Introduction to UNIX:
Lecture Eight

8.1 Objectives
This chapter covers:
8.2 Shells and Shell Scripts
A shell is a program which reads and executes commands for the user. Shells also usually provide features such job control, input and output redirection and a command language for writing shell scripts. A shell script is simply an ordinary text file containing a series of commands in a shell command language (just like a "batch file" under MS-DOS).

There are many different shells available on UNIX systems (e.g. sh, bash, csh, ksh, tcsh etc.), and they each support a different command language. Here we will discuss the command language for the Bourne shell sh since it is available on almost all UNIX systems (and is also supported under bash and ksh).

8.3 Shell Variables and the Environment
A shell lets you define variables (like most programming languages). A variable is a piece of data that is given a name. Once you have assigned a value to a variable, you access its value by prepending a $ to the name:

    $ bob='hello world'
    $ echo $bob
    hello world

Variables created within a shell are local to that shell, so only that shell can access them. The set command will show you a list of all variables currently defined in a shell. If you wish a variable to be accessible to commands outside the shell, you can export it into the environment:

    $ export bob

(under csh you used setenv). The environment is the set of variables that are made available to commands (including shells) when they are executed. UNIX commands and programs can read the values of environment variables, and adjust their behaviour accordingly. For example, the environment variable PAGER  is used by the man command (and others) to see what command should be used to display multiple pages. If you say:

    $ export PAGER=cat

and then try the man command (say man pwd), the page will go flying past without stopping. If you now say:

    $ export PAGER=more

normal service should be resumed (since now more will be used to display the pages one at a time). Another environment variable that is commonly used is the EDITOR variable which specifies the default editor to use (so you can set this to vi or emacs or which ever other editor you prefer). To find out which environment variables are used by a particular command, consult the man pages for that command.

Another interesting environment variable is PS1, the main shell prompt string which you can use to create your own custom prompt. For example:

    $ export PS1="(\h) \w> "
    (lumberjack) ~>

The shell often incorporates efficient mechanisms for specifying common parts of the shell prompt (e.g. in bash you can use \h for the current host, \w for the current working directory, \d for the date, \t for the time, \u for the current user and so on - see the bash man page).

Another important environment variable is PATH. PATH is a list of directories that the shell uses to locate executable files for commands. So if the PATH is set to:


and you typed ls, the shell would look for /bin/ls, /usr/bin/ls etc. Note that the PATH contains'.', i.e. the current working directory. This allows you to create a shell script or program and run it as a command from your current directory without having to explicitly say "./filename".

Note that PATH has nothing to with filenames that are specified as arguments to commands (e.g. cat myfile.txt would only look for ./myfile.txt, not for /bin/myfile.txt, /usr/bin/myfile.txt etc.)

8.4 Simple Shell Scripting
Consider the following simple shell script, which has been created (using an editor) in a text file called simple:

# this is a comment
echo "The number of arguments is $#"
echo "The arguments are $*"
echo "The first is $1"
echo "My process number is $$"
echo "Enter a number from the keyboard: "
read number
echo "The number you entered was $number"

The shell script begins with the line "#!/bin/sh" . Usually "#" denotes the start of a comment, but #! is a special combination that tells UNIX to use the Bourne shell (sh) to interpret this script. The #! must be the first two characters of the script. The arguments passed to the script can be accessed through $1, $2, $3 etc. $* stands for all the arguments, and $# for the number of arguments. The process number of the shell executing the script is given by $$. the read number statement assigns keyboard input to the variable number.

To execute this script, we first have to make the file simple executable:

    $ ls -l simple
    -rw-r--r--    1 will  finance  175  Dec 13  simple
    $ chmod +x simple
    $ ls -l simple
    -rwxr-xr-x    1 will  finance  175  Dec 13  simple
    $ ./simple hello world
    The number of arguments is 2
    The arguments are hello world
    The first is hello
    My process number is 2669
    Enter a number from the keyboard:
    The number you entered was 5

We can use input and output redirection in the normal way with scripts, so:

    $ echo 5 | simple hello world

would produce similar output but would not pause to read a number from the keyboard.

8.5 More Advanced Shell Scripting
8.6 Start-up Shell Scripts
When you first login to a shell, your shell runs a systemwide start-up script (usually called /etc/profile under sh, bash and ksh and /etc/.login under csh). It then looks in your home directory and runs your personal start-up script (.profile under sh, bash and ksh and .cshrc under csh and tcsh). Your personal start-up script is therefore usually a good place to set up environment variables such as PATH, EDITOR etc. For example with bash, to add the directory ~/bin to your PATH, you can include the line:

    export PATH=$PATH:~/bin

in your .profile. If you subsequently modify your .profile and you wish to import the changes into your current shell, type:

    $ source .profile
    $ . ./profile

The source command is built into the shell. It ensures that changes to the environment made in .profile affect the current shell, and not the shell that would otherwise be created to execute the .profile script.

With csh, to add the directory ~/bin to your PATH, you can include the line:

    set path = ( $PATH $HOME/bin )

in your .cshrc.


© September 2001 William Knottenbelt (