Summary: awk question

From: Ju-Lien Lim (
Date: Tue Mar 03 1998 - 07:57:50 CST

My thanks to everyone (too many to list) who
responded to my "silly" awk syntax question. Below
is the summary:

Original Question:
I'm not too familiar with awk, and I'm just trying to
get the home directory of a given user (to put it in
a shell script)
using awk (or sed... it doesn't matter which).
Anyhow, I'm trying to use split or FS to tell it to
user ":" as the
delimiter, but neither seems to work for me:

#grep oracle /etc/passwd


homedir=`grep $user /etc/passwd | awk '
      BEGIN {
      split($0, field, ":")
      print field[5]

echo "$user (home directory) = $homedir"

# I also tried the following:
# grep $user /etc/passwd | awk '{ FS = ":"; print


I made 3 mistakes:

1. The syntax for awk is something like
    reg-expr { commands}

BEGIN is a special regular expression that excutes
it's associated commands BEFORE any input lines are

2. FS needs to be set on the command line as an
arguemnt to awk. (or as one of the commands executed
by the BEGIN section.)

3. It's field 6 not 5... (I'm thinking in C, i.e.
things start from subscript 0! Oh well...)

The following are some of the solutions people

OPTION 1: Using getent, you will get the entry line
for the user
no matter if it is a local user, or defined in nis or
nis+ tables.

getent passwd $user | cut -f 6 -d ":"

OPTION 2: If you preffer using grep, you still can do
but you restrict your search to local users:

grep $user /etc/passwd | cut -f6 -d ":"

OPTION 3: using awk and getent command

getent passwd $user | awk -F: '{print $6}'

OPTION 4: Using awk and checking directly in the file

grep $user /etc/passwd | awk -F: '{print $6}'


awk -F: ' /^oracle/ {print $6}' </etc/passwd


grep ^$username: /etc/passwd |awk -F: '{print $5}'

You want the ^ at the front of $username so that it
will match only at the START of a line, and the : at
the end so it
matches the ":" separator. This way it will match
bin: rather than foobin or foobin1 for the username...


nawk -F\: '{print $6}' /etc/passwd


awk -F: '$1 == user {print $6}' user=$user <

The -F: tells awk to use ":" as delimiter.
The "user=$user" as last argument imports the shell
variable $user as
awk variable user.
<You could also program in ksh, and just use ~user.>

homedir=`grep $user /etc/passwd | awk '
      BEGIN { FS = ":" }
      { print $5 }'`
echo "$user (home directory) = $homedir"


 grep "^oracle:" /etc/passwd | awk 'BEGIN { FS=":" }
{ print $6 }'

 grep $user /etc/passwd | awk '{FS=":"; print $6}'


you must use BEGIN when you try to declare that your
field separator is
so this is the complete script that you want

set user=oracle
set homedir=`grep $user /etc/passwd|nawk 'BEGIN
{FS=":"} {print $6}'`


both csh and ksh understand ~<uid> (which they expand
to home of <uid>). One way that this will work
transparently with nis is:

homedir=`/bin/csh -fc "echo ~$user"`


# grep $user /etc/passwd | nnawk ' BEGIN { FS=":" }
{ print $5 } '

this will print the fifth field of the password file.
where fields (FS) are separated by a colon. you need
to specify the value of FS in the begin section,
otherwise it gets overwritten in each line.

you can also use the -v option to just have one
command instead of two (grep and awk)

awk -vUSR=$user ' BEGIN { FS=":" } { if ($1==USR)
print $5 } '

this is fairly well documented in the man page for


Get your free address at

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:32 CDT