[selowitch]

My hard drive is quite full and I'm not sure why. Accordingly, I'd like to display the top, say, 30 largest files on my hard drive in descending order by size. I'm figuring on using the "du" command in the Terminal but I'm finding the precise syntax a little hard to glean from the man pages.

Anyone know how to form this command?

EDIT: I found a good link here that helps a lot. Thought I'd let you know how this issue progressed for me. Thanks!

[Gavin]

You could also check out OmniDiskSweeper - not quite what you want but it helps you drill in to find where your disk space went.

or use finder to search for files > xxxx Mb

[Agent69]

I back up my computer using the command prompt and before I start, I like to know how much data each directory in my home folder is using. I do this, I issue the following command:

Code:
du -h -d 0

The -h means human readable, so it will print sizes in megabytes, gigabytes, and so on. The -d 0 is the depth level, which allows you to control how far down it dig from the current directory. There is also a -c option for grand total but I have never used it.

Hope this helps.

[selowitch]

Originally Posted by Agent69

Code:
du -h -d 0

That is helpful, but what I'm really looking for is to find the largest twenty or thirty files on my machine sorted by size. Is there a way to do that?

There is a program called DooBar that might do the trick, but I'd prefer to do it via the command line if possible.

[OreoCookie]

Code:
du -h -d 1

is more helpful. It'll show you how much space each subdirectory of the current directory takes.

[selowitch]

Code:
du -h -d 1 *|sort

Better yet, I think. The only trouble is it sorts alphabetically, so "k" comes before "mb" and so it doesn't truly order by size. Any ideas how to further refine it?

[Agent69]

Code:
du -h -d 1

is more helpful. It'll show you how much space each subdirectory of the current directory takes.

You're right. That's what I get for working for memory (I'm at work).

If you're wanting to sort by size, you would probably NOT want human readable output. It would probably be better to use the default or -k for consistancy's sake.

You could probably do something along the line of:

Code:
du -d 1 * | sort | head -n 30

However, when I tried this with a secure shell account on a OpenBSD machine, files with zero bytes showed up first, so you would probably need to prevent those lines from printing. (I believe you could use Grep to accomplish this but I am not sure.)

[Agent69]

Actually, it looks like I was remembering correctly about "-d 0". Here is a copy of what happens when I do so in my home directory:

Code:
Goot:~ goo$ du -h -d 0 *       
181M    Applications
121M    Audio
 16K    Desktop
 63M    Documents
201M    Library
291M    Movies
121M    Music
286M    Pictures
582M    Zoe
Goot:~ goo$

Note that this command returns the sizes of the directories in Home. Here is what happens with a "-d 1":

Code:
Goot:~ goo$ du -h -d 1 *
244K    Applications/Backdrop.app
7.2M    Applications/BitTorrent.app
1.8M    Applications/Bookdog.app
 22M    Applications/Camino.app
4.4M    Applications/Cocktail.app
4.9M    Applications/Cog.app
552K    Applications/ExifRenamer.app
 19M    Applications/GraphicConverter
 36M    Applications/MacMAME
 11M    Applications/NetNewsWire Lite.app
1.7M    Applications/NicePlayer.app
 25M    Applications/OmniOutliner Professional.app
 23M    Applications/RealPlayer.app
2.4M    Applications/Safari Enhancer.app
156K    Applications/SafariSpeed.app
732K    Applications/SMARTReporter.app
 10M    Applications/SubEthaEdit.app
 72K    Applications/Twemco.wdgt
3.5M    Applications/ViewIt.app
6.7M    Applications/XinePlayer.app
181M    Applications
 65M    Audio/Gwen Sefani - Love Angel Music Baby
 56M    Audio/Wang Chung - To Live And Die In LA
121M    Audio
 16K    Desktop
 36K    Documents/Comic Books
 68K    Documents/Federal Taxes
7.9M    Documents/Flash
632K    Documents/Mac Weekly Journal
 41M    Documents/MacMAME
 13M    Documents/Miscellaneous
 63M    Documents
  0B    Library/Address Book Plug-Ins
216K    Library/Application Enhancers
4.6M    Library/Application Support
  0B    Library/Assistants
  0B    Library/Audio
  0B    Library/Autosave Information
191M    Library/Caches
  0B    Library/Classic
  0B    Library/ColorPickers
4.0K    Library/Colors
8.0K    Library/ColorSync
  0B    Library/Contextual Menu Items
216K    Library/Cookies
  0B    Library/Documentation
4.0K    Library/Favorites
 40K    Library/FontCollections
680K    Library/Fonts
 36K    Library/Icons
  0B    Library/iMovie
8.0K    Library/Internet Plug-Ins
  0B    Library/iTunes
  0B    Library/Keyboard Layouts
 32K    Library/Keychains
236K    Library/Logs
428K    Library/Mail
  0B    Library/Mail Downloads
  0B    Library/Phones
468K    Library/PreferencePanes
2.7M    Library/Preferences
  0B    Library/Printers
440K    Library/Receipts
4.0K    Library/Recent Servers
 32K    Library/Safari
  0B    Library/Saved Searches
  0B    Library/Screen Savers
172K    Library/Services
  0B    Library/Sounds
4.0K    Library/Spelling
 12K    Library/Syndication
201M    Library
291M    Movies
121M    Music/iTunes
121M    Music
1012K   Pictures/BeOS-Icons
3.5M    Pictures/Cgchannel
 38M    Pictures/Deviantart
1.3M    Pictures/FR4NK
396K    Pictures/Frank-Cho
796K    Pictures/Katrine-Ewert
 13M    Pictures/Mitch-Bird
6.5M    Pictures/My Junk
159M    Pictures/Photosig
3.0M    Pictures/Photosig - REQs
 24M    Pictures/Rakaz.nl
4.2M    Pictures/Rampant-Mac
 23M    Pictures/Screenshots
4.4M    Pictures/Sxc.hu
3.2M    Pictures/Wallpapers
286M    Pictures
582M    Zoe
Goot:~ goo$

[Mithras]

I use the following script to sort the output of du . by descending size, and convert the sizes to human-readable.
Use like this:
du | usagereport.pl all > usage.txt

#!/usr/bin/env perl
use warnings;
use strict;

my ($doall) = @ARGV;

my @dirs;

while (<STDIN>)
{
chomp;
if (m/^(\d+)\s+(.*)$/)
{
my ($size, $dir) = ($1, $2);
my $is_topdir = 0;
if ($dir =~ m/^(\.\/)?([^\/]*)$/)
{
$is_topdir = 1;
}

my %this_dir;
$this_dir{'dir'} = $dir;
$this_dir{'size'} = $size;
if ($doall)
{
push( @dirs, \%this_dir );
}
else
{
if ($is_topdir)
{
push( @dirs, \%this_dir );
}
}
}
}

# sort
@dirs = sort { $$b{'size'} <=> $$a{'size'} } @dirs;

#
# output
#
foreach my $dir_ref (@dirs)
{
my %this_dir = %$dir_ref;
print size_to_human($this_dir{'size'}) . "\t" . $this_dir{'dir'} . "\n";
}

sub size_to_human
{
my ($size) = @_;

my $prec = 2;

my $size_kb = $size;
my $size_mb = round( $size / (1024 **1), 2);
my $size_gb = round( $size / (1024 **2), 2);

my $size_text = "";
if ($size_gb > 1)
{
$size_text = "$size_gb GB";
}
elsif ($size_mb > 1)
{
$size_text = "$size_mb MB";
}
else
{
$size_text = "$size_kb kB";
}
}

sub round
{
my ($x, $prec) = @_;
return ( (int ( $x * (10 ** $prec) ) ) / (10 ** $prec) );
}

Hope that helps.
-mithras

[gorm@cbs.dtu.dk]

Originally Posted by selowitch

Code:
du -h -d 1 *|sort

Better yet, I think. The only trouble is it sorts alphabetically, so "k" comes before "mb" and so it doesn't truly order by size. Any ideas how to further refine it?

The following commandline will give you disk usage in kilobytes for any item (file or dir) below the current dir. The results are then sorted numerically (-n) and in reverse (-r) so largest files are reported first. Finally the 10 first on the list are shown

Code:
du -k . | sort -nr | head -10

Listing everything in kilobytes (du -k) instead of in human readable format (du -h) makes sorting more relevant