This article is mainly notes I have taken for CSCB09/CSC209 at UofT.

File Expansion

This is like a lite version of Regex.

Program Structure

Scripts usually start with #!/bin/bash or #!/bin/sh.

Basically, #! followed by your script interpreter, not the same on every machine. This will tell the machine how to execute the script.

It is followed by shell commands, anything you can do with the interactive version of it, you can put it into the file.

Note: You must mark the file as executable, use chmod.

Run the program by typing ./ The file does not have to end in .sh.


To assign a variable, use the following syntax

VAR=value # No space before/after =

To get a value, use the following syntax


You don’t need to declare type, if you try to access a variable that is not assigned, it will just give you an empty string.

For example:

[email protected]:~$ FOO=test
[email protected]:~$ echo $FOO
[email protected]:~$ echo $bar

[email protected]:~$

Built-in Variables

Use printenv to see a list of variables that is built-in to your system.

One of the very useful one is $PATH. I won’t go to details here, since you should know how to use $PATH.


Variables defined in script are lost when it ends.

Subshell does not have access to parent shell’s variables.

To let subshell access parent shell’s variables, you have to export it.

$ export FOO="test"
$ ./ # now have access to $FOO

To maintain variables after running a script, you have to source it.

$ source
$ # parent shell now have access to variables within


Quotes are essential to shell programming. There are three types of quotes:

' Single Quote

Everything within a single quote are literal, for example, if you type echo '$FOO', it will literally output $FOO.

" Double Quote

Double quote will expand variables within it, for example the following script:

echo "FOO's value is $FOO"

It will expand $FOO and output FOO's value is test.

` Backtick

Backtick will create a new shell process, run the command within it, and store the output value.

For example the following script:

FOO=`ls -a | grep *.c`
echo "FOO's value is $FOO"

It should output all .c files within the current directory.

Note: Since backtick will create a new shell process, avoid it if you can.

Reading From STDIN

To read from STDIN, we use read command.

Consider the following script

echo "Please enter two words:"
read FOO BAR
echo "$FOO, $BAR"

When you run it and enter hello world, you should get the following output

[email protected]:~$ ./
Please enter two words:
hello world
hello, world

If you enter hi my name is jun, you will get the following output

[email protected]:~$ ./
Please enter two words:
hi my name is jun
hi, my name is jun

Commandline Arguments

All commandline arguments are placed in positional paramaters.

Basically it just means, $1 is the first argument, $2 is the second, and so on.

After $9, use curly brackets to wrap the number ${10} sh only allows 9 positional arguments, bash allows more.

Consider the following script

echo "The script is called: $0"
echo "First argument is: $1"
echo "Second argument is: $2"
echo "We have $# arguments in total"
echo "All arguments [email protected]"

Running it with ./ hello world my name is jun will have the following output

The script is called: ./
First argument is: hello
Second argument is: world
We have 6 arguments in total
All arguments hello world my name is jun


There is only one “array” within bourne shell, which is the positional parameters.

Use the set command to assign parameters to positional parameters.

For example:

set hello world

Will set $1 to hello, and $2 to world.

The previous arguments will be lost!

It is very useful to get command outputs, for example set `date`

Shifting Arguments

shift command will shift all positional arguments to the left.

For example

set hello world

It will result in hello being lost, and world become $1. It is useful to loop through all arguments.

If Statement

In bash, if statement will proceed only if the command returns 0, otherwise it will run else.

Return value of a command is not the output of the command.

UNIX returns 0 if a command runs successfully, other numbers for failure.

You can check the last process’ return value with $?.

For example, the following scripts

if ls
    echo Exit code is $?
    echo Failure, exit code is $?

It will print Exit code is 0 if ls runs successfully.

Test Command

test is a command you can use to test stuff, it will return 0 if true, 1 if false.

For example you can do this:

if test $FOO = $BAR
    echo Strings are identical
    echo Strings are different

Short form of test is []

if [ $FOO = $BAR ]
    echo Strings are identical
    echo Strings are different

There are many things you can test on, following is some of them


expr is used to do integer arithmetics. It can also be used for strings, but we usually don’t do that.

For example

x=`expr $x + 1`
# If you want to do multiplication, you have to escape
x=`expr $x \* 5`

While Loop

While loop will keep running as long as the command returns 0.

The basic structure is

while command
    more commands

You can use while to read from a file, line by line


while read line
    echo $line
done < $file

For Loop

For loop will loop though all provided values

for i in 1 2 "test" $FOO
    echo $i

One example for for loop would be to rename all files within a directory to filename.txt.

for i in `ls`
    mv $i $i.txt

You can also use for loop to loop through all positional arguments

for word in $*
    echo $word


You can also create your own functions like this

func() {
    echo $arg1 $arg2
    return 1
func hello world

Arguements are passed as positional parameters. There is no variable scope, you can access variables inside or outside of the subroutine.

Return value is in $?, since subroutine is just a command.