Sorting files according to size recursively

Question

I need to find the largest files in a folder.
How do I scan a folder recursively and sort the contents by size?

I have tried using ls -R -S, but this lists the directories as well.
I also tried using find.

Solution 1

You can also do this with just du. Just to be on the safe side I'm using this version of du:

$ du --version
du (GNU coreutils) 8.5

The approach:

$ du -ah ..DIR.. | grep -v "/$" | sort -rh

Breakdown of approach

The command du -ah DIR will produce a list of all the files and directories in a given directory DIR. The -h will produce human readable sizes which I prefer. If you don't want them then drop that switch. I'm using the head -6 just to limit the amount of output!

$ du -ah ~/Downloads/ | head -6
4.4M    /home/saml/Downloads/kodak_W820_wireless_frame/W820_W1020_WirelessFrames_exUG_GLB_en.pdf
624K    /home/saml/Downloads/kodak_W820_wireless_frame/easyshare_w820.pdf
4.9M    /home/saml/Downloads/kodak_W820_wireless_frame/W820_W1020WirelessFrameExUG_GLB_en.pdf
9.8M    /home/saml/Downloads/kodak_W820_wireless_frame
8.0K    /home/saml/Downloads/bugs.xls
604K    /home/saml/Downloads/netgear_gs724t/GS7xxT_HIG_5Jan10.pdf

Easy enough to sort it smallest to biggest:

$ du -ah ~/Downloads/ | sort -h | head -6
0   /home/saml/Downloads/apps_archive/monitoring/nagios/nagios-check_sip-1.3/usr/lib64/nagios/plugins/check_ldaps
0   /home/saml/Downloads/data/elasticsearch/nodes/0/indices/logstash-2013.04.06/0/index/write.lock
0   /home/saml/Downloads/data/elasticsearch/nodes/0/indices/logstash-2013.04.06/0/translog/translog-1365292480753
0   /home/saml/Downloads/data/elasticsearch/nodes/0/indices/logstash-2013.04.06/1/index/write.lock
0   /home/saml/Downloads/data/elasticsearch/nodes/0/indices/logstash-2013.04.06/1/translog/translog-1365292480946
0   /home/saml/Downloads/data/elasticsearch/nodes/0/indices/logstash-2013.04.06/2/index/write.lock

Reverse it, biggest to smallest:

$ du -ah ~/Downloads/ | sort -rh | head -6
10G /home/saml/Downloads/
3.8G    /home/saml/Downloads/audible/audio_books
3.8G    /home/saml/Downloads/audible
2.3G    /home/saml/Downloads/apps_archive
1.5G    /home/saml/Downloads/digital_blasphemy/db1440ppng.zip
1.5G    /home/saml/Downloads/digital_blasphemy

Don't show me the directory, just the files:

$ du -ah ~/Downloads/ | grep -v "/$" | sort -rh | head -6 
3.8G    /home/saml/Downloads/audible/audio_books
3.8G    /home/saml/Downloads/audible
2.3G    /home/saml/Downloads/apps_archive
1.5G    /home/saml/Downloads/digital_blasphemy/db1440ppng.zip
1.5G    /home/saml/Downloads/digital_blasphemy
835M    /home/saml/Downloads/apps_archive/cad_cam_cae/salome/Salome-V6_5_0-LGPL-x86_64.run

If you want to exclude all directories from the output, you can use a trick with the presence of a dot character. This assumes that your directory names do not contain dots, and that the files you are looking for do. Then you can filter out the directories with grep -v '\s/[^.]*$':

$ du -ah ~/Downloads/ | grep -v '\s/[^.]*$' | sort -rh | head -2
1.5G    /home/saml/Downloads/digital_blasphemy/db1440ppng.zip
835M    /home/saml/Downloads/apps_archive/cad_cam_cae/salome/Salome-V6_5_0-LGPL-x86_64.run

If you just want the list of smallest to biggest, but the top 6 offending files you can reverse the sort switch, drop (-r), and use tail -6 instead of the head -6.

$ du -ah ~/Downloads/ | grep -v "/$" | sort -h | tail -6
835M    /home/saml/Downloads/apps_archive/cad_cam_cae/salome/Salome-V6_5_0-LGPL-x86_64.run
1.5G    /home/saml/Downloads/digital_blasphemy
1.5G    /home/saml/Downloads/digital_blasphemy/db1440ppng.zip
2.3G    /home/saml/Downloads/apps_archive
3.8G    /home/saml/Downloads/audible
3.8G    /home/saml/Downloads/audible/audio_books

Solution 2

If you want to find all files in the current directory and its sub directories and list them according to their size (without considering their path), and assuming none of the file names contain newline characters, with GNU find, you can do this:

find . -type f -printf "%s\t%p\n" | sort -n

From man find on a GNU system:

   -printf format
          True; print format  on  the  standard  output,
          interpreting  `\'  escapes and `%' directives.
          Field widths and precisions can  be  specified
          as  with the `printf' C function.  Please note
          that many of the  fields  are  printed  as  %s
          rather  than  %d, and this may mean that flags
          don't work as you  might  expect.   This  also
          means  that  the `-' flag does work (it forces
          fields to be  left-aligned).   Unlike  -print,
          -printf  does  not add a newline at the end of
          the string.  The escapes and directives are:

          %p     File's name.
          %s     File's size in bytes.

From man sort:

   -n, --numeric-sort
          compare according to string numerical value

Solution 3

Try the following command:

ls -1Rhs | sed -e "s/^ *//" | grep "^[0-9]" | sort -hr | head -n20

It'll list top-20 biggest files in the current directory recursively.

Note: The 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 local bin path to PATH, e.g.

export PATH="/usr/local/opt/coreutils/libexec/gnubin:$PATH" # Add a "gnubin" for coreutils.

Alternatively use:

ls -1Rs | sed -e "s/^ *//" | grep "^[0-9]" | sort -nr | head -n20

---

For the biggest directories use du, e.g.:

du -ah . | sort -rh | head -20

or:

du -a . | sort -rn | head -20

Solution 4

This will find all files recursively, and sort them by size. It prints out all file sizes in kb, and rounds down so you may see 0 KB files, but it was close enough for my uses, and works on OSX.

find . -type f -print0 | xargs -0 ls -la | awk '{print int($5/1000) " KB\t" $9}' | sort -n -r -k1

Solution 5

Simple solution for Mac/Linux which skips directories:

find . -type f -exec du -h {} \; | sort -h

View more solutions

Comments

[Admin]

Do you want to list the files in each subdirectory separately or do you want to find all files in all subdirs and list them by size irrespective of which subdir they are in? Also, what do you mean by "directory" and "folder"? You seem to be using them to describe different things.

[Admin]

Are you saying that you just want to list the files in a given directory as well as the files in its sub-directories without showing just the sub-directories? Please try and clean up you question, it's not very clear.

[Admin]

Related unix.stackexchange.com/questions/158289/…

[Jan Warchoł]

The grep -v "/$" part doesn't seem to be doing what you expected, as the directories don't have a slash appended. Does anyone know how to exclude directories from results?

[slm]

@JanekWarchol - what version of coreutils are you using?

[Jan Warchoł]

I'm on 8.13. But anyway, the output in your answer doesn't have trailing /s either - for example /home/saml/Downloads/audible seems to be a directory, but it doesn't have a slash. Only /home/saml/Downloads/ has a slash, but that's probably because you wrote it with a slash when specifying the argument for initial du.

[slm]

@JanekWarchol - Look at the 5th text box. The grep is just to filter the ~/Downloads/ bit out. As you've stated, it's just to filter out the argument of ~/Downloads when du processes it. I changed the word directories to directory since I think that's ultimately what was causing the confusion. Thanks for the feedback!

[slm]

@JanekWarchol - incidentally to omit the directories you'll have to change tactics and use find to generate a list of files only and then have du tally them up.

[Artem]

worked on Ubuntu 14.04 too!

[ekerner]

This finds dirs also

[Roman Gaufman]

This doesn't list just files, but also lists directories :(

[Roman Gaufman]

Doesn't work on Mac unfortunately, shows: find: -printf: unknown primary or operator

[Roman Gaufman]

This lists directories, not just files :(

[Roman Gaufman]

Perfect, this is the first solution that works on Mac and doesn't show directories :) - thank you!

[terdon]

@RomanGaufman yes, that's why the answer specifies GNU find. If you install the GNU tools on your Mac, it will work there too.

[Brad Parks]

@RomanGaufman - thanks for the feedback! from my tests, find . -type f finds files... it works recursively, you're right, but it lists all the files it finds, not the directories themselves

[schily]

Xargs has been used in the 1980s. It is a bad idea since 1989 when execplus has been introduced by David Korn.

[Matrix]

how filter to show only file with number of lines >= X ? (X = 0 for exemple)

[Ryan Shirley]

du --version in not an option for me, however the approach does work

[mircealungu]

there's no need for the -s in sort. or?

[flochtililoch]

building on that solution, and the solution offered on this post: unix.stackexchange.com/questions/22432/…, I was able to yield a result with files only with the following command: find . -type f -exec du -ah {} + | grep -v "/$" | sort -rh

[Student]

@flochtililoch Super answer! It works for me :)

[nyanpasu64]

Use du --apparent-size to view the length of the file in bytes (not the size taken on disk).

[odony]

@JanWarchoł I could not find any way for du to provide a different output for directory (-S changes their size, not the output format). So I resorted to a workaround that assumes that your directory names do not have any dots in them, and that the files you care about do. In that case you can replace the non-working grep -v '/$' by grep -v '\s/[^.]*$'. And you have to make sure you specify DIR as an absolute path, so there is no dot in there.

[Kamil Dziedzic]

How about listing only directories, I'm not interested in files.

[Erikw]

Great command but one problem: It does not handle filenames well with spaces. Only the part up until the first space will be printed as of $9. I tried to improve on this by printing ranges $9-end. I ended up with: find "$path" -type f -print0 | xargs -0 ls -la | awk '{for (i=5; i<NF; i++) { if (i == 5) { printf int($i/1024) " KiB\t" } else if (i >= 9 ) { printf $i " " } }; if (NF >= 5) print $NF; }' | sort -n -r -k1. It seems to work, but feedback is much welcome to make this a bit more concise

[Admin]

You can also use "%P" (uppercase) which is: "File's name with the name of the starting-point under which it was found removed". In this case it would remove the "./" at the beginning of each path.