[bash] An example of how to use getopts in bash

I want to call myscript file in this way:

$ ./myscript -s 45 -p any_string

or

$ ./myscript -h  #should display help
$ ./myscript     #should display help

My requirements are:

  • getopt here to get the input arguments
  • check that -s exists, if not return an error
  • check that the value after the -s is 45 or 90
  • check that the -p exists and there is an input string after
  • if the user enters ./myscript -h or just ./myscript then display help

I tried so far this code:

#!/bin/bash
while getopts "h:s:" arg; do
  case $arg in
    h)
      echo "usage" 
      ;;
    s)
      strength=$OPTARG
      echo $strength
      ;;
  esac
done

But with that code I get errors. How to do it with Bash and getopt?

This question is related to bash shell getopts

The answer is


#!/bin/bash

usage() { echo "Usage: $0 [-s <45|90>] [-p <string>]" 1>&2; exit 1; }

while getopts ":s:p:" o; do
    case "${o}" in
        s)
            s=${OPTARG}
            ((s == 45 || s == 90)) || usage
            ;;
        p)
            p=${OPTARG}
            ;;
        *)
            usage
            ;;
    esac
done
shift $((OPTIND-1))

if [ -z "${s}" ] || [ -z "${p}" ]; then
    usage
fi

echo "s = ${s}"
echo "p = ${p}"

Example runs:

$ ./myscript.sh
Usage: ./myscript.sh [-s <45|90>] [-p <string>]

$ ./myscript.sh -h
Usage: ./myscript.sh [-s <45|90>] [-p <string>]

$ ./myscript.sh -s "" -p ""
Usage: ./myscript.sh [-s <45|90>] [-p <string>]

$ ./myscript.sh -s 10 -p foo
Usage: ./myscript.sh [-s <45|90>] [-p <string>]

$ ./myscript.sh -s 45 -p foo
s = 45
p = foo

$ ./myscript.sh -s 90 -p bar
s = 90
p = bar

I know that this is already answered, but for the record and for anyone with the same requeriments as me I decided to post this related answer. The code is flooded with comments to explain the code.

Updated answer:

Save the file as getopt.sh:

#!/bin/bash

function get_variable_name_for_option {
    local OPT_DESC=${1}
    local OPTION=${2}
    local VAR=$(echo ${OPT_DESC} | sed -e "s/.*\[\?-${OPTION} \([A-Z_]\+\).*/\1/g" -e "s/.*\[\?-\(${OPTION}\).*/\1FLAG/g")

    if [[ "${VAR}" == "${1}" ]]; then
        echo ""
    else
        echo ${VAR}
    fi
}

function parse_options {
    local OPT_DESC=${1}
    local INPUT=$(get_input_for_getopts "${OPT_DESC}")

    shift
    while getopts ${INPUT} OPTION ${@};
    do
        [ ${OPTION} == "?" ] && usage
        VARNAME=$(get_variable_name_for_option "${OPT_DESC}" "${OPTION}")
            [ "${VARNAME}" != "" ] && eval "${VARNAME}=${OPTARG:-true}" # && printf "\t%s\n" "* Declaring ${VARNAME}=${!VARNAME} -- OPTIONS='$OPTION'"
    done

    check_for_required "${OPT_DESC}"

}

function check_for_required {
    local OPT_DESC=${1}
    local REQUIRED=$(get_required "${OPT_DESC}" | sed -e "s/\://g")
    while test -n "${REQUIRED}"; do
        OPTION=${REQUIRED:0:1}
        VARNAME=$(get_variable_name_for_option "${OPT_DESC}" "${OPTION}")
                [ -z "${!VARNAME}" ] && printf "ERROR: %s\n" "Option -${OPTION} must been set." && usage
        REQUIRED=${REQUIRED:1}
    done
}

function get_input_for_getopts {
    local OPT_DESC=${1}
    echo ${OPT_DESC} | sed -e "s/\([a-zA-Z]\) [A-Z_]\+/\1:/g" -e "s/[][ -]//g"
}

function get_optional {
    local OPT_DESC=${1}
    echo ${OPT_DESC} | sed -e "s/[^[]*\(\[[^]]*\]\)[^[]*/\1/g" -e "s/\([a-zA-Z]\) [A-Z_]\+/\1:/g" -e "s/[][ -]//g"
}

function get_required {
    local OPT_DESC=${1}
    echo ${OPT_DESC} | sed -e "s/\([a-zA-Z]\) [A-Z_]\+/\1:/g" -e "s/\[[^[]*\]//g" -e "s/[][ -]//g"
}

function usage {
    printf "Usage:\n\t%s\n" "${0} ${OPT_DESC}"
    exit 10
}

Then you can use it like this:

#!/bin/bash
#
# [ and ] defines optional arguments
#

# location to getopts.sh file
source ./getopt.sh
USAGE="-u USER -d DATABASE -p PASS -s SID [ -a START_DATE_TIME ]"
parse_options "${USAGE}" ${@}

echo ${USER}
echo ${START_DATE_TIME}

Old answer:

I recently needed to use a generic approach. I came across with this solution:

#!/bin/bash
# Option Description:
# -------------------
#
# Option description is based on getopts bash builtin. The description adds a variable name feature to be used
# on future checks for required or optional values.
# The option description adds "=>VARIABLE_NAME" string. Variable name should be UPPERCASE. Valid characters
# are [A-Z_]*.
#
# A option description example:
#   OPT_DESC="a:=>A_VARIABLE|b:=>B_VARIABLE|c=>C_VARIABLE"
#
# -a option will require a value (the colon means that) and should be saved in variable A_VARIABLE.
# "|" is used to separate options description.
# -b option rule applies the same as -a.
# -c option doesn't require a value (the colon absense means that) and its existence should be set in C_VARIABLE
#
#   ~$ echo get_options ${OPT_DESC}
#   a:b:c
#   ~$
#


# Required options 
REQUIRED_DESC="a:=>REQ_A_VAR_VALUE|B:=>REQ_B_VAR_VALUE|c=>REQ_C_VAR_FLAG"

# Optional options (duh)
OPTIONAL_DESC="P:=>OPT_P_VAR_VALUE|r=>OPT_R_VAR_FLAG"

function usage {
    IFS="|"
    printf "%s" ${0}
    for i in ${REQUIRED_DESC};
    do
        VARNAME=$(echo $i | sed -e "s/.*=>//g")
    printf " %s" "-${i:0:1} $VARNAME"
    done

    for i in ${OPTIONAL_DESC};
    do
        VARNAME=$(echo $i | sed -e "s/.*=>//g")
        printf " %s" "[-${i:0:1} $VARNAME]"
    done
    printf "\n"
    unset IFS
    exit
}

# Auxiliary function that returns options characters to be passed
# into 'getopts' from a option description.
# Arguments:
#   $1: The options description (SEE TOP)
#
# Example:
#   OPT_DESC="h:=>H_VAR|f:=>F_VAR|P=>P_VAR|W=>W_VAR"
#   OPTIONS=$(get_options ${OPT_DESC})
#   echo "${OPTIONS}"
#
# Output:
#   "h:f:PW"
function get_options {
    echo ${1} | sed -e "s/\([a-zA-Z]\:\?\)=>[A-Z_]*|\?/\1/g"
}

# Auxiliary function that returns all variable names separated by '|'
# Arguments:
#       $1: The options description (SEE TOP)
#
# Example:
#       OPT_DESC="h:=>H_VAR|f:=>F_VAR|P=>P_VAR|W=>W_VAR"
#       VARNAMES=$(get_values ${OPT_DESC})
#       echo "${VARNAMES}"
#
# Output:
#       "H_VAR|F_VAR|P_VAR|W_VAR"
function get_variables {
    echo ${1} | sed -e "s/[a-zA-Z]\:\?=>\([^|]*\)/\1/g"
}

# Auxiliary function that returns the variable name based on the
# option passed by.
# Arguments:
#   $1: The options description (SEE TOP)
#   $2: The option which the variable name wants to be retrieved
#
# Example:
#   OPT_DESC="h:=>H_VAR|f:=>F_VAR|P=>P_VAR|W=>W_VAR"
#   H_VAR=$(get_variable_name ${OPT_DESC} "h")
#   echo "${H_VAR}"
#
# Output:
#   "H_VAR"
function get_variable_name {
    VAR=$(echo ${1} | sed -e "s/.*${2}\:\?=>\([^|]*\).*/\1/g")
    if [[ ${VAR} == ${1} ]]; then
        echo ""
    else
        echo ${VAR}
    fi
}

# Gets the required options from the required description
REQUIRED=$(get_options ${REQUIRED_DESC})

# Gets the optional options (duh) from the optional description
OPTIONAL=$(get_options ${OPTIONAL_DESC})

# or... $(get_options "${OPTIONAL_DESC}|${REQUIRED_DESC}")

# The colon at starts instructs getopts to remain silent
while getopts ":${REQUIRED}${OPTIONAL}" OPTION
do
    [[ ${OPTION} == ":" ]] && usage
    VAR=$(get_variable_name "${REQUIRED_DESC}|${OPTIONAL_DESC}" ${OPTION})
    [[ -n ${VAR} ]] && eval "$VAR=${OPTARG}"
done

shift $(($OPTIND - 1))

# Checks for required options. Report an error and exits if
# required options are missing.

# Using function version ...
VARS=$(get_variables ${REQUIRED_DESC})
IFS="|"
for VARNAME in $VARS;
do
    [[ -v ${VARNAME} ]] || usage
done
unset IFS

# ... or using IFS Version (no function)
OLDIFS=${IFS}
IFS="|"
for i in ${REQUIRED_DESC};
do
    VARNAME=$(echo $i | sed -e "s/.*=>//g")
    [[ -v ${VARNAME} ]] || usage
    printf "%s %s %s\n" "-${i:0:1}" "${!VARNAME:=present}" "${VARNAME}"
done
IFS=${OLDIFS}

I didn't test this roughly, so I could have some bugs in there.


POSIX 7 example

It is also worth checking the example from the standard: http://pubs.opengroup.org/onlinepubs/9699919799/utilities/getopts.html

aflag=
bflag=
while getopts ab: name
do
    case $name in
    a)    aflag=1;;
    b)    bflag=1
          bval="$OPTARG";;
    ?)   printf "Usage: %s: [-a] [-b value] args\n" $0
          exit 2;;
    esac
done
if [ ! -z "$aflag" ]; then
    printf "Option -a specified\n"
fi
if [ ! -z "$bflag" ]; then
    printf 'Option -b "%s" specified\n' "$bval"
fi
shift $(($OPTIND - 1))
printf "Remaining arguments are: %s\n" "$*"

And then we can try it out:

$ sh a.sh
Remaining arguments are: 
$ sh a.sh -a
Option -a specified
Remaining arguments are: 
$ sh a.sh -b
No arg for -b option
Usage: a.sh: [-a] [-b value] args
$ sh a.sh -b myval
Option -b "myval" specified
Remaining arguments are: 
$ sh a.sh -a -b myval
Option -a specified
Option -b "myval" specified
Remaining arguments are: 
$ sh a.sh remain
Remaining arguments are: remain
$ sh a.sh -- -a remain
Remaining arguments are: -a remain

Tested in Ubuntu 17.10, sh is dash 0.5.8.


The example packaged with getopt (my distro put it in /usr/share/getopt/getopt-parse.bash) looks like it covers all of your cases:

#!/bin/bash

# A small example program for using the new getopt(1) program.
# This program will only work with bash(1)
# An similar program using the tcsh(1) script language can be found
# as parse.tcsh

# Example input and output (from the bash prompt):
# ./parse.bash -a par1 'another arg' --c-long 'wow!*\?' -cmore -b " very long "
# Option a
# Option c, no argument
# Option c, argument `more'
# Option b, argument ` very long '
# Remaining arguments:
# --> `par1'
# --> `another arg'
# --> `wow!*\?'

# Note that we use `"$@"' to let each command-line parameter expand to a 
# separate word. The quotes around `$@' are essential!
# We need TEMP as the `eval set --' would nuke the return value of getopt.
TEMP=`getopt -o ab:c:: --long a-long,b-long:,c-long:: \
     -n 'example.bash' -- "$@"`

if [ $? != 0 ] ; then echo "Terminating..." >&2 ; exit 1 ; fi

# Note the quotes around `$TEMP': they are essential!
eval set -- "$TEMP"

while true ; do
    case "$1" in
        -a|--a-long) echo "Option a" ; shift ;;
        -b|--b-long) echo "Option b, argument \`$2'" ; shift 2 ;;
        -c|--c-long) 
            # c has an optional argument. As we are in quoted mode,
            # an empty parameter will be generated if its optional
            # argument is not found.
            case "$2" in
                "") echo "Option c, no argument"; shift 2 ;;
                *)  echo "Option c, argument \`$2'" ; shift 2 ;;
            esac ;;
        --) shift ; break ;;
        *) echo "Internal error!" ; exit 1 ;;
    esac
done
echo "Remaining arguments:"
for arg do echo '--> '"\`$arg'" ; done

Use getopt

Why getopt?

To parse elaborated command-line arguments to avoid confusion and clarify the options we are parsing so that reader of the commands can understand what's happening.

What is getopt?

getopt is used to break up (parse) options in command lines for easy parsing by shell procedures, and to check for legal options. It uses the GNU getopt(3) routines to do this.

getopt can have following types of options.

  1. No-value options
  2. key-value pair options

Note: In this document, during explaining syntax:

  • Anything inside [ ] is optional parameter in the syntax/examples.
  • is a place holder, which mean it should be substituted with an actual value.

HOW TO USE getopt?

Syntax: First Form

getopt optstring parameters

Examples:

# This is correct
getopt "hv:t::" "-v 123 -t123"  
getopt "hv:t::" "-v123 -t123"  # -v and 123 doesn't have whitespace

# -h takes no value.
getopt "hv:t::" "-h -v123"


# This is wrong. after -t can't have whitespace.
# Only optional params cannot have whitespace between key and value
getopt "hv:t::" "-v 123 -t 123"

# Multiple arguments that takes value.
getopt "h:v:t::g::" "-h abc -v 123 -t21"

# Multiple arguments without value
# All of these are correct
getopt "hvt" "-htv"
getopt "hvt" "-h -t -v"
getopt "hvt" "-tv -h"

Here h,v,t are the options and -h -v -t is how options should be given in command-line.

  1. 'h' is a no-value option.
  2. 'v:' implies that option -v has value and is a mandatory option. ':' means has a value.
  3. 't::' implies that option -t has value but is optional. '::' means optional.

In optional param, value cannot have whitespace separation with the option. So, in "-t123" example, -t is option 123 is value.

Syntax: Second Form

getopt [getopt_options] [--] [optstring] [parameters]

Here after getopt is split into five parts

  • The command itself i.e. getopt
  • The getopt_options, it describes how to parse the arguments. single dash long options, double dash options.
  • --, separates out the getopt_options from the options you want to parse and the allowed short options
  • The short options, is taken immediately after -- is found. Just like the Form first syntax.
  • The parameters, these are the options that you have passed into the program. The options you want to parse and get the actual values set on them.

Examples

getopt -l "name:,version::,verbose" -- "n:v::V" "--name=Karthik -version=5.2 -verbose"

Syntax: Third Form

getopt [getopt_options] [-o options] [--] [optstring] [parameters]

Here after getopt is split into five parts

  • The command itself i.e. getopt
  • The getopt_options, it describes how to parse the arguments. single dash long options, double dash options.
  • The short options i.e. -o or --options. Just like the Form first syntax but with option "-o" and before the "--" (double dash).
  • --, separates out the getopt_options from the options you want to parse and the allowed short options
  • The parameters, these are the options that you have passed into the program. The options you want to parse and get the actual values set on them.

Examples

getopt -l "name:,version::,verbose" -a -o "n:v::V" -- "-name=Karthik -version=5.2 -verbose"

GETOPT_OPTIONS

getopt_options changes the way command-line params are parsed.

Below are some of the getopt_options

Option: -l or --longoptions

Means getopt command should allow multi-character options to be recognised. Multiple options are separated by comma.

For example, --name=Karthik is a long option sent in command line. In getopt, usage of long options are like

getopt "name:,version" "--name=Karthik"

Since name: is specified, the option should contain a value

Option: -a or --alternative

Means getopt command should allow long option to have a single dash '-' rather than double dash '--'.

Example, instead of --name=Karthik you could use just -name=Karthik

getopt "name:,version" "-name=Karthik"

A complete script example with the code:

#!/bin/bash

# filename: commandLine.sh
# author: @theBuzzyCoder

showHelp() {
# `cat << EOF` This means that cat should stop reading when EOF is detected
cat << EOF  
Usage: ./installer -v <espo-version> [-hrV]
Install Pre-requisites for EspoCRM with docker in Development mode

-h, -help,          --help                  Display help

-v, -espo-version,  --espo-version          Set and Download specific version of EspoCRM

-r, -rebuild,       --rebuild               Rebuild php vendor directory using composer and compiled css using grunt

-V, -verbose,       --verbose               Run script in verbose mode. Will print out each step of execution.

EOF
# EOF is found above and hence cat command stops reading. This is equivalent to echo but much neater when printing out.
}


export version=0
export verbose=0
export rebuilt=0

# $@ is all command line parameters passed to the script.
# -o is for short options like -v
# -l is for long options with double dash like --version
# the comma separates different long options
# -a is for long options with single dash like -version
options=$(getopt -l "help,version:,verbose,rebuild,dryrun" -o "hv:Vrd" -a -- "$@")

# set --:
# If no arguments follow this option, then the positional parameters are unset. Otherwise, the positional parameters 
# are set to the arguments, even if some of them begin with a ‘-’.
eval set -- "$options"

while true
do
case $1 in
-h|--help) 
    showHelp
    exit 0
    ;;
-v|--version) 
    shift
    export version=$1
    ;;
-V|--verbose)
    export verbose=1
    set -xv  # Set xtrace and verbose mode.
    ;;
-r|--rebuild)
    export rebuild=1
    ;;
--)
    shift
    break;;
esac
shift
done

Running this script file:

# With short options grouped together and long option
# With double dash '--version'

bash commandLine.sh --version=1.0 -rV
# With short options grouped together and long option
# With single dash '-version'

bash commandLine.sh -version=1.0 -rV

# OR with short option that takes value, value separated by whitespace
# by key

bash commandLine.sh -v 1.0 -rV

# OR with short option that takes value, value without whitespace
# separation from key.

bash commandLine.sh -v1.0 -rV

# OR Separating individual short options

bash commandLine.sh -v1.0 -r -V

The problem with the original code is that:

  • h: expects parameter where it shouldn't, so change it into just h (without colon)
  • to expect -p any_string, you need to add p: to the argument list

Basically : after the option means it requires the argument.


The basic syntax of getopts is (see: man bash):

getopts OPTSTRING VARNAME [ARGS...]

where:

  • OPTSTRING is string with list of expected arguments,

    • h - check for option -h without parameters; gives error on unsupported options;
    • h: - check for option -h with parameter; gives errors on unsupported options;
    • abc - check for options -a, -b, -c; gives errors on unsupported options;
    • :abc - check for options -a, -b, -c; silences errors on unsupported options;

      Notes: In other words, colon in front of options allows you handle the errors in your code. Variable will contain ? in the case of unsupported option, : in the case of missing value.

  • OPTARG - is set to current argument value,

  • OPTERR - indicates if Bash should display error messages.

So the code can be:

#!/usr/bin/env bash
usage() { echo "$0 usage:" && grep " .)\ #" $0; exit 0; }
[ $# -eq 0 ] && usage
while getopts ":hs:p:" arg; do
  case $arg in
    p) # Specify p value.
      echo "p is ${OPTARG}"
      ;;
    s) # Specify strength, either 45 or 90.
      strength=${OPTARG}
      [ $strength -eq 45 -o $strength -eq 90 ] \
        && echo "Strength is $strength." \
        || echo "Strength needs to be either 45 or 90, $strength found instead."
      ;;
    h | *) # Display help.
      usage
      exit 0
      ;;
  esac
done

Example usage:

$ ./foo.sh 
./foo.sh usage:
    p) # Specify p value.
    s) # Specify strength, either 45 or 90.
    h | *) # Display help.
$ ./foo.sh -s 123 -p any_string
Strength needs to be either 45 or 90, 123 found instead.
p is any_string
$ ./foo.sh -s 90 -p any_string
Strength is 90.
p is any_string

See: Small getopts tutorial at Bash Hackers Wiki


"getops" and "getopt" are very limited. While "getopt" is suggested not to be used at all, it does offer long options. Where as "getopts" does only allow single character options such as "-a" "-b". There are a few more disadvantages when using either one.

So i've written a small script that replaces "getopts" and "getopt". It's a start, it could probably be improved a lot.

Update 08-04-2020: I've added support for hyphens e.g. "--package-name".

Usage: "./script.sh package install --package "name with space" --build --archive"

# Example:
# parseArguments "${@}"
# echo "${ARG_0}" -> package
# echo "${ARG_1}" -> install
# echo "${ARG_PACKAGE}" -> "name with space"
# echo "${ARG_BUILD}" -> 1 (true)
# echo "${ARG_ARCHIVE}" -> 1 (true)
function parseArguments() {
  PREVIOUS_ITEM=''
  COUNT=0
  for CURRENT_ITEM in "${@}"
  do
    if [[ ${CURRENT_ITEM} == "--"* ]]; then
      printf -v "ARG_$(formatArgument "${CURRENT_ITEM}")" "%s" "1" # could set this to empty string and check with [ -z "${ARG_ITEM-x}" ] if it's set, but empty.
    else
      if [[ $PREVIOUS_ITEM == "--"* ]]; then
        printf -v "ARG_$(formatArgument "${PREVIOUS_ITEM}")" "%s" "${CURRENT_ITEM}"
      else
        printf -v "ARG_${COUNT}" "%s" "${CURRENT_ITEM}"
      fi
    fi

    PREVIOUS_ITEM="${CURRENT_ITEM}"
    (( COUNT++ ))
  done
}

# Format argument.
function formatArgument() {
  ARGUMENT="${1^^}" # Capitalize.
  ARGUMENT="${ARGUMENT/--/}" # Remove "--".
  ARGUMENT="${ARGUMENT//-/_}" # Replace "-" with "_".
  echo "${ARGUMENT}"
}