release/src/router/libvorbis/vq/train.c

   1 /********************************************************************
   2  *                                                                  *
   3  * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
   4  * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
   5  * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
   6  * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
   7  *                                                                  *
   8  * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2001             *
   9  * by the Xiph.Org Foundation http://www.xiph.org/                  *
  10  *                                                                  *
  11  ********************************************************************
  12
  13  function: utility main for training codebooks
  14  last mod: $Id: train.c 16037 2009-05-26 21:10:58Z xiphmont $
  15
  16  ********************************************************************/
  17
  18 #include <stdlib.h>
  19 #include <stdio.h>
  20 #include <math.h>
  21 #include <string.h>
  22 #include <errno.h>
  23 #include <signal.h>
  24 #include "vqgen.h"
  25 #include "vqext.h"
  26 #include "bookutil.h"
  27
  28 static char *rline(FILE *in,FILE *out,int pass){
  29   while(1){
  30     char *line=get_line(in);
  31     if(line && line[0]=='#'){
  32       if(pass)fprintf(out,"%s\n",line);
  33     }else{
  34       return(line);
  35     }
  36   }
  37 }
  38
  39 /* command line:
  40    trainvq  vqfile [options] trainfile [trainfile]
  41
  42    options: -params     entries,dim,quant
  43             -subvector  start[,num]
  44             -error      desired_error
  45             -iterations iterations
  46 */
  47
  48 static void usage(void){
  49   fprintf(stderr, "\nOggVorbis %s VQ codebook trainer\n\n"
  50           "<foo>vqtrain vqfile [options] [datasetfile] [datasetfile]\n"
  51           "options: -p[arams]     <entries,dim,quant>\n"
  52           "         -s[ubvector]  <start[,num]>\n"
  53           "         -e[rror]      <desired_error>\n"
  54           "         -i[terations] <maxiterations>\n"
  55           "         -d[istance]   quantization mesh spacing for density limitation\n"
  56           "         -b <dummy>    eliminate cell size biasing; use normal LBG\n\n"
  57           "         -c <dummy>    Use centroid (not median) midpoints\n"
  58
  59           "examples:\n"
  60           "   train a new codebook to 1%% tolerance on datafile 'foo':\n"
  61           "      xxxvqtrain book -p 256,6,8 -e .01 foo\n"
  62           "      (produces a trained set in book-0.vqi)\n\n"
  63           "   continue training 'book-0.vqi' (produces book-1.vqi):\n"
  64           "      xxxvqtrain book-0.vqi\n\n"
  65           "   add subvector from element 1 to <dimension> from files\n"
  66           "      data*.m to the training in progress, prodicing book-1.vqi:\n"
  67           "      xxxvqtrain book-0.vqi -s 1,1 data*.m\n\n",vqext_booktype);
  68 }
  69
  70 int exiting=0;
  71 void setexit(int dummy){
  72   fprintf(stderr,"\nexiting... please wait to finish this iteration\n");
  73   exiting=1;
  74 }
  75
  76 int main(int argc,char *argv[]){
  77   vqgen v;
  78
  79   int entries=-1,dim=-1;
  80   int start=0,num=-1;
  81   float desired=.05f,mindist=0.f;
  82   int iter=1000;
  83   int biasp=1;
  84   int centroid=0;
  85
  86   FILE *out=NULL;
  87   char *line;
  88   long i,j,k;
  89   int init=0;
  90   q.quant=-1;
  91
  92   argv++;
  93   if(!*argv){
  94     usage();
  95     exit(0);
  96   }
  97
  98   /* get the book name, a preexisting book to continue training */
  99   {
 100     FILE *in=NULL;
 101     char *filename=alloca(strlen(*argv)+30),*ptr;
 102
 103     strcpy(filename,*argv);
 104     in=fopen(filename,"r");
 105     ptr=strrchr(filename,'-');
 106     if(ptr){
 107       int num;
 108       ptr++;
 109       num=atoi(ptr);
 110       sprintf(ptr,"%d.vqi",num+1);
 111     }else
 112       strcat(filename,"-0.vqi");
 113
 114     out=fopen(filename,"w");
 115     if(out==NULL){
 116       fprintf(stderr,"Unable to open %s for writing\n",filename);
 117       exit(1);
 118     }
 119
 120     if(in){
 121       /* we wish to suck in a preexisting book and continue to train it */
 122       float a;
 123
 124       line=rline(in,out,1);
 125       if(strcmp(line,vqext_booktype)){
 126         fprintf(stderr,"wrong book type; %s!=%s\n",line,vqext_booktype);
 127         exit(1);
 128       }
 129
 130       line=rline(in,out,1);
 131       if(sscanf(line,"%d %d %d",&entries,&dim,&vqext_aux)!=3){
 132         fprintf(stderr,"Syntax error reading book file\n");
 133         exit(1);
 134       }
 135
 136       vqgen_init(&v,dim,vqext_aux,entries,mindist,
 137                  vqext_metric,vqext_weight,centroid);
 138       init=1;
 139
 140       /* quant setup */
 141       line=rline(in,out,1);
 142       if(sscanf(line,"%ld %ld %d %d",&q.min,&q.delta,
 143                 &q.quant,&q.sequencep)!=4){
 144         fprintf(stderr,"Syntax error reading book file\n");
 145         exit(1);
 146       }
 147
 148       /* quantized entries */
 149       i=0;
 150       for(j=0;j<entries;j++){
 151         for(k=0;k<dim;k++){
 152           line=rline(in,out,0);
 153           sscanf(line,"%f",&a);
 154           v.entrylist[i++]=a;
 155         }
 156       }
 157       vqgen_unquantize(&v,&q);
 158
 159       /* bias */
 160       i=0;
 161       for(j=0;j<entries;j++){
 162         line=rline(in,out,0);
 163         sscanf(line,"%f",&a);
 164         v.bias[i++]=a;
 165       }
 166
 167       v.seeded=1;
 168       {
 169         float *b=alloca((dim+vqext_aux)*sizeof(float));
 170         i=0;
 171         while(1){
 172           for(k=0;k<dim+vqext_aux;k++){
 173             line=rline(in,out,0);
 174             if(!line)break;
 175             sscanf(line,"%f",b+k);
 176           }
 177           if(feof(in))break;
 178           vqgen_addpoint(&v,b,b+dim);
 179         }
 180       }
 181
 182       fclose(in);
 183     }
 184   }
 185
 186   /* get the rest... */
 187   argv=argv++;
 188   while(*argv){
 189     if(argv[0][0]=='-'){
 190       /* it's an option */
 191       if(!argv[1]){
 192         fprintf(stderr,"Option %s missing argument.\n",argv[0]);
 193         exit(1);
 194       }
 195       switch(argv[0][1]){
 196       case 'p':
 197         if(sscanf(argv[1],"%d,%d,%d",&entries,&dim,&q.quant)!=3)
 198           goto syner;
 199         break;
 200       case 's':
 201         if(sscanf(argv[1],"%d,%d",&start,&num)!=2){
 202           num= -1;
 203           if(sscanf(argv[1],"%d",&start)!=1)
 204             goto syner;
 205         }
 206         break;
 207       case 'e':
 208         if(sscanf(argv[1],"%f",&desired)!=1)
 209           goto syner;
 210         break;
 211       case 'd':
 212         if(sscanf(argv[1],"%f",&mindist)!=1)
 213           goto syner;
 214         if(init)v.mindist=mindist;
 215         break;
 216       case 'i':
 217         if(sscanf(argv[1],"%d",&iter)!=1)
 218           goto syner;
 219         break;
 220       case 'b':
 221         biasp=0;
 222         break;
 223       case 'c':
 224         centroid=1;
 225         break;
 226       default:
 227         fprintf(stderr,"Unknown option %s\n",argv[0]);
 228         exit(1);
 229       }
 230       argv+=2;
 231     }else{
 232       /* it's an input file */
 233       char *file=strdup(*argv++);
 234       FILE *in;
 235       int cols=-1;
 236
 237       if(!init){
 238         if(dim==-1 || entries==-1 || q.quant==-1){
 239           fprintf(stderr,"-p required when training a new set\n");
 240           exit(1);
 241         }
 242         vqgen_init(&v,dim,vqext_aux,entries,mindist,
 243                    vqext_metric,vqext_weight,centroid);
 244         init=1;
 245       }
 246
 247       in=fopen(file,"r");
 248       if(in==NULL){
 249         fprintf(stderr,"Could not open input file %s\n",file);
 250         exit(1);
 251       }
 252       fprintf(out,"# training file entry: %s\n",file);
 253
 254       while((line=rline(in,out,0))){
 255         if(cols==-1){
 256           char *temp=line;
 257           while(*temp==' ')temp++;
 258           for(cols=0;*temp;cols++){
 259             while(*temp>32)temp++;
 260             while(*temp==' ')temp++;
 261           }
 262
 263           fprintf(stderr,"%d colums per line in file %s\n",cols,file);
 264
 265         }
 266         {
 267           int i;
 268           float b[cols];
 269           if(start+num*dim>cols){
 270             fprintf(stderr,"ran out of columns reading %s\n",file);
 271             exit(1);
 272           }
 273           while(*line==' ')line++;
 274           for(i=0;i<cols;i++){
 275
 276             /* static length buffer bug workaround */
 277             char *temp=line;
 278             char old;
 279             while(*temp>32)temp++;
 280
 281             old=temp[0];
 282             temp[0]='\0';
 283             b[i]=atof(line);
 284             temp[0]=old;
 285
 286             while(*line>32)line++;
 287             while(*line==' ')line++;
 288           }
 289           if(num<=0)num=(cols-start)/dim;
 290           for(i=0;i<num;i++)
 291             vqext_addpoint_adj(&v,b,start+i*dim,dim,cols,num);
 292
 293         }
 294       }
 295       fclose(in);
 296     }
 297   }
 298
 299   if(!init){
 300     fprintf(stderr,"No input files!\n");
 301     exit(1);
 302   }
 303
 304   vqext_preprocess(&v);
 305
 306   /* train the book */
 307   signal(SIGTERM,setexit);
 308   signal(SIGINT,setexit);
 309
 310   for(i=0;i<iter && !exiting;i++){
 311     float result;
 312     if(i!=0){
 313       vqgen_unquantize(&v,&q);
 314       vqgen_cellmetric(&v);
 315     }
 316     result=vqgen_iterate(&v,biasp);
 317     vqext_quantize(&v,&q);
 318     if(result<desired)break;
 319   }
 320
 321   /* save the book */
 322
 323   fprintf(out,"# OggVorbis VQ codebook trainer, intermediate file\n");
 324   fprintf(out,"%s\n",vqext_booktype);
 325   fprintf(out,"%d %d %d\n",entries,dim,vqext_aux);
 326   fprintf(out,"%ld %ld %d %d\n",
 327           q.min,q.delta,q.quant,q.sequencep);
 328
 329   /* quantized entries */
 330   fprintf(out,"# quantized entries---\n");
 331   i=0;
 332   for(j=0;j<entries;j++)
 333     for(k=0;k<dim;k++)
 334       fprintf(out,"%d\n",(int)(rint(v.entrylist[i++])));
 335
 336   fprintf(out,"# biases---\n");
 337   i=0;
 338   for(j=0;j<entries;j++)
 339     fprintf(out,"%f\n",v.bias[i++]);
 340
 341   /* we may have done the density limiting mesh trick; refetch the
 342      training points from the temp file */
 343
 344   rewind(v.asciipoints);
 345   fprintf(out,"# points---\n");
 346   {
 347     /* sloppy, no error handling */
 348     long bytes;
 349     char buff[4096];
 350     while((bytes=fread(buff,1,4096,v.asciipoints)))
 351       while(bytes)bytes-=fwrite(buff,1,bytes,out);
 352   }
 353
 354   fclose(out);
 355   fclose(v.asciipoints);
 356
 357   vqgen_unquantize(&v,&q);
 358   vqgen_cellmetric(&v);
 359   exit(0);
 360
 361   syner:
 362     fprintf(stderr,"Syntax error in argument '%s'\n",*argv);
 363     exit(1);
 364 }