daiduo/TurboPFor-Integer-Compression
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

TurboPFor: TurboVByte - Variable byte

Browse Source
This commit is contained in:
x
2023-03-10 20:28:32 +01:00
parent 269a503806
commit 46001f9856
1 changed files with 72 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

148
vint.c
View File

@ -1,5 +1,5 @@
/**
Copyright (C) powturbo 2013-2019
Copyright (C) powturbo 2013-2023
GPL v2 License
This program is free software; you can redistribute it and/or modify
@ -21,19 +21,25 @@
- twitter : https://twitter.com/powturbo
- email : powturbo [_AT_] gmail [_DOT_] com
**/
// vint.c - "Integer Compression" variable byte
#ifndef USIZE
#include <stdio.h>
#include <string.h>
// vint.c - "Integer Compression" variable byte
#ifndef USIZE
#pragma warning( disable : 4005)
#pragma warning( disable : 4090)
#pragma warning( disable : 4068)
#define BITUTIL_IN
#define VINT_IN
#include "conf.h"
#include "vint.h"
#include "bitutil.h"
#include <stdio.h>
#include <string.h>
#include "include_/conf.h"
#include "include_/bitutil.h"
#include "include_/vint.h"
#define VB_MAX 0xfd // reserved : 0xfe=all zeros, 0xff=overflow
#include "include_/vlcbyte.h"
#include "include_/bitutil_.h"
//----------------------------------------------------------------
#define OVERFLOWD(in,n,out) if(*in == 0xff) { memcpy(out, in+1, n*(USIZE/8)); return in+1+n*(USIZE/8); }
#define OVERFLOWE(in,n,out,op, _goto_) if(op+32 > out + (uint64_t)n*(USIZE/8)) { *out = 0xff; memcpy(out+1, in, n*(USIZE/8)); op = out+1+n*(USIZE/8); _goto_; }
#define UN 8 // 4 //
@ -80,22 +86,17 @@
#define USIZE 8
#include "vint.c"
#undef USIZE
#else
#define uint_t TEMPLATE3(uint, USIZE, _t)
#define uint_t T3(uint, USIZE, _t)
#if VDELTA == 0
//#if USIZE < 64
#define OVERFLOWD(in,n,out,vbmax) if(*in == (vbmax)) { memcpy(out, in+1, n*(USIZE/8)); return in+1+n*(USIZE/8); }
#define OVERFLOWE(in,n,out,op,vbmax) if(op > out + n*(USIZE/8)) { *out = (vbmax); memcpy(out+1, in, n*(USIZE/8)); op = out+1+n*(USIZE/8); }
//#else
//#define OVERFLOWD(in,n,out,vbmax)
//#define OVERFLOWE(in,n,out,op,vbmax)
//#endif
unsigned char *TEMPLATE2(vbdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out) {
register uint_t x, *op;
OVERFLOWD(in, n, out, VB_MAX+1);
#define VBE(_i_) TEMPLATE2(_vbget, USIZE)(in, x, op[_i_] = x)
unsigned char *T2(vbdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out) {
register uint_t x, *op;
OVERFLOWD(in, n, out);
#define VBE(_i_) T2(_vbget, USIZE)(in, x, op[_i_] = x)
for(op = out; op != out+(n&~(UN-1)); op += UN) {
VBE(0); VBE(1); VBE(2); VBE(3);
#if UN > 4
@ -104,16 +105,16 @@ unsigned char *TEMPLATE2(vbdec, USIZE)(unsigned char *__restrict in, unsigned n
PREFETCH(in+16*USIZE, 0);
}
while(op != out+n)
TEMPLATE2(_vbget, USIZE)(in, x, *op++ = x );
T2(_vbget, USIZE)(in, x, *op++ = x );
return in;
}
#undef VBE
unsigned char *TEMPLATE2(vbenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out) {
unsigned char *T2(vbenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out) {
uint_t x, *ip;
unsigned char *op = out;
#define VBD(_i_) x = ip[_i_]; TEMPLATE2(_vbput, USIZE)(op, x, ;);
#define VBD(_i_) x = ip[_i_]; T2(_vbput, USIZE)(op, x, ;);
for(ip = in; ip != in+(n&~(UN-1)); ip += UN) { PREFETCH(ip+USIZE*8, 0);
VBD(0); VBD(1); VBD(2); VBD(3);
#if UN > 4
@ -122,37 +123,37 @@ unsigned char *TEMPLATE2(vbenc, USIZE)(uint_t *__restrict in, unsigned n, unsign
}
while(ip != in+n) {
x = *ip++;
TEMPLATE2(_vbput, USIZE)(op, x, ;);
T2(_vbput, USIZE)(op, x, ;);
}
OVERFLOWE(in,n,out,op,VB_MAX+1);
OVERFLOWE(in,n,out,op,;);
return op;
}
#undef VBD
uint_t TEMPLATE2(vbgetx, USIZE)(unsigned char *__restrict in, unsigned idx) {
uint_t T2(vbgetx, USIZE)(unsigned char *__restrict in, unsigned idx) {
unsigned char *ip;
unsigned i;
uint_t x;
if(*in == 255)
return TEMPLATE2(ctou, USIZE)(in+1+idx*(USIZE/8));
return T2(ctou, USIZE)(in+1+idx*(USIZE/8));
for(ip = in,i = 0; i <= idx; i++)
ip += TEMPLATE2(_vbvlen, USIZE)(*ip);
TEMPLATE2(_vbget, USIZE)(in, x, ;);
ip += T2(vbvlen, USIZE)(*ip);
T2(_vbget, USIZE)(in, x, ;);
return x;
}
/*unsigned TEMPLATE2(vbgeteq, USIZE)(unsigned char *__restrict in, unsigned n, uint_t key, unsigned char **__restrict _ip) {
/*unsigned T2(vbgeteq, USIZE)(unsigned char *__restrict in, unsigned n, uint_t key, unsigned char **__restrict _ip) {
unsigned i;
unsigned char *ip;
uint_t x;
if(*in == 255) {
for(ip = (*_ip==in)?in:*ip; ip < in+n; ip+USIZE/8) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
if((x = TEMPLATE2(ctou, USIZE)(ip)) == key) break;
T2(_vbget, USIZE)(ip, x, ;);
if((x = T2(ctou, USIZE)(ip)) == key) break;
}
} else for(ip = *_ip,i=idx; i < n; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
if(x == key) break;
}
*_ip = ip;
@ -162,21 +163,21 @@ uint_t TEMPLATE2(vbgetx, USIZE)(unsigned char *__restrict in, unsigned idx) {
#define BIT_SET( p, n) (p[(n)/BITS] |= (0x80>>((n)%BITS)))
#define BIT_CLEAR(p, n) (p[(n)/BITS] &= ~(0x80>>((n)%BITS)))
#define BIT_ISSET(p, n) (p[(n)/BITS] & (0x80>>((n)%BITS)))
unsigned char *TEMPLATE2(vbddenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
unsigned char *T2(vbddenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
uint_t *ip,v,pd=0,*p;
unsigned char *bp=out; out += (n+7)/8;
#define VBDDE(i) { uint_t x; start = ip[i]-start; x = start-pd; pd = start; \
if(!x) BIT_CLEAR(bp,i); else { BIT_SET(bp,i); x = TEMPLATE2(zigzagenc, USIZE)(x); TEMPLATE2(_vbput, USIZE)(out, x, ;); } start = ip[i]; }
if(!x) BIT_CLEAR(bp,i); else { BIT_SET(bp,i); x = T2(zigzagenc, USIZE)(x); T2(_vbput, USIZE)(out, x, ;); } start = ip[i]; }
for(ip = in; ip != in+(n&~(8-1)); ip+=8,bp++) { VBDDE(0);VBDDE(1);VBDDE(2);VBDDE(3);VBDDE(4);VBDDE(5);VBDDE(6);VBDDE(7);}
for(p=ip; p != in+n; p++) VBDDE(p-ip);
return out;
}
unsigned char *TEMPLATE2(vbdddec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
unsigned char *T2(vbdddec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
uint_t *op,pd=0,*p;
unsigned i;
#define VBDDD(i) { uint_t x=0; if(BIT_ISSET(bp,i)) { TEMPLATE2(_vbget, USIZE)(in, x, ;); pd += TEMPLATE2(zigzagdec, USIZE)(x); } op[i] = (start += pd); }
#define VBDDD(i) { uint_t x=0; if(BIT_ISSET(bp,i)) { T2(_vbget, USIZE)(in, x, ;); pd += T2(zigzagdec, USIZE)(x); } op[i] = (start += pd); }
unsigned char *bp=in; in+=(n+7)/8;
for(op = out; op != out+(n&~(8-1)); op+=8,bp++) {
if(!bp[0]) { op[0]=(start+=pd); op[1]=(start+=pd); op[2]=(start+=pd); op[3]=(start+=pd); op[4]=(start+=pd); op[5]=(start+=pd); op[6]=(start+=pd); op[7]=(start+=pd); continue; }
@ -188,21 +189,21 @@ unsigned char *TEMPLATE2(vbdddec, USIZE)(unsigned char *__restrict in, unsigned
#undef VBDDE
#undef VBDDD
unsigned char *TEMPLATE2(vbzenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
unsigned char *T2(vbzenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
uint_t *ip,v;
unsigned char *op = out;
#define VBZE { v = TEMPLATE2(zigzagenc, USIZE)((TEMPLATE3(int, USIZE, _t))(*ip)-(TEMPLATE3(int, USIZE, _t))start); start=*ip++; TEMPLATE2(_vbput, USIZE)(op, v, ;); }
#define VBZE { v = T2(zigzagenc, USIZE)((T3(int, USIZE, _t))(*ip)-(T3(int, USIZE, _t))start); start=*ip++; T2(_vbput, USIZE)(op, v, ;); }
for(ip = in; ip != in+(n&~(UN-1)); ) { VBZE;VBZE;VBZE;VBZE; }
while(ip != in+n) VBZE; //OVERFLOWE(in,n,out,op);
return op;
}
#undef VBZE
unsigned char *TEMPLATE2(vbzdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
unsigned char *T2(vbzdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
uint_t x,*op;
#define VBZD { TEMPLATE2(_vbget, USIZE)(in, x, ;); *op++ = (start += TEMPLATE2(zigzagdec, USIZE)(x)); }
#define VBZD { T2(_vbget, USIZE)(in, x, ;); *op++ = (start += T2(zigzagdec, USIZE)(x)); }
for(op = out; op != out+(n&~(UN-1)); ) { VBZD; VBZD; VBZD; VBZD;
#if UN > 4
VBZD; VBZD; VBZD; VBZD;
@ -215,24 +216,24 @@ unsigned char *TEMPLATE2(vbzdec, USIZE)(unsigned char *__restrict in, unsigned n
}
#undef VBZD
uint_t TEMPLATE2(vbzgetx, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
uint_t T2(vbzgetx, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
unsigned char *ip;
unsigned i;
uint_t x;
for(ip = in,i = 0; i <= idx; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
start += x+VDELTA;
}
return start;
}
unsigned TEMPLATE2(vbzgeteq, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t key, uint_t start ) {
unsigned T2(vbzgeteq, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t key, uint_t start ) {
unsigned i;
unsigned char *ip;
uint_t x;
for(ip = *in,i=idx; i < n; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
if((start += x+VDELTA) == key)
break;
}
@ -240,21 +241,21 @@ unsigned TEMPLATE2(vbzgeteq, USIZE)(unsigned char **__restrict in, unsigned n, u
return i;
}
unsigned char *TEMPLATE2(vbxenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
unsigned char *T2(vbxenc, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
uint_t *ip,v;
unsigned char *op = out;
#define VBXE { v = (*ip)^start; start=*ip++; TEMPLATE2(_vbput, USIZE)(op, v, ;); }
#define VBXE { v = (*ip)^start; start=*ip++; T2(_vbput, USIZE)(op, v, ;); }
for(ip = in; ip != in+(n&~(UN-1)); ) { VBXE;VBXE;VBXE;VBXE; }
while(ip != in+n) VBXE; //OVERFLOWE(in,n,out,op);
return op;
}
#undef VBZE
unsigned char *TEMPLATE2(vbxdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
unsigned char *T2(vbxdec, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
uint_t x,*op;
#define VBXD { TEMPLATE2(_vbget, USIZE)(in, x, ;); *op++ = (start ^= x); }
#define VBXD { T2(_vbget, USIZE)(in, x, ;); *op++ = (start ^= x); }
for(op = out; op != out+(n&~(UN-1)); ) { VBXD; VBXD; VBXD; VBXD;
#if UN > 4
VBXD; VBXD; VBXD; VBXD;
@ -267,24 +268,24 @@ unsigned char *TEMPLATE2(vbxdec, USIZE)(unsigned char *__restrict in, unsigned n
}
#undef VBZD
uint_t TEMPLATE2(vbxgetx, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
uint_t T2(vbxgetx, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
unsigned char *ip;
unsigned i;
uint_t x;
for(ip = in,i = 0; i <= idx; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
start ^= x;
}
return start;
}
unsigned TEMPLATE2(vbxgeteq, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t key, uint_t start ) {
unsigned T2(vbxgeteq, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t key, uint_t start ) {
unsigned i;
unsigned char *ip;
uint_t x;
for(ip = *in,i=idx; i < n; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
if((start ^= x) == key)
break;
}
@ -293,45 +294,40 @@ unsigned TEMPLATE2(vbxgeteq, USIZE)(unsigned char **__restrict in, unsigned n, u
}
#endif
unsigned char *TEMPLATE2(VBDENC, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
unsigned char *T2(VBDENC, USIZE)(uint_t *__restrict in, unsigned n, unsigned char *__restrict out, uint_t start) {
unsigned char *op = out;
uint_t *ip, b = 0,v;
if(!n) return out;
#if USIZE == 64
#define VB_MX 255
#else
#define VB_MX VB_MAX
#endif
#define VBDE { v = ip[0]-start-VDELTA; start = *ip++; TEMPLATE2(_vbput, USIZE)(op, v, ;); b |= (v /*^ x*/); }
#define VBDE { v = ip[0]-start-VDELTA; start = *ip++; T2(_vbput, USIZE)(op, v, ;); b |= (v /*^ x*/); }
for(ip = in; ip != in + (n&~(UN-1)); ) { VBDE; VBDE; VBDE; VBDE;
#if UN > 4
VBDE; VBDE; VBDE; VBDE;
#endif
}
while(ip != in+n) VBDE;
if(!b) { op = out; *op++ = VB_MX; } // if (x) { op = out; *op++ = VB_MAX-2; TEMPLATE2(_vbput, USIZE)(op, x, ;); }
if(!b) { op = out; *op++ = 0xfe; } // if (x) { op = out; *op++ = VB_MAX-2; T2(_vbput, USIZE)(op, x, ;); }
#if USIZE < 64
OVERFLOWE(in,n,out,op,VB_MAX-1);
OVERFLOWE(in,n,out,op, ;);
#endif
return op;
}
#undef VBDE
unsigned char *TEMPLATE2(VBDDEC, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
unsigned char *T2(VBDDEC, USIZE)(unsigned char *__restrict in, unsigned n, uint_t *__restrict out, uint_t start) {
uint_t x,*op;
if(!n) return in;
#if USIZE < 64
OVERFLOWD(in,n,out,VB_MAX-1);
OVERFLOWD(in,n,out);
#endif
if(in[0] == VB_MX) {
if(in[0] == 0xfe) {
in++;
#if (defined(__SSE2__) || defined(__ARM_NEON)) && USIZE == 32
#if VDELTA == 0
if(n) TEMPLATE2(BITZERO, USIZE)(out, n, start);
if(n) T2(BITZERO, USIZE)(out, n, start);
#else
if(n) TEMPLATE2(BITDIZERO,USIZE)(out, n, start, VDELTA);
if(n) T2(BITDIZERO,USIZE)(out, n, start, VDELTA);
#endif
#else
#if VDELTA == 0
@ -343,9 +339,9 @@ unsigned char *TEMPLATE2(VBDDEC, USIZE)(unsigned char *__restrict in, unsigned n
return in;
}
#if 0 //USIZE < 64
else if(in[0] == VB_MAX-2) { in++;
else if(in[0] == 0xfe) { in++;
uint_t z;
TEMPLATE2(_vbget, USIZE)(in, z, ;);
T2(_vbget, USIZE)(in, z, ;);
#if VDELTA == 0
for(x = 0; x < n; x++) out[x] = start+z;
#else
@ -354,7 +350,7 @@ unsigned char *TEMPLATE2(VBDDEC, USIZE)(unsigned char *__restrict in, unsigned n
return in;
}
#endif
#define VBDD(i) { TEMPLATE2(_vbget, USIZE)(in, x, x+=VDELTA); op[i] = (start += x); }
#define VBDD(i) { T2(_vbget, USIZE)(in, x, x+=VDELTA); op[i] = (start += x); }
for(op = out; op != out+(n&~(UN-1)); op+=UN) {
VBDD(0); VBDD(1); VBDD(2); VBDD(3);
#if UN > 4
@ -367,7 +363,7 @@ unsigned char *TEMPLATE2(VBDDEC, USIZE)(unsigned char *__restrict in, unsigned n
}
#undef VBDD
uint_t TEMPLATE2(VBDGETX, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
uint_t T2(VBDGETX, USIZE)(unsigned char *__restrict in, unsigned idx, uint_t start) {
unsigned char *ip;
unsigned i=0;
uint_t x;
@ -378,13 +374,13 @@ uint_t TEMPLATE2(VBDGETX, USIZE)(unsigned char *__restrict in, unsigned idx, ui
if(u & 1) return start + VDELTA;
#endif
for(ip = in; i <= idx; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
start += x+VDELTA;
}
return start;
}
unsigned TEMPLATE2(VBDGETGEQ, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t *key, uint_t start ) {
unsigned T2(VBDGETGEQ, USIZE)(unsigned char **__restrict in, unsigned n, unsigned idx, uint_t *key, uint_t start ) {
unsigned i=0;
unsigned char *ip=*in;
uint_t x;
@ -396,7 +392,7 @@ unsigned TEMPLATE2(VBDGETGEQ, USIZE)(unsigned char **__restrict in, unsigned n,
}
#endif
for(ip = *in; i < n; i++) {
TEMPLATE2(_vbget, USIZE)(ip, x, ;);
T2(_vbget, USIZE)(ip, x, ;);
if((start += x+VDELTA) == *key)
break;
}