#!/usr/bin/perl -w
use strict;
use Getopt::Std;
use Cwd;

# don't bother displaying info for cumulative sizes smaller than total
# divided by this:
my $detail = 30;

#< options 
my %opts;
getopts('bkmghd:', \%opts); 
#(b)ytes, (k)ilobytes, (m)egabytes, (g)igabytes, (h)uman-readable
#(d)etail
if ($opts{h}) {
    print qq(usage: duke [-bkmg] [-d n] [dir]
             -bkmg   -  show sizes in bytes, kb, Mb, or Gb, respectively.
                        (default is human-readable)
             -d n    -  amount of detail: show entities at most total/n in size
                        (default $detail -- by necessity this is a bit more 
                         than the number of items displayed)\n);
    exit 0;
}
$detail = $opts{d} if $opts{d} and $opts{d} =~ /^\d+$/ and $opts{d} > 1; 

#< size output format
my $divfactor=0;
if ($opts{b}) {
    $divfactor=1;
} elsif ($opts{k}) {
    $divfactor=1024;
} elsif ($opts{m}) {
    $divfactor = 1024 * 1024;
} elsif ($opts{g}) {
    $divfactor = 1024 * 1024 * 1024;
} else {
    $opts{human}=1;
}
sub human_readable {
    my $bytes = shift;
    unless ($opts{human}) {
        return sprintf "%1.0f", ($bytes / $divfactor);
    }
    my $kb = $bytes / 1024;
    if ($bytes < 1000) {
        return sprintf " %d",$bytes;
    } elsif ($bytes < 1024) {
        return sprintf "%0.2fk", $kb;
    } elsif ($kb < 10) {
        return sprintf "%1.1fk", $kb;
    } else {
        my $mb = $kb / 1024;
        if ($kb < 1000) {
            return sprintf "%3.0fk", $kb;
        } elsif ($kb < 1024) {
            return sprintf "%0.2fM", $mb;
        } elsif ($mb < 10) {
            return sprintf "%1.1fM", $mb;
        } else {
            my $gb = $mb / 1024;
            if ($mb < 1000) {
                return sprintf "%3.0fM", $mb
            } elsif ($mb < 1024) {
                return sprintf "%0.2fG", $gb;
            } elsif ($gb < 10) {
                return sprintf "%1.1fG", $gb;
            } else {
                return sprintf "%3.0fG", $gb;
            }
        }
    }
}
#>
#> 
#< walk(dir)
sub walk {  
    # for each directory I'm given, I want to return an arrayref of the form
    # [ sum_of_content_sizes, name, age, 
    #   [biggest_constituent], [next_biggest_constituent], ...
    #       ... [smallest_constituent]
    # ]

    my $dirname = shift;
    my $dirsize = -s $dirname;
    my $dirage = -M _;
    
    local *DIR;
    opendir DIR, $dirname or do {
        warn "WARNING: cannot open directory [$dirname]: $!\n";
        warn "         this directory will be skipped: the report will be wrong.
\n";
        return [$dirsize, $dirage, $dirname];
    };

    my @sizes = ();
    my @files = readdir(DIR);
    closedir DIR;
    shift @files; shift @files; # skip . and ..
    foreach my $file (@files) {
        # get the stat info:
        my ($dev, $ino, $mode, $nlink, $uid, $gid, $rdev, $size, 
            $atime, $mtime, $ctime, $blksize, $blocks) = lstat qq($dirname/$file
);
        # we won't use most of it.  The real size is returned in $size,
        # but this isn't what actually matters to counting disk usage,
        $size = $blksize*(1+(int($size / $blksize)));

        if (not -l _ and -d _) {
            my $subdir = walk(qq($dirname/$file));
            push @sizes, $subdir;
            $dirsize += $subdir->[0];
            $dirage = $subdir->[1] if $subdir->[1] < $dirage;
        } else {
            my $age = -M _;
            push @sizes, [ $size, $age, $file ];
            $dirsize += $size;
            $dirage = $age if $age < $dirage;
        }
    }
    return [ $dirsize, $dirage, $dirname,
             sort { $b->[0] <=> $a->[0] } @sizes ];
}
#>
#< display(sizes) 
sub display {
    my $sizes = shift;
    my $depth = shift;
    my $total = shift;
    $depth ||= 0;
    my ($size, $age, $name, @contents) = @$sizes;
    $name =~ s|/$||;
    $name =~ s|.*/|| if scalar @contents;
    $total ||= $size;
    return if ($size * $detail < $total);
    print "   "x$depth;
    printf "%s  %s (%1.1f)\n", human_readable($size), $name, $age;
    foreach my $ref (@contents) {
        last unless display($ref, $depth+1, $total);
    }
    return 1;
}
#>

print "#size  name (age in days)\n";
display(walk(shift(@ARGV) or cwd()));