#!/usr/bin/perl -w

my $num_words;

%wordscores = qw(advertising 40
				service 20
				solution 15
				guarantee 30
				visa 20
				mastercard 30
				american 10
				express 20
				e-commerce 50
				commerce 50
				free 30
				website 15
				pay 20
				buy 20
				cpan -100
				perl -100
				mlm 100
				xxx 100
				inconvenience 50
				homeowner 30
				cash 30
				loan 30
				income 20
				rich 40
				financial 20
				millennium 50 
				millenium 50
				executive 30
				opportunity 40
				legal 30
				illegal 30
				viagra 70
				exotic 70
				payment 30
				satisfaction 30
				friend 20
				copyright 20 
				invest 40
				freedom 20
				quick 20
				descrambler 70
				cable 40
				dollar 30
				dollars 30
				chain 30
				adult 40
				spam 40
				very 10
				extremely 10);
				

while ($filename = shift @ARGV) {

  open(FILE, $filename) or die "Unable to open $filename. This is not our fault: $!\n";
  @message = <FILE>;
  $spam_score = spam_score(@message);
#  print "$filename\t$spam_score\t";
  if ($spam_score > 0.8) {
	 print "YES\n";
  } else {
	 print "NO\n";
  }
}

sub spam_score {
  my $score = 0;
  my @message = @_;
  my $headers = '';
 HEADER:
  while ($line = shift @message) {
	 chomp $line;
	 last HEADER if ($line eq '');
	 $headers .= "$line ";
  }
  my $body = join(' ', @message);
  
  # deal with headers
  
  # deal with body
  
  $seen = word_counts($body);
  foreach (keys %wordscores) {
	 if ($seen{$_}){
		$score += sqrt($seen{$_}) * $wordscores{$_};
	 }
  }

  if (phone_number($body)) {$score += 40;} # in first 2/3 of file
  if (toll_free($body)) {$score += 70;}
  
#  $score /= $num_words;
  $score /= scalar(keys %seen);
  return ($score);
}

sub phone_number {
  my $str = shift @_;
  my $start = substr($str, 0, int(0.66 * length($str)));
  return ($start =~ /\d{3}.\d{4}/);
}

sub toll_free {
  my $str = shift @_;
  return ($str =~ /8[087]{2}.{1,3}\d{3}.\d{4}/)
}

sub word_counts {
  undef %seen;
  $num_words = 0;
  $str = shift @_;
  while ($str =~ /(\w[\'\w-]*)/g ) {
	 $seen{lc $1}++;
	 $num_words += 1;
  } 
  return \%seen;
}