How do I get from this:


to this:


using sed?

  • 2
    A helpful tip: when you are not sure what a command does, type in man the_command_name and you will get a helpful instruction manual for the command. You can also go to www.google.com and type in "command_name tutorial" and you will find plenty of step-by-side guides. – DBedrenko Sep 5 '14 at 7:05
  • 2
    For tutorial on sed visit : grymoire.com/Unix/Sed.html – Pandya Sep 5 '14 at 10:29

10 Answers 10


I'd use a simple grep to look for user90:

$ echo "randomcollege-nt\user90" | grep -o user90

If user90 is not constant, prefer this command:

$ echo "randomcollege-nt\user90" | grep -oP '(?<=randomcollege-nt\\)\w+'

Finally using sed to edit the file in place:

$ sed -ri 's/randomcollege-nt\\(user[0-9]+)/\1/' my_file

Or to match all possible user accounts:

$ sed -ri 's/randomcollege-nt\\(\w+)/\1/' my_file
  • Funny thing is I tried that and it nothing returned. I'm using 12.04 if that matters. It only shows up if I grep -F the whole string. – Roboman1723 Sep 4 '14 at 15:12
  • @Roboman1723 tested on 12.04, my first commands works. Be sure that you don't have a grep alias defined somewhere. BTW as it seems you're looking for inline replacement with sed I've added such command. – Sylvain Pineau Sep 4 '14 at 15:41
  • Is there a way you could make that sed command cut out all "randomcollege-nt\" but display more than just user 90? Example: randomcollege-nt\user90 randomcollege-nt\user91 randomcollege-nt\user92 Output: user90 user91 user92 – Roboman1723 Sep 4 '14 at 15:46
  • 1
    @Roboman1723 You should attach your original file and your desired output to your question so that everyone willing to help you can see it. – Sylvain Pineau Sep 4 '14 at 15:49
  • 1
    @SylvainPineau i think the first command is not an appropriate one. – Avinash Raj Sep 5 '14 at 11:57

You're parsing some text to extract the username from a domain\username string, most likely from Windows. Most of the above answers are only addressing your specific example string.

The best way to do this is using regex in sed to extract whatever comes after \. Here's how you would do it:

sed 's|.*\\\(.*\)|\1|'

That will match everything (.*) until a backslash (here, we're escaping it, so it's \\), then match everything after the backslash (.*), but making it a capture group (i.e. wrap brackets around it, but we also have to escape them, so \(.*\)). Now that we have whatever comes after the \ in the string as a capture group, we print it by referencing it with \1.

You can use the above sed command with any domain name, not necessarily randomcollege-nt.

$ echo "randomcollege-nt\user90" | sed 's|.*\\\(.*\)|\1|'

$ echo "domain\username" | sed 's|.*\\\(.*\)|\1|'

$ echo "anydomainname\roboman1723" | sed 's|.*\\\(.*\)|\1|'

Another sed:

$ echo "randomcollege-nt\user90" | LC_ALL=C sed -e 's/.*\\//'

or POSIXly:

$ a='randomcollege-nt\user90'
$ printf '%s\n' "${a##*\\}"

I know you want to use sed, but I'd use something different...

echo "randomcollege-nt\user90" | cut -d'\' -f2
  • 1
    Knowing how to use a tool is just as important as knowing which tool to use. – Rafał Cieślak Sep 4 '14 at 16:10

Is this the question ?

$ echo randomcollege-nt\user90| sed -e s,randomcollege-nt\,,

if the sting randomcollege-nt is not contant use the awk commande above/below.

  • 1
    when I run that I get /user90 not user90 – Roboman1723 Sep 4 '14 at 15:08
  • 2
    No that's alright,I got user90. – saptarshi nag Sep 4 '14 at 15:41

Rather you use 'awk' to filter "user90":

echo "randomcollege-nt\user90" | awk -F\\ {'print $2'}
  • Is they're a way with awk to edit the file in place and not have to output it to a new file? – Roboman1723 Sep 4 '14 at 15:12
  • @Roboman1723 see stackoverflow.com/a/16531920/2072269 – muru Sep 4 '14 at 15:16
  • Yes you have to use -f option to awk.For details see man page. – saptarshi nag Sep 4 '14 at 15:23
  • Note that the argument to -F can be a regex, which allows for removing prefixes that match a pattern (and thus don't have to be exactly known strings). Also, when the input has multiple lines (commonly the case with a file), you can use 'NF > 1 {print $2}' to avoid awk printing empty lines. For example, awk -F "^Some Key *: " 'NF > 1 {print $2}' will get you everything after lines starting with "Some Key : " where the amount of spaces before the colon can vary. – Matthias Braun Apr 28 at 21:09

This simple grep command will do the job,

$ echo 'randomcollege-nt\user90' | grep -oP '[^\\]*$'

With sed delete everything in a string before a specific character (define into double bracket [Specific char]).

echo "randomcollege-nt\user90" | sed 's/.*[\]//'

Means replace all (.*[\]) characters before a \ char with whitespace character(//)

If you have a file and want to inplace replace use -i flag in sed command like this:

sed -i 's/.*[\]//' /path/to/FileName

The original question asked for sed, but I see that alternatives are popular here.

If you are using Bash, parameter expansion is by far the simplest:

echo "${ORIGIN#*\\}"

If you are potentially expecting more than one backslash, double the hash signs:

echo "${ORIGIN##*\\}"

For more information, man bash and search for Parameter Expansion.


In case anyone is trying to remove a commented out # server_tokens off; from nginx automatically to help with auto dev-ops:

sudo sed -ri 's/#\s(server_tokens off;)/\1/' /etc/nginx/nginx.conf.

Tested and working for nginx/1.12.1 on Ubuntu 16.04 LTS. Related answer to locking down NGINX by turning off server tokens here.

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.