Find files in one directory not in another

Question

Are there any Linux/Unix tools which find all the files in one directory not present in another? Basically I'm looking for diff which works on the output of ls.

Short and sweet scripts are also appreciated.

score 23 · Answer 1 · answered Sep 01 '10 at 15:50

23

diff does this already:

diff dir1 dir2

Example output:

Only in dir1: some_file.txt
Only in dir1: some_other_file.txt
Only in dir2: third_file.txt

answered Sep 01 '10 at 15:50

tobym

6,271
3
17
9

5

This is good. One gripe: diff is actually running on each of the files that are in both. Is there an obscure option to just run against file names (I may have missed it)? Otherwise, I suggest `diff dir1 dir2 | grep "Only"` – Willi Ballenthin Sep 01 '10 at 16:12
wow this just save me many minutes of bash scripting thanks – user5359531 Feb 18 '20 at 12:51

Dennis Williamson · Answer 2 · 2010-09-02T00:56:29.527

14

Bash:

diff <(cd dir1; ls) <(cd dir2; ls)

Compare only the filenames - not the contents of the files.

edited Sep 02 '10 at 00:56

answered Sep 01 '10 at 16:32

Dennis Williamson

62,149
16
116
151

Also ksh and zsh. – Dennis Williamson Sep 01 '10 at 18:37
use rsync for sync dirs – c4f4t0r Sep 04 '17 at 13:02
1

Do not parse `ls`, use `find` instead! – val - disappointed in SE Feb 11 '20 at 20:39
@valisstillwithMonica: This isn't an example of parsing `ls` since it's only getting a list of files and not looking at fields in output such as `ls -l`. It's also not parsing the names themselves so special characters aren't a problem except for newlines. But diff is going to treat parts of file names surrounding newlines separately anyway (even [this answer](https://serverfault.com/a/177012/1293) with `grep Only` fails in this regard). – Dennis Williamson Jun 28 '22 at 21:56
can this be made to be recursive? – majorgear May 23 '23 at 14:57
@majorgear: These answers offer ways to do that: [with `--brief`](https://serverfault.com/a/839992/1293) or [with `find`](https://serverfault.com/a/871967/1293) (use `find .` if on a Mac). – Dennis Williamson May 23 '23 at 16:11

score 4 · Answer 3 · answered Sep 01 '10 at 16:36

Like people told you here, you can use DIFF in various usage variations. Or you just use dirdiff instead, which is meant for what you're trying! :-)

But if you want to keep some directories in sync then you really should take a look on rsync.

Regards

score 4 · Answer 4 · answered Mar 22 '17 at 19:29

4

If you are wanting to do this through all sub directories as well, a good way to do it is:

diff --brief -r dir1/ dir2/

I prefer using brief, but you can leave that out if you want.

answered Mar 22 '17 at 19:29

trueCamelType

1,086
5
20
42

This is exactly what I was looking for but found WAY more files than I anticipated. Is there any way to copy the files found in `dir2/` to a new folder? `dir3/` – trex005 Aug 20 '22 at 19:33

score 1 · Answer 5 · answered Sep 01 '10 at 15:42

1

for i in $(ls -1 directory1); do if (test -f directory2/$i) then echo $i; fi; done

answered Sep 01 '10 at 15:42

James L

6,025
1
22
26

score 1 · Answer 6 · answered Sep 04 '17 at 11:58

Dennis Williamson had a good answer, but I needed to do this recursively. GNU findutils 4.7.0 doesn't sort its output, so here's what I used

diff <(cd $dir1; find | sort) <(cd $dir2; find | sort)

To do this only one way, and produce a list of files, I used this:

diff <(cd $dir1; find | sort) <(cd $dir2; find | sort) \
| grep '< ./' | sed "s,< ./,$dir1/,"

For this to work properly, neither $dir1 nor $dir2 should include the trailing slash.

score 0 · Answer 7 · answered Sep 01 '10 at 15:39

0

This works..

ls -a1 /dir1 | sort > /tmp/1
ls -a1 /dir2 | sort > /tmp/2
diff /tmp/1 /tmp/2

answered Sep 01 '10 at 15:39

Warner

23,756
2
59
69

score 0 · Answer 8 · answered Sep 01 '10 at 15:41

0

untested:

find /dir/A -printf "%P" | while read f; do
  if [ ! -e "/dir/B/$f" ]; then
    echo $f
  fi
done

answered Sep 01 '10 at 15:41

Javier

9,268
2
24
24

Find files in one directory not in another

8 Answers8