use strict;
use warnings;

#Part one of the code, the wordlist is a file with one word per line and I transform this into a hash

	my $filename = "wordlist.txt";
	open(INPUT, $filename) or die "Cannot open $filename";

	my $line = <INPUT>;
	while($line = <INPUT>){
	chomp($line);
	my @words = split(/\s+/, $line);
	
my %unique = map {$_ => 1 } @words;
my @unique = %unique;

#part two of code, open the text file and extract words only (because the file also includes many numbers), and count the number of occurence stored in a hash

open (DATA, "4.txt") or die;

my @UnNum;
my $x;
my %dict;

while (<DATA>) {
    chomp;
      $_ = lc;   
      s/ -- / /g; 
      s/ - / /g; 
      s/ +/ /g;   
      s/[.,:;?"!_()\[\]]//g; 
  
   my @UnNum = split(/\s+/);

foreach $x(@UnNum){
if ($x =~  /(([a-zA-Z']+-)*[a-zA-Z']+)/ ){
    ++$dict{$x};
}}}

#part 3 of the code, I try to compare the two different hashes and add the total number of occurrences

while ((my $words,my $number) = each (%dict))
{my $total+= $number; 
if (exists($unique{$words})){
my $corresponding +=$number;

print "There are $corresponding corresponding words of in total $total words";}}

}