#!/usr/bin/perl -w

# package size for gentoo
# by bobrik <ibobrik@gmail.com>, under GPLv3

my $pkgdir = '/var/db/pkg/';
opendir(PKGDIR, $pkgdir) or die "Can't open package directory: $pkgdir (${!})";
my @categories = ();

while ($dir = readdir(PKGDIR)) {
	if ($dir !~ /^\./) {
		push(@categories, $dir);
	}
}

closedir(PKGDIR);

my @packages = ();
for $category (@categories) {
	opendir(CATEGORY, $pkgdir.$category) or die "Can't open ${pkgdir}${category}: $1";
	while ($package = readdir(CATEGORY)) {
		if ($package !~ /^\./) {
			push(@packages, $category.'/'.$package);
		}
	}
	closedir(CATEGORY);
}

my %sizes;
my %packages;
my @result_sizes;
my @result_packages;

my $i = 0;
for $package (@packages) {
	open(PACKAGE, $pkgdir.$package."/CONTENTS") or (warn "Can't open package fillist for $package (${!})" && next);
	my @lines = <PACKAGE>;
	close(PACKAGE);
	my $size = 0;
	my %md5sums;
	for $line (@lines) {
		if ($line =~ /obj (.*) ([a-z0-9]{32}) \d+/) {
			if (-e $1) {
				$md5sums{$2} = $1;
			}
		}
	}
	for $md5 (keys %md5sums) {
		$size += (-s $md5sums{$md5});
	}
	$result_sizes[$i] = $size;
	$result_packages[$i] = $package;
	$sizes{$size} = $i;
	$packages{$package} = $i;
	++$i;
	warn "Processed $package...\n";
}

for (sort {$a <=> $b} @result_sizes) {
	next if $_ == 0;
	print (int(${_}/1024/1024*1000)/1000);
	print " Mb ".$result_packages[$sizes{$_}]."\n";
}