-
Notifications
You must be signed in to change notification settings - Fork 5
/
rcutil.c
1173 lines (1059 loc) · 58.9 KB
/
rcutil.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/**
Copyright (C) powturbo 2013-2023
GPL v3 License
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
- homepage : https://sites.google.com/site/powturbo/
- github : https://github.com/powturbo
- twitter : https://twitter.com/powturbo
- email : powturbo [_AT_] gmail [_DOT_] com
**/
// TurboRC: Range Coder - misc. functions
#include <stdio.h>
#include <string.h>
#include <ctype.h>
#include <math.h> //isnan
#include "include_/conf.h"
#include <float.h> //DBL_MAX
#include "include/turborc.h"
#include "include_/rcutil.h"
#include "rcutil_.h"
//-------------------------------- malloc ----------------------------------------
#ifdef _WIN32
#include <windows.h>
static SIZE_T largePageSize = 0, vinit_ = 0;
static void vinit(int enable) {
if(vinit_) return; vinit_++;
HANDLE hToken = 0;
if(OpenProcessToken(GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &hToken)) {
TOKEN_PRIVILEGES tp = {0};
if(LookupPrivilegeValue(NULL, SE_LOCK_MEMORY_NAME, &tp.Privileges[0].Luid)) {
tp.PrivilegeCount = 1;
tp.Privileges[0].Attributes = enable?SE_PRIVILEGE_ENABLED:0;
if(AdjustTokenPrivileges(hToken, FALSE, &tp, 0, NULL, 0)) {
#ifndef NDEBUG
unsigned rc = GetLastError();
if(rc == ERROR_SUCCESS)
largePageSize = GetLargePageMinimum();
else printf("AdjustTokenPrivileges.rc=%d ", rc);
#endif
}
}
CloseHandle(hToken);
}
HMODULE hKernel;
if(hKernel = GetModuleHandle(TEXT("kernel32.dll"))) {
typedef SIZE_T (WINAPI * GetLargePageMinimumProcT)();
GetLargePageMinimumProcT largePageMinimumProc = (GetLargePageMinimumProcT)GetProcAddress(hKernel, "GetLargePageMinimum");
if(largePageMinimumProc) {
largePageSize = largePageMinimumProc();
if ((largePageSize & (largePageSize - 1)) != 0) largePageSize = 0;
}
} //printf("LP=%d ", (int)largePageSize);
}
#endif
void *vmalloc(size_t size) {
#ifdef _WIN32
vinit(1);
if(largePageSize /*&& largePageSize <= (1 << 30)*/ && size >= (size_t)(1 << 18)) {
void *rc = VirtualAlloc(0, (size + largePageSize - 1) & (~(largePageSize - 1)), MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE);
//if(!rc) printf("VAlloc failed rc=%d ", (int)GetLastError()); else printf("LP=%lld/%zd ", largePageSize, size);
if(rc) return rc; //printf("LP=%d rc=%d ", (int)largePageSize, (int)rc);
}
return VirtualAlloc(0, size, MEM_COMMIT, PAGE_READWRITE);
#else
return malloc(size);
#endif
}
void vfree(void *address) {
#ifdef _WIN32
VirtualFree(address, 0, MEM_RELEASE);
#else
free(address);
#endif
}
//--------------------------- convert i/o to big endian -----------------------------
#if defined(__BYTE_ORDER__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
#define BSWAP16(a) a
#define BSWAP32(a) a
#define BSWAP64(a) a
#else
#define BSWAP16(a) bswap16(a)
#define BSWAP32(a) bswap32(a)
#define BSWAP64(a) bswap64(a)
#endif
//--------------------------- lzp preprocessor (lenmin >= 32) ------------------------------------------------------------------------
#ifndef NCOMP
#define H_BITS 16 // hash table size
#define emitmatch(_l_,_op_) { unsigned _l = _l_-lenmin+1; *_op_++ = 255; while(_l >= 254) { *_op_++ = 254; _l -= 254; OVERFLOW(in,inlen,out,op,goto end); } *_op_++ = _l; }
#define emitch(_ch_,_op_) { *_op_++ = _ch_; if(_ch_ == 255) *_op_++ = 0; OVERFLOW(in,inlen, out, op, goto end); }
#define LZPINI(_n_) if(lenmin < 32) lenmin = 32;\
if(_n_ < lenmin) { memcpy(out, in, _n_); return _n_;}\
if(!hbits) { hbits = _n_ >= (1<<24)?21:H_BITS; }\
hbits = hbits>12?hbits:12;\
if(hbits > H_BITS && !(htab = calloc(1<<hbits, 4))) { htab = _htab; hbits = H_BITS; }
size_t lzpenc(unsigned char *__restrict in, size_t inlen, unsigned char *__restrict out, unsigned lenmin, unsigned hbits) { //printf("m=%u ", lenmin);
unsigned _htab[1<<H_BITS] = {0}, *htab = _htab, cl, cx, h4 = 0;
unsigned char *ip = in, *cp, *op = out;
LZPINI(inlen); //unsigned cnt = 1<<hbits, lmin = lenmin, ocnt=0;
for(cx = ctou32(ip), ctou32(op) = cx, cx = BSWAP32(cx), op += 4, ip += 4; ip < in+inlen-lenmin;) {
h4 = HASH(h4, cx);
cp = in + htab[h4]; //cnt -= htab[h4] == 0; //if(cnt*100/(1<<hbits) != ocnt) { printf("%d ", cnt); ocnt=cnt*100/(1<<hbits); } //unsigned p = cnt * 100 / (1<<hbits); if(lmin > 32 && lmin <= lenmin && ip - in > (1<<20)) { if(p < 70) lmin--; else lmin++; printf("l=%u ", lmin);}
htab[h4] = ip - in;
if(ctou64(ip) == ctou64(cp) && ctou64(ip+8) == ctou64(cp+8) && ctou64(ip+16) == ctou64(cp+16) && ctou64(ip+24) == ctou64(cp+24)) { // match
for(cl = 32;;) {
if(ip+cl >= in+inlen-32) break;
if(ctou64(ip+cl) != ctou64(cp+cl)) break; cl += 8;
if(ctou64(ip+cl) != ctou64(cp+cl)) break; cl += 8;
if(ctou64(ip+cl) != ctou64(cp+cl)) break; cl += 8;
if(ctou64(ip+cl) != ctou64(cp+cl)) break; cl += 8;
} //unsigned x = PREDLEN(avg, cl); printf("%u,", x);
if(cl >= lenmin) {
for(; ip+cl < in+inlen && ip[cl] == cp[cl]; cl++);
emitmatch(cl, op);
ip += cl;
cx = BSWAP32(ctou32(ip-4));
continue;
}
}
unsigned ch = *ip++; emitch(ch, op); cx = cx<<8 | ch; // literal
}
while(ip < in+inlen) { unsigned c = *ip++; emitch(c, op); } //unsigned cnt = 0; for(int i=0; i < (1<<hbits); i++) cnt += htab[i] != 0; printf("c=%u %.2f", cnt, (double)cnt*100.0/(1<<hbits) );
end:if(htab != _htab) free(htab);
return op - out;
}
#endif
#ifndef NDECOMP
size_t lzpdec(unsigned char *in, size_t outlen, unsigned char *out, unsigned lenmin, unsigned hbits) {
unsigned _htab[1<< H_BITS] = {0}, *htab = _htab, cx, h4 = 0;
unsigned char *ip = in, *op = out;
LZPINI(outlen);
for(cx = ctou32(ip), ctou32(op) = cx, cx = BSWAP32(cx), op += 4, ip += 4; op < out+outlen;) {
unsigned c; h4 = HASH(h4, cx);
unsigned char *cp = out + htab[h4],*op_;
htab[h4] = op - out;
if((c = *ip++) == 255)
if(*ip) {
c = 0; do c += *ip; while(*ip++ == 254);
for(op_ = op+c+lenmin-1; op < op_; *op++ = *cp++);
cx = BSWAP32(ctou32(op-4));
continue;
} else ip++, c = 255;
cx = cx << 8 | (*op++ = c);
}
if(htab != _htab) free(htab);
return ip - in;
}
#endif
#ifndef NCOMP
// --------------------------------- QLFC - Quantized Local Frequency Coding ------------------------------------------
// QLFC=MTF-Backward: number of different symbols until the next occurrence (number of symbols will be seen before the next one)
// MTF: number of different symbols since the last occurence (number of symbols were seen until the current one)
// References: https://ieeexplore.ieee.org/document/1402216
uint8_t *rcqlfc(uint8_t *__restrict in, size_t n, uint8_t *__restrict out, uint8_t *__restrict r2c) {
unsigned char f[1<<8] = {0};
uint8_t *ip, *op = out;
int m;
for(m = 0; m < (1<<8); m++) r2c[m] = m; //if(!in[n-1]) r2c[0] = 1, r2c[1] = 0;
for(m = -1,ip = in+n; ip > in; ) {
uint8_t *p, c = *--ip; //--------- run length ---------------------------
MEMDEC8(cv, c); MEMRUNR8(in,ip,cv,c,goto a);
a:;
//-------------- mtf: move to front -----------------------------------
#if 0 //combined search + move to front.
unsigned c0, c1;
for(c0 = r2c[0], r2c[0] = c,p = &r2c[1];;p += 4) {
c1 = p[0]; p[0] = c0; if(c1 == c) break;
c0 = p[1]; p[1] = c1; if(c0 == c) { p++; break; }
c1 = p[2]; p[2] = c0; if(c1 == c) { p += 2; break; }
c0 = p[3]; p[3] = c1; if(c0 == c) { p += 3; break; }
}
#else
uint8_t *pb=NULL,*q=NULL; p = r2c; // search
#ifdef __AVX2__
for(;;) { unsigned m = _mm256_movemask_epi8(_mm256_cmpeq_epi8(_mm256_loadu_si256((__m256i*)p), cv)); if(m) { p += ctz32(m); break; } p += 32;}
#elif defined(__SSE__)
for(;;) { uint16_t m = _mm_movemask_epi8( _mm_cmpeq_epi8( _mm_loadu_si128((__m128i*)p), cv)); if(m) { p += ctz16(m); break; } p += 16; }
#else
while(*p != c) p++;
#endif
for(q = p; q > r2c; ) { // move to front
#ifdef __AVX2__
q -= 32; _mm256_storeu_si256(q+1,_mm256_loadu_si256((__m256i*)q));
#elif defined(__SSE__)
q -= 16; _mm_storeu_si128(q+1, _mm_loadu_si128((__m128i*)q));
#else
q -= 16; ctou64(q+1+8) = ctou64(q+8); ctou64(q+1) = ctou64(q);
#endif
}
r2c[0] = c;
#endif
if(!f[c]) *op++ = m++,f[c]++; else *op++ = p-r2c-1;
}
return op;
}
#endif
//------------------------------------------- utf8 preprocessing -----------------------------------------------
#ifndef NCOMP
#include "include_/vlcbyte.h" // vsput/vsget
//-- utf-8 to code point ----------
#define UTF8_INV 0x10fe00 // with 3 bytes private area
#define UTF8LENMASK(_c_, _m_, _l_) \
if ((_c_ & 0xe0) == 0xc0) { _l_ = 2; _m_ = 0x1f; }\
else if ((_c_ & 0xf0) == 0xe0) { _l_ = 3; _m_ = 0x0f; }\
else if ((_c_ & 0xf8) == 0xf0) { _l_ = 4; _m_ = 0x07; }\
else if ((_c_ & 0xfc) == 0xf8) { _l_ = 5; _m_ = 0x03; }\
else if ((_c_ & 0xfe) == 0xfc) { _l_ = 6; _m_ = 0x01; }\
else { _l_ = 0; _m_ = 0xff; } /*invalid*/
#define UTF8GET(_ip_, _m_, _l_, _c_) do { int _ui;\
_c_ &= (_m_);\
for(_ui = 1; _ui < (_l_); _ui++) {\
if(((_ip_)[_ui] & 0xc0) != 0x80) { _l_ = 0; break; }\
_c_ = (_c_) << 6 | ((_ip_)[_ui] & 0x3f);\
}\
} while(0)
#define UCGET(_ip_, _c_, _l_) {\
if((_c_ = *_ip_) < 128) { _ip_++; _l_ = 1; }\
else {\
unsigned _mask, _c = _c_;\
UTF8LENMASK(_c_, _mask, _l_);\
UTF8GET(_ip_, _mask, _l_, _c_); _ip_ += _l_;\
if(!_l_) { _c_ = UTF8_INV + _c; _ip_++; } /*invalid -> private area*/\
}\
}
//------------------------- utf-8 (1-gramm)) preprocessor ---------------------------------------------------------
#pragma pack(1)
typedef struct { unsigned c, cnt; /*unsigned short v;*/ } _PACKED sym_t; // c=key, cnt=count, v=value (=cid function)
#pragma pack()
static ALWAYS_INLINE unsigned cid(unsigned c) { // utf-8 classification
if(c >= 0 && c <= 0xff) return 0; // 1 byte
if(c >= 0x2E80 && c <= 0x2F00) return 2; // cjk radicals supplement
if(c >= 0x3400 && c <= 0x4DC0) return 2; // 2 bytes
if(c >= 0x4E00 && c <= 0xA000) return 2;
if(c >= 0x9FA6 && c <= 0x9FCC) return 2;
if(c >= 0xF900 && c <= 0xFB00) return 2; // compatibility ideographs
if(c > 0xff && c <= 0xffff) return 1;
if(c >= 0x20000 && c <= 0x2A6E0) return 3; // 3 bytes
if(c >= 0x2A700 && c < 0x2B740) return 3;
if(c >= 0x2B740 && c < 0x2B820) return 3;
if(c >= 0x2B820 && c <= 0x2CEB0) return 3; //included as of Unicode 8.0
if(c >= 0x2F800 && c <= 0x2FA20) return 3;
if(c > 0xffff && c <= 0xffffff) return 4;
return 5; // 4 bytes
}
// qsort compare functions --------------------------------
#define SC(_x_) cid(_x_)
#define CMPA(_a_,_b_,_t_) ((*(_t_ *)(_a_) > *(_t_ *)(_b_)) - (*(_t_ *)(_a_) < *(_t_ *)(_b_)))
#define CMPD(_a_,_b_,_t_) ((*(_t_ *)(_a_) < *(_t_ *)(_b_)) - (*(_t_ *)(_a_) > *(_t_ *)(_b_)))
#define CMPSA(_a_,_b_, _t_, _v_) (((((_t_ *)_a_)->_v_) > (((_t_ *)_b_)->_v_)) - ((((_t_ *)_a_)->_v_) < (((_t_ *)_b_)->_v_)))
#define CMPSD(_a_,_b_, _t_, _v_) (((((_t_ *)_a_)->_v_) < (((_t_ *)_b_)->_v_)) - ((((_t_ *)_a_)->_v_) > (((_t_ *)_b_)->_v_)))
#define CMPSFA(_a_,_b_, _t_, _v_) (( SC(((_t_ *)_a_)->_v_) > SC(((_t_ *)_b_)->_v_)) - ( SC(((_t_ *)_a_)->_v_) < SC(((_t_ *)_b_)->_v_)))
// qsort compare: a=ascending d=descending
// unsigned
static int cmpua(const void *a, const void *b) { return CMPA(a,b,uint32_t); }
static int cmpud(const void *a, const void *b) { return CMPD(a,b,uint32_t); }
// struct sym_t : n = count, c = key. na:count ascending, nd:count descending, ca: key c ascending
static int cmpsna(const void *a, const void *b) { return CMPSA(a,b,sym_t,cnt); }
static int cmpsnd(const void *a, const void *b) { return CMPSD(a,b,sym_t,cnt); }
static int cmpsca(const void *a, const void *b) { return CMPSA(a,b,sym_t,c ); }
// struct sym_t : n = count, c = key, v = value
static int cmpsndca(const void *a, const void *b) { int c; if(c = CMPSD(a,b,sym_t,cnt)) return c; return CMPSA(a,b,sym_t,c ); }
static int cmpscand(const void *a, const void *b) { int c; if(c = CMPSFA(a,b,sym_t,c)) return c; return CMPSD(a,b,sym_t,cnt); }
// symbol table output
static unsigned char *symsput(sym_t *stab, unsigned stabn, unsigned char *out, unsigned flag) {
unsigned char *op = out+1, *p;
out[0] = flag;
ctou16(op) = stabn; op += 2; p = op; op += 1+2+2;
unsigned i=0, v=0,u;
while(i < stabn) { if(stab[i].c > 0xff) break; *op++ = stab[i++].c; } *p = i;
v = 0; while(i < stabn) { if(stab[i].c > 0xffff) break; ctou16(op) = BSWAP16(stab[i++].c); op += 2; } ctou16(p+1) = i;
v = 0; while(i < stabn) { if(stab[i].c > 0xffffff) break; u = stab[i++].c; *op++ = u>>16; ctou16(op) = BSWAP16(u); op += 2; } ctou16(p+3) = i;
v = 0; while(i < stabn) { ctou32(op) = BSWAP32(stab[i++].c); op += 4; }
return op;
}
#define NOHASH
#ifdef NOHASH
#define CGET( _a_, _htab_,_hbits_,_hmask_, _c_, _i_) { _i_ = _htab_[_c_]-1; } // get index i in a for a previously added key c
#define CFIND( _a_, _htab_,_hbits_,_hmask_, _c_, _i_, _h_) { _i_ = _htab_[_h_ = _c_]-1; } // return index i of key c if found, otherwise -1
#define CREHASH( _a_, _htab_,_hbits_,_hmask_, _n_) { unsigned _i; memset(_htab_, 0, (1<<_hbits_)*sizeof(_htab_[0])); for(_i = 0; _i < _n_; _i++) { unsigned _h = _a_[_i].c; _htab_[_h] = _i+1;} }
#endif
size_t utf8enc(unsigned char *__restrict in, size_t inlen, unsigned char *__restrict out, unsigned flag) {
#define SYMBITS 16
unsigned char *ip, *op = out, cinv = 0;
sym_t stab[1<<SYMBITS] = {0};
#ifdef NOHASH
#define HBITS 24 // 16*2 = 32MB
unsigned short *stabh = calloc((size_t)(1<<HBITS), sizeof(stabh[0])); if(!stabh) die("utf8enc: calloc failed size=%z\n", 1<<HBITS);
#else
#define HBITS (SYMBITS+1)
#define HMASK ((1<<HBITS)-1)
unsigned short stabh[1<< HBITS] = {0};
#endif
unsigned stabn = 0, xprep8 = flag & BWT_PREP8, verbose = flag & BWT_VERBOSE, itmax = (flag>>10) & 0xf, xsort = (flag>>14) & 0x3, cnt;
if(verbose) { printf("utf8:sort=%u prep8=%d ", xsort, xprep8); fflush(stdout); } //unsigned st_crd=0;
for(ip = in; ip < in+inlen;) { // build the symbol dictionary
unsigned c, ci, l = 1, h;
UCGET(ip, c, l);
#ifdef UTF8INV
if(!l) { op = out+inlen; if(verbose) { printf("invalid utf8 ");fflush(stdout); }
goto e;
} // convert to code point + utf-8 validity check //if(c > cmax) cmax = c;
#else
if(!l) { if(verbose) { printf("#"); fflush(stdout); }
if(++cinv > 16) { op = out+inlen; if(verbose) { printf("invalid utf8 symbols"); fflush(stdout); }
goto e;
}
} else if(c >= UTF8_INV && c <= UTF8_INV+0xff) { op = out+inlen; if(verbose) { printf("?"); fflush(stdout); } goto e; } // symbol not allowed
#endif
CFIND(stab, stabh,HBITS,HMASK, c, ci, h);
if(ci != -1) stab[ci].cnt++;
else {
if(stabn >= (1<<SYMBITS)-1
#ifdef NOHASH
|| c >= (1<<HBITS)
#endif
) { // max. number of symbols is 64k
op = out+inlen; if(verbose) { if(c >= (1<<HBITS)) printf("utf-8 overflow\n");else printf("number of symbol > 64k\n");fflush(stdout); }
goto e;
}
CADD(stab, stabh, stabn, c, h, cid(c));
}
}
if(verbose) { printf("sym=%u sort='%s' ", stabn, xsort?"freq":"cat");fflush(stdout); }
switch(xsort) {
case 0: qsort(stab, stabn, sizeof(sym_t), cmpscand); break; // sort by code group + count (bwt mode)
case 1: qsort(stab, stabn, sizeof(sym_t), cmpsnd); break; // sort by count
}
unsigned cnt8 = 0; cnt = 0;
for(int i = 0; i < stabn; i++) {
if(stab[i].c <= 0xff) cnt8 += stab[i].cnt;
cnt += stab[i].cnt;
}
cnt8 = (uint64_t)cnt8*128 / cnt; if(verbose) { printf("ratio=%u ", cnt8);fflush(stdout); }
if(cnt8 > 64 && !(flag & BWT_RATIO)) { op = out+inlen; goto e; } // enough saving for converting to 16-bits?
CREHASH(stab, stabh,HBITS,HMASK, stabn); // rehash after sort
op = out+4;
if(itmax <= 1 || !xprep8) op = symsput(stab, stabn, op, xprep8?1:0); // output the dictionary
unsigned hdlen = op - out;
if(hdlen & 1) *op++ = 0; // offset to data must be even for 16 bits bwt
if(itmax>1 || !xprep8) { if(verbose) { printf("'16 bits output' "); fflush(stdout); }
if(op + cnt*2 >= out+(inlen*255)/256-8) { op = out+inlen; goto e;}; // check overflow in case of 16 bits
for(ip = in; ip < in+inlen;) { // fixed length encoding: 16-bits
unsigned c, l, ci;
UCGET(ip, c, l);
CGET(stab, stabh,HBITS,HMASK, c, ci);
ctou16(op) = BSWAP16(ci); op += 2;
}
} else { if(verbose) { printf("'8-16 bits output' ");fflush(stdout); }
for(ip = in; ip < in+inlen;) { // variable length encoding: 8-16 bits
unsigned c, l, ci;
UCGET(ip, c, l);
CGET(stab, stabh,HBITS,HMASK, c, ci);
vsput20(op, ci); OVERFLOW(in,inlen,out, op, goto e);
}
} ctou32(out) = op - out; if(verbose) { printf("len='%u' ", (unsigned)(op-out));fflush(stdout); }
e:
#ifdef NOHASH
if(stabh) free(stabh);
#endif
if(op >= out+inlen) {
op = out+inlen;
if(flag & BWT_COPY) memcpy(out,in, inlen);
}
return op - out;
}
#endif
#ifndef NDECOMP
//--------------------------- decode ---------------------
#define ctoutf8(_u_, _c_, _l_) {\
unsigned _c = _c_;\
if(likely(_c <= 0x7f)) { _u_ = _c; _l_ = 1;}\
else if( _c <= 0x7ff) { _u_ = 0x80c0 | (_c & 0x3f) << 8 | _c >> 6; _l_ = 2;}\
else if( _c <= 0xffff) { _u_ = 0x8080e0 | (_c & 0x3f) <<16 | ((_c & 0xfc0) >> 6) << 8 | _c >> 12; _l_ = 3;}\
else if( _c >= UTF8_INV && _c <= UTF8_INV+0xff) { _u_ = _c - UTF8_INV; _l_ = 3;}\
else { _u_ = 0x808080f0u | (_c & 0x3f) <<24 | (_c & 0xfc0) << 10 | (_c & 0x3f000) >> 4 | _c >> 18; _l_ = 4;}\
}
unsigned symsget(unsigned *sym, unsigned char **_in, unsigned *flag) {
unsigned char *in = *_in, *ip = in;
unsigned m0,m1,m2,stabn,i = 0;
*flag = *ip++;
stabn = ctou16(ip); ip+=2;
m0 = *ip++;
m1 = ctou16(ip); ip += 2;
m2 = ctou16(ip); ip += 2;
while(i < m0) sym[i++] = *ip++;
while(i < m1) sym[i++] = BSWAP16(ctou16(ip)), ip+=2;
while(i < m2) sym[i++] = (unsigned)ip[0]<<16 | BSWAP16(ctou16(ip+1)), ip+=3;
while(i < stabn) sym[i++] = BSWAP32(ctou32(ip)), ip+=4;
if((ip-in) & 1) ip++; // align to even offset for 16-bits bwt.
*_in = ip;
return stabn;
}
size_t utf8dec(unsigned char *__restrict in, size_t outlen, unsigned char *__restrict out) {
unsigned char* op = out, * ip = in + 4, len[1 << 16] = { 0 };
unsigned i,flag, sym[1<<16], inlen = ctou32(in),stabn = symsget(sym, &ip, &flag);
for(i = 0; i < stabn; i++) {
unsigned u,l;
ctoutf8(u, sym[i], l);
sym[i] = u;
len[i] = l;
}
if(flag & 1) { // variable length encoding: 8-16 bits
#define ST { unsigned u; vsget20(ip, u); ctou32(op) = sym[u]; op+=len[u]; }
while(ip < in+inlen-8-4) { ST; ST; ST; ST; }
while(ip < in+inlen) {
unsigned i,u;
vsget20(ip, i);
u = sym[i];
switch(len[i]) {
case 4: ctou32(op) = u; op+=4; break;
case 3: op[0] = (uint8_t)u; op[1] = (uint16_t)u>>8; op[2] = (uint8_t)(u>>16); op+=3; break;
case 2: ctou16(op) = u; op+=2; break;
case 1: *op++ = u;
}
}
} else { // fixed length encoding: 16-bits
#define ST(_i_) { unsigned u = BSWAP16(ctou16(ip+_i_*2)); ctou32(op) = sym[u]; op+=len[u]; }
for(; ip < in+(inlen-8-4); ip += 8) { ST(0); ST(1); ST(2); ST(3); }
for(; ip < in+inlen ; ip += 2) {
unsigned i = BSWAP16(ctou16(ip)), u = sym[i];
switch(len[i]) {
case 4: ctou32(op) = u; op += 4; break;
case 3: op[0] = (uint8_t)u; op[1] = (uint16_t)u>>8; op[2] = (uint8_t)(u>>16); op += 3; break;
case 2: ctou16(op) = u; op += 2; break;
case 1: *op++ = u;
}
}
}
return 0;
}
#endif
//--------------------------- Turbohist: https://github.com/powturbo/Turbo-Histogram ----------
#define CSIZE (256 + 8)
typedef unsigned cnt_t;
#define HISTEND(_c_,_cn_,_cnt_) { int _i,_j;\
memset(_cnt_, 0, 256*sizeof(_cnt_[0]));\
for(_i=0; _i < 256; _i++)\
for(_j=0; _j < _cn_;_j++) _cnt_[_i] += _c_[_j][_i];\
}
#define HISTEND8(_c_,_cnt_) HISTEND(_c_,8,_cnt_)
#define HISTEND4(_c_,_cnt_) HISTEND(_c_,4,_cnt_)
//-------------------- 64 bits ------------------------------------
#if defined(__i386__) || defined(__x86_64__)
#define CU64(_u_,_o_,_c_) { unsigned _x = _u_;\
c[0 ][(unsigned char )_x ]+=_c_;\
c[1 ][(unsigned short)_x>> 8]+=_c_; _x>>=16;\
c[2 ][(unsigned char )_x ]+=_c_;\
c[3 ][(unsigned short)_x>> 8]+=_c_; _x=(_u_)>>=32;\
c[0+_o_][(unsigned char )_x ]+=_c_;\
c[1+_o_][(unsigned short)_x>> 8]+=_c_; _x>>=16;\
c[2+_o_][(unsigned char )_x ]+=_c_;\
c[3+_o_][(unsigned short)_x>> 8]+=_c_;\
}
#else
#define CU64(_u_,_o_,_c_) { unsigned _x = _u_;\
c[0 ][(unsigned char) _x ]+=_c_;\
c[1 ][(unsigned char)(_x>> 8)]+=_c_;\
c[2 ][(unsigned char)(_x>>16)]+=_c_;\
c[3 ][ _x>>24 ]+=_c_; _x=(_u_)>>=32;\
c[0+_o_][(unsigned char) _x ]+=_c_;\
c[1+_o_][(unsigned char)(_x>> 8)]+=_c_;\
c[2+_o_][(unsigned char)(_x>>16)]+=_c_;\
c[3+_o_][ _x>>24 ]+=_c_;\
}
#endif
#define UZ 8 // Load size 1x 64 bits = 8 bytes
#define I164(_i_,_o_) { uint64_t u1 = ctou64(ip+UZ+_i_*UZ*2+ 0); CU64(u0, _o_, 1);\
u0 = ctou64(ip+UZ+_i_*UZ*2+ 8); CU64(u1, _o_, 1);\
}
#define N64 64
unsigned histcalc8(unsigned char *__restrict in, unsigned inlen, unsigned *__restrict cnt) {
#define IC 4
cnt_t c[8][CSIZE] = {0};
unsigned char *ip = in;
if(inlen >= UZ+N64) {
uint64_t u0 = ctou64(ip);
for(; ip <= in+inlen-(UZ+N64); ip += N64) {
I164(0,IC); I164(1,IC); I164(2,IC); I164(3,IC);
PREFETCH(ip+512, 0);
}
}
while(ip != in+inlen) c[0][*ip++]++;
HISTEND8(c, cnt);
unsigned a = 256; while(a > 1 && !cnt[a-1]) a--;
return a;
}
#define UZ 16 // Load size 2x 64 bits = 2*8 bytes
#define CR64(u,v,_o_,_c_) if(likely(u!=v)) { CU64(u,_o_,1); CU64(v,_o_,1); } else if((u^(v<<8)) < (1<<8)) c[_c_][(unsigned char)u]+=UZ; else CU64(u, _o_,2)
#define I2R64(_i_,_o_) { uint64_t u1 = ctou64(ip+UZ+_i_*UZ*2+ 0), v1 = ctou64(ip+UZ+_i_*UZ*2+ 8); CR64(u0,v0,_o_,_i_);\
u0 = ctou64(ip+UZ+_i_*UZ*2+16); v0 = ctou64(ip+UZ+_i_*UZ*2+24); CR64(u1,v1,_o_,_i_);\
}
unsigned histrcalc8(unsigned char *__restrict in, unsigned inlen, unsigned *__restrict cnt) {
#define IC 4
cnt_t c[8][CSIZE] = {0};
unsigned char *ip = in;
if(inlen >= UZ+N64) {
uint64_t u0 = ctou64(ip), v0 = ctou64(ip+8);
for(; ip <= in+inlen-(UZ+N64); ip += N64) {
I2R64(0,IC); I2R64(1,IC);
PREFETCH(ip+512, 0);
}
}
while(ip != in+inlen) c[0][*ip++]++;
HISTEND8(c, cnt);
unsigned a = 256; while(a > 1 && !cnt[a-1]) a--;
return a;
}
void memrev(unsigned char a[], unsigned n) {
size_t i=0, j;
#if defined(__AVX2__)
__m256i cv = _mm256_set_epi8( 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,15, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,15 );
for(j = i; j < n >> (1+5); ++j,i += 32) {
__m256i lo = _mm256_shuffle_epi8(_mm256_loadu_si256((__m256i*)&a[i ]),cv),
hi = _mm256_shuffle_epi8(_mm256_loadu_si256((__m256i*)&a[n-i-32]),cv);
_mm256_storeu_si256((__m256i*)&a[i], _mm256_permute2x128_si256(hi,hi,1));
_mm256_storeu_si256((__m256i*)&a[n - i - 32], _mm256_permute2x128_si256(lo,lo,1));
}
//for( ; i < (n >> 1); ++i ) { unsigned char t = a[i]; a[i] = a[n - i - 1]; a[n - i - 1] = t; }
#elif defined(__SSSE3__)
__m128i cv = _mm_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);
for(j = i; j < n >>(1+4); ++j, i += 16) {
__m128i hi = _mm_shuffle_epi8(_mm_loadu_si128((__m128i*)&a[n-i-16]), cv),
lo = _mm_shuffle_epi8(_mm_loadu_si128((__m128i*)&a[i ]), cv);
_mm_storeu_si128((__m128i*)&a[i ], hi);
_mm_storeu_si128((__m128i*)&a[n - i - 16], lo);
}
//for( ; i < n >> 1; ++i ) { unsigned char t = a[i]; a[i] = a[n - i - 1]; a[n - i - 1] = t; }
#else
//unsigned i; for(--n,i=0;i < n; ++i) { unsigned char t = a[i]; a[i] = a[n]; a[n--] = t; }
i=0;
#endif
for(;i < n/2; ++i) { unsigned char t = a[i]; a[i] = a[n-i-1]; a[n-i-1] = t; }
}
#ifndef NCOMP
size_t bitenc(unsigned char *__restrict in, size_t inlen, unsigned char *__restrict out) {
unsigned cnt[256] = {0}, map[256] = {0}, a = histrcalc8(in, inlen, cnt), c = 0;
unsigned char *ip = in, *op = out;
*op++ = 0;
for(int i = 0; i < 256; i++) if(cnt[i]) { map[i] = c++; *op++ = i; }
if(c > 16) { *out = 0xff; memcpy(out+1, in, inlen); return inlen; }
*out = c;
if(c <= 2) { for(; ip < in+(inlen&~(8-1)); ip+=8) *op++ = map[ip[7]]<<7 | map[ip[6]]<<6 | map[ip[5]]<<5 | map[ip[4]]<<4 | map[ip[3]]<<3 | map[ip[2]]<<2 | map[ip[1]]<<1 | map[ip[0]]; }
else if(c <= 4) { for(; ip < in+(inlen&~(4-1)); ip+=4) *op++ = map[ip[3]]<<6 | map[ip[2]]<<4 | map[ip[1]]<<2 | map[ip[0]]; }
else if(c <= 16) { for(; ip < in+(inlen&~(2-1)); ip+=2) *op++ = map[ip[1]]<<4 | map[ip[0]]; }
while(ip < in+inlen) *op++ = *ip++;
return op - out;
}
#endif
#ifndef NDECOMP
size_t bitdec(unsigned char *__restrict in, size_t outlen, unsigned char *__restrict out) {
unsigned char *ip = in, *op = out, *p;
if(*in == 0xff) { memcpy(out, in+1, outlen); return outlen+1; }
unsigned c = *ip++; p = ip; ip += c;
if(c <= 2) { for(op = out; op < out+(outlen&~(8-1)); op+=8) { unsigned u = *ip++; op[0] = p[(u )& 1]; op[1] = p[(u>>1)&1]; op[2] = p[(u>>2)&1]; op[3] = p[(u>>3)&1];
op[4] = p[(u>>4)& 1]; op[5] = p[(u>>5)&1]; op[6] = p[(u>>6)&1]; op[7] = p[(u>>7)&1]; }}
else if(c <= 4) { for(op = out; op < out+(outlen&~(4-1)); op+=4) { unsigned u = *ip++; op[0] = p[(u )& 3]; op[1] = p[(u>>2)&3]; op[2] = p[(u>>4)&3]; op[3] = p[(u>>6)&3]; }}
else if(c <=16) { for(op = out; op < out+(outlen&~(2-1)); op+=2) { unsigned u = *ip++; op[0] = p[(u )&15]; op[1] = p[ u &15]; op[2] = p[u>>4]; }}
while(op < out+outlen) *op++ = *ip++;
return ip - in;
}
#endif
#ifndef _NDELTA
#define XORENC( _u_, _pu_, _usize_) ((_u_)^(_pu_)) // xor predictor
#define XORDEC( _u_, _pu_, _usize_) ((_u_)^(_pu_))
#define ZZAGENC(_u_, _pu_, _usize_) T2(zigzagenc,_usize_)((_u_)-(_pu_)) //zigzag predictor
#define ZZAGDEC(_u_, _pu_, _usize_) (T2(zigzagdec,_usize_)(_u_)+(_pu_))
#define ECE while(ip != in+n) *out++ = *ip++
#define EDE ECE
//#define ELE for(;ip != in+n; ip++) l += (bsr8(ip[0])<<1)+1; return l;
//#define EL(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); l += (T2(bsr,_sz_)(T2(zigzagenc,_sz_)(z - u[_i_]))<<1)+1; u[_i_] = z; }
#define ELE for(;ip != in+n; ip++) l += bsr8(ip[0]); return l;
#if 1
#define EL(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); l += T2(bsr,_sz_)(T2(zigzagenc,_sz_)(z - u[_i_])); u[_i_] = z; }
#define EC(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = /*T2(zigzagenc,_sz_)*/(z - u[_i_]); u[_i_] = z; }
#define ED(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = (u[_i_] += /*T2(zigzagdec,_sz_)*/(z)); }
#elif 1 //best
#define EL(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); l += T2(bsr,_sz_)(T2(zigzagenc,_sz_)(z - u[_i_])); u[_i_] = z; }
#define EC(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = T2(zigzagenc,_sz_)(z - u[_i_]); u[_i_] = z; }
#define ED(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = (u[_i_] += T2(zigzagdec,_sz_)(z)); }
#elif 0
#define EL(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); l += T2(bsr,_sz_)(T2(zigzagenc,_sz_)(z ^ u[_i_])); u[_i_] = z; }
#define EC(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = z ^ u[_i_]; u[_i_] = z; }
#define ED(_i_,_sz_) { z = T2(ctou,_sz_)(ip+_i_*_sz_/8); T2(ctou,_sz_)(out+_i_*_sz_/8) = (u[_i_] ^= z); }
#endif
//----------- 24 -----------------
#ifndef NCOMP
//----------- 16 -----------------
unsigned delta8l16(uint8_t *in, size_t n) { size_t l=0; uint8_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~1); ip+= 2 ) { EL(0, 8);EL(1,8); } ELE; }
void delta8e16(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~1); ip+= 2,out+= 2) { EC(0, 8);EC(1,8); } ECE; }
void delta8d16(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~1); ip+= 2,out+= 2) { ED(0, 8);ED(1,8); } EDE; }
unsigned delta16l16(uint8_t *in, size_t n) { size_t l = 0; uint16_t u[1]={0},z; uint8_t *ip; for(ip = in; ip < in+(n&~1); ip+= 2 ) EL(0,16); ELE; }
void delta16e16(uint8_t *in, size_t n, uint8_t *out) { uint16_t u[1]={0},z; uint8_t *ip; for(ip = in; ip < in+(n&~1); ip+= 2,out+= 2) EC(0,16); ECE; }
void delta16d16(uint8_t *in, size_t n, uint8_t *out) { uint16_t u[1]={0},z; uint8_t *ip; for(ip = in; ip < in+(n&~1); ip+= 2,out+= 2) ED(0,16); EDE; }
unsigned delta8l24(uint8_t *in, size_t n) { size_t l = 0; uint8_t u[3]={0},z; uint8_t *ip; for(ip = in; ip < in+(n-3); ip+= 3 ) { EL(0, 8);EL(1,8);EL(2,8); } ELE; }
void delta8e24(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[3]={0},z; uint8_t *ip; for(ip = in; ip < in+(n-3); ip+= 3,out+= 3) { EC(0, 8);EC(1,8);EC(2,8); } ECE; }
//----------- 32 ----------------
unsigned delta8l32(uint8_t *in, size_t n) { size_t l = 0; uint8_t u[4]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4 ) { EL(0, 8);EL(1,8);EL(2,8);EL(3,8);} ELE; } //4D
void delta8e32(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[4]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) { EC(0, 8);EC(1,8);EC(2,8);EC(3,8);} ECE; }
void delta8d32(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[4]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) { ED(0, 8);ED(1,8);ED(2,8);ED(3,8);} EDE; }
unsigned delta16l32(uint8_t *in, size_t n) { size_t l = 0; uint16_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4 ) { EL(0,16);EL(1,16); } ELE; }
void delta16e32(uint8_t *in, size_t n, uint8_t *out) { uint16_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) { EC(0,16);EC(1,16); } ECE; }
void delta16d32(uint8_t *in, size_t n, uint8_t *out) { uint16_t u[2]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) { ED(0,16);ED(1,16); } EDE; }
unsigned delta32l32(uint8_t *in, size_t n) { size_t l = 0; uint32_t u[1]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4 ) EL(0,32); ELE; }
void delta32e32(uint8_t *in, size_t n, uint8_t *out) { uint32_t u[1]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) EC(0,32); ECE; }
void delta32d32(uint8_t *in, size_t n, uint8_t *out) { uint32_t u[1]={0},z; uint8_t *ip; for(ip = in; ip != in+(n&~3); ip+= 4,out+= 4) ED(0,32); EDE; }
void xorenc16(uint8_t *in, size_t inlen, uint8_t *out) {
uint16_t u0 = 0; uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip+=2,out+=2) {
ctou16(out) = ctou16(ip) ^ u0; u0 = ctou16(ip);
}
while(ip < in+inlen) *out++ = *ip++;
}
void xordec16(uint8_t *in, size_t inlen, uint8_t *out) {
uint16_t u0 = 0; uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip += 2,out += 2) {
ctou16(out) = u0 ^= ctou16(ip);
}
while(ip < in+inlen) *out++ = *ip++;
}
void zzagenc16(uint8_t *in, size_t inlen, uint8_t *out) {
int16_t s = 0,o;
uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip += 2, out += 2)
o = ctou16(ip), ctou16(out) = zigzagenc16((short)ctou16(ip) - s), s = o;
while(ip < in+inlen) *out++ = *ip++;
}
void zzagdec16(uint8_t *in, size_t inlen, uint8_t *out) {
int16_t s = 0; uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip += 2,out += 2) {
uint16_t x = ctou16(ip);
ctou16(out) = (s += zigzagdec16(x));
}
while(ip < in+inlen) *out++ = *ip++;
}
void nbenc16(uint8_t *in, size_t inlen, uint8_t *out) {
uint16_t u0 = 0,x; uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip+=2,out+=2) {
x = (short)ctou16(ip)-(short)u0;
ctou16(out) = nb_enc16(x); u0 = ctou16(ip);
}
while(ip < in+inlen) *out++ = *ip++;
}
void nbdec16(uint8_t *in, size_t inlen, uint8_t *out) {
uint16_t u0 = 0,x; uint8_t *ip;
for(ip = in; ip < in+(inlen&~1); ip += 2,out += 2) {
x = ctou16(ip);
ctou16(out) = (u0 += nb_dec16(x));
}
while(ip < in+inlen) *out++ = *ip++;
}
#endif
#ifndef NDECOMP
void delta8d24(uint8_t *in, size_t n, uint8_t *out) { uint8_t u[3]={0},z; uint8_t *ip; for(ip = in; ip < in+(n-3); ip+= 3,out+= 3) { ED(0, 8);ED(1,8);ED(2,8); } EDE; }
#endif
#endif
#ifndef _NQUANT
extern int verbose;
#ifndef min
#define min(x,y) (((x)<(y)) ? (x) : (y))
#define max(x,y) (((x)>(y)) ? (x) : (y))
#endif
//------------------------ Floating point statistics ------------------------------------------------------------------
#define BR(b) ((b/8)*100.0)/(double)(n*esize)
#define CMPA(_a_,_b_,_t_) ((*(_t_ *)(_a_) > *(_t_ *)(_b_)) - (*(_t_ *)(_a_) < *(_t_ *)(_b_)))
static int cmpua16(const void *a, const void *b) { return CMPA(a,b,uint16_t); }
static int cmpua32(const void *a, const void *b) { return CMPA(a,b,uint32_t); }
static int cmpua64(const void *a, const void *b) { return CMPA(a,b,uint64_t); }
void fpstat(unsigned char *in, size_t n, unsigned char *out, int s, unsigned char *_tmp) { if(verbose>0) printf("\nFloating point statistics\n");
double imin = DBL_MAX, imax = DBL_MIN, isum = 0, //original data (input) : minimum,maximum,sum
eamin = DBL_MAX, eamax = DBL_MIN, easum = 0, easumsqr = 0, //absolute error : abs(input-output)
ermin = DBL_MAX, ermax = DBL_MIN, ersum = 0, ersumsqr = 0, //relative error : abs(input-output)/abs(input)
osum = 0; //transformed lossy data (output) : sum
long long xtb = 0, xlb = 0, zlb = 0, tb = 0, lb = 0, elb = 0, mtb = 0, itb = 0;
size_t idn = 0;
unsigned char *ip, *op;
unsigned esize = s<0?-s:s, t = 0, uni = 0, zero=0;
long long mant = 0;
int expo = 0,e;
if(_tmp || verbose > 4) {
unsigned char *tmp = _tmp;
if(!tmp) { tmp = malloc(n*esize); if(!tmp) die("malloc failed\n"); } memcpy(tmp, out, n*esize);
switch(esize) {
case 2: { uint16_t *p,*t = tmp; qsort(tmp, n, 2, cmpua16); for(uni=zero=0,p = t; p < t+n-1; p++) { if(p[0] != p[1]) uni++; if(!p[0]) zero++; } } break;
case 4: { uint32_t *p,*t = tmp; qsort(tmp, n, 4, cmpua32); for(uni=zero=0,p = t; p < t+n-1; p++) { if(p[0] != p[1]) uni++; if(!p[0]) zero++; } } break;
case 8: { uint64_t *p,*t = tmp; qsort(tmp, n, 8, cmpua64); for(uni=zero=0,p = t; p < t+n-1; p++) { if(p[0] != p[1]) uni++; if(!p[0]) zero++; } } break;
default: die("#fpstat");
}
if(!_tmp) free(tmp);
}
for(ip = in, op = out; ip < in+n*esize; ip += esize, op += esize)
switch(s) {
#if defined(HAVE_FLOAT16)
case -2: isum += ctof16(ip); osum += ctof16(op); break;
#endif
case -4: isum += ctof32(ip); osum += ctof32(op); break;
case -8: isum += ctof64(ip); osum += ctof64(op); break;
case 1: isum += ctou8( ip); osum += ctou8( op); break;
case 2: isum += ctou16(ip); osum += ctou16(op); break;
case 4: isum += ctou32(ip); osum += ctou32(op); break;
case 8: isum += ctou64(ip); osum += ctou64(op); break;
}
double iavg = isum/n, oavg = osum/n, isumpavg = 0, osumpavg = 0, iosumpavg = 0; uint64_t xstart = 0, zstart = 0;
#define EXPO16(u) ((u>>10 & 0x1f) - 15 )
#define EXPO32(u) ((u>>23 & 0xff) - 0x7e )
#define EXPO64(u) ((u>>52 & 0x7ff) - 0x3fe)
#define MANT16(u) (u & 0x83ffu) //SeeeeeMMMMMMMMMM
#define MANT32(u) (u & 0x807fffffu)
#define MANT64(u) (u & 0x800fffffffffffffull)
#define U(s) T3(uint, s, _t) u = T2(ctou,s)(op), v = T2(ctou,s)(ip);\
itb += v?T2(ctz,s)(v):s; tb += u?T2(ctz,s)(u):s; lb += u?T2(clz,s)(u):s; AC(t<=s,"Fatal t=%d ", t); \
xstart ^= u; xtb += xstart?T2(ctz,s)(xstart):s; xlb += xstart?T2(clz,s)(xstart):0; xstart = u;\
zstart = T2(zigzagenc,s)(u - zstart); zlb += zstart?T2(clz,s)(zstart):s; zstart = u
for(ip = in, op = out; ip < in+n*esize; ip += esize, op += esize) {
double id, od;
unsigned e; uint64_t m;
switch(s) {
#if defined(HAVE_FLOAT16)
case -2: { unsigned e; uint16_t m;id = ctof16(ip); od = ctof16(op); U(16); e = EXPO16(u); expo = clz16(zigzagenc16(e-expo))/*-(16-(16-MANTF16-1))*/; elb+=expo; expo = e;
m = MANT16(u); mant = ctz16( m^mant) ; mtb+=mant; mant = m;//ctz16(zigzagenc16(m-mant))
} break;
#endif
case -4: { unsigned e; uint32_t m;id = ctof32(ip); od = ctof32(op); U(32); e = EXPO32(u); expo = clz32(zigzagenc32(e-expo))/*-(32-(32-MANTF32-1))*/; elb+=expo; expo = e;
m = MANT32(u); mant = ctz32( m^mant) ; mtb+=mant; mant = m;//ctz32(zigzagenc32(m-mant))
} break;
case -8: { unsigned e; uint64_t m;id = ctof64(ip); od = ctof64(op); U(64); e = EXPO64(u); expo = clz32(zigzagenc32(e-expo))/*-(32-(64-MANTF64-1))*/; elb+=expo; expo = e;
m = MANT64(u); mant = ctz64( m^mant) ; mtb+=mant; mant = m;//ctz64(zigzagenc64(m-mant))
} break;
case 1: { id = ctou8( ip); od = ctou8( op); U( 8);} break;
case 2: { id = ctou16(ip); od = ctou16(op); U(16);} break;
case 4: { id = ctou32(ip); od = ctou32(op); U(32);} break;
case 8: { id = ctou64(ip); od = ctou64(op); U(64);} break;
}
imax = max(id, imax);
imin = min(id, imin);
double ea = fabs(id - od); eamax = max(eamax,ea); eamin = min(eamin,ea); easum += ea; easumsqr += ea*ea; // absolute error
if(id) { idn++;
double er = ea/fabs(id); ermax = max(ermax,er); ermin = min(ermin,er); ersum += er; ersumsqr += er*er; // relative error
}
isumpavg += (id - iavg)*(id - iavg);
osumpavg += (od - oavg)*(od - oavg);
iosumpavg += (id - iavg)*(od - oavg); //bits += ctz64(ctou64(&od)) - ctz64(ctou64(&id));
}
double fb = 0;
if(s == -2) fb = (double)elb*100/((double)n*5);
else if(s == -4) fb = (double)elb*100/((double)n*8);
else if(s == -8) fb = (double)elb*100/((double)n*11);
double mse = easumsqr/n, irange = imax - imin;
if(verbose >= 2) printf("\n");
//printf("Leading/Trailing bits [%.2f%%,%.2f%%=%.2f%%]. XOR[%.2f%%,%.2f%%=%.2f%%] Zigzag[%.2f%%]\n", BR(lb), BR(tb), BR(lb+tb), BR(xlb), BR(xtb), BR(xlb+xtb), BR(zlb)/*BR(elb), BR(mtb), BR(elb+mtb)*/ );
if(verbose >= 2) printf("Range: [min=%g / max=%g] = %g\n", imin, imax, irange);
if(verbose > 3 || _tmp) printf("zeros=[%u,%.2f%%], Distinct=[%u=%.4f%%] ctz=%.1f%%\n", zero,(double)zero*100.0/(double)n, uni, (double)uni*100.0/(double)n, (double)((tb-itb)/8)*100.0/(double)(n*esize));
//if(verbose > 2) printf("Min error: Absolute = %g, Relative = %g, pointwise relative(PWE) = %g\n", eamin, eamin/irange, eamax/irange, ermax);
//if(verbose > 2) printf("Avg error: Absolute = %g, Relative = %g, pointwise relative(PWE) = %g\n", easum/idn, (easum/idn)/irange, ersum/idn);
if(verbose > 2) printf("Max error: Absolute = %g, Relative = %g, pointwise relative(PWE) = %g\n", eamax, eamax/irange, ermax); else if(verbose==2) printf("e=%g ", ermax);
double psnr=20*log10(irange)-10*log10(mse);
if(verbose > 2) printf("Peak Signal-to-Noise Ratio: PSNR = %.1f\n", psnr); else if(verbose==2) printf("PSNR=%.0f ", psnr);
if(verbose > 2) printf("Normalized Root Mean Square Error: NRMSE = %g\n", sqrt(mse)/irange); else if(verbose==2) printf("NRMSE=%g ", sqrt(mse)/irange);
double std1 = sqrt(isumpavg/n), std2 = sqrt(osumpavg/n), ee = iosumpavg/n, acEff = (iosumpavg/n)/sqrt(isumpavg/n)/sqrt(osumpavg/n);
if(verbose > 2) printf("Pearson Correlation Coefficient = %f\n", (iosumpavg/n)/sqrt(isumpavg/n)/sqrt(osumpavg/n));
}
//----------- Quantization -----------------------------------
#define ROUND16(x) roundf(x)
#define ROUND32(x) roundf(x)
#define ROUND64(x) round(x) //#include "include_/vlcbyte.h"
#define QUANTE(t_s, _x_, _fmin_, _delta_) T2(ROUND,t_s)(((_x_) - _fmin_)*_delta_) //T2(ROUND,t_s)((_x_) * 100)
#define _FPQUANTE( t_s, _op_, _x_, _fmin_, _delta_) *_op_++ = QUANTE(t_s, _x_, _fmin_, _delta_) //#define _FPQUANTVE(t_s, _op_, _x_, _fmin_, _delta_) { uint16_t _u = QUANTE(t_s, _x_, _fmin_, _delta_); vsput20(_op_,_u); }
#define FPQUANTE(t_t, _in_, inlen, _op_, qmax, t_s, pfmin, pfmax, _zmin_, _fpquante_) {\
t_t fmin = _in_[0], fmax = _in_[0], *_ip;\
for(_ip = _in_; _ip < _in_+(inlen/(t_s/8)); _ip++)\
if(*_ip > fmax) fmax = *_ip; else if(*_ip < fmin) fmin = *_ip;\
*pfmin = fmin; *pfmax = fmax;\
t_t _delta = (fmax - fmin <= _zmin_)?(t_t)0.0:qmax/(fmax - fmin);\
for(_ip = _in_; _ip < _in_+(inlen/(t_s/8)); _ip++)\
_fpquante_(t_s, _op_, _ip[0], fmin, _delta);\
}
#define FPQUANTE8(t_t, _in_, _inlen_, _out_, qmax, t_s, pfmin, pfmax, _zmin_, _fpquante_) {\
t_t fmin = *pfmin, fmax = *pfmax, *_ip;\
unsigned char *_op = _out_, *_ep_ = _out_ + _inlen_, *_ep = _ep_; /*unsigned cm = 0,cx = 0;*/\
if(fmin == 0.0 && fmax == 0.0) {\
fmax = _in_[0], fmin = _in_[0];\
for(_ip = _in_; _ip < _in_ + (_inlen_/(t_s/8)); _ip++)\
if(*_ip > fmax) fmax = *_ip; else if(*_ip < fmin) fmin = *_ip;\
*pfmin = fmin; *pfmax = fmax;\
} else qmax--;\
t_t _delta = (fmax - fmin <= _zmin_)?(t_t)0.0:qmax/(fmax - fmin);\
\
for(_ip = _in_; _ip < _in_+(_inlen_/(t_s/8)); _ip++) {\
t_t _f = _ip[0];\
if(_f < fmin || _f > fmax) { *_op++ = qmax+1; _ep -= (t_s/8); T2(ctof,t_s)(_ep) = _f; /*_f > fmax?cx++:cm++;*/ } /*store outliers w/o compression at the buffer end*/\
else _fpquante_(t_s, _op, _f, fmin, _delta); \
if(_op+8 >= _ep) goto ovr;\
} /*if(verbose > 2) printf("qmax=%u outliers:%u+%u=%u ",qmax, cm, cx, cm+cx);*/\
unsigned _l = _ep_ - _ep; if(_op+_l >= _ep_) goto ovr;\
memcpy(_op, _ep, _l); _op += _l;\
return _op - _out_;\
ovr: if(verbose>2) printf("overflow:%zu ", _inlen_); \
memcpy(_out_, _in_, _inlen_); return _inlen_;\
}
#define QUANTD(_x_) fmin + (_x_) * fmax
#define _FPQUANTD(t_t, _ip_, _x_) _x_ = fmin + (t_t)(*_ip_++) * fmax //#define _FPQUANTVD(_ip_, _x_) { unsigned _u; vsget20(_ip_,_u); _x_ = QUANTD(_u); }
#define FPQUANTD(t_t, _ip_, outlen, _out_, qmax, fmin, fmax, _fpquantd_) { t_t *_op;\
fmax = (fmax - fmin) / qmax;\
for(_op = _out_; _op < _out_+(outlen/sizeof(_out_[0])); _op++) _fpquantd_(t_t, _ip_, _op[0]);\
}
#define FPQUANTD8(t_t, _ip_, outlen, _out_, qmax, t_s, fmin, fmax, _fpquantd_) { \
t_t *_op; unsigned char *_ep = _ip_ + inlen;\
qmax--; fmax = (fmax - fmin) / qmax;\
for(_op = _out_; _op < _out_+(outlen/(t_s/8)); _op++)\
if(*_ip_ == qmax+1) { _ip_++; _ep -= t_s/8; *_op = T2(ctof,t_s)(_ep); }\
else _fpquantd_(t_t, _ip_, _op[0]);\
}
#if defined(HAVE_FLOAT16)
size_t fpquant8e16( _Float16 *in, size_t inlen, uint8_t *out, unsigned qmax, _Float16 *pfmin, _Float16 *pfmax, _Float16 zmin) { FPQUANTE8(_Float16, in, inlen, out, qmax, 16, pfmin, pfmax, zmin,_FPQUANTE); }
size_t fpquant16e16(_Float16 *in, size_t inlen, uint16_t *out, unsigned qmax, _Float16 *pfmin, _Float16 *pfmax, _Float16 zmin) { FPQUANTE( _Float16, in, inlen, out, qmax, 16, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
//size_t fpquantv8e16(_Float16 *in, size_t inlen, uint8_t *out, unsigned qmax, _Float16 *pfmin, _Float16 *pfmax, _Float16 zmin) { unsigned char *op = out; FPQUANTE(_Float16, in, inlen, op, qmax, 16, pfmin, pfmax, zmin,_FPQUANTVE); return op - out; }
#endif
size_t fpquant8e32( float *in, size_t inlen, uint8_t *out, unsigned qmax, float *pfmin, float *pfmax, float zmin) { FPQUANTE8( float, in, inlen, out, qmax, 32, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant16e32( float *in, size_t inlen, uint16_t *out, unsigned qmax, float *pfmin, float *pfmax, float zmin) { FPQUANTE( float, in, inlen, out, qmax, 32, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant32e32( float *in, size_t inlen, uint32_t *out, unsigned qmax, float *pfmin, float *pfmax, float zmin) { FPQUANTE( float, in, inlen, out, qmax, 32, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant8e64( double *in, size_t inlen, uint8_t *out, unsigned qmax, double *pfmin, double *pfmax, double zmin) { FPQUANTE8( double, in, inlen, out, qmax, 64, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant16e64( double *in, size_t inlen, uint16_t *out, unsigned qmax, double *pfmin, double *pfmax, double zmin) { FPQUANTE( double, in, inlen, out, qmax, 64, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant32e64( double *in, size_t inlen, uint32_t *out, unsigned qmax, double *pfmin, double *pfmax, double zmin) { FPQUANTE( double, in, inlen, out, qmax, 64, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
size_t fpquant64e64( double *in, size_t inlen, uint64_t *out, unsigned qmax, double *pfmin, double *pfmax, double zmin) { FPQUANTE( double, in, inlen, out, qmax, 64, pfmin, pfmax, zmin,_FPQUANTE); return inlen; }
#if defined(HAVE_FLOAT16)
size_t fpquant8d16( uint8_t *in, size_t outlen, _Float16 *out, unsigned qmax, _Float16 fmin, _Float16 fmax, size_t inlen) { FPQUANTD8(_Float16, in, outlen, out, qmax, 16, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant16d16( uint16_t *in, size_t outlen, _Float16 *out, unsigned qmax, _Float16 fmin, _Float16 fmax) { FPQUANTD(_Float16, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
//size_t fpquantv8d16( uint8_t *in, size_t outlen, _Float16 *out, unsigned qmax, _Float16 fmin, _Float16 fmax) { unsigned char *ip = in; FPQUANTD(_Float16, in, outlen, out, qmax, fmin, fmax, _FPQUANTVD); return ip - in; }
#endif
size_t fpquant8d32( uint8_t *in, size_t outlen, float *out, unsigned qmax, float fmin, float fmax, size_t inlen) { FPQUANTD8( float, in, outlen, out, qmax, 32, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant16d32( uint16_t *in, size_t outlen, float *out, unsigned qmax, float fmin, float fmax) { FPQUANTD( float, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant32d32( uint32_t *in, size_t outlen, float *out, unsigned qmax, float fmin, float fmax) { FPQUANTD( float, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant8d64( uint8_t *in, size_t outlen, double *out, unsigned qmax, double fmin, double fmax, size_t inlen) { FPQUANTD8( double, in, outlen, out, qmax, 64, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant16d64( uint16_t *in, size_t outlen, double *out, unsigned qmax, double fmin, double fmax) { FPQUANTD( double, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant32d64( uint32_t *in, size_t outlen, double *out, unsigned qmax, double fmin, double fmax) { FPQUANTD( double, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
size_t fpquant64d64( uint64_t *in, size_t outlen, double *out, unsigned qmax, double fmin, double fmax) { FPQUANTD( double, in, outlen, out, qmax, fmin, fmax, _FPQUANTD); return outlen;}
//----------- Lossy floating point conversion: pad the trailing mantissa bits with zero bits according to the relative error e (ex. 0.00001) ----------
#if defined(HAVE_FLOAT16)
// https://clang.llvm.org/docs/LanguageExtensions.html#half-precision-floating-point
_Float16 _fprazor16(_Float16 d, float e, int lg2e) {
uint16_t du = ctou16(&d), sign, u;
int b = (du>>10 & 0x1f) - 15; // exponent=[5 bits,bias=15], mantissa=10 bits SeeeeeMMMMMMMMMM
_Float16 ed;
if ((b = 12 - b - lg2e) <= 0)
return d;
b = b > 10?10:b;
sign = du & (1<<15);
du &= 0x7fff;
for(d = ctof16(&du), ed = e * d;;) {
u = du & (~((1u<<(--b))-1)); if(d - ctof16(&u) <= ed) break;
u = du & (~((1u<<(--b))-1)); if(d - ctof16(&u) <= ed) break;
}
u |= sign;
return ctof16(&u);
}
void fprazor16(_Float16 *in, unsigned n, _Float16 *out, float e) {
int lg2e = -log(e)/log(2.0); _Float16 *ip;
for (ip = in; ip < in+n; ip++,out++)
*out = _fprazor16(*ip, e, lg2e);
}
#endif
float _fprazor32(float d, float e, int lg2e) {
uint32_t du = ctou32(&d), sign, u;
int b = (du>>23 & 0xff) - 0x7e;
float ed;
if((b = 25 - b - lg2e) <= 0)
return d; AS(!isnan(d), "_fprazor32: isnan");
b = b > 23?23:b;
sign = du & (1<<31);
du &= 0x7fffffffu;
for(d = ctof32(&du), ed = e * d;;) {
u = du & (~((1u<<(--b))-1)); if(d - ctof32(&u) <= ed) break;
u = du & (~((1u<<(--b))-1)); if(d - ctof32(&u) <= ed) break;
u = du & (~((1u<<(--b))-1)); if(d - ctof32(&u) <= ed) break;
}
u |= sign;
return ctof32(&u);
}
void fprazor32(float *in, unsigned n, float *out, float e) {
int lg2e = -log(e)/log(2.0);
float *ip;
for(ip = in; ip < in+n; ip++,out++)
*out = _fprazor32(*ip, e, lg2e);
}
double _fprazor64(double d, double e, int lg2e) { //if(isnan(d)) return d;
uint64_t du = ctou64(&d), sign, u;
int b = (du>>52 & 0x7ff) - 0x3fe;
double ed;
if((b = 54 - b - lg2e) <= 0)
return d;
b = b > 52?52:b;
sign = du & (1ull<<63);