added some extra spaced out swearwords

[spider.git] / perl / BadWords.pm

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index e7d1169e3a390abd5845f5922bb299eaa79cefdc..9814e3fadd5ef712d25315887d1582e2c509d338 100644 (file)
--- a/perl/BadWords.pm
+++ b/perl/BadWords.pm
@@ -56,7 +56,33 @@ sub load
 # check the text against the badwords list
 sub check
 {
-       return grep { $badword->in($_) } split(/\b/, lc shift);
+       my $s = uc shift;
+       
+       for (split(/\s+/, $s)) {
+               s/[^\w]//g;
+               return $_ if $badword->in($_);
+               s/\'?S$//;
+               return $_ if $badword->in($_);
+       }
+       
+       # look for a few of the common ones with spaces and stuff
+       if ($s =~ /F[\s\W]*U[\s\W]*C[\s\W]*K/) {
+               return "FUCK";
+       } elsif ($s =~ /C[\s\W]*U[\s\W]*N[\s\W]*T/) {
+               return "CUNT";
+       } elsif ($s =~ /W[\s\W]*A[\s\W]*N[\s\W]*K/) {
+               return "WANK";
+       } elsif ($s =~ /C[\s\W]*[0O][\s\W]*C[\s\W]*K/) {
+               return "COCK";
+       } elsif ($s =~ /S[\s\W]*H[\s\W]*[I1][\s\W]*T/) {
+               return "SHIT";
+       } elsif ($s =~ /P[\s\W]*[I1][\s\W]*S[\s\W]*S/) {
+               return "PISS";
+       } elsif ($s =~ /B[\s\W]*[O0][\s\W]*L[\s\W]*L[\s\W]*[O0][\s\W]*[CK]/) {
+               return "BOLLOCKS";
+       }
+       
+       return ();
 }
 
 1;