filtering output of who with grep and cut - linux

Question

I have this exercice :
Create a bash script that check if the user passed as a parameter is
connected and if he is display when he connected. Indications : use the command who, the grep filter and the
command cut.
But i have some trouble to solve it.
#!/bin/bash
who>who.txt;
then
grep $1 who.txt
for a in who.txt
do
echo "$a"
done
else
echo "$1 isnt connected"
fi
So first of all i want to only keep the line where i can find the user in a .txt and then i want to cut each part with a loop in the who command to keep only the date but the problem is that i don't know how to cut here because it's seperated with multiple spaces.
So i'am really blocked and i don't see where to go to do this. I'am a beginner with bash.

Answer 1

If I understand you simply want to check to see if a user is logged in, then that is what the users command is for. If you want to wrap it in a short script, then you could do something like the following:
#!/bin/bash
[ -z "$1" ] && { ## validate 1 argument given on command line
printf "error: insufficient input, usage: %s username.\n" "${0##*/}" >&2
exit 1
}
## check if that argument is among the logged in users
if $(users | grep -q "$1" >/dev/null 2>&1) ; then
printf " user: %s is logged in.\n" "$1"
else
printf " user: %s is NOT logged in.\n" "$1"
fi
Example/Use
$ bash chkuser.sh dog
user: dog is NOT logged in.
$ bash chkuser.sh david
user: david is logged in.

Answer 2

cut is a rather awkward tool for parsing who's output, unless you use fixed column positions. In delimiter mode, with -d ' ', each space makes a separate empty field. It's not like awk where fields are separated by a run of spaces.
who(1) output looks like this (and GNU who has no option to cut it down to just the username/time):
$ who
peter tty1 2015-11-13 18:53
john pts/13 2015-11-12 08:44 (10.0.0.1)
john pts/14 2015-11-12 08:44 (10.0.0.1)
john pts/15 2015-11-12 08:44 (10.0.0.1)
john pts/16 2015-11-12 08:44 (10.0.0.1)
peter pts/9 2015-11-14 16:09 (:0)
I didn't check what happens with very long usernames, whether they're truncated or whether they shift the rest of the line over. Parsing it with awk '{print $3, $4} would feel much safer, since it would still work regardless of exact column position.
But since you need to use cut, let's assume that those exact column positions (time starting from 23 and running until 38) are constant across all systems where we want this script to work, and all terminal widths. (who doesn't appear to vary its output for $COLUMNS or the tty-driver column width (the columns field in stty -a output)).
Putting all that together:
#!/bin/sh
who | grep "^$1 " | cut -c 23-38
The regex on the grep command line will only match at the beginning of the line, and has to match a space following the username (to avoid substring matches). Then those lines that match are filtered through cut, to extract only the columns containing the timestamp.
With an empty cmdline arg, will print the login time for every logged-in user. If the pattern doesn't match anything, the output will be empty. To explicitly detect this and print something else, capture the pipeline output with var=$(pipeline), and test if it's the empty-string or not.
This will print a time for every separate login from the same user. You could use grep's count limit arg (see the man page) to stop after one match, but it might not be the most recent time. You might use sort -n | head -1 or something.
If you don't have to write a loop in the shell, don't. It's much better to write a pipeline that makes one pass over the data. The shell itself is slow, but as long as it doesn't have to parse every line of what you're dealing with, that doesn't matter.
Also note how I quoted the expansion of $1 with double quotes, to avoid the shell applying word splitting and glob expansion to it.
For more shell stuff, see the Wooledge Bash FAQ and guide. That's a good place to get started learning idioms that don't suck (i.e. don't break when you have filenames and directories with spaces in them, or filenames containing a ?, or lines with trailing spaces that you want to not munge...).