bash: how to split a string, assign tokens to an array

Hi brettmjohnson,
what will the entire script do maybe i can bring you a better solution.
/Rob

> ... awk 'BEGIN{FS=":";OFS=","; ...

Unfortunately, the GECOS field can contain commas (it is actually a comma-separated list in itself).
I have already used awk for splitting, but I was looking for a slightly more elegant solution.


> what will the entire script do maybe i can bring you a better solution.

I am writing a set of adminstration scripts for a client that has a Linux box
acting as a web server for 2 dozen virtual domains and a mail server (using Qmail)
for both local users (shell acct, POP3/SMTP) and "virtual users" (forwarding all
mail for a virtual domain to an external party).

I am using the username, UID, full name extracted from the GECOS field, and
home directory field for various administrative tasks.

> Just curious, what version of  "id" are you using ?

Excellent observation.  Although the target machine is running RH 7.1,
I was prototyping the utility on Mac OS X, which uses id v1.19 from FreeBSD.
I will probably have to change it to grep the /etc/passwd file for the appropriate fields.

I like the idea of using 'cut' to split the string, 'tho.

But I will going to give an A grade to a good solution that uses a built-in  
or external command that doesn't need to iterate (reparsing the string each time).

brettmjohnson,
Made a function for this purpose with the same syntax as explode in php

# Function
myexplode () {
array=`echo $2 | awk "BEGIN{FS=\"$1\";OFS=\"NISSEPISSE\";}{print \\$1,\\$2,\\$3,\\$4,\\$5,\\$6,\\$7}" | sed 's/ /KALLEKOSKIT/g' | sed 's/NISSEPISSE/ /g'`
echo $array
i=0
for ar in $array
do
   tmp[$i]=`echo "$ar" | sed 's/KALLEKOSKIT/ /g'`
   i=`expr $i + 1`
done
return $tmp
}
# End func

#test script
tmp= myexplode ":" "`tail -n 1 /etc/passwd`"
echo ${tmp[4]}
exit
# End

/Rob

This script does everything internally.  You might want to tweak the way it loops to more robustly handle short input strings.  I'm not sure what your restriction about iteration means.

#!/bin/bash

set -f
VAR="cyrus:*:77:6::0:0:Cyrus User:/var/imap:/usr/bin/false"

TMP=$VAR
for i in 0 1 2 3 4 5 6 7 8 9; do
    RES[$i]=${TMP%%:*}
    TMP=${TMP#*:}
done
echo res[3] is ${RES[3]}

echo "\$VAR is $VAR"
echo -n "\$RES is:"
for i in ${RES[@]} ; do
    echo -n " '$i'"
done
echo

My script mistakenly prints out 'Cyrus' and 'User' as two separate entries, but that is just a bug in my display loop.  ${RES[7]} correctly contains "Cyrus User".

To print out correctly, use this line for the second for loop:
for i in "${RES[@]}" ; do