What is Shell Programming?

• Putting UNIX commands in a file
• Almost always special-purpose code
• Often one-time code
• Seldom used where speed is important
• Often used to manipulate files

Shell Programming Functions

By encapsulating common functionality into easy to manage units called shell functions, you can simplify your scripts, making them easier to maintain. This chapter shows you how to use shell functions in your scripts.

Shell functions provide a way of mapping a name to a list of commands. Functions are similar to subroutines and procedures in other programming languages. You can also think of them as miniature shell scripts, complete with exit codes and arguments. The main difference between a script and a function is that a new instance of the shell is started for a shell script, whereas functions run in the current instance of the shell.

This chapter is divided into the following two sections:

• Using functions
• Understanding scope, recursion, return codes, and data sharing

The first section introduces the syntax for defining functions and illustrates their use, whereas the second section covers more advanced topics relating to the interaction of scripts and functions.

Using Functions

Functions are defined as follows:

name () { list ; }

Here, name is the name of the function and list is a list of commands. The list of commands, list, is referred to as the body of the function. The parentheses, ( and ), that follow name are required.

The job of a function is to bind name to list, so that whenever name is specified list is executed. When a function is defined, list is not executed; the shell parses list to ensure that there are no syntax errors and stores name in its list of commands. The following example illustrates a basic function definition:

lsl() { ls –l ; }

Here you define the function lsl and specify list as ls –l.

An alternative form of function definition is available in ksh, bash, and zsh:

function name { list ; }

Here, name is the name of the function and list is the list of commands to be executed. This form of function definition is not available in the Bourne shell. Scripts that need to be ported to older systems should not use this form for function definition.

Executing Functions

You can execute or call a function that has been defined by specifying its name. For example, you can execute the function lsl, defined in the previous example, as follows:

$ lsl

This causes the shell to execute the body of the function, in this case the command ls –l, and output the result. The output will be similar to the following:

total 6
drwxrwxrwt  3 root wheel 512 Oct 29 08:59 ./
drwxr-xr-x 25 root wheel 512 Oct 29 00:02 ../
drwxrwxrwt  2 root wheel 512 Nov 3 17:49 vi.recover/

Functions are normally defined on the command line or within a script. Once defined, the function acts as a valid command in all the sub-shells started by that shell or script. For example, if you enter the command:

$ lsl() { ls –l ; }

The function lsl becomes a valid command name that can be accessed by specifying lsl. It is accessible in sub-shells as well:

$ ( lsl )
total 6 drwxrwxrwt 3 root wheel 512 Oct 29 08:59 ./ drwxr-xr-x 25 root wheel 512 Oct 29 00:02 ../ drwxrwxrwt 2 root wheel 512 Nov 3 17:49 vi.recover/

A function defined in a script is accessible within that script and any sub-shells started by that script. For example, consider the following script:

#!/bin/sh
lsl() { ls –l ; }
cd "$1" && lsl

The function lsl is only available in that script.

Arguments

Just as you can execute commands with arguments, you can also execute functions with arguments. The general syntax for invoking a function is as follows:

name arg1 ... argN

Here, name is the name of the function and arg1 ... argN are the arguments to the function. The arguments specified to a function are accessed in the same way as arguments specified to a shell script; the individual arguments are available as $1, $2, and so on, whereas the set of all the arguments is available as $@.

The following function illustrates the use of individual arguments:

printMsg () { echo "$1: $2" ; }

This function uses echo to print a message with a colon, :, which separates the first two arguments when it’s executed as follows:

printMsg Error Failed
the output is

Error: Failed

As defined, this function can handle only two arguments; it ignores all the others. In order to make the function a bit more useful, it needs to be able to handle an arbitrary number of arguments. Because all of the arguments specified to a function are available in the variable $@, you can use it as follows:

printMsg() {
  PREFIX="$1"
  shift 
  echo "$PREFIX: $@"
}

Here, you have redefined the function printMsg. It saves its first argument in $PREFIX and then usesecho to print the message in the desired format. You use shift to remove the first argument from $@before calling echo. Now you can execute the function with any number of arguments and the message will be printed properly. For example, if printMsg is executed as follows:

printMsg Info All Quiet on the Western Front

the output is

Info: All Quiet on the Western Front

Function Chaining

Function chaining is the process of calling a function from another function. The following script illustrates function chaining:

#!/bin/sh

orange () {
  echo "Now in orange"
  banana          # call func2()
}

banana () {
  echo "Now in banana"
}

orange

This script defines two functions, orange and banana, and then executes orange. The first function,orange, outputs a message and then calls the function banana. The second function, banana, just outputs a message. The output from this script is

Now in orange
Now in banana

Common Errors

Two common errors with declaring and using functions are

• Omitting the parentheses, (), in a function definition.
• Specifying the parentheses, (), in a function invocation.

The following example illustrates the first type of error:

lsl { ls -l ; }

Here, the parentheses are missing after lsl. This is an invalid function definition and will result in an error message similar to the following:

sh: syntax error: '}' unexpected

The following command illustrates the second type of error:

$ lsl()

Here, the function lsl is executed along with the parentheses, (). This will not work because the shell interprets it as a redefinition of the function with the name lsl. Usually such an invocation results in a prompt similar to the following:

>

This is a prompt produced by the shell when it expects you to provide more input. The input it expects is the body of the function lsl.

Aliases Versus Functions

An alias is an abbreviation or an alternative name, usually mnemonic, for a command. Aliases were first introduced in csh and were later adopted by ksh, bash, and zsh. They are not supported in the Bourne shell.

Aliases are defined using the alias command:

alias name="cmd"

Here name is the name of the alias and cmd is the command to execute when name is specified. Aliases are similar to functions in that they associate a command with a name. Two key differences are

• In an alias, cmd cannot be a compound command or a list.
• In an alias, there is no way to manipulate the argument list ($@).

Due to their limited capabilities, aliases are not commonly used in shell programs. They are discussed here for the sake of completeness.

As an example, the following command defines the alias lsl and specifies that the command ls –lshould be executed when the command lsl is specified:

alias lsl="ls –l"

This alias is equivalent to the function:

lsl () { ls –l "$@" ; }

A common use for aliases is to specify a default set of options to a command. For example, say you have the following alias:

alias ls="ls –a"

When the ls command is given, the shell executes ls –a instead of plain ls without options. It is possible to mimic this behavior with a function such as:

name () { path "$@" ; }

Here, name is the name of the command to be “aliased” and path is the fully qualified path to the command. For example, the following function is equivalent to the alias given in the previous example:

ls () { /bin/ls –a "$@" ; }

Unalias

Once an alias has been defined, it can be unset using the unalias command:

unalias name

Here, name is the name of the alias to be unset. For example, the following command unsets the alias lsl:

unalias lsl

Unsetting Functions

Once a function has been defined, it can be undefined via the unset command:

unset name

Here, name is the name of the function you want to unset. For example, the following command unsets the previously defined function lsl():

unset lsl

After a function has been unset it cannot be executed.