#!/apps/bin/perl -w



# this scrips computes the cosine between two vectors
# depending on the value of a parameter it will use counts and idf
# two parameters
# -text1: a string
# -text2: a string
# elements are separated by one space

# text1 = x a x 
# text2 = b x

# overlap is 1

use IDF;
use COSINE;

my $eng_idf_word = "HK-WORD-enidf";
my $eng_idf_lemma = "HK-LEMMA-enidf";

my $option = shift;

die "Lemmas or Words???\n" unless (($option eq "-l") || ($option eq "-w"));

my $idf_file = ($option eq "-l") ? $eng_idf_lemma : $eng_idf_word; 
my $OPT =  ($option eq "-l")  ? "LEX":"ORIGIN";

my $text1 = shift;

my $text2 = shift;

open_nidf($idf_file);

# 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 $key (keys %set1) {
    $idf = get_nidf("$key");
    print "KEY: $key, VALUE: $idf\n";
    $set1{$key}=$set1{$key}*$idf;
}


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

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

foreach $key (keys %set2) {
    $idf = get_nidf("$key");
    print "KEY: $key, VALUE: $idf\n";
    $set2{$key}=$set2{$key}*$idf;
}


my $result1 = simple_cosine(\%set1,\%set2);

my $result2 = complete_cosine(\%set1,\%set2);

my $result3 = cosine_with_idf($text1,$text2,"ENG",$OPT);

print "Cosinr 0/1) $result1\n";
print "Cosine a): $result2\n";
print "Cosine b): $result3\n";

# simple cosine assumes weith 1 for presence of the word

sub simple_cosine {
    my $s1 = shift;
    my $s2 = shift;
   
 
    
    foreach $key (keys %$s1) {

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

    $cos = $cos/(sqrt((scalar keys %$s1)*(scalar keys %$s2)));
    return $cos;

}



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";
	
    }
    

}

sub complete_cosine {

    my $s1 = shift;
    my $s2 = shift;
    my $cos = 0;
    my $norm1 = 0;
    my $norm2 = 0;

    foreach $key (keys %$s1) {
	
	$val = $$s1{$key};
	print "==> $key , $val\n";
	$norm1 = $norm1 + $val*$val;

	if(exists($$s2{$key})) {
	   $cos = $cos + ($val)*($$s2{$key}); 
	}
    }

    foreach $key (keys %$s2) {
	
	$val = $$s2{$key};
	print "++> $key, $val\n";
	$norm2 = $norm2 + $val*$val;
    }
    if($cos==0) {return $cos;}
#    print "COS: $cos, NORM1: $norm1, NORM2: $norm2\n";
    $cos = $cos/sqrt($norm1*$norm2);
    return $cos;
    
}