Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Does any one have a template shell script for doing something with ls for a list of directory names and looping through each one and doing something?

I'm planning to do ls -1d */ to get the list of directory names.

share|improve this question

6 Answers 6

up vote 17 down vote accepted

Edited not to use ls where a glob would do, as @shawn-j-goff and others suggested.

Just use a for..do..done loop:

for f in *; do
  echo "File -> $f"

You can replace the * with *.txt or any other glob that returns a list (of files, directories, or anything for that matter), a command that generates a list, e.g., $(cat filelist.txt), or actually replace it with a list.

Within the do loop, you just refer to the loop variable with the dollar sign prefix (so $f in the above example). You can echo it or do anything else to it you want.

For example, to rename all the .xml files in the current directory to .txt:

for x in *.xml; do 
  t=$(echo $x | sed 's/\.xml$/.txt/'); 
  mv $x $t && echo "moved $x -> $t"

Or even better, if you are using Bash you can use Bash parameter expansions rather than spawning a subshell:

for x in *.xml; do 
  mv $x $t && echo "moved $x -> $t"
share|improve this answer
What if the filename has a space in it? –  Daniel A. White Aug 28 '09 at 17:17
I'm iterating all folders. –  Daniel A. White Aug 28 '09 at 17:18
Unfortunately, as Daniel said, the code in this answer will break if any of the files or folders contains a space or newline in their name. It shows a very common misuse of for loops, and a typical pitfall of trying to parse the output of ls. @DanielA.White, you might consider unaccepting this answer if it wasn't helpful (or is potentially misleading), since like you said, you're acting on directories. Shawn J. Goff's answer should provide a more robust and working solution to your issue. –  slhck Jan 23 '13 at 17:48
-∞ You shouldn't parse ls output, you shouldn't read output in a for loop, you should use $() instead of `` and Use More Quotes™. I'm sure @CoverosGene meant well, but this is just terrible. –  l0b0 Apr 17 at 19:31
A better alternative if you really want to use ls is ls -1 | while read line; do stuff; done. At least that one won't break for whitespaces. –  Emmanuel Joubaud Aug 1 at 8:47

Using the output of ls to get filenames is a bad idea. It can lead to malfunctioning and even dangerous scripts. This is because a filename can contain any character except / and the nullcharacter, and ls does not use either of those characters as delimiters, so if a filename has a space or a newline, you will get unexpected results.

There are two very good ways of iterating over files. Here, I've used simply echo to demonstrate doing something with the filename; you can use anything, though.

The first is to use the shell's native globbing features.

for dir in */; do
  echo "$dir"

The shell expands */ into separate arguments that the for loop reads; even if there is a space, newline, or any other strange character in the filename, for will see each complete name as an atomic unit; it's not parsing the list in any way.

If you want to go recursively into subdirectories, then this won't do unless your shell has some extended globbing features (such as bash's globstar. If your shell doesn't have these features, or if you want to ensure that your script will work on a variety of systems, then the next option is to use find.

find . -type d -exec echo '{}' \;

Here, the find command will call echo and pass it an argument of the filename. It does this once for each file it finds. As with the previous example, there is no parsing of a list of filenames; instead, a fileneame is sent completely as an argument.

The syntax of the -exec argument looks a little funny. find takes the first argument after -exec and treats that as the program to run, and every subsequent argument, it takes as an argument to pass to that program. There are two special arguments that -exec needs to see. The first one is {}; this argument gets replaced with a filename that the previous parts of find generates. The second one is ;, which lets find know this is the end of the list of arguments to pass to the program; find needs this because you can continue with more arguments that are intended for find and not intended for the exec'd program. The reason for the \ is that the shell also treats ; specially - it represents the end of a command, so we need to escape it so that the shell gives it as an argument to find rather than consuming it for itself; another way of getting the shell to not treat it specially is to put it in in quotes: ';' works just as well as \; for this purpose.

share|improve this answer
+1 this is definitely the way to go when you need to generate a file list and use it in a command. find -exec is limited by only being able to run a single commands. With the loop, you can pipe to your heart's content. –  MaQleod May 10 '12 at 22:45
+1. I wish more people would use find. There is magic that can be done, and even the perceived limitations of -exec can be worked around. The -print0 option is also valuable for use with xargs. –  ghoti Oct 17 at 22:24

Why not set IFS to a newline, then capture the output of ls in an array? Setting IFS to newline should resolve issues with funny characters in file names; using ls can be nice because it has a built-in sort facility.

(In testing I was having trouble setting IFS to \n but setting it to newline backspace works, as suggested elsewhere here):

E.g. (assuming desired ls search pattern passed in $1):

IFS=$(echo -en "\n\b")

FILES=($(/bin/ls "$1"))

for AFILE in ${FILES[@]}
    ... do something with a file ...


This is especially handy on OS X, e.g., to capture a list of files sorted by creation date (oldest to newest), the ls command is ls -t -U -r.

share|improve this answer
Filenames can also contain newlines, and they often do when users are permitted to name their own files. Breaking on \n is insufficient. The only valid solutions use a for loop with pathname expansion, or find. –  ghoti Oct 17 at 22:21

If you have GNU Parallel http://www.gnu.org/software/parallel/ installed you can do this:

ls | parallel echo {} is in this dir

To rename all .txt to .xml:

ls *.txt | parallel mv {} {.}.xml

Watch the intro video for GNU Parallel to learn more: http://www.youtube.com/watch?v=OpaiGYxkSuQ

share|improve this answer

This is how I do it, but there are probably more efficient ways.

ls > filelist.txt

while read filename; do echo filename: "$filename"; done < filelist.txt
share|improve this answer
Stick to pipes in place of the file: >ls | while read i; do echo filename: $i; done –  Nerdling Aug 28 '09 at 18:46
Cool. I should say that you can also use $EDITOR filelist.txt in between the two commands. Lots of stuff you can do in an editor that is easier than on the command line. Not relevant to this question, though. –  TREE Aug 31 '09 at 18:45

For files with spaces in you will have to make sure to quote the variable like:

 for i in $(ls); do echo "$i"; done;

or, you can change the input field separator (IFS) environment variable:

 IFS=$'\n';for file in $(ls); do echo $i; done

Finally, depending on what you're doing, you may not even need the ls:

 for i in *; do echo "$i"; done;
share|improve this answer
nice use of a subshell in the first example –  Nerdling Aug 28 '09 at 18:46
Why does IFS need a $ after the assignment and before the new character? –  Andy Ibanez Nov 29 '12 at 18:57
Filenames can also contain newlines. Breaking on \n is insufficient. It's never a good idea to recommend parsing the output of ls. –  ghoti Oct 17 at 22:18

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.