Bash Parameters and Variables

From NovaOrdis Knowledge Base
Jump to navigation Jump to search

Internal

Parameters and Variables

Shell stores values internally using entities named parameters.

Parameters can be designated by a number (e.g. 1), a special character (e.g. *), or a name (e.g. blue). Parameters designated by a name are referred to as variables. The variables that have a special meaning to the shell (for example "IFS") are known as internal, keyword or built-in variables. Non-variable parameters can be further categorized in positional parameters, which are the shell's command line arguments, retrievable as $1, $2, etc. and special parameters, that are denoted by special characters.

Parameter and Variable Expansion - Accessing a Parameter's or Variable's Value

A dollar sign ($) that is NOT followed by an open parenthesis initiates parameter or variable expansion. The simplest case of expansion is retrieving the parameter or variable value: the associated value is retrieved by prefixing the parameter/variable's number, character or name with $ or enclosing it in ${...}. bash supports complex expansion rules, documented here:

Parameter and Variable Expansion


Difference between $ and ${...}

TODO

Positional Parameters $1, $2, ...

A positional parameter is a parameter denoted by one or more digits, other than the single digit 0 (the single digit 0 denotes the special parameter $0).

bash allows only nine parameters to be referenced directly (n = 1–9) as in:

$1

$1 is equivalent with ${1}.

When a positional parameter consisting of more than a single digit is expanded, it must be enclosed in braces. For n values greater than 9, the command line arguments must be specified as ${n}:

${10}

Positional parameters are assigned from the shell’s arguments when it is invoked. They may not be assigned to with assignment statements, but they can be reassigned with set builtin command and they can be shifted left with the bash built-in command shift.

The positional parameters are temporarily replaced when a shell function is executed.

Iterating over Positional Paramenters

Solution 1

while [ -n "$1" ]; do
  ...
  shift
done

Solution 2

Use an Array

Special Parameters

For a list of special parameters, see:

Special Parameters

Variables

A variable is a shell parameter designated by a name, not a number or a special character. Variables can be user-created, or built-in.

Variable Assignment

Variable Assignment

Unlike positional parameters and special parameters, variable can be assigned values by using the "=" character, with no spaces before or after:

blue=10

Built-in Variables

The variables that have a special meaning to the shell (for example "IFS") are known as internal, keyword or built-in variables. For a list of built-in variables, see:

Built-in Variables

Parameter and Variable Expansion

Parameters include:

  • Command line (positional) parameters: $1, $2, ...
  • Special parameters:

Variables include:

  • User-created variables
  • Keyword variables

Parameters and variables are not expanded if the string they are included in are enclosed within single quotes or if the leading dollar sign is escaped. Parameters and variables are expanded if the string they are included in is enclosed within double quotes.

Parameter and variable expansion is a particular case of bash command line expansion:

bash Command Line Expansion

Expand to a Default Value

http://wiki.bash-hackers.org/syntax/pe#use_a_default_value
${var:-alternative}

The :- form : if the variable 'var' referred in the expression was not defined, or it is the empty string, the expression expands the alternative.

${var-alternative}

The - form: if the colon is omitted, the expression is expanded to the alternative value only if the variable is not defined, not when it was empty.

Assign a Default Value

http://wiki.bash-hackers.org/syntax/pe#assign_a_default_value
 ${var:=word}
${var=word}

This expression works like expansion to a default value, but the word is not only expanded, but also assigned to the variable, if it was unset or null. The first form work if the variable is unset or an empty string, the second only if the parameter was unset.

Use an Alternate Value

http://wiki.bash-hackers.org/syntax/pe#use_an_alternate_value
${var:+alternative}

This form expands to nothing if the parameter is unset or empty.

${var+alternative}

This form expands to nothing if the variable is unset, and to the alternative if the variable is empty.

Variable Expansion in an Arbitrary File

If an arbitrary text file contains environment variable declarations, they can be expanded with envsubst. No other shell command line expansions will be performed. More details:

envsubst

String Length

The length of a variable's value:

${#var}

$!... Variable Indirection, Indirect Expansion

If the first character of parameter is an exclamation point (!), a level of variable indirection is introduced. Bash uses the value of the variable formed from the rest of parameter as the name of the variable; this variable is then expanded and that value is used in the rest of the substitution, rather than the value of parameter itself. This is known as indirect expansion. The exceptions to this are the expansions of ${!prefix*} and ${!name[@]} described below. The exclamation point must immediately follow immediately after the left brace in order to introduce indirection. Example:

a="b"
b="blah"
echo ${!a}

produces "blah".

Note that assigning a value to a indirect variable does not need to use "${!...}". For the above example, we can simply:

${a}=blah

This will assign "blah" to a variable named "b".

In each of the cases below, word is subject to tilde expansion, parameter expansion, command substitution, and arithmetic expansion.

When not performing substring expansion, using the forms documented below, bash tests for a parameter that is unset or null. Omitting the colon results in a test only for a parameter that is unset.

Undeclared Variable

An undeclared variable is equivalent with an empty string when referenced. There is no error when such a variable is referenced in a bash program:

echo ">${no_such_var}<" 
><

Uninitialized Variable

An uninitialized variable is equivalent with an empty string when referenced. There is no error when such a variable is referenced in a bash program:

local some_var
echo ">${some_var}<" 
><

Global Variable

Global variables are only maintained within the context of the current shell. Once the shell exits, the global variables are discarded.

The global variables declared in a shell are visible inside the functions executed within that shell and inside functions invoked from function invoked from the shell, recursively.

The value of a global variable is not available to sub-shells, unless the variable is exported, with the #export keyword.

export

export VAR=VALUE

Declaring a global variable to be exported causes the variable and its value to be copied in the environment of a sub-shell. The sub-shell gets a copy of the variable, not a reference. The sub-shell has no access to the parent process's environment whatsoever. Modifying the variable from the sub-shell does not reflect into the value of the global variable maintained by the invoking shell. When the shell sub-process terminates any changes made to its environment are lost. There is no way to modify directly the parent's environment.

Variables exported by Sub-Shells

If a sub-shell invoked from another shell exports a global variable, once the sub-shell exits, the exported variables are discarded, and the invoking shell does not sees them.

Local Variable

If a variable is declared inside a function, without any qualifier, it automatically becomes a global variable, and it is visible to the entire shell after the function execution, even after the function exits.

In order to prevent a variable declared inside of a function to become global, it must be declared as local: the local keyword designates a local variable. Local variables can only be declared inside a function.

Example:

local a=b

Local Variable Assignment and Failure


Do not do a local variable assignment from a function that may return non-zero and expect to use the return value to exit

Do not do something like this:

local a=$(do_something) || exit 1 # hoping to exit if 'do_something' returns a non-zero value

Even if do_something returns a non-zero error code, that is not detected and exit is not executed.

Do this instead:

local a;
a=$(do_something) || exit

Listing Declared Variables

Use declare or typeset:

declare -p

A way of obtaining the value for a specific variable is:

local var_name="HISTSIZE"
declare -p | grep "declare .. ${var_name}" | sed -e 's/.*=//'

Booleans

Recommended usage pattern:

Declaration:

some_var=true

Usage:

${some_var} && echo "this will execute"

if ${some_var}; then
   echo "this will execute"
fi

Do not use:

if [ ${some_var} ]; then ...

This will always evaluate to "true" regardless of the some_var value.