#!/usr/bin/perl

use Date::Manip;

# for gathering stats on a gentoo rsync server..
# version 0.01 - feb 15 2006
# Travis Morgan, travis@bigfiber.net

if (@ARGV != 1) {
	print "Usage: ".$0." logfile\n";
	exit;
}
$logfile = $ARGV[0];

&Date_Init("Internal=1");

$startdate = 99999999999999;
$enddate = 0;
$totcon = 0;
$totbw = 0;

$tz = Date_TimeZone;

$numxfer = 20;
$numhits = 20;
$numspeed = 20;

unless (open LOG, $logfile) {
	die("Could not open log file $logfile");
}
@log = <LOG>;
close LOG;

foreach $line (@log) {
	chomp $line;
	@line = split / /, $line;
	$pid = $line[2];
	$pid =~ s/\[(.*)\]/\1/;

	$datetime = $line[0].$line[1];
	$datetime =~ s/\///g;
	$datetime =~ s/://g;

	# start of a sync 
	if ($line =~ / \(([0-9]+\.){3,3}[0-9]+\)$/) {
		# figure out the "hostname ip" hostid
		$hostid = $line[$#line - 1]." ".$line[$#line];
		# save a sync for $hostid at the time for the dir synced
		$hosts{$hostid}{$datetime}{'dir'} = $line[6];
		# set the pid to this hostid
		$pids{$pid}{'hostid'} = $hostid;
		$pids{$pid}{'datetime'} = $datetime;
	# end of a sync
	} elsif ($line =~ /] sent /) {
		# date will be in UTC - adjust
		$datetime = Date_ConvTZ(ParseDate($datetime),"UTC",$tz);
		
		if ($pids{$pid}) {
			$hosts{ $pids{$pid}{'hostid'} }{ $pids{$pid}{'datetime'} }{'sent'} = $line[4];
			$hosts{ $pids{$pid}{'hostid'} }{ $pids{$pid}{'datetime'} }{'rcvd'} = $line[8];
			$hosts{ $pids{$pid}{'hostid'} }{ $pids{$pid}{'datetime'} }{'finish'} = $datetime;
			delete($pids{$pid});
		}

	# other info like errors, etc
	} else {
	}

	# find the start date time
	if ($startdate > $datetime) {
		$startdate = $datetime;
	}

	# find the end date and time
	if ($enddate < $datetime) {
		$enddate = $datetime;
	}

}

foreach $hostid ( sort keys %hosts ) {
	foreach $date ( keys %{$hosts{$hostid}} ) {
	$totcon++;
		if ($hosts{$hostid}{$date}{'sent'}) {
			$transfer = $hosts{$hostid}{$date}{'sent'} + $hosts{$hostid}{$date}{'rcvd'};
			$totbw += $transfer;
			if ($hoststat{$hostid}) {
				$hoststat{$hostid}{'xfer'} = $hoststat{$hostid}{'xfer'} + $transfer;
				$hoststat{$hostid}{'num'} = $hoststat{$hostid}{'num'} + 1;
				$hoststat{$hostid}{'time'} = $hoststat{$hostid}{'time'} + &UnixDate($hosts{$hostid}{$date}{'finish'},"%s") - &UnixDate($date,"%s");
			} else {
				$hoststat{$hostid}{'xfer'} = $transfer;
				$hostnum{$hostid}{'num'} = 1;
				$hoststat{$hostid}{'time'} = &UnixDate($hosts{$hostid}{$date}{'finish'},"%s") - &UnixDate($date,"%s");
			}
		}
	}
}

$logstart = &UnixDate($startdate,"%s");
$logend = &UnixDate($enddate,"%s");
$logtime = ($logend - $logstart)/60/60/24;

printf "Log spans %.2f days starting at %s and ending at %s.\n",$logtime,&UnixDate($startdate,"%c"),&UnixDate($enddate,"%c");
print "Total connections during this time: $totcon\n";
printf "Total bandwidth during this time: %.3f gigabytes\n",$totbw/1024/1024/1024;
print "\n\n";

print "TOP HOSTS BY BANDWIDTH USE\n";
print "==========================\n";
printf "%4s   %11s   %s\n","Num","Transfer MB","Host";
printf "%4s   %11s   %s\n","---","-----------","----";
$num = 1;
foreach $hostid ( sort hashDescendingXfer (keys (%hoststat))) {
	if ($num <= $numxfer) {
		printf "%4d   %11.2f   %s\n",$num,$hoststat{$hostid}{'xfer'}/1024/1024,$hostid;
	}
	$num++;
}
print "\n\n";

print "TOP HOSTS BY NUMBER OF SYNCS\n";
print "============================\n";
printf "%4s   %11s   %s\n","Num","Connections","Host";
printf "%4s   %11s   %s\n","---","-----------","----";
$num = 1;
foreach $hostid ( sort hashDescendingHits (keys (%hoststat))) {
	if ($num <= $numxfer) {
		printf "%4d   %11d   %s\n",$num,$hoststat{$hostid}{'num'},$hostid;
	}
	$num++;
}
print "\n\n";

foreach $hostid ( keys (%hoststat)) {
	if ($hoststat{$hostid}{'time'} == 0) {
		$hoststat{$hostid}{'speed'} = 0;
	} else {
		$hoststat{$hostid}{'speed'} = $hoststat{$hostid}{'xfer'} / 1024 / $hoststat{$hostid}{'time'};
	}
}
		
print "TOP HOSTS BY TRANSFER SPEED\n";
print "============================\n";
printf "%4s   %11s   %s\n","Num","Speed kB/s","Host";
printf "%4s   %11s   %s\n","---","----------","----";
$num = 1;
foreach $hostid ( sort hashDescendingSpeed (keys (%hoststat))) {
	if ($num <= $numspeed) {
		# the timing info isn't very accurate so use only hosts that transferred over 2 megs
		if ($hoststat{$hostid}{'xfer'} > (2*1024*1024)) {
			printf "%4d   %11.2f   %s\n",$num,$hoststat{$hostid}{'speed'},$hostid;
			$num++;
		}
	}
}
print "\n\n";

sub hashDescendingXfer {
	$hoststat{$b}{'xfer'} <=> $hoststat{$a}{'xfer'};
}

sub hashDescendingHits {
	$hoststat{$b}{'num'} <=> $hoststat{$a}{'num'};
}

sub hashDescendingSpeed {
	$hoststat{$b}{'speed'} <=> $hoststat{$a}{'speed'};
}