#!/usr/bin/perl
#
# Break down the bytes of this access.log.


my %uri_bytes;
my $bytes_total;

while (defined (my $line = <STDIN>)) {
    # For PEx, the pid is included.
    my ($pid, $host, $rfc931, $authuser, $timestamp, $request, $status,
	$bytes) =
	($line =~ /^(\d+) (\S+) (\S+) (\S+) \[([^\]]*)\] \"([^\"]*)\" (\S+) (\S+)/);

    my ($uri) = ($request =~ /^GET\s(.*?)\s/);

    if ($uri) {
	my $s;

	# xxx Suboptimal!
	if ($uri =~ /\?/) {
	    ($s) = ($uri =~ /^(.*?)\?/);
	}
	else {
	    $s = $uri;
	}
	
	$uri_bytes{$s} += $bytes;
	$bytes_total += $bytes;
    }
}

my @a;
foreach my $k (keys (%uri_bytes)) {
    push (@a, [$k, $uri_bytes{$k}]);
}
@a = sort {-($a->[1] <=> $b->[1])} @a;

foreach my $e (@a) {
    printf "%12d %3d %s\n", $e->[1], $e->[1] * 1000 / $bytes_total, $e->[0];
}