fix badword, ipv6 address detect, add CTY

[spider.git] / perl / BadWords.pm

diff --git a/perl/BadWords.pm b/perl/BadWords.pm
index 3d7404387664d34ffe7e44cda5d7380a31f2ccd2..312a04082c705fac550f9b0fcb065388cca2e564 100644 (file)
--- 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+)/  || (0,0));
-$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}";
@@ -107,7 +87,6 @@ sub check
        return @out if @out;
        
        for (split(/\b/, $s)) {
-               s/\'?S$//;
                push @out, $_ if $badword->in($_);
        }