Linux delete file with size 0

Question

How do I delete a certain file in linux if its size is 0. I want to execute this in an crontab without any extra script.

l filename.file | grep 5th-tab | not eq 0 | rm

Something like this?

score 332 · Accepted Answer · edited Feb 28 '22 at 13:35

332

This will delete all the files in a directory (and below) that are size zero.

find /tmp -size 0 -print -delete

If you just want a particular file;

if [ ! -s /tmp/foo ] ; then
  rm /tmp/foo
fi

edited Feb 28 '22 at 13:35

4wk_

2,458
3
34
46

answered Mar 29 '11 at 16:41

Paul Tomblin

179,021
58
319
408

8

shortcut: `[ -s /tmp/foo ] || rm /tmp/foo` (test if size is zero, else remove). Also note the `xargs` is unsafe if file/directory names contain spaces; `find ... -exec rm '{}' \;` is safe in that situation. – FrankH. May 13 '11 at 10:25
11

@Frank, you are incorrect about `xargs`. The '-print0` and `xargs -0` corrects for the spaces. – Paul Tomblin May 13 '11 at 10:49
1

@FrankH: Plus, even if using `find -exec`, always favour `+` over `;` in cases where you can (and this is one such case). – C. K. Young Jun 10 '11 at 20:47
5

Would `rm --` (note the trailing `--` characters) be safer here than simply `rm` to prevent rogue filenames? http://serverfault.com/questions/337082/how-do-i-prevent-accidental-rm-rf – OnlineCop Oct 26 '15 at 23:20
Is there also a command for running a dry run, to see which files would get removed? – Simon Baars Apr 21 '18 at 17:29
2

@SimonBaars `find /tmp -size 0 -print0 |xargs -0 echo rm` – Paul Tomblin Apr 22 '18 at 20:26
Why not simply `-delete`? – Ruslan Apr 04 '19 at 06:36
@Ruslan because `-delete` is new, and my version works with every version of `find` produced in the last 40 years. – Paul Tomblin Apr 04 '19 at 16:32
Is `-print0` really 40 years old and ubiquitous? – Ruslan Apr 04 '19 at 16:35
@Rusian pretty sure it was in BSD 4.3, although apparently it didn't get into strict SysV Unices like early AIX. – Paul Tomblin Apr 04 '19 at 16:42

score 273 · Answer 2 · edited Sep 04 '15 at 07:33

273

you would want to use find:

 find . -size 0 -delete

edited Sep 04 '15 at 07:33

Antonio

19,451
13
99
197

answered Mar 29 '11 at 16:44

James.Xu

8,249
5
25
36

41

I would add "-type f", as also directory are marked to be of size zero. The "dot" is optional. – Antonio Aug 02 '13 at 10:28
7

The "dot" is optional for Linux, but not optional for Mac OS – Jakub M. Jan 31 '14 at 11:22
9

There is an `-empty` option :-) – lilydjwg Mar 20 '16 at 12:42
11

@lilydjwg Exactly, if we're allowed to use non POSIX flags, `find . -empty -delete` is the coolest. :-) – Anne van Rossum Oct 19 '16 at 19:32
2

You can add `-maxdepth 1` for the current folder. – aloisdg Feb 02 '17 at 16:31

Antonio · Answer 3 · 2017-03-17T12:53:49.977

150

To search and delete empty files in the current directory and subdirectories:

find . -type f -empty -delete

-type f is necessary because also directories are marked to be of size zero.

The dot . (current directory) is the starting search directory. If you have GNU find (e.g. not Mac OS), you can omit it in this case:

find -type f -empty -delete

From GNU find documentation:

If no files to search are specified, the current directory (.) is used.

edited Mar 17 '17 at 12:53

answered Aug 05 '13 at 07:24

Antonio

19,451
13
99
197

2

I would add -name '*.SomeFileExtension' for example: if you wanted to delete just text files then I would use: `find . -name '*.txt' -type f -empty -delete` – jspek Apr 18 '19 at 16:30
@jspek, well, that depends if you have that specific use... Usually when you are after empty files you are up to kill them all. :) – Antonio Apr 18 '19 at 20:45
2

Had to grab a coffee after running this command on a directory with 2.2 million files. :P Had worked like a charm when I came back, 350.000 remained. Thanks! – ArendE Feb 26 '20 at 19:42

score 17 · Answer 4 · edited Feb 29 '16 at 01:41

17

You can use the command find to do this. We can match files with -type f, and match empty files using -size 0. Then we can delete the matches with -delete.

find . -type f -size 0 -delete

edited Feb 29 '16 at 01:41

David Maust

8,080
3
32
36

answered Feb 28 '16 at 13:03

PYK

3,674
29
17

1

find . -maxdepth 1 -type f -size 0 -delete This finds the empthy files in the current directory without going into sub-directories. – user7194913 Jan 12 '18 at 08:56

score 5 · Answer 5 · answered Feb 18 '16 at 19:31

5

This works for plain BSD so it should be universally compatible with all flavors. Below.e.g in pwd ( . )

find . -size 0 |  xargs rm

answered Feb 18 '16 at 19:31

user1874594

2,277
1
25
49

Doesn't BSD support `-delete` option? https://www.freebsd.org/cgi/man.cgi?find(1) – Antonio Mar 17 '17 at 13:06
@Antonio OpenBSD 6.0 at least doesn't. – Colin 't Hart Feb 04 '19 at 13:37

score 5 · Answer 6 · answered May 13 '11 at 09:35

On Linux, the stat(1) command is useful when you don't need find(1):

(( $(stat -c %s "$filename") )) || rm "$filename"

The stat command here allows us just to get the file size, that's the -c %s (see the man pages for other formats). I am running the stat program and capturing its output, that's the $( ). This output is seen numerically, that's the outer (( )). If zero is given for the size, that is FALSE, so the second part of the OR is executed. Non-zero (non-empty file) will be TRUE, so the rm will not be executed.

score 1 · Answer 7 · answered Sep 02 '14 at 18:27

1

For a non-recursive delete (using du and awk):

rm `du * | awk '$1 == "0" {print $2}'`

answered Sep 02 '14 at 18:27

Harrison

830
1
15
28

-bash: /usr/bin/du: Argument list too long – Anthony Jul 11 '17 at 06:30

score 1 · Answer 8 · edited Jun 28 '15 at 14:25

1

find . -type f -empty -exec rm -f {} \;

edited Jun 28 '15 at 14:25

josliber

43,891
12
98
133

answered Jun 28 '15 at 09:49

Diệu Thu Lê

25
1

1

you can just use the -delete flag - saves an exec. – Yehosef Jul 07 '15 at 09:07

Linux delete file with size 0

8 Answers8

Linked

Related