#  Access log slicer by A.L. Mudd (perl@tonymudd.co.uk)
print "\nAccess log slicer.\n------------------\n";

# open the old refer.txt file and read it in.
open (REFS,"refer.txt")
    or print "Could not open refer.txt\n";
while ($line = <REFS>)
{
    chomp $line;
    $refs{$line} = 1;
}
close REFS;

# open the old view.txt.
open (VIEW,"view.txt")
    or print "Could not open view.txt\n";
while ($line = <VIEW>)
{
    chomp $line;
    $views{$line} = 1;
}
close VIEW;

open (INFILE, "access.log")
    or die "Could not open file $!\n";
while ($line = <INFILE>)
{
#  s08103.vdo.com.au - - [16/Feb/1998:10:05:08 +0000] home.clara.net "GET /alm/antigrav.html HTTP/1.0" 200 2086 "http://home.clara.net/alm/" "Mozilla/4.0 (compatible; MSIE 4.0; Windows 95)"
    if ($line=~ /^([\w\d\.-]+)\s-\s-\s\[(.+)\s.+\]\s(.+)\s\"\w+\s(.+)\sHTTP\/.+\"\s\d+\s[\d-]+\s\"(.+)\"\s\".+\"\n$/)
    {
	$remote_host = $1;
	$time = $2;
	# $3 is local hostname
	$localhost = $3;
	$request = $4;
	$referrer = $5;
	if ( ($localhost =~ /^home\.clara\.net$/) ||
	     ($localhost =~ /^-$/ ))
	{   # that was the hostname used in the request, so remove the
	    # first part of the request - it's the hostname
	    $request =~ /^\/[^\/]+(.+)$/;
	    $request = $1;
	}
	print "$request\t$referrer\n";
	$refs{$referrer} = 1;
	$views{$remote_host} = 1;
    }
    else
    {
        print "line didn't match $line";
    }
}
close INFILE;

# now write out the new refs file.
open (REFS, ">refer.txt")
    or die "Could not open refer.txt $!\n";
foreach $ref (sort keys %refs)
{
    print REFS "$ref\n";
}
close REFS;

# now write out the new view file.
open (VIEW, ">view.txt")
    or die "Could not open view.txt $!\n";
foreach $view (sort keys %views)
{
    print VIEW "$view\n";
}
close VIEW;