#!/usr/bin/awk

# ajoute des valeurs manquantes
# à une base au format attribut-valeur

BEGIN{
  if(ARGC != 4){
    print "# ajoute des valeurs manquantes a une base au format attribut-valeur";
    print ARGV[0] ": usage <startcol> <% missing values> [data]";
    exit;
  }

  startcol = ARGV[1];
  prop = ARGV[2] / 100;
#  print "# adding " prop * 100 " % of random missing values starting from column " startcol " on " ARGV[3];
  ARGV[1] = ARGV[3];
  ARGC = 2;
  srand(0);
  FS = ",";
  OFS = ","
}

/\#/ {
  print $0;
  next;
}

{ 
  for(i = 1; i < NF; i ++){
    if(i >= startcol &&  rand() < prop)
      printf("?," );
    else
      printf("%s,", $i);
  }
  if(rand() < prop)
    printf("?" );
  else
    printf("%s", $NF);
  printf("\n");
}