#!/usr/bin/gawk -f

BEGIN{
  FS=OFS=",";
  IGNORECASE=1;
  bins=10;
  dataLine=0;
  attr=0;
  data=0;
  min=10**32;;
  max=-10**32;
  classField=0;
}

{sub(/\%.*/,"")}
/^[ \t]*$/ {next}
/%/ {next;}

/@relation/{ print $0; }
/@attribute/{ attrs[attr++] = $0; }
/@data/{ 
  dataLine=1;  
  attr--;
  for ( i = 0; i < attr; i++ )
    print attrs[i];

  if ( attrs[attr] ~ /real|continuous|integer|number/ )
  {
    sub( /real|continuous|integer|number/, "", attrs[attr] );
    attrs[attr]=attrs[attr] "_1_";
    for( i = 2; i <= bins; i++ )
      attrs[attr] = attrs[attr] ",_" i "_";
  }

  print attrs[attr];
  print $0;
  next;
}

dataLine==1{
  datas[data] = $1;
  classes[data] = $NF;
  for( d = 2; d < NF; d++ )
    datas[data] = datas[data] "," $d;

  if ( $NF !~ /\?/ )
  {
    max =  max < $NF ? $NF : max;
    min = min > $NF ? $NF : min;
  }
  classField=NF;
  data++;
}

END
  step = int(( max - min )/bins);
  for ( d = 0; d < data; d++ )
  {
    binNo = int(classes[d]/step);
    binNo = binNo < 1 ? 1 : binNo;
    binNo = binNo > bins ? bins : binNo;
    print datas[d] ",_" binNo  "_";
  }
}