Chapter 23. Functions

Like "real" programming languages, Bash has functions, though in a somewhat limited implementation. A function is a subroutine, a code block that implements a set of operations, a "black box" that performs a specified task. Wherever there is repetitive code, when a task repeats with only slight variations in procedure, then consider using a function.

function function_name {
command...
}

or

function_name () {
command...
}

This second form will cheer the hearts of C programmers (and is more portable).

As in C, the function's opening bracket may optionally appear on the second line.

function_name ()
{
command...
}

Note

A function may be "compacted" into a single line.

   1 fun () { echo "This is a function"; echo; }
   2 #                                 ^     ^

In this case, however, a semicolon must follow the final command in the function.

   1 fun () { echo "This is a function"; echo } # Error!
   2 #                                       ^

Functions are called, triggered, simply by invoking their names. A function call is equivalent to a command.


Example 23-1. Simple functions

   1 #!/bin/bash
   2 
   3 JUST_A_SECOND=1
   4 
   5 funky ()
   6 { # This is about as simple as functions get.
   7   echo "This is a funky function."
   8   echo "Now exiting funky function."
   9 } # Function declaration must precede call.
  10 
  11 
  12 fun ()
  13 { # A somewhat more complex function.
  14   i=0
  15   REPEATS=30
  16 
  17   echo
  18   echo "And now the fun really begins."
  19   echo
  20 
  21   sleep $JUST_A_SECOND    # Hey, wait a second!
  22   while [ $i -lt $REPEATS ]
  23   do
  24     echo "----------FUNCTIONS---------->"
  25     echo "<------------ARE-------------"
  26     echo "<------------FUN------------>"
  27     echo
  28     let "i+=1"
  29   done
  30 }
  31 
  32   # Now, call the functions.
  33 
  34 funky
  35 fun
  36 
  37 exit 0

The function definition must precede the first call to it. There is no method of "declaring" the function, as, for example, in C.
   1 f1
   2 # Will give an error message, since function "f1" not yet defined.
   3 
   4 declare -f f1      # This doesn't help either.
   5 f1                 # Still an error message.
   6 
   7 # However...
   8 
   9 	  
  10 f1 ()
  11 {
  12   echo "Calling function \"f2\" from within function \"f1\"."
  13   f2
  14 }
  15 
  16 f2 ()
  17 {
  18   echo "Function \"f2\"."
  19 }
  20 
  21 f1  #  Function "f2" is not actually called until this point,
  22     #+ although it is referenced before its definition.
  23     #  This is permissible.
  24     
  25     # Thanks, S.C.

It is even possible to nest a function within another function, although this is not very useful.
   1 f1 ()
   2 {
   3 
   4   f2 () # nested
   5   {
   6     echo "Function \"f2\", inside \"f1\"."
   7   }
   8 
   9 }  
  10 
  11 f2  #  Gives an error message.
  12     #  Even a preceding "declare -f f2" wouldn't help.
  13 
  14 echo    
  15 
  16 f1  #  Does nothing, since calling "f1" does not automatically call "f2".
  17 f2  #  Now, it's all right to call "f2",
  18     #+ since its definition has been made visible by calling "f1".
  19 
  20     # Thanks, S.C.

Function declarations can appear in unlikely places, even where a command would otherwise go.
   1 ls -l | foo() { echo "foo"; }  # Permissible, but useless.
   2 
   3 
   4 
   5 if [ "$USER" = bozo ]
   6 then
   7   bozo_greet ()   # Function definition embedded in an if/then construct.
   8   {
   9     echo "Hello, Bozo."
  10   }
  11 fi  
  12 
  13 bozo_greet        # Works only for Bozo, and other users get an error.
  14 
  15 
  16 
  17 # Something like this might be useful in some contexts.
  18 NO_EXIT=1   # Will enable function definition below.
  19 
  20 [[ $NO_EXIT -eq 1 ]] && exit() { true; }     # Function definition in an "and-list".
  21 # If $NO_EXIT is 1, declares "exit ()".
  22 # This disables the "exit" builtin by aliasing it to "true".
  23 
  24 exit  # Invokes "exit ()" function, not "exit" builtin.
  25 
  26 
  27 
  28 # Or, similarly:
  29 filename=file1
  30 
  31 [ -f "$filename" ] &&
  32 foo () { rm -f "$filename"; echo "File "$filename" deleted."; } ||
  33 foo () { echo "File "$filename" not found."; touch bar; }
  34 
  35 foo
  36 
  37 # Thanks, S.C. and Christopher Head

Note

What happens when different versions of the same function appear in a script?
   1 #  As Yan Chen points out,
   2 #  when a function is defined multiple times,
   3 #  the final version is what is invoked.
   4 #  This is not, however, particularly useful.
   5 
   6 func ()
   7 {
   8   echo "First version of func ()."
   9 }
  10 
  11 func ()
  12 {
  13   echo "Second version of func ()."
  14 }
  15 
  16 func   # Second version of func ().
  17 
  18 exit $?
  19 
  20 #  It is even possible to use functions to override
  21 #+ or preempt system commands.
  22 #  Of course, this is *not* advisable.

23.1. Complex Functions and Function Complexities

Functions may process arguments passed to them and return an exit status to the script for further processing.

   1 function_name $arg1 $arg2

The function refers to the passed arguments by position (as if they were positional parameters), that is, $1, $2, and so forth.


Example 23-2. Function Taking Parameters

   1 #!/bin/bash
   2 # Functions and parameters
   3 
   4 DEFAULT=default                             # Default param value.
   5 
   6 func2 () {
   7    if [ -z "$1" ]                           # Is parameter #1 zero length?
   8    then
   9      echo "-Parameter #1 is zero length.-"  # Or no parameter passed.
  10    else
  11      echo "-Param #1 is \"$1\".-"
  12    fi
  13 
  14    variable=${1-$DEFAULT}                   #  What does
  15    echo "variable = $variable"              #+ parameter substitution show?
  16                                             #  ---------------------------
  17                                             #  It distinguishes between
  18                                             #+ no param and a null param.
  19 
  20    if [ "$2" ]
  21    then
  22      echo "-Parameter #2 is \"$2\".-"
  23    fi
  24 
  25    return 0
  26 }
  27 
  28 echo
  29    
  30 echo "Nothing passed."   
  31 func2                          # Called with no params
  32 echo
  33 
  34 
  35 echo "Zero-length parameter passed."
  36 func2 ""                       # Called with zero-length param
  37 echo
  38 
  39 echo "Null parameter passed."
  40 func2 "$uninitialized_param"   # Called with uninitialized param
  41 echo
  42 
  43 echo "One parameter passed."   
  44 func2 first           # Called with one param
  45 echo
  46 
  47 echo "Two parameters passed."   
  48 func2 first second    # Called with two params
  49 echo
  50 
  51 echo "\"\" \"second\" passed."
  52 func2 "" second       # Called with zero-length first parameter
  53 echo                  # and ASCII string as a second one.
  54 
  55 exit 0

Important

The shift command works on arguments passed to functions (see Example 33-16).

But, what about command-line arguments passed to the script? Does a function see them? Well, let's clear up the confusion.


Example 23-3. Functions and command-line args passed to the script

   1 #!/bin/bash
   2 # func-cmdlinearg.sh
   3 #  Call this script with a command-line argument,
   4 #+ something like $0 arg1.
   5 
   6 
   7 func ()
   8 
   9 {
  10 echo "$1"
  11 }
  12 
  13 echo "First call to function: no arg passed."
  14 echo "See if command-line arg is seen."
  15 func
  16 # No! Command-line arg not seen.
  17 
  18 echo "============================================================"
  19 echo
  20 echo "Second call to function: command-line arg passed explicitly."
  21 func $1
  22 # Now it's seen!
  23 
  24 exit 0

In contrast to certain other programming languages, shell scripts normally pass only value parameters to functions. Variable names (which are actually pointers), if passed as parameters to functions, will be treated as string literals. Functions interpret their arguments literally.

Indirect variable references (see Example 34-2) provide a clumsy sort of mechanism for passing variable pointers to functions.


Example 23-4. Passing an indirect reference to a function

   1 #!/bin/bash
   2 # ind-func.sh: Passing an indirect reference to a function.
   3 
   4 echo_var ()
   5 {
   6 echo "$1"
   7 }
   8 
   9 message=Hello
  10 Hello=Goodbye
  11 
  12 echo_var "$message"        # Hello
  13 # Now, let's pass an indirect reference to the function.
  14 echo_var "${!message}"     # Goodbye
  15 
  16 echo "-------------"
  17 
  18 # What happens if we change the contents of "hello" variable?
  19 Hello="Hello, again!"
  20 echo_var "$message"        # Hello
  21 echo_var "${!message}"     # Hello, again!
  22 
  23 exit 0

The next logical question is whether parameters can be dereferenced after being passed to a function.


Example 23-5. Dereferencing a parameter passed to a function

   1 #!/bin/bash
   2 # dereference.sh
   3 # Dereferencing parameter passed to a function.
   4 # Script by Bruce W. Clare.
   5 
   6 dereference ()
   7 {
   8      y=\$"$1"   # Name of variable.
   9      echo $y    # $Junk
  10 
  11      x=`eval "expr \"$y\" "`
  12      echo $1=$x
  13      eval "$1=\"Some Different Text \""  # Assign new value.
  14 }
  15 
  16 Junk="Some Text"
  17 echo $Junk "before"    # Some Text before
  18 
  19 dereference Junk
  20 echo $Junk "after"     # Some Different Text after
  21 
  22 exit 0


Example 23-6. Again, dereferencing a parameter passed to a function

   1 #!/bin/bash
   2 # ref-params.sh: Dereferencing a parameter passed to a function.
   3 #                (Complex Example)
   4 
   5 ITERATIONS=3  # How many times to get input.
   6 icount=1
   7 
   8 my_read () {
   9   #  Called with my_read varname,
  10   #+ outputs the previous value between brackets as the default value,
  11   #+ then asks for a new value.
  12 
  13   local local_var
  14 
  15   echo -n "Enter a value "
  16   eval 'echo -n "[$'$1'] "'  #  Previous value.
  17 # eval echo -n "[\$$1] "     #  Easier to understand,
  18                              #+ but loses trailing space in user prompt.
  19   read local_var
  20   [ -n "$local_var" ] && eval $1=\$local_var
  21 
  22   # "And-list": if "local_var" then set "$1" to its value.
  23 }
  24 
  25 echo
  26 
  27 while [ "$icount" -le "$ITERATIONS" ]
  28 do
  29   my_read var
  30   echo "Entry #$icount = $var"
  31   let "icount += 1"
  32   echo
  33 done  
  34 
  35 
  36 # Thanks to Stephane Chazelas for providing this instructive example.
  37 
  38 exit 0

Exit and Return

exit status

Functions return a value, called an exit status. The exit status may be explicitly specified by a return statement, otherwise it is the exit status of the last command in the function (0 if successful, and a non-zero error code if not). This exit status may be used in the script by referencing it as $?. This mechanism effectively permits script functions to have a "return value" similar to C functions.

return

Terminates a function. A return command [1] optionally takes an integer argument, which is returned to the calling script as the "exit status" of the function, and this exit status is assigned to the variable $?.


Example 23-7. Maximum of two numbers

   1 #!/bin/bash
   2 # max.sh: Maximum of two integers.
   3 
   4 E_PARAM_ERR=250    # If less than 2 params passed to function.
   5 EQUAL=251          # Return value if both params equal.
   6 #  Error values out of range of any
   7 #+ params that might be fed to the function.
   8 
   9 max2 ()             # Returns larger of two numbers.
  10 {                   # Note: numbers compared must be less than 257.
  11 if [ -z "$2" ]
  12 then
  13   return $E_PARAM_ERR
  14 fi
  15 
  16 if [ "$1" -eq "$2" ]
  17 then
  18   return $EQUAL
  19 else
  20   if [ "$1" -gt "$2" ]
  21   then
  22     return $1
  23   else
  24     return $2
  25   fi
  26 fi
  27 }
  28 
  29 max2 33 34
  30 return_val=$?
  31 
  32 if [ "$return_val" -eq $E_PARAM_ERR ]
  33 then
  34   echo "Need to pass two parameters to the function."
  35 elif [ "$return_val" -eq $EQUAL ]
  36   then
  37     echo "The two numbers are equal."
  38 else
  39     echo "The larger of the two numbers is $return_val."
  40 fi  
  41 
  42   
  43 exit 0
  44 
  45 #  Exercise (easy):
  46 #  ---------------
  47 #  Convert this to an interactive script,
  48 #+ that is, have the script ask for input (two numbers).

Tip

For a function to return a string or array, use a dedicated variable.
   1 count_lines_in_etc_passwd()
   2 {
   3   [[ -r /etc/passwd ]] && REPLY=$(echo $(wc -l < /etc/passwd))
   4   #  If /etc/passwd is readable, set REPLY to line count.
   5   #  Returns both a parameter value and status information.
   6   #  The 'echo' seems unnecessary, but . . .
   7   #+ it removes excess whitespace from the output.
   8 }
   9 
  10 if count_lines_in_etc_passwd
  11 then
  12   echo "There are $REPLY lines in /etc/passwd."
  13 else
  14   echo "Cannot count lines in /etc/passwd."
  15 fi  
  16 
  17 # Thanks, S.C.


Example 23-8. Converting numbers to Roman numerals

   1 #!/bin/bash
   2 
   3 # Arabic number to Roman numeral conversion
   4 # Range: 0 - 200
   5 # It's crude, but it works.
   6 
   7 # Extending the range and otherwise improving the script is left as an exercise.
   8 
   9 # Usage: roman number-to-convert
  10 
  11 LIMIT=200
  12 E_ARG_ERR=65
  13 E_OUT_OF_RANGE=66
  14 
  15 if [ -z "$1" ]
  16 then
  17   echo "Usage: `basename $0` number-to-convert"
  18   exit $E_ARG_ERR
  19 fi  
  20 
  21 num=$1
  22 if [ "$num" -gt $LIMIT ]
  23 then
  24   echo "Out of range!"
  25   exit $E_OUT_OF_RANGE
  26 fi  
  27 
  28 to_roman ()   # Must declare function before first call to it.
  29 {
  30 number=$1
  31 factor=$2
  32 rchar=$3
  33 let "remainder = number - factor"
  34 while [ "$remainder" -ge 0 ]
  35 do
  36   echo -n $rchar
  37   let "number -= factor"
  38   let "remainder = number - factor"
  39 done  
  40 
  41 return $number
  42        # Exercises:
  43        # ---------
  44        # 1) Explain how this function works.
  45        #    Hint: division by successive subtraction.
  46        # 2) Extend to range of the function.
  47        #    Hint: use "echo" and command-substitution capture.
  48 }
  49    
  50 
  51 to_roman $num 100 C
  52 num=$?
  53 to_roman $num 90 LXXXX
  54 num=$?
  55 to_roman $num 50 L
  56 num=$?
  57 to_roman $num 40 XL
  58 num=$?
  59 to_roman $num 10 X
  60 num=$?
  61 to_roman $num 9 IX
  62 num=$?
  63 to_roman $num 5 V
  64 num=$?
  65 to_roman $num 4 IV
  66 num=$?
  67 to_roman $num 1 I
  68 # Successive calls to conversion function!
  69 # Is this really necessary??? Can it be simplified?
  70 
  71 echo
  72 
  73 exit

See also Example 10-28.

Important

The largest positive integer a function can return is 255. The return command is closely tied to the concept of exit status, which accounts for this particular limitation. Fortunately, there are various workarounds for those situations requiring a large integer return value from a function.


Example 23-9. Testing large return values in a function

   1 #!/bin/bash
   2 # return-test.sh
   3 
   4 # The largest positive value a function can return is 255.
   5 
   6 return_test ()         # Returns whatever passed to it.
   7 {
   8   return $1
   9 }
  10 
  11 return_test 27         # o.k.
  12 echo $?                # Returns 27.
  13   
  14 return_test 255        # Still o.k.
  15 echo $?                # Returns 255.
  16 
  17 return_test 257        # Error!
  18 echo $?                # Returns 1 (return code for miscellaneous error).
  19 
  20 # ======================================================
  21 return_test -151896    # Do large negative numbers work?
  22 echo $?                # Will this return -151896?
  23                        # No! It returns 168.
  24 #  Version of Bash before 2.05b permitted
  25 #+ large negative integer return values.
  26 #  Newer versions of Bash plug this loophole.
  27 #  This may break older scripts.
  28 #  Caution!
  29 # ======================================================
  30 
  31 exit 0

A workaround for obtaining large integer "return values" is to simply assign the "return value" to a global variable.
   1 Return_Val=   # Global variable to hold oversize return value of function.
   2 
   3 alt_return_test ()
   4 {
   5   fvar=$1
   6   Return_Val=$fvar
   7   return   # Returns 0 (success).
   8 }
   9 
  10 alt_return_test 1
  11 echo $?                              # 0
  12 echo "return value = $Return_Val"    # 1
  13 
  14 alt_return_test 256
  15 echo "return value = $Return_Val"    # 256
  16 
  17 alt_return_test 257
  18 echo "return value = $Return_Val"    # 257
  19 
  20 alt_return_test 25701
  21 echo "return value = $Return_Val"    #25701

A more elegant method is to have the function echo its "return value to stdout," and then capture it by command substitution. See the discussion of this in Section 33.8.


Example 23-10. Comparing two large integers

   1 #!/bin/bash
   2 # max2.sh: Maximum of two LARGE integers.
   3 
   4 #  This is the previous "max.sh" example,
   5 #+ modified to permit comparing large integers.
   6 
   7 EQUAL=0             # Return value if both params equal.
   8 E_PARAM_ERR=-99999  # Not enough params passed to function.
   9 #           ^^^^^^    Out of range of any params that might be passed.
  10 
  11 max2 ()             # "Returns" larger of two numbers.
  12 {
  13 if [ -z "$2" ]
  14 then
  15   echo $E_PARAM_ERR
  16   return
  17 fi
  18 
  19 if [ "$1" -eq "$2" ]
  20 then
  21   echo $EQUAL
  22   return
  23 else
  24   if [ "$1" -gt "$2" ]
  25   then
  26     retval=$1
  27   else
  28     retval=$2
  29   fi
  30 fi
  31 
  32 echo $retval        # Echoes (to stdout), rather than returning value.
  33                     # Why?
  34 }
  35 
  36 
  37 return_val=$(max2 33001 33997)
  38 #            ^^^^             Function name
  39 #                 ^^^^^ ^^^^^ Params passed
  40 #  This is actually a form of command substitution:
  41 #+ treating a function as if it were a command,
  42 #+ and assigning the stdout of the function to the variable "return_val."
  43 
  44 
  45 # ========================= OUTPUT ========================
  46 if [ "$return_val" -eq "$E_PARAM_ERR" ]
  47   then
  48   echo "Error in parameters passed to comparison function!"
  49 elif [ "$return_val" -eq "$EQUAL" ]
  50   then
  51     echo "The two numbers are equal."
  52 else
  53     echo "The larger of the two numbers is $return_val."
  54 fi
  55 # =========================================================
  56   
  57 exit 0
  58 
  59 #  Exercises:
  60 #  ---------
  61 #  1) Find a more elegant way of testing
  62 #+    the parameters passed to the function.
  63 #  2) Simplify the if/then structure at "OUTPUT."
  64 #  3) Rewrite the script to take input from command-line parameters.

Here is another example of capturing a function "return value." Understanding it requires some knowledge of awk.
   1 month_length ()  # Takes month number as an argument.
   2 {                # Returns number of days in month.
   3 monthD="31 28 31 30 31 30 31 31 30 31 30 31"  # Declare as local?
   4 echo "$monthD" | awk '{ print $'"${1}"' }'    # Tricky.
   5 #                             ^^^^^^^^^
   6 # Parameter passed to function  ($1 -- month number), then to awk.
   7 # Awk sees this as "print $1 . . . print $12" (depending on month number)
   8 # Template for passing a parameter to embedded awk script:
   9 #                                 $'"${script_parameter}"'
  10 
  11 #  Needs error checking for correct parameter range (1-12)
  12 #+ and for February in leap year.
  13 }
  14 
  15 # ----------------------------------------------
  16 # Usage example:
  17 month=4        # April, for example (4th month).
  18 days_in=$(month_length $month)
  19 echo $days_in  # 30
  20 # ----------------------------------------------

See also Example A-7 and Example A-39.

Exercise: Using what we have just learned, extend the previous Roman numerals example to accept arbitrarily large input.

Redirection

Redirecting the stdin of a function

A function is essentially a code block, which means its stdin can be redirected (as in Example 3-1).


Example 23-11. Real name from username

   1 #!/bin/bash
   2 # realname.sh
   3 #
   4 # From username, gets "real name" from /etc/passwd.
   5 
   6 
   7 ARGCOUNT=1       # Expect one arg.
   8 E_WRONGARGS=65
   9 
  10 file=/etc/passwd
  11 pattern=$1
  12 
  13 if [ $# -ne "$ARGCOUNT" ]
  14 then
  15   echo "Usage: `basename $0` USERNAME"
  16   exit $E_WRONGARGS
  17 fi  
  18 
  19 file_excerpt ()    #  Scan file for pattern,
  20 {                  #+ then print relevant portion of line.
  21   while read line  # "while" does not necessarily need [ condition ]
  22   do
  23     echo "$line" | grep $1 | awk -F":" '{ print $5 }'
  24     # Have awk use ":" delimiter.
  25   done
  26 } <$file  # Redirect into function's stdin.
  27 
  28 file_excerpt $pattern
  29 
  30 # Yes, this entire script could be reduced to
  31 #       grep PATTERN /etc/passwd | awk -F":" '{ print $5 }'
  32 # or
  33 #       awk -F: '/PATTERN/ {print $5}'
  34 # or
  35 #       awk -F: '($1 == "username") { print $5 }' # real name from username
  36 # However, it might not be as instructive.
  37 
  38 exit 0

There is an alternate, and perhaps less confusing method of redirecting a function's stdin. This involves redirecting the stdin to an embedded bracketed code block within the function.
   1 # Instead of:
   2 Function ()
   3 {
   4  ...
   5  } < file
   6 
   7 # Try this:
   8 Function ()
   9 {
  10   {
  11     ...
  12    } < file
  13 }
  14 
  15 # Similarly,
  16 
  17 Function ()  # This works.
  18 {
  19   {
  20    echo $*
  21   } | tr a b
  22 }
  23 
  24 Function ()  # This doesn't work.
  25 {
  26   echo $*
  27 } | tr a b   # A nested code block is mandatory here.
  28 
  29 
  30 # Thanks, S.C.

Notes

[1]

The return command is a Bash builtin.