#!/usr/bin/perl
#
# Quick scrip to scrub the referrers section of the awstats file. 
# By: David A. Flanigan
# No Warrnety provided or implied, use at your own risk. 
# Free for any use. Please send improvements back to dave@flanigan.net
# E-mail dave@flanigan.net with questions on modifying this script. 
# http://www.flanigan.net/scripts
#

# blacklist file format, domain or strings, 1 per line. No regular expressesion. 
my $blfile = "/etc/awstats/blacklist.txt";

# Name of Domain to clear - should as specified in awstats config. 
# default domain if nothing is specified in the command line
if ($ARGV[0] ne "") { $awdom = $ARGV[0]; }
else { $awdom = "www.flanigan.net"; }

# init varables
use File::Copy;
$found = 0;
$removed = 0;
$date = `/bin/date`;
chop $date;

#Step 1 - Creat name of the awstats lib file. 
my $awdate = `/bin/date +%m%Y`;
chop $awdate;

$awlib = "/var/lib/awstats/awstats$awdate.$awdom.txt";
$awnew = "/var/lib/awstats/awstats$awdate.$awdom.txt.new";
$awbak = "/var/lib/awstats/awstats$awdate.$awdom.txt.awcback";

#check to verify the $awlib file exists before doing anything else
if (!( -e $awlib)) { die "Error Opening Lib file: $awlib file not found.\n"; }

#creat copy of the awstats file
copy($awlib, $awbak);

#step 3 open awstats.txt and blacklist file
open (AWTXT, "$awlib");
open (AWNEW, ">$awnew");

# Array the BL
open (BL, "$blfile");
@blacklist = <BL>;	
close BL;


# Step 3 - main processing loop
foreach $line( <AWTXT> )
{
	if ($start == 1)
	{
		foreach $item(@blacklist)
		{
			chomp $item;
			#print "debug:$item | $_ | found: $found\n";
			if ($line =~ /$item/)
			{
				$removed = $removed +1;
				#print "removed entry: $line";
				$found = 1;
			}
		}
		if ($line =~ /END_PAGEREFS/) { $start = 0; print AWNEW "$line"; }
		if (($found == 0) && ($start == 1)) { print AWNEW "$line"; }
		$found = 0;
	}
	elsif ($line =~ /BEGIN_PAGEREFS/)
	{
		$start = 1;
		print AWNEW "$line";
	}
	else { print AWNEW "$line"; }
}
close AWTXT;
close AWNEW;

#replace current file with corrected file
copy ($awnew, $awlib);
unlink($awnew);

#spit out a quick log
if ($removed > 0) 
{
	open (LOGOUT, ">>/var/log/referspam.log");
	print LOGOUT "$date|$removed\n";
	close LOGOUT;

	#priout output to be reutnred by cron
	#print "\n\nAwstat Clean Script removed $removed referrer spam entries from $awlib\nScript run time: $date\n";
}
exit(0);