#!/usr/bin/perl

# simple squid log "billing" for czuby.net

use Time::localtime;

if (@ARGV < 1)
{
	print "Usage: $0 <ip>\n";
	exit 0;
}

@urle = @daty = undef;
%redutant = ();
$i = $j = $k = $fromdate = $enddate = 0;

while(<STDIN>)
{
	chomp;
	$j++;

	if (m!(\d+)\.\d+\s+\d+ (\d+\.\d+\.\d+\.\d+) (TCP_HIT|TCP_MISS|TCP_REFRESH_HIT|TCP_REF_FAIL_HIT|TCP_REFRESH_MISS|TCP_CLIENT_REFRESH_MISS|TCP_IMS_HIT|TCP_MEM_HIT)/\d{3} (\d+) GET (?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))? (\w+|-) .*!)
	{
		if ($2 eq $ARGV[0])
		{
			$k++;
			$tm = localtime($1);
			$data = sprintf("%02d-%02d-%04d %02d:%02d:%02d",
				$tm->mday, $tm->mon+1, $tm->year+1900,
				$tm->hour, $tm->min, $tm->sec);

			if ($fromdate == 0)
			{
				$fromdate = $data;
			}

			$url = $5 . "://" . $6 . $7;

			if ($1 - $redutant{$6} > 60 && $6 ne "adserver.gadu-gadu.pl" && $6 ne "appmsg.gadu-gadu.pl" && $host ne $6)
			{
				$urle[$i] = $url;
				$daty[$i++] = $data;
				$host = $6;
			}

			$redutant{$6} = $1;
		}
	}
}

$enddate = $daty[$i-1];
$ile = @urle;

print << "EOF";

[anal by venglin\@freebsd.lublin.pl]

Przetworzono $j linii logów dla $ARGV[0]
$k dopasowanych, $ile po eliminacji
($fromdate do $enddate)

Oto spis używanych adresów WWW:

EOF

for(1..$ile)
{
format STDOUT =
@>>>>>>>>>>>>>>>>>>>>		@<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
$daty[$_],			$urle[$_]
.

write;
}