Blame vq/distribution.c

Packit 06404a
/********************************************************************
Packit 06404a
 *                                                                  *
Packit 06404a
 * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
Packit 06404a
 * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
Packit 06404a
 * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
Packit 06404a
 * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
Packit 06404a
 *                                                                  *
Packit 06404a
 * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2001             *
Packit 06404a
 * by the Xiph.Org Foundation http://www.xiph.org/                  *
Packit 06404a
 *                                                                  *
Packit 06404a
 ********************************************************************
Packit 06404a
Packit 06404a
 function: utility for finding the distribution in a data set
Packit 06404a
 last mod: $Id: distribution.c 16037 2009-05-26 21:10:58Z xiphmont $
Packit 06404a
Packit 06404a
 ********************************************************************/
Packit 06404a
Packit 06404a
#include <stdlib.h>
Packit 06404a
#include <stdio.h>
Packit 06404a
#include <math.h>
Packit 06404a
#include <string.h>
Packit 06404a
#include <errno.h>
Packit 06404a
#include "bookutil.h"
Packit 06404a
Packit 06404a
/* command line:
Packit 06404a
   distribution file.vqd
Packit 06404a
*/
Packit 06404a
Packit 06404a
int ascend(const void *a,const void *b){
Packit 06404a
  return(**((long **)a)-**((long **)b));
Packit 06404a
}
Packit 06404a
Packit 06404a
int main(int argc,char *argv[]){
Packit 06404a
  FILE *in;
Packit 06404a
  long lines=0;
Packit 06404a
  float min;
Packit 06404a
  float max;
Packit 06404a
  long bins=-1;
Packit 06404a
  int flag=0;
Packit 06404a
  long *countarray;
Packit 06404a
  long total=0;
Packit 06404a
  char *line;
Packit 06404a
Packit 06404a
  if(argv[1]==NULL){
Packit 06404a
    fprintf(stderr,"Usage: distribution {data.vqd [bins]| book.vqh} \n\n");
Packit 06404a
    exit(1);
Packit 06404a
  }
Packit 06404a
  if(argv[2]!=NULL)
Packit 06404a
    bins=atoi(argv[2])-1;
Packit 06404a
Packit 06404a
  in=fopen(argv[1],"r");
Packit 06404a
  if(!in){
Packit 06404a
    fprintf(stderr,"Could not open input file %s\n",argv[1]);
Packit 06404a
    exit(1);
Packit 06404a
  }
Packit 06404a
Packit 06404a
  if(strrchr(argv[1],'.') && strcmp(strrchr(argv[1],'.'),".vqh")==0){
Packit 06404a
    /* load/decode a book */
Packit 06404a
Packit 06404a
    codebook *b=codebook_load(argv[1]);
Packit 06404a
    static_codebook *c=(static_codebook *)(b->c);
Packit 06404a
    float delta;
Packit 06404a
    int i;
Packit 06404a
    fclose(in);
Packit 06404a
Packit 06404a
    switch(c->maptype){
Packit 06404a
    case 0:
Packit 06404a
      printf("entropy codebook only; no mappings\n");
Packit 06404a
      exit(0);
Packit 06404a
      break;
Packit 06404a
    case 1:
Packit 06404a
      bins=_book_maptype1_quantvals(c);
Packit 06404a
      break;
Packit 06404a
    case 2:
Packit 06404a
      bins=c->entries*c->dim;
Packit 06404a
      break;
Packit 06404a
    }
Packit 06404a
Packit 06404a
    max=min=_float32_unpack(c->q_min);
Packit 06404a
    delta=_float32_unpack(c->q_delta);
Packit 06404a
Packit 06404a
    for(i=0;i
Packit 06404a
      float val=c->quantlist[i]*delta+min;
Packit 06404a
      if(val>max)max=val;
Packit 06404a
    }
Packit 06404a
Packit 06404a
    printf("Minimum scalar value: %f\n",min);
Packit 06404a
    printf("Maximum scalar value: %f\n",max);
Packit 06404a
Packit 06404a
    switch(c->maptype){
Packit 06404a
    case 1:
Packit 06404a
      {
Packit 06404a
        /* lattice codebook.  dump it. */
Packit 06404a
        int j,k;
Packit 06404a
        long maxcount=0;
Packit 06404a
        long **sort=calloc(bins,sizeof(long *));
Packit 06404a
        long base=c->lengthlist[0];
Packit 06404a
        countarray=calloc(bins,sizeof(long));
Packit 06404a
Packit 06404a
        for(i=0;i<bins;i++)sort[i]=c->quantlist+i;
Packit 06404a
        qsort(sort,bins,sizeof(long *),ascend);
Packit 06404a
Packit 06404a
        for(i=0;i<b->entries;i++)
Packit 06404a
          if(c->lengthlist[i]>base)base=c->lengthlist[i];
Packit 06404a
Packit 06404a
        /* dump a full, correlated count */
Packit 06404a
        for(j=0;j<b->entries;j++){
Packit 06404a
          if(c->lengthlist[j]){
Packit 06404a
            int indexdiv=1;
Packit 06404a
            printf("%4d: ",j);
Packit 06404a
            for(k=0;k<b->dim;k++){        
Packit 06404a
              int index= (j/indexdiv)%bins;
Packit 06404a
              printf("%+3.1f,", c->quantlist[index]*_float32_unpack(c->q_delta)+
Packit 06404a
                     _float32_unpack(c->q_min));
Packit 06404a
              indexdiv*=bins;
Packit 06404a
            }
Packit 06404a
            printf("\t|");
Packit 06404a
            for(k=0;k<base-c->lengthlist[j];k++)printf("*");
Packit 06404a
            printf("\n");
Packit 06404a
          }
Packit 06404a
        }
Packit 06404a
Packit 06404a
        /* do a rough count */
Packit 06404a
        for(j=0;j<b->entries;j++){
Packit 06404a
          int indexdiv=1;
Packit 06404a
          for(k=0;k<b->dim;k++){
Packit 06404a
            if(c->lengthlist[j]){
Packit 06404a
              int index= (j/indexdiv)%bins;
Packit 06404a
              countarray[index]+=(1<<(base-c->lengthlist[j]));
Packit 06404a
              indexdiv*=bins;
Packit 06404a
            }
Packit 06404a
          }
Packit 06404a
        }
Packit 06404a
Packit 06404a
        /* dump the count */
Packit 06404a
Packit 06404a
        {
Packit 06404a
          long maxcount=0,i,j;
Packit 06404a
          for(i=0;i
Packit 06404a
            if(countarray[i]>maxcount)maxcount=countarray[i];
Packit 06404a
      
Packit 06404a
          for(i=0;i
Packit 06404a
            int ptr=sort[i]-c->quantlist;
Packit 06404a
            int stars=rint(50./maxcount*countarray[ptr]);
Packit 06404a
            printf("%+08f (%8ld) |",c->quantlist[ptr]*delta+min,countarray[ptr]);
Packit 06404a
            for(j=0;j
Packit 06404a
            printf("\n");
Packit 06404a
          }
Packit 06404a
        }
Packit 06404a
      }
Packit 06404a
      break;
Packit 06404a
    case 2:
Packit 06404a
      {
Packit 06404a
        /* trained, full mapping codebook. */
Packit 06404a
        printf("Can't do probability dump of a trained [type 2] codebook (yet)\n");
Packit 06404a
      }
Packit 06404a
      break;
Packit 06404a
    }
Packit 06404a
  }else{
Packit 06404a
    /* load/count a data file */
Packit 06404a
Packit 06404a
    /* do it the simple way; two pass. */
Packit 06404a
    line=setup_line(in);
Packit 06404a
    while(line){      
Packit 06404a
      float code;
Packit 06404a
      char buf[80];
Packit 06404a
      lines++;
Packit 06404a
Packit 06404a
      sprintf(buf,"getting min/max (%.2f::%.2f). lines...",min,max);
Packit 06404a
      if(!(lines&0xff))spinnit(buf,lines);
Packit 06404a
      
Packit 06404a
      while(!flag && sscanf(line,"%f",&code)==1){
Packit 06404a
        line=strchr(line,',');
Packit 06404a
        min=max=code;
Packit 06404a
        flag=1;
Packit 06404a
      }
Packit 06404a
      
Packit 06404a
      while(line && sscanf(line,"%f",&code)==1){
Packit 06404a
        line=strchr(line,',');
Packit 06404a
        if(line)line++;
Packit 06404a
        if(code
Packit 06404a
        if(code>max)max=code;
Packit 06404a
      }
Packit 06404a
      
Packit 06404a
      line=setup_line(in);
Packit 06404a
    }
Packit 06404a
    
Packit 06404a
    if(bins<1){
Packit 06404a
      if((int)(max-min)==min-max){
Packit 06404a
        bins=max-min;
Packit 06404a
      }else{
Packit 06404a
        bins=25;
Packit 06404a
      }
Packit 06404a
    }
Packit 06404a
    
Packit 06404a
    printf("\r                                                     \r");
Packit 06404a
    printf("Minimum scalar value: %f\n",min);
Packit 06404a
    printf("Maximum scalar value: %f\n",max);
Packit 06404a
Packit 06404a
    if(argv[2]){
Packit 06404a
      
Packit 06404a
      printf("\n counting hits into %ld bins...\n",bins+1);
Packit 06404a
      countarray=calloc(bins+1,sizeof(long));
Packit 06404a
      
Packit 06404a
      rewind(in);
Packit 06404a
      line=setup_line(in);
Packit 06404a
      while(line){      
Packit 06404a
        float code;
Packit 06404a
        lines--;
Packit 06404a
        if(!(lines&0xff))spinnit("counting distribution. lines so far...",lines);
Packit 06404a
        
Packit 06404a
        while(line && sscanf(line,"%f",&code)==1){
Packit 06404a
          line=strchr(line,',');
Packit 06404a
          if(line)line++;
Packit 06404a
          
Packit 06404a
          code-=min;
Packit 06404a
          code/=(max-min);
Packit 06404a
          code*=bins;
Packit 06404a
          countarray[(int)rint(code)]++;
Packit 06404a
          total++;
Packit 06404a
        }
Packit 06404a
        
Packit 06404a
        line=setup_line(in);
Packit 06404a
      }
Packit 06404a
    
Packit 06404a
      /* make a pretty graph */
Packit 06404a
      {
Packit 06404a
        long maxcount=0,i,j;
Packit 06404a
        for(i=0;i
Packit 06404a
          if(countarray[i]>maxcount)maxcount=countarray[i];
Packit 06404a
        
Packit 06404a
        printf("\r                                                     \r");
Packit 06404a
        printf("Total scalars: %ld\n",total);
Packit 06404a
        for(i=0;i
Packit 06404a
          int stars=rint(50./maxcount*countarray[i]);
Packit 06404a
          printf("%08f (%8ld) |",(max-min)/bins*i+min,countarray[i]);
Packit 06404a
          for(j=0;j
Packit 06404a
          printf("\n");
Packit 06404a
        }
Packit 06404a
      }
Packit 06404a
    }
Packit 06404a
Packit 06404a
    fclose(in);
Packit 06404a
Packit 06404a
  }
Packit 06404a
  printf("\nDone.\n");
Packit 06404a
  exit(0);
Packit 06404a
}