vq/train.c

   1 /********************************************************************
   2  *                                                                  *
   3  * THIS FILE IS PART OF THE Ogg Vorbis SOFTWARE CODEC SOURCE CODE.  *
   4  * USE, DISTRIBUTION AND REPRODUCTION OF THIS SOURCE IS GOVERNED BY *
   5  * THE GNU PUBLIC LICENSE 2, WHICH IS INCLUDED WITH THIS SOURCE.    *
   6  * PLEASE READ THESE TERMS DISTRIBUTING.                            *
   7  *                                                                  *
   8  * THE OggSQUISH SOURCE CODE IS (C) COPYRIGHT 1994-2000             *
   9  * by Monty <monty@xiph.org> and The XIPHOPHORUS Company            *
  10  * http://www.xiph.org/                                             *
  11  *                                                                  *
  12  ********************************************************************
  13
  14  function: utility main for training codebooks
  15  last mod: $Id: train.c,v 1.13 2000/01/05 15:05:00 xiphmont Exp $
  16
  17  ********************************************************************/
  18
  19 #include <stdlib.h>
  20 #include <stdio.h>
  21 #include <math.h>
  22 #include <string.h>
  23 #include <errno.h>
  24 #include <signal.h>
  25 #include "vqgen.h"
  26 #include "vqext.h"
  27 #include "bookutil.h"
  28
  29 static char *rline(FILE *in,FILE *out,int pass){
  30   while(1){
  31     char *line=get_line(in);
  32     if(line && line[0]=='#'){
  33       if(pass)fprintf(out,"%s\n",line);
  34     }else{
  35       return(line);
  36     }
  37   }
  38 }
  39
  40 /* command line:
  41    trainvq  vqfile [options] trainfile [trainfile]
  42
  43    options: -params     entries,dim,quant
  44             -subvector  start[,num]
  45             -error      desired_error
  46             -iterations iterations
  47 */
  48
  49 static void usage(void){
  50   fprintf(stderr, "\nOggVorbis %s VQ codebook trainer\n\n"
  51           "<foo>vqtrain vqfile [options] [datasetfile] [datasetfile]\n"
  52           "options: -p[arams]     <entries,dim,quant>\n"
  53           "         -s[ubvector]  <start[,num]>\n"
  54           "         -e[rror]      <desired_error>\n"
  55           "         -i[terations] <maxiterations>\n\n"
  56           "examples:\n"
  57           "   train a new codebook to 1%% tolerance on datafile 'foo':\n"
  58           "      xxxvqtrain book -p 256,6,8 -e .01 foo\n"
  59           "      (produces a trained set in book-0.vqi)\n\n"
  60           "   continue training 'book-0.vqi' (produces book-1.vqi):\n"
  61           "      xxxvqtrain book-0.vqi\n\n"
  62           "   add subvector from element 1 to <dimension> from files\n"
  63           "      data*.m to the training in progress, prodicing book-1.vqi:\n"
  64           "      xxxvqtrain book-0.vqi -s 1,1 data*.m\n\n",vqext_booktype);
  65 }
  66
  67 int exiting=0;
  68 void setexit(int dummy){
  69   fprintf(stderr,"\nexiting... please wait to finish this iteration\n");
  70   exiting=1;
  71 }
  72
  73 int main(int argc,char *argv[]){
  74   vqgen v;
  75
  76   int entries=-1,dim=-1;
  77   int start=0,num=-1;
  78   double desired=.05;
  79   int iter=1000;
  80
  81   FILE *out=NULL;
  82   char *line;
  83   long i,j,k;
  84   int init=0;
  85   q.quant=-1;
  86
  87   argv++;
  88   if(!*argv){
  89     usage();
  90     exit(0);
  91   }
  92
  93   /* get the book name, a preexisting book to continue training */
  94   {
  95     FILE *in=NULL;
  96     char *filename=alloca(strlen(*argv)+30),*ptr;
  97
  98     strcpy(filename,*argv);
  99     in=fopen(filename,"r");
 100     ptr=strrchr(filename,'-');
 101     if(ptr){
 102       int num;
 103       ptr++;
 104       num=atoi(ptr);
 105       sprintf(ptr,"%d.vqi",num+1);
 106     }else
 107       strcat(filename,"-0.vqi");
 108
 109     out=fopen(filename,"w");
 110     if(out==NULL){
 111       fprintf(stderr,"Unable to open %s for writing\n",filename);
 112       exit(1);
 113     }
 114
 115     if(in){
 116       /* we wish to suck in a preexisting book and continue to train it */
 117       double a;
 118
 119       line=rline(in,out,1);
 120       if(strcmp(line,vqext_booktype)){
 121         fprintf(stderr,"wrong book type; %s!=%s\n",line,vqext_booktype);
 122         exit(1);
 123       }
 124
 125       line=rline(in,out,1);
 126       if(sscanf(line,"%d %d %d",&entries,&dim,&vqext_aux)!=3){
 127         fprintf(stderr,"Syntax error reading book file\n");
 128         exit(1);
 129       }
 130
 131       vqgen_init(&v,dim,vqext_aux,entries,vqext_metric,vqext_weight);
 132       init=1;
 133
 134       /* quant setup */
 135       line=rline(in,out,1);
 136       if(sscanf(line,"%ld %ld %d %d",&q.min,&q.delta,
 137                 &q.quant,&q.sequencep)!=4){
 138         fprintf(stderr,"Syntax error reading book file\n");
 139         exit(1);
 140       }
 141
 142       /* quantized entries */
 143       i=0;
 144       for(j=0;j<entries;j++){
 145         for(k=0;k<dim;k++){
 146           line=rline(in,out,0);
 147           sscanf(line,"%lf",&a);
 148           v.entrylist[i++]=a;
 149         }
 150       }
 151       vqgen_unquantize(&v,&q);
 152
 153       /* bias, points */
 154       i=0;
 155       for(j=0;j<entries;j++){
 156         line=rline(in,out,0);
 157         sscanf(line,"%lf",&a);
 158         v.bias[i++]=a;
 159       }
 160
 161       {
 162         double *b=alloca((dim+vqext_aux)*sizeof(double));
 163         i=0;
 164         v.entries=0; /* hack to avoid reseeding */
 165         while(1){
 166           for(k=0;k<dim+vqext_aux;k++){
 167             line=rline(in,out,0);
 168             if(!line)break;
 169             sscanf(line,"%lf",b+k);
 170           }
 171           if(feof(in))break;
 172           vqgen_addpoint(&v,b,b+dim);
 173         }
 174         v.entries=entries;
 175       }
 176
 177       fclose(in);
 178     }
 179   }
 180
 181   /* get the rest... */
 182   argv=argv++;
 183   while(*argv){
 184     if(argv[0][0]=='-'){
 185       /* it's an option */
 186       if(!argv[1]){
 187         fprintf(stderr,"Option %s missing argument.\n",argv[0]);
 188         exit(1);
 189       }
 190       switch(argv[0][1]){
 191       case 'p':
 192         if(sscanf(argv[1],"%d,%d,%d",&entries,&dim,&q.quant)!=3)
 193           goto syner;
 194         break;
 195       case 's':
 196         if(sscanf(argv[1],"%d,%d",&start,&num)!=2){
 197           num= -1;
 198           if(sscanf(argv[1],"%d",&start)!=1)
 199             goto syner;
 200         }
 201         break;
 202       case 'e':
 203         if(sscanf(argv[1],"%lf",&desired)!=1)
 204           goto syner;
 205         break;
 206       case 'i':
 207         if(sscanf(argv[1],"%d",&iter)!=1)
 208           goto syner;
 209         break;
 210       default:
 211         fprintf(stderr,"Unknown option %s\n",argv[0]);
 212         exit(1);
 213       }
 214       argv+=2;
 215     }else{
 216       /* it's an input file */
 217       char *file=strdup(*argv++);
 218       FILE *in;
 219       int cols=-1;
 220
 221       if(!init){
 222         if(dim==-1 || entries==-1 || q.quant==-1){
 223           fprintf(stderr,"-p required when training a new set\n");
 224           exit(1);
 225         }
 226         vqgen_init(&v,dim,vqext_aux,entries,vqext_metric,vqext_weight);
 227         init=1;
 228       }
 229
 230       in=fopen(file,"r");
 231       if(in==NULL){
 232         fprintf(stderr,"Could not open input file %s\n",file);
 233         exit(1);
 234       }
 235       fprintf(out,"# training file entry: %s\n",file);
 236
 237       while((line=rline(in,out,0))){
 238         if(cols==-1){
 239           char *temp=line;
 240           while(*temp==' ')temp++;
 241           for(cols=0;*temp;cols++){
 242             while(*temp>32)temp++;
 243             while(*temp==' ')temp++;
 244           }
 245         }
 246         {
 247           int i;
 248           double b[cols];
 249           if(start*num+dim>cols){
 250             fprintf(stderr,"ran out of columns reading %s\n",file);
 251             exit(1);
 252           }
 253           while(*line==' ')line++;
 254           for(i=0;i<cols;i++){
 255
 256             /* static length buffer bug workaround */
 257             char *temp=line;
 258             char old;
 259             while(*temp>32)temp++;
 260
 261             old=temp[0];
 262             temp[0]='\0';
 263             b[i]=atof(line);
 264             temp[0]=old;
 265
 266             while(*line>32)line++;
 267             while(*line==' ')line++;
 268           }
 269           if(num<=0)num=(cols-start)/dim;
 270           for(i=0;i<num;i++)
 271             vqext_addpoint_adj(&v,b,start+i*dim,dim,cols);
 272
 273         }
 274       }
 275       fclose(in);
 276     }
 277   }
 278
 279   if(!init){
 280     fprintf(stderr,"No input files!\n");
 281     exit(1);
 282   }
 283
 284   vqext_preprocess(&v);
 285
 286   /* train the book */
 287   signal(SIGTERM,setexit);
 288   signal(SIGINT,setexit);
 289
 290   for(i=0;i<iter && !exiting;i++){
 291     double result;
 292     if(i!=0)vqgen_unquantize(&v,&q);
 293     result=vqgen_iterate(&v);
 294     vqgen_quantize(&v,&q);
 295     if(result<desired)break;
 296   }
 297
 298   /* save the book */
 299
 300   fprintf(out,"# OggVorbis VQ codebook trainer, intermediate file\n");
 301   fprintf(out,"%s\n",vqext_booktype);
 302   fprintf(out,"%d %d %d\n",entries,dim,vqext_aux);
 303   fprintf(out,"%ld %ld %d %d\n",q.min,q.delta,q.quant,q.sequencep);
 304
 305   /* quantized entries */
 306   fprintf(out,"# quantized entries---\n");
 307   i=0;
 308   for(j=0;j<entries;j++)
 309     for(k=0;k<dim;k++)
 310       fprintf(out,"%d\n",(int)(rint(v.entrylist[i++])));
 311
 312   fprintf(out,"# biases---\n");
 313   i=0;
 314   for(j=0;j<entries;j++)
 315     fprintf(out,"%f\n",v.bias[i++]);
 316
 317   fprintf(out,"# points---\n");
 318   i=0;
 319   for(j=0;j<v.points;j++)
 320     for(k=0;k<dim+vqext_aux;k++)
 321       fprintf(out,"%f\n",v.pointlist[i++]);
 322
 323   fclose(out);
 324   exit(0);
 325
 326   syner:
 327     fprintf(stderr,"Syntax error in argument '%s'\n",*argv);
 328     exit(1);
 329 }