X-Git-Url: http://dxcluster.net/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FBadWords.pm;h=141b3e9a729b6a933149cb0bf71bf4a942938285;hb=b06fd1d447d5411d8d441e807f93efa897b68aaf;hp=36db8ffba44384d76a3bc8acbc9789d0447cbbef;hpb=888290f339e2ee00894445fecb14f0b506d12368;p=spider.git

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index 36db8ffb..141b3e9a 100644
--- a/perl/BadWords.pm
+++ b/perl/BadWords.pm
@@ -3,7 +3,7 @@
 #
 # Copyright (c) 2000 Dirk Koopman
 #
-# $Id$
+#
 #
 
 package BadWords;
@@ -17,24 +17,18 @@ use DXDebug;
 
 use IO::File;
 
-use vars qw($badword @regex);
+use vars qw($badword $regexcode);
 
-my $oldfn = "$main::data/badwords";
-my $regex = "$main::data/badw_regex";
-my $bwfn = "$main::data/badword";
+my $oldfn = localdata("badwords");
+my $regex = localdata("badw_regex");
+my $bwfn = localdata("badword");
 
 # copy issue ones across
-filecopy("$regex.issue", $regex) unless -e $regex;
+filecopy("$regex.gb.issue", $regex) unless -e $regex;
 filecopy("$bwfn.issue", $bwfn) unless -e $bwfn;
 
 $badword = new DXHash "badword";
 
-use vars qw($VERSION $BRANCH);
-$VERSION = sprintf( "%d.%03d", q$Revision$ =~ /(\d+)\.(\d+)/ );
-$BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/ ) || 0;
-$main::build += $VERSION;
-$main::branch += $BRANCH;
-
 # load the badwords file
 sub load
 {
@@ -61,11 +55,10 @@ sub load
 sub create_regex
 {
 	my @out;
-	@regex = ();
-	
 	my $fh = new IO::File $regex;
 	
 	if ($fh) {
+		my $s = "sub { my \$str = shift; my \@out; \n";
 		while (<$fh>) {
 			chomp;
 			next if /^\s*\#/;
@@ -74,13 +67,19 @@ sub create_regex
 				# create a closure for each word so that it matches stuff with spaces/punctuation
 				# and repeated characters in it
 				my $w = uc $_;
-				my @l = map { $_ eq 'I' ? '[I1]' : ($_ eq 'O' ? '[O0]' : $_) }split //, $w;
-				my $e = join '+[\s\W]+', @l;
-				my $s = eval qq{sub { return \$_[0] =~ /$e+/ ? '$w' : () } };
-				push @regex, $s unless $@;
-				dbg("create_regex: $@") if $@;
+				my @l = split //, $w;
+				my $e = join '+[\s\W]*', @l;
+				$s .= "push \@out, \$1 if \$str =~ /\\b($e)/;\n";
 			}
 		}
+		$s .= "return \@out;\n}";
+		$regexcode = eval $s;
+		dbg($s) if isdbg('badword');
+		if ($@) {
+			@out = ($@);
+			dbg($@);
+			return @out;
+		}
 		$fh->close;
 	} else {
 		my $l = "can't open $regex $!";
@@ -96,17 +95,12 @@ sub check
 {
 	my $s = uc shift;
 	my @out;
-	
-	for (@regex) {
-		push @out, &$_($s);
-	}
+
+	push @out, &$regexcode($s) if $regexcode;
 	
 	return @out if @out;
 	
-	for (split(/\s+/, $s)) {
-		s/[^\w]//g;
-		push @out, $_ if $badword->in($_);
-		s/\'?S$//;
+	for (split(/\b/, $s)) {
 		push @out, $_ if $badword->in($_);
 	}