X-Git-Url: http://dxcluster.net/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FBadWords.pm;h=312a04082c705fac550f9b0fcb065388cca2e564;hb=6ec22e78a4a344ce675645fabf18b2a1971f364a;hp=d0572cc2de4cb9f99b60b02461383d56c8117e6f;hpb=299bbec5a5628c3333cdf4ed45fed3e0c21ca753;p=spider.git

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index d0572cc2..312a0408 100644
--- a/perl/BadWords.pm
+++ b/perl/BadWords.pm
@@ -3,7 +3,7 @@
 #
 # Copyright (c) 2000 Dirk Koopman
 #
-# $Id$
+#
 #
 
 package BadWords;
@@ -19,47 +19,27 @@ use IO::File;
 
 use vars qw($badword $regexcode);
 
-my $oldfn = "$main::data/badwords";
-my $regex = "$main::data/badw_regex";
-my $bwfn = "$main::data/badword";
-
-# copy issue ones across
-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;
+our $regex;
 
 # load the badwords file
 sub load
 {
+	my $bwfn = localdata("badword");
+	filecopy("$main::data.issue", $bwfn) unless -e $bwfn;
+	
 	my @out;
-	my $fh = new IO::File $oldfn;
+
+	$badword = new DXHash "badword";
 	
-	if ($fh) {
-		while (<$fh>) {
-			chomp;
-			next if /^\s*\#/;
-			my @list = split " ";
-			for (@list) {
-				$badword->add($_);
-			}
-		}
-		$fh->close;
-		$badword->put;
-		unlink $oldfn;
-	}
 	push @out, create_regex(); 
 	return @out;
 }
 
 sub create_regex
 {
+	$regex = localdata("badw_regex");
+	filecopy("$regex.gb.issue", $regex) unless -e $regex;
+	
 	my @out;
 	my $fh = new IO::File $regex;
 	
@@ -75,7 +55,7 @@ sub create_regex
 				my $w = uc $_;
 				my @l = split //, $w;
 				my $e = join '+[\s\W]*', @l;
-				$s .= "push \@out, \$1 if \$str =~ /\\b($e)/;\n";
+				$s .= qq{push \@out, \$1 if \$str =~ m|\\b($e+)|;\n};
 			}
 		}
 		$s .= "return \@out;\n}";
@@ -106,8 +86,7 @@ sub check
 	
 	return @out if @out;
 	
-	for (split(/\s+/, $s)) {
-		s/\'?S$//;
+	for (split(/\b/, $s)) {
 		push @out, $_ if $badword->in($_);
 	}