#!/usr/bin/perl
#
# usage:
#   cat access_log | this_script > access_log.without_spam
#


my @reject_referrers = (
	'.*stories-site\.com$',
	'.*stories\.org$',
	'.*girls\.name$',
	'.*tits\.name$',
	'.*service\.name$',
	'.*hole\.name$',
	'.*shemale\.name$',
	'.*gang-bang\.name$',
	'.*movies\.us$',
	'.*videos\.us$',
	'.*xxx.*',
	'.*beast.*',
	'.*pussy.*',
	'.*fetish.*',
	'.*sex.*',
	'.*porn.*',
	'.*nudity.*',
	'.*cumfiesta.*',
	'.*hand-job.*',
	'.*handjob.*',
	'.*webcam.*',
	'.*hardcore.*',
	'.*asian.*',
	'.*anime.*'
);

while (my $input = <STDIN>) {
  $input =~ s/\r//;
  $input =~ s/\n//;

  my $matched = 0;
  if ($input =~ /^(.+) (.+) (.+) \[(.+)\] \"([^"]+)\" (\d+) (\d+) \"([^"]*)\" \"([^"]*)\"$/) {
    my $referrer = $8;
    foreach my $r (@reject_referrers) {
      $matched = 1 if ($referrer =~ /$r/);
    }
  }

  print "$input\n" if ($matched == 0);
}