X-Git-Url: http://dxcluster.org/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FBadWords.pm;h=81e7ec90eee67c57a93bb52cb08b4079dea100c3;hb=7346d1f8129457eee2e602b36e27435141533049;hp=2336bb0468e22b1b95e3b517f5750c6397dd9fc2;hpb=2b58ccdf81685a1167a43c38705a0d84b9d8d661;p=spider.git

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index 2336bb04..81e7ec90 100644
--- a/perl/BadWords.pm
+++ b/perl/BadWords.pm
@@ -12,12 +12,13 @@ use strict;
 
 use DXUtil;
 use DXVars;
+use DXHash;
 use IO::File;
 
-use vars qw(%badwords $fn);
+use vars qw($badword);
 
-$fn = "$main::data/badwords";
-%badwords = ();
+my $oldfn = "$main::data/badwords";
+$badword = new DXHash "badword";
 
 use vars qw($VERSION $BRANCH);
 $VERSION = sprintf( "%d.%03d", q$Revision$ =~ /(\d+)\.(\d+)/ );
@@ -29,23 +30,24 @@ $main::branch += $BRANCH;
 sub load
 {
 	my @out;
-	return unless -e $fn;
-	my $fh = new IO::File $fn;
+	return unless -e $oldfn;
+	my $fh = new IO::File $oldfn;
 	
 	if ($fh) {
-		%badwords = ();
 		while (<$fh>) {
 			chomp;
 			next if /^\s*\#/;
 			my @list = split " ";
 			for (@list) {
-				$badwords{lc $_}++;
+				$badword->add($_);
 			}
 		}
 		$fh->close;
+		$badword->put;
+		unlink $oldfn;
 	} else {
-		my $l = "can't open $fn $!";
-		dbg('err', $l);
+		my $l = "can't open $oldfn $!";
+		dbg($l);
 		push @out, $l;
 	}
 	return @out;
@@ -54,7 +56,29 @@ sub load
 # check the text against the badwords list
 sub check
 {
-	return grep { $badwords{$_} } 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;