What command do I use to find the size of all the files (recursively) in a Linux or Mac OS X directory?
5 Answers
The BSD version of du
used in OS X reports size with 512-byte blocks -- the sizes are essentially rounded up to the next 512-byte value. This tells you the space on disk, which is larger than the amount of data. If you have a lot of small files, the difference can be large.
Here's an example.
This is the value with regular du
. It's in 512-byte blocks:
$ du -s
248 .
The -h
flag results in a more readable number, in kilobytes. As expected, it's half the number of 512-byte blocks:
$ du -hs
124K .
Finally, you can use find
and awk
to give you the sum of actual bytes in the files. This is kind of slow, but it works:
$ find . -type f -exec ls -l {} \; | awk '{sum += $5} END {print sum}'
60527
This value matches exactly the number reported by Finder's Get Info window. (There are no weird forks or xattrs in this set of files.) It's significantly smaller than the value reported by du
.
Here's how it works: it gets a list of all the files, and passes them to ls -l
; then awk
is used to count up the bytes. The -type f
flag is there so that only files (and not directories) get sent to ls
. Without that flag, it'll also send directory names to ls
, and each file will be listed twice : once as an individual file, and once as an item in the directory.
The GNU version of du
can give values in actual bytes instead of blocks. It's unfortunate that the BSD version of du
is not as flexible.
-
Great explanation. Interestingly when using
du
from GNU coreutils with the-b
option I get a different result than with yourfind
pipe. For a 36 GB directory I get a 82 KB difference. Jul 25, 2012 at 22:15 -
is there any way to print this out in a more human readable format? Apr 26, 2017 at 21:42
Show the size of a single file
du -h path_to_a_file
Show the size of the contents of a directory, each sub-directory, and each individual file:
du -h path_to_a_directory
Show the size of the contents of a directory:
du -sh path_to_a_directory
-
25
-
@Johannes, I guess you should post this comment as an answer, so it can be accepted?– ArjanAug 15, 2009 at 17:15
-
I find the -c option redundant (at least on Mac OS X) and du -sh supresses the subtotals. Aug 17, 2009 at 18:35
-
Aha, I didn't know one could accept one's own answer. Now at least people know the question is solved.– ArjanAug 18, 2009 at 9:41
-
I'll happily accept a new answer that is an improvement on mine. Aug 18, 2009 at 16:10
You can use du -ah .
which displays sizes of all files and directories recursively.
This can be combined with sort
, so you'll see the top-20 biggest directories in the current folder:
du -ah . | sort -rh | head -20
Note: Option -h
for sort
is not available on OSX/BSD, so you've to install sort
from coreutils
(e.g. via brew
) and apply the bin path to PATH
, e.g.
export PATH="/usr/local/opt/coreutils/libexec/gnubin:$PATH" # Add a "gnubin" for coreutils.
Otherwise use:
du -a . | sort -rn | head -20
-
First command not work on my macOS 10.12.1 #du -ah . | sort -rh | head -20 sort: invalid option -- h Try `sort --help' for more information. Nov 24, 2016 at 15:11
-
3@nucleartux Check the note below the command. This works in GNU sort which you can install on macOS via:
brew install coreutils
, otherwise use the BSD sort command at the end without-h
.– kenorbNov 24, 2016 at 15:44
du - tells the disk use not the file size.
find . -type f -print0 | xargs -0 stat -f%z | awk '{b+=$1} END {print b}'
above terminal code (im on osx 10.6) offers for me the best result and is waaay faster than "find ... -exec"
a quick benchmark
time find . -type f -print0 | xargs -0 stat -f'%z' | awk '{b+=$1} END {print b}'
4744010970
real 0m0.086s
user 0m0.029s
sys 0m0.073s
time find . -type f -exec ls -l {} \; | awk '{sum += $5} END {print sum}'
4744010970
real 0m18.515s
user 0m2.929s
sys 0m9.339s
-
1If you want to do the same on Ubuntu:
find . -type f -print0 | xargs -0 stat -c%s | awk '{b+=$1} END {print b}'
May 4, 2015 at 12:51 -
is there a way to get this result in a human-readable format? Apr 26, 2017 at 21:46
I combined all your approuches and combined it with a human readable output the result is:
#!/bin/sh
find $1 -type f -print0 | xargs -0 stat -f'%z' | awk '{b+=$1} END {print b}' | awk '{ sum=$1 ; hum[1024**3]="Gb";hum[1024**2]="Mb";hum[1024]="Kb"; for (x=1024**3; x>=1024; x/=1024){ if (sum>=x) { printf "%.2f %s\n",sum/x,hum[x];break } }}'
Link to the gist: https://gist.github.com/mlegenhausen/9365461
-k
option to use block size of 1K-bytedu
, I usencdu
, you can find the reference here, remysharp.com/2018/08/23/cli-improved