#!/usr/bin/perl
# This program rebuilds the cdb database. If called from the web server it
# will add the remote IP address to the blocklist before doing the rebuild.
# This should only be pointed to from web pages that use meta tags to
# tell robots not to follow links off the page.
use CDB_File;
umask 022;
chdir '/var/www/html/robot';
if ($ENV{REMOTE_ADDR} =~ m/^\d+\.\d+\.\d+\.\d+$/) {
if (open(IP,">ip1/$ENV{REMOTE_ADDR}")) {
# Save referer header in case some bozo links to this page. We will want to
# know that and block access from any server that does this.
# Later we will want to add a meta robot tag to keep valid robots who are
# misdirected here from remembering the page which will make them come back
# repeatedly without using a referer (sic) header which we can block on.
print IP "$ENV{HTTP_REFERER}\n";
close(IP);
}
print "Content-type: text/html\n\n";
$title = "Your IP address ($ENV{REMOTE_ADDR}) has been blocked!";
print << "EOF";
$title
$title
All future requests from your IP address ($ENV{REMOTE_ADDR}) will result
in a 403 forbidden result.
Most likely this message is being read by a robot that does not pay
attention to meta robot tags. Robots that ignore these tags
are not permitted here. for more information see
http://www.robotstxt.org/wc/exclusion.html
.
EOF
}
else {
print "Rebuilding the blocklist.\n";
}
# Use a format compatible with rbldns.
opendir(IP, 'ip1');
while ($ip = readdir(IP)) {
if ($ip =~ m/^(\d+)\.(\d+)\.(\d+)\.(\d+)$/) {
$key = pack 'CCCCC', $1, $2, $3, $4, 32;
$ip{$key} = '';
}
elsif ($ip =~ m/^(\d+)\.(\d+)\.(\d+)$/) {
$key = pack 'CCCCC', $1, $2, $3, 0, 24;
$ip{$key} = '';
}
elsif ($ip =~ m/^(\d+)\.(\d+)$/) {
$key = pack 'CCCCC', $1, $2, 0, 0, 16;
$ip{$key} = '';
}
}
$ip{''} = (pack 'CCCC', 127, 0, 0, 10) . 'Spam block';
CDB_File::create %ip, 'data1/data.cdb', "data1/data.cdb.$$";