*/
109a28cd43dSSascha Wildner         if (buffSize < pos + size0) {
110a28cd43dSSascha Wildner             memset(buffPtr+pos, 0, buffSize-pos);
111a28cd43dSSascha Wildner             return;
112a28cd43dSSascha Wildner         }
113a28cd43dSSascha Wildner         memset(buffPtr+pos, 0, size0);
114a28cd43dSSascha Wildner         pos += size0;
115a28cd43dSSascha Wildner         buffPtr[pos-1] = RDG_genChar(seedPtr, ldt);
116a28cd43dSSascha Wildner         continue;
117a28cd43dSSascha Wildner     }
118a28cd43dSSascha Wildner 
119a28cd43dSSascha Wildner     /* init */
120a28cd43dSSascha Wildner     if (pos==0) buffPtr[0] = RDG_genChar(seedPtr, ldt), pos=1;
121a28cd43dSSascha Wildner 
122a28cd43dSSascha Wildner     /* Generate compressible data */
123a28cd43dSSascha Wildner     while (pos < buffSize) {
124a28cd43dSSascha Wildner         /* Select : Literal (char) or Match (within 32K) */
125a28cd43dSSascha Wildner         if (RDG_rand15Bits(seedPtr) < matchProba32) {
126a28cd43dSSascha Wildner             /* Copy (within 32K) */
127a28cd43dSSascha Wildner             U32 const length = RDG_randLength(seedPtr) + 4;
128a28cd43dSSascha Wildner             U32 const d = (U32) MIN(pos + length , buffSize);
129a28cd43dSSascha Wildner             U32 const repeatOffset = (RDG_rand(seedPtr) & 15) == 2;
130a28cd43dSSascha Wildner             U32 const randOffset = RDG_rand15Bits(seedPtr) + 1;
131a28cd43dSSascha Wildner             U32 const offset = repeatOffset ? prevOffset : (U32) MIN(randOffset , pos);
132a28cd43dSSascha Wildner             size_t match = pos - offset;
133a28cd43dSSascha Wildner             while (pos < d) { buffPtr[pos++] = buffPtr[match++];   /* correctly manages overlaps */ }
134a28cd43dSSascha Wildner             prevOffset = offset;
135a28cd43dSSascha Wildner         } else {
136a28cd43dSSascha Wildner             /* Literal (noise) */
137a28cd43dSSascha Wildner             U32 const length = RDG_randLength(seedPtr);
138a28cd43dSSascha Wildner             U32 const d = (U32) MIN(pos + length, buffSize);
139a28cd43dSSascha Wildner             while (pos < d) { buffPtr[pos++] = RDG_genChar(seedPtr, ldt); }
140a28cd43dSSascha Wildner     }   }
141a28cd43dSSascha Wildner }
142a28cd43dSSascha Wildner 
143a28cd43dSSascha Wildner 
RDG_genBuffer(void * buffer,size_t size,double matchProba,double litProba,unsigned seed)144a28cd43dSSascha Wildner void RDG_genBuffer(void* buffer, size_t size, double matchProba, double litProba, unsigned seed)
145a28cd43dSSascha Wildner {
146a28cd43dSSascha Wildner     U32 seed32 = seed;
147a28cd43dSSascha Wildner     BYTE ldt[LTSIZE];
148a28cd43dSSascha Wildner     memset(ldt, '0', sizeof(ldt));  /* yes, character '0', this is intentional */
149a28cd43dSSascha Wildner     if (litProba<=0.0) litProba = matchProba / 4.5;
150a28cd43dSSascha Wildner     RDG_fillLiteralDistrib(ldt, (fixedPoint_24_8)(litProba * 256 + 0.001));
151a28cd43dSSascha Wildner     RDG_genBlock(buffer, size, 0, matchProba, ldt, &seed32);
152a28cd43dSSascha Wildner }
153a28cd43dSSascha Wildner 
154a28cd43dSSascha Wildner 
RDG_genStdout(unsigned long long size,double matchProba,double litProba,unsigned seed)155a28cd43dSSascha Wildner void RDG_genStdout(unsigned long long size, double matchProba, double litProba, unsigned seed)
156a28cd43dSSascha Wildner {
157a28cd43dSSascha Wildner     U32 seed32 = seed;
158a28cd43dSSascha Wildner     size_t const stdBlockSize = 128 KB;
159a28cd43dSSascha Wildner     size_t const stdDictSize = 32 KB;
160a28cd43dSSascha Wildner     BYTE* const buff = (BYTE*)malloc(stdDictSize + stdBlockSize);
161a28cd43dSSascha Wildner     U64 total = 0;
162a28cd43dSSascha Wildner     BYTE ldt[LTSIZE];   /* literals distribution table */
163a28cd43dSSascha Wildner 
164a28cd43dSSascha Wildner     /* init */
165a28cd43dSSascha Wildner     if (buff==NULL) { perror("datagen"); exit(1); }
166a28cd43dSSascha Wildner     if (litProba<=0.0) litProba = matchProba / 4.5;
167a28cd43dSSascha Wildner     memset(ldt, '0', sizeof(ldt));   /* yes, character '0', this is intentional */
168a28cd43dSSascha Wildner     RDG_fillLiteralDistrib(ldt, (fixedPoint_24_8)(litProba * 256 + 0.001));
169a28cd43dSSascha Wildner     SET_BINARY_MODE(stdout);
170a28cd43dSSascha Wildner 
171a28cd43dSSascha Wildner     /* Generate initial dict */
172a28cd43dSSascha Wildner     RDG_genBlock(buff, stdDictSize, 0, matchProba, ldt, &seed32);
173a28cd43dSSascha Wildner 
174a28cd43dSSascha Wildner     /* Generate compressible data */
175a28cd43dSSascha Wildner     while (total < size) {
176a28cd43dSSascha Wildner         size_t const genBlockSize = (size_t) (MIN (stdBlockSize, size-total));
177a28cd43dSSascha Wildner         RDG_genBlock(buff, stdDictSize+stdBlockSize, stdDictSize, matchProba, ldt, &seed32);
178a28cd43dSSascha Wildner         total += genBlockSize;
179a28cd43dSSascha Wildner         { size_t const unused = fwrite(buff, 1, genBlockSize, stdout); (void)unused; }
180a28cd43dSSascha Wildner         /* update dict */
181a28cd43dSSascha Wildner         memcpy(buff, buff + stdBlockSize, stdDictSize);
182a28cd43dSSascha Wildner     }
183a28cd43dSSascha Wildner 
184a28cd43dSSascha Wildner     /* cleanup */
185a28cd43dSSascha Wildner     free(buff);
186a28cd43dSSascha Wildner }
187