57

I want to grep the output of my ls -l command:

-rw-r--r-- 1 root root 1866 Feb 14 07:47 rahmu.file -rw-r--r-- 1 rahmu user 95653 Feb 14 07:47 foo.file -rw-r--r-- 1 rahmu user 1073822 Feb 14 21:01 bar.file

I want to run grep rahmu on column $3 only, so the output of my grep command should look like this:

-rw-r--r-- 1 rahmu user 95653 Feb 14 07:47 foo.file -rw-r--r-- 1 rahmu user 1073822 Feb 14 21:01 bar.file

What's the simplest way to do it? The answer must be portable across many Unices, preferably focusing on Linux and Solaris.

NB: I'm not looking for a way to find all the files belonging to a given user. This example was only given to make my question clearer.

Improve this question
2
  • 6
    Note that parsing the output of ls is inherently fragile (think what happens if a user name contains whitespace — and yes, this happens on some platforms). Use find instead. Commented Feb 16, 2012 at 1:47
  • This doesn't answer the title question, but I'll mention that the --printf option to stat may come in handy in places where you might otherwise consider parsing ls. But usually, find is what you want (as @Gilles mentioned). Commented Jan 17, 2018 at 9:52

5 Answers 5

Reset to default
71

One more time awk saves the day!

Here's a straightforward way to do it, with a relatively simple syntax:

ls -l | awk '{if ($3 == "rahmu") print $0;}'

or even simpler: (Thanks to Peter.O in the comments)

ls -l | awk '$3 == "rahmu"'
Improve this answer
4
  • 10
    Or: ls -l | awk '$3=="rahmu"' ... and by the way, the opening ( bracket should be after if ... awk '{if($3 ...' Commented Feb 15, 2012 at 15:11
  • 3
    @Peter.O +1 from me. Your command is more succinct, correct, since print is the default for an awk match. Commented Feb 15, 2012 at 16:06
  • To avoid localization issues (date format affects column order) it is often suggested to ensure standard format and language: LANG=C ls -l | awk ... Commented Feb 16, 2012 at 10:26
  • Not what OP asked. Commented Jun 17 at 20:28
29

If you are looking to match only part of the string on a given column, you can use advice from https://stackoverflow.com/questions/17001849/awk-partly-string-match-if-column-partly-matches

some_command | awk '$6 ~ /string/'
Improve this answer
2
  • 2
    This should be the top answer Commented Nov 7, 2018 at 8:49
  • work great. thanks. Commented Sep 23, 2022 at 18:34
12

If by column, you mean fixed-size column, you could:

ls -l | grep "^.\{15\}rahmu"

where ^ means the beginning of the line, . means any character and \{15\} means exactly 15 occurrences of the previous character (any character in this case).

Improve this answer
1
  • One can also do ls -l | grep -E '^.{15}rahmu' because -E is defined by POSIX anyway and then you don't have to escape the command group character. Commented May 14, 2020 at 13:03
6

I know I'm late to the party, but I believe a very general solution to this problem is to use awk with regular expressions like so:

awk '$3~/rahmu/{print}' input_file

The part enclosed by '//' can be any kind of regular expression.

Improve this answer
1

To grep based on single username column (and print only files), you can try:

ls -la | grep "^-\S\+\s\+\S\+\s\+rahmu"

For directories, change - into d, for any type remove -.

It basically selects each column by \S\+\s\+ which matches a non-space characters followed by a space characters, so we're matching three first columns and then the username.

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.