#!/usr/bin/perl -w
# httpsum.pl,v 1.4 1999/09/25 06:23:12 wessels Exp

$cutoff=50;

$curtime=time();		# get current time stamp

$hostname = shift || die "usage: $0 servername\n";

$thcount = $tdoccount = $uniqdoccnt = 0;

while(<>) {
 next unless (/(\S+) \S+ \S+ \[[^\]]+\] "([^"]+)" \S+ (\S+)/);
 $from = $1;
 $request = $2;
 $size = $3;
 next unless ($request =~ /\S+\s+(\S+)/);
 $what = $1;
 $doccnt{$what}++;
 if(!$dochst{$what,$from}){$dochstcnt{$what}++;}
 $dochst{$what,$from}=1;
 $hcount{$from}++;
 $tdoccount++;
 &directory_bytes($what,$size);
 $client_bytes_hash{$from} += $size unless ($size eq '-');
}

foreach $key (keys %hcount){$thcount++;}
foreach $key (keys %doccnt){$uniqdoccnt++;}

($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime($curtime);
printf "HTTP server summary on $hostname at %d/%02d/%02d %.2d:%.2d:%.2d\n\n",
       $year+1900,$mon+1,$mday, $hour,$min,$sec;
$yday = $wday = $isdst = 0;	# shut up perl -w

printf"Total of $tdoccount documents ($uniqdoccnt unique) accessed by $thcount hosts\n\n";

printf "Top $cutoff accessed documents:\n\n";

$cutoffcnt = 0;
foreach $key ( sort cnumerically_access ( keys %doccnt)) {
 printf "%7d times from %5d hosts for %s\n",
       $doccnt{$key},$dochstcnt{$key},$key;
 if($cutoffcnt++ > $cutoff){last;}
}

printf "\n\n$cutoff most active hosts:\n\n";

$cutoffcnt = 0;
foreach $key ( sort cnumerically_host ( keys %hcount)) {
 last if $cutoffcnt++ >= $cutoff;
 printf "%7d times from %s\n",
       $hcount{$key},$key;
 last if $hcount{$key} <= 1;
}

$cutoffcnt = 0;
printf "\n\n$cutoff most popular directories by volume:\n\n";
foreach $p ( sort cmp_directory_bytes keys %directory_bytes_hash) {
	printf "%9d %s\n",  $directory_bytes_hash{$p}, $p;
	last if (++$cutoffcnt == $cutoff);
}

$cutoffcnt = 0;
printf "\n\n$cutoff most popular clients by volume:\n\n";
foreach $p ( sort cmp_client_bytes keys %client_bytes_hash) {
	printf "%9d %s\n",  $client_bytes_hash{$p}, $p;
	last if (++$cutoffcnt == $cutoff);
}

sub cnumerically_access { $doccnt{$b} <=> $doccnt{$a}; }
sub cnumerically_host   { $hcount{$b} <=> $hcount{$a}; }
sub cmp_directory_bytes	{ $directory_bytes_hash{$b} <=> $directory_bytes_hash{$a}; }
sub cmp_client_bytes	{ $client_bytes_hash{$b} <=> $client_bytes_hash{$a}; }

sub directory_bytes {
	local($U,$S) = @_;
	local @C = split(/\//, $U);
	local ($u) = '';
	local ($x) = $#C + 0;
	local ($c);
	return if ($S eq '-');
	foreach $c (@C) {
		$u .= $c;
		$u .= '/' unless ($c eq $C[$x]);
		$directory_bytes_hash{$u} += $S;
	}
}