I have a file that only uses \n for new lines, but I need it to have \r\n for each new line. How can I do this?

For example, I solved it in Vim using :%s/\n/\r\n/g, but I would like to use a script or command-line application. Any suggestions?

I tried looking this up with sed or grep, but I got immediately confused by the escape sequence workarounds (I am a bit green with these commands).

If interested, the application is related to my question/answer here


You can use unix2dos (which found on Debian):

unix2dos file

Note that this implementation won't insert a CR before every LF, only before those LFs that are not already preceded by one (and only one) CR and will skip binary files (those that contain byte values in the 0x0 -> 0x1f range other than LF, FF, TAB or CR).

or use sed:

CR=$(printf '\r')
sed "s/\$/$CR/" file

or use awk:

awk '{printf "%s\r\n", $0}' file

or use perl:

perl -pe 's|\n|\r\n|' file
  • Note that unix2dox won't insert a CR before every LF, only before those LFs that are not already preceded by one (and only one) CR. – Stéphane Chazelas Sep 1 '14 at 16:34
  • @StéphaneChazelas: Thanks for the point, added it to my answer. – cuonglm Sep 1 '14 at 16:47
  • 1
    The key piece of information for me here was using '\r' with sed does the trick. – rkersh Jun 16 '16 at 17:43

This is exactly what unix2dos does:

$ unix2dos file.txt

That will replace file.txt in-place with a version with CRLF line endings.

If you want to do it with sed, you can insert a carriage return at the end of every line:

sed -e 's/$/\r/' file.txt

This replaces (s) the zero-size area right before the end of the line ($) with \r. To do in-place replacement (like unix2dos does), use sed -i.bak, although that is a non-standard extension - if you don't have it, use a temporary file.

  • 3
    Use \r in replacement is not portable. At least, it won't work in Mac OSX. – cuonglm Sep 1 '14 at 6:43
  • Note that there are several implementations of unix2dos. You may want to give a link to the home page of a specific implementation (like waterlan.home.xs4all.nl/dos2unix.html as found on Debian) instead of linux.die.net/man which doesn't make it clear which implementation (or version) is documented. Note that there are some u2d, ux2dos, todos commands around as well. – Stéphane Chazelas Sep 1 '14 at 16:36

If you are doing the conversion on a Unix machine (OS X, Linux), open the file with vi or vim:

$ vim my-file

press the ESC key to ensure that you are not in insert mode, then type

:set ff=dos


:set fileformat=dos

This does conversion in place by setting the file format.

To write the file and quit the editor use


On the command line you can do

$ vi +':w ++ff=dos' +':q' my-file
  • thanks for explaining all the steps - for a newbie it makes it much easier to understand what is going on. – user3616725 Apr 19 '17 at 12:01

Doing this with POSIX is tricky:

  • POSIX Sed does not support \r or \15. Even if it did, the in place option -i is not POSIX

  • POSIX Awk does support \r and \15, however the -i inplace option is not POSIX

  • d2u and dos2unix are not POSIX utilities, but ex is

  • POSIX ex does not support \r, \15, \n or \12

To remove carriage returns:

awk 'BEGIN{RS="\1";ORS="";getline;gsub("\r","");print>ARGV[1]}' file

To add carriage returns:

awk 'BEGIN{RS="\1";ORS="";getline;gsub("\n","\r&");print>ARGV[1]}' file

A portable shell function that will do this:

u2dos() (set -f; IFS='
'; printf '%s\r\n' $(cat "$1"))

With that you can do:

u2dos file >dosfile

In awk you can try

awk '{print $0 "\r"}'


awk -v r=$'\r' '{print $0 r}'

The $'\r' is an example of ANSI-C style quoting as supported by a few shells like ksh93, bash, zsh, mksh and FreeBSD sh and likely to be included in issue 8 of the Single Unix Specification.

It offers a general way to express weird characters, try this, for example:

awk -v r=$'\U0001F608' '{print $0 r}'

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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