#!/usr/bin/perl -w

(@ARGV==2) || die("usage: compute-init-scores inits-total inits-used\n");
($rand,$real)=@ARGV;

$totrand=0;

open(RAND,$rand) || die("No $rand!\n");
while(<RAND>) {
    ($num,$pat)=split(" ",$_);
    $random{$pat}=$num;
    $totrand+=$num;
}
close(RAND);

open(REAL,$real) || die("No $real!\n");
while(<REAL>) {
    @fields=split(" ",$_);
    $pat=$fields[$#fields]; # pattern is always in last field
    $real{$pat}++;
    $totreal++;
}
close(REAL);

while(($pat,$rand)=each %random) {
    $rl=$real{$pat};
    $freqreal=$rl/$totreal;
    $freqrand=$rand/$totrand;
    if($freqreal>0) {
      $score=log($freqreal/$freqrand);
    }
    else {
      $score=0;
    }
    printf("%s %d %d %f %f %f\n",$pat,$rl,$rand,$freqreal,$freqrand,$score);
}
  
