#!/apps/bin/perl -w

# this scrips computes the overlap between two text units
# each text unit is treated as a set of elements and so it
# is simetric.
# two parameters
# -text1: a string
# -text2: a string
# elements are separated by one space

# text1 = x a x 
# text2 = b x

# overlap is 1

my $text1 = shift;

my $text2 = shift;



# store the tokens in hash tables, the key is the token, the value is the number of
# times the token appears in the text

my @tokens1 = split(/ /,$text1);
my @tokens2 = split(/ /,$text2);

my %set1 = ();
my %set2 = ();

my $set1_count = 0;
my $tokens1_count = 0;
my $set2_count = 0;
my $tokens2_count = 0;

foreach $token (@tokens1) {
    $tokens1_count++;

    if(exists($set1{$token})) {

	$set1{$token} = $set1{$token}+1;
    } else {$set1{$token} =1; $set1_count++ }
} 


foreach $token (@tokens2) {
    $tokens2_count++;
    if(exists($set2{$token})) {

	$set2{$token} = $set2{$token}+1;
    } else {$set2{$token} =1; $set2_count++ }
} 


my $overlap = set_overlap(\%set1,\%set2);

my $normalized = $overlap/($set1_count+$set2_count-$overlap);


print "OVERLAP: $overlap\n";

print "NORMALIZED: $normalized\n";

# set overlap uses pointers to the sets, so you must invoke it like
# set_overlap(\%set1,\%set2)

sub set_overlap {
    my $s1 = shift;
    my $s2 = shift;
    my $overlap = 0;
 
    foreach $key (keys %$s1) {

	if(exists($$s2{$key})) {
	    $overlap++;
	}
    }

    return $overlap;

}

sub show_tokens {
# I will use the @_ array  
    print "TOKENS: $_[0]\n";
    print "DIFFERENTS: $_[1]\n";
    shift @_;
    shift @_;
    my %table = @_;
    foreach $key (keys %table) {
	print "$key\n";
	
    }
    

}