Friday, November 21, 2008

Another Simple Scriptlet To Make The Unix And Linux CLI More User Friendly

Hey there,

Today we're going to shoot out another quick scriptlet that might be useful (or distracting ;) from time to time. It’s a bit of a follow up on our post from earlier this week on using bash to produce fancy user names for folks logged into your machine, although it’s a little bit longer. Once again, if you go by the 65 character rule, this isn’t a one liner, even though it’s all going on one line ;)

Our little foray into variable interpolation today has to do with making /etc/passwd just a bit more accessible. Most folks are familiar with the 7 colon-separated fields of this file (if not, they’re: The user’s Username, an encrypted password (or usually an “x” if you’re using a shadow password system), the user’s UID, the user’s GID, the user’s gecos information (Also referred to as the “comment field,” usually either empty or populated with a short description to make the account more easily recognizable) , the user’s home directory and, finally, the user’s default shell)

Now, say what you will about the structure and setup of the passwd file, but it’s fairly easy to follow. Almost everything in it is the opposite of cryptic. The password field (field number 2) is, of course, not going to be tailored for your ease of understanding ;) If you get really lucky and you hop on a machine that actually lists the encrypted password in field two of /etc/passwd, you can copy the file to any remote machine and go to work on it. We've posted quite a few scripts to do automated password cracking. You should just see an "x" for a standard user account. If you don’t, have a talk with your SysAdmin or, if you’re him, please move on to a shadowed password system. Too much information exists in the public domain to make the old-style setup anything more than a trifling hassle to someone who wants to exploit your system. Especially a disgruntled employee!

Other than that, the user ID (field number 3) is an easy relation to make, since the username is in the same file (field number 1). The only thing in there that doesn’t practically explain itself is field number 4; the GID (group id) field. It’s not that big of a deal to look a user up in /etc/passwd, glean that group id information and find out what group the user is in by looking at /etc/group. It’s as simple as this:

host # grep user1 /etc/passwd
user1:x:37527:501::/export/home/user1:/bin/bash
host # grep 501 /etc/group
people::501:


and, just like that, we know that the user "user1"'s primary group membership is in the group "people." There are also secondary groups to consider, but finding this out is trivial as well. Just do the following:

host # grep -w user1 /etc/group
guys::600:user1
folks::601:user1


And, yes, I can hear you out there ;) I'm aware of commands like "groups" and "id" ;) They're very good at getting this information, as well (which is why we're going to use "groups" in our one-long-liner. Your output may look slightly different). Note that some versions don't have the -a option for "id" and some have either -g, -G, any combination of the three or all of them - it depends on your distro:

host # groups <-- This is supposed to list groups in order, from left to right, with the leftmost group being the "primary" and every group after that being a "secondary." The ordering of the "secondary" groups in the list is generally "as they appear" in /etc/group. So, since the line with "guys" is on a line closer to the beginning of the /etc/group file than the line with "folks," they're printed in that order. No other relationship exists between them that I'm aware of.
people guys folks
host # id -a
uid=37527(user1) gid=501(people) groups=501(people)
<-- On Solaris; not quite as helpful as "id" on other flavours of Linux or Unix.

Anyway, I think I've met my word-count quota ;) Just kidding. I have no idea how much I've typed. I'm not even sure I want to know...

Here's a nice little command line (and, consequently, the topic of this post ;) to print every user's group(s) in the (modified) output of /etc/passwd. Just type it out, like so:

host # IFSBAK=$IFS;export IFS=":";while read one two three four five;do groupnames=`groups $one|sed 's/ /,/g'`;echo "${one}:${two}:${three}:${groupnames}:${five}";done </etc/passwd;export IFS=$IFSBAK


...lots of output left out for reasons of brevity and security :) ...
noaccess:x:60002:noaccess:No Access User:/
user1:x:37527:people,guys,folks::/export/home/user1:/bin/bash
...


And, there we have it. All nice and tidy :) Here's the same thing in script format (below) to make it a little easier to read. Be sure (now matter how you run it) that you don't clip the beginning and end, as you'll probably want to reset your shell's IFS (field separator) to what it was before we changed it to the colon (:) to make parsing /etc/password possible without using awk or cut, etc... Also, if it's of any interest and you noted that I only counted my variables up to "five" it's because "while read" will swallow up the rest of a line in its last variable. So, instead of reading and writing five, six and seven, I just read five, since that would include from field five all the way to the end of the line (at the end of field 7), and field 4 (the group field) was the last field on every line that we were actually interested in :)

#!/bin/bash

IFSBAK=$IFS
export IFS=":"

while read one two three four five
do
groupnames=`groups $one|sed 's/ /,/g'`
echo "${one}:${two}:${three}:${groupnames}:${five}"
done </etc/passwd

export IFS=$IFSBAK


Have a great weekend!

, Mike




Please note that this blog accepts comments via email only. See our Mission And Policy Statement for further details.