[SOLVED] having only one instance of my search results

henrtm05 · 10-06-2010, 12:22 AM

This script has been solved thanks.

Dark_Helmet · 10-06-2010, 02:00 AM

I looked up sort's options on my man page, and to be honest, I don't know what is meant by "output only the first of an equal run." Or more specifically, what an "equal run" is.

Regardless, the way I've always done what you're trying to do is to pipe into uniq. Such as:

Code:

finaldata=`echo "$whole" | sort | uniq`

EDIT: Well, after thinking about it, I guess "equal run" is a strange way of saying "repeated sequence." Anyways...

henrtm05 · 10-06-2010, 02:21 AM

Quote:

Originally Posted by Dark_Helmet

I looked up sort's options on my man page, and to be honest, I don't know what is meant by "output only the first of an equal run." Or more specifically, what an "equal run" is.

Regardless, the way I've always done what you're trying to do is to pipe into uniq. Such as:

Code:

finaldata=`echo "$whole" | sort | uniq`

EDIT: Well, after thinking about it, I guess "equal run" is a strange way of saying "repeated sequence." Anyways...

I have actually tried that the result is as follows this is my whole data

one
duplicate
two
two
three
three
two
three

and it gives me the following output with the finaldata=`echo "$whole" | sort | uniq` command.

duplicate
one
three
two
three
two

my command the finaldata=`echo "$whole" | sort -u` does the same thing as finaldata=`echo "$whole" | sort | uniq` I have tried it both ways with the same result.

Dark_Helmet · 10-06-2010, 03:05 AM

I'm about to fall asleep, but I don't think there's a problem with what you're trying to do with sort. There is very likely a problem with the script's logic that's causing weird output. Here's why I say that.

I made a simple script, using the test data you gave:

Code:

#!/bin/bash

echo -e "one\nduplicate\ntwo\ntwo\nthree\nthree\ntwo\nthree" | sort -u

the output:

Code:

duplicate
one
three
two

So, sort works as you expect. So the problem must be in the script logic. I can't debug it completely, but I do see a potential problem: you're doing no error-checking at all.

To emphasize the point, what happens if one of the groups in the comma-separated list given by the user doesn't exist, or there's a typo? That can wreak havoc on your assignment to firstgrp. If no match is found, then you'll grep for an empty string between to colons (e.g. "::") on /etc/passwd. I'd be willing to bet that will turn up some matches for non-human accounts, but the script will just keep on processing.

sag47 · 10-06-2010, 09:36 AM

I notice your $whole contains users (letters) from /etc/passwd and group IDs (numbers) from /etc/group. Can you tell me why you're doing that?

If I were you I would commend out the following lines

Code:

groupadd $newgroup
usermod -G $newgroup "$finaldata"

and echo the variables so you know how the script is behaving. Also your system won't go haywire from a buggy script. I hope you're not running this as root with these serious bugs. I also think that usermod command needs to be in a for or while loop because $finaldata contains newline chars. I'm not sure usermod can handle that (I may be wrong).

Like Dark Helmet says you need error checking. Before running

Code:

whole=`echo -e "$firstgrp\n$remaingrpsmod"`

You should check that there aren't any zero length strings.

Code:

if [ ! -z $firstgrp ];then
  whole=`echo -e "$firstgrp\n$remaingrpsmod"`
fi;

This is all quick info off the top of my head so be sure to research my statements. Also please make use of the forum formatting such as the code tags (which can be viewed when you go into the advanced post editor by clicking 'Go Advanced'). Gotta run!

henrtm05 · 10-06-2010, 11:46 PM

I am taking users from multiple groups and adding them into one and not duplicating the process if a user is in two of those groups.