X-Git-Url: http://dxcluster.org/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FBadWords.pm;h=81e7ec90eee67c57a93bb52cb08b4079dea100c3;hb=7346d1f8129457eee2e602b36e27435141533049;hp=e7d1169e3a390abd5845f5922bb299eaa79cefdc;hpb=1fd948e7dfd70949e959dd0a633d4e5b762d07e0;p=spider.git

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index e7d1169e..81e7ec90 100644
--- a/perl/BadWords.pm
+++ b/perl/BadWords.pm
@@ -56,7 +56,29 @@ 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";
+	}
+	
+	return undef;
 }
 
 1;