CodeSnippets: cmdparser - parse command line arguments [regex] [shell] [osx] [mac] [bash] [command] [line] [option] [parse] [basic] [parameter] [parser] [switch] [argument] [flag] [getopts] [getopt] [posix] [gnu] [cmdparser]

Author: jv
License: The MIT License, Copyright (c) 2008 jv
Description: a basic regex-based command line parser for use in bash scripts (Mac OS X); an alternative to the builtin getopts command (cf. help getopts); use at your own risk
Usage: /path/to/script_with_cmdparser -a -b -c -f file

Related links:
- cmdparser - parse command line arguments with Bash built-in regex matching
- Process positional parameters non-destructively in Bash
- ws - search the web from the command line (an example of using cmdparser)

Version 1: modifies (the number of) command line arguments ($# and $@).


#!/bin/bash

export PATH=/usr/bin:/bin:/usr/sbin:/sbin
export IFS=$' \t\n'

# create a fake command line
set -- -abcc -c -zz -flag1="" -flag2=arg -flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' -flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces  / * + ` \ ! ' -flag9 ~/Desktop/*.txt filename1 filename2 filename3

#set -- -abcc -c -zz -flag1="" -flag2=arg -flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' -flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ ! ' -flag9 '~/Desktop/*.txt' filename1 filename2 filename3


printf "%s\n" "$@" | nl
#printf "%s" "$@"$'\n' | nl
#printf "%s" "${@/%/ }" | nl


: <<-'COMMENT'

# copy & paste examples for the command line

echo "filename1" "filename2" "filename3" | ~/Desktop/cmdparser.txt -abcc -c -zz -flag1 arg -flag2=arg --flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' --flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ ! ' -flag9 ~/Desktop/*.txt -

echo "filename1" "filename2" "filename3" | ~/Desktop/cmdparser.txt -abcc -c -zz -flag1 arg -flag2=arg --flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' --flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ ! ' -flag9 '~/Desktop/*.txt' -

COMMENT



# cmdparser

usage="usage: $(/usr/bin/basename "$0") [-a] [-b] [-c] [-cc] [-zz] [-flag1 arg] [-flag2 'arg1 arg2 ...'] [-flag3=arg] [-flag4=\"arg1 arg2\"] ..."


# Note: names of flags and switches may not contain the characters "-", "." or "=".

# define the names of flags as a regular expression
# flags are command line options that require arguments

flags="(flag1|flag2|flag3|flag4|flag5|flag6|flag7|flag8|flag9)"


# define the names of switches as a regular expression
# Switches are command line options that do not take arguments.
# Make sure multi-char switches precede single-char switches in the regular expression.
# Note that the regular expression contains neither the special read-from-stdin switch "-" 
# nor the special end-of-options switch "--".

switches="(cc|zz|a|b|c)"  


declare flag1 flag2 flag3 flag4 flag5 flag6 flag7 flag8 flag9                # flags
declare -i a=0 b=0 c=0 cc=0 zz=0                                            # switches
                         
declare argstr argvar argvar_escaped char flagvar optstr piped pipedstr       # script variables
declare -i optid pipedvar

# piped="piped" will be used for variable creation 
# example: piped="piped"; pipedstr="piped arg"; eval $piped='"$(echo "$pipedstr")"'; echo "$piped"

piped="piped"

# default value is set to "no pipe"
pipedvar=0
pipedstr=""

# if /dev/stdin has a size greater than zero ...
if [[ -s /dev/stdin ]]; then pipedstr="$(</dev/stdin)"; fi 

if [[ $# -eq 0 ]] && [[ -z "$pipedstr" ]]; then
  printf "\n%s\n\n%s\n\n" 'No arguments specified!' "$usage" 1>&2
  exit 1
fi 

if [[ $# -eq 0 ]] && [[ -n "$pipedstr" ]]; then
  eval $piped='"${pipedstr}"'  
  pipedvar=1
fi 

# if there are command line arguments ...
# Note that $pipedvar may still be set to 1 below if the special read-from-stdin switch "-" is given.

if [[ $pipedvar -eq 0 ]]; then

   optstr=" "  
   optid=0

   while [[ -n "$optstr" ]]; do     

      # try to extract valid flags or switches from positional parameter $1
      # $1 gets shifted afterwards (cf. help shift)

      optstr="$(printf "%s" "$1" | /usr/bin/egrep -e "^--?${flags}$")"

      if [[ -n "$optstr" ]]; then optid=1; fi
      if [[ -z "$optstr" ]]; then optid=2; optstr="$(printf "%s" "$1" | /usr/bin/egrep -e "^--?${switches}$")"; fi
      if [[ -z "$optstr" ]]; then optid=3; optstr="$(printf "%s" "$1" | /usr/bin/egrep -e "^--?${switches}+$")"; fi
      if [[ -z "$optstr" ]]; then optid=4; optstr="$(printf "%s" "$1" | /usr/bin/egrep -e "^--?(${flags}=.*|${flags}[^[:space:]]+)$")"; fi


      if [[ -z "$optstr" ]]; then  
         if [[ "$1" = "-" ]] && [[ "$@" = "-" ]]; then  
            optid=5
            optstr="-"
         fi
      fi


      if [[ -z "$optstr" ]]; then

         # create argstr by appending a space to each command line argument
         argstr="$(printf "%s" "${@/%/ }")"
         argstr="$(printf "%s" "${argstr}" | /usr/bin/egrep -o -m 1 -e "[[:space:]]--?(${flags}|${switches})" | /usr/bin/head -n 1  )"

         if [[ -n "${argstr}" ]]; then 
            printf "\n%s\x21\n\n%s\n\n%s\n\n" "Undefined non-option string: ${1:0:1000} is followed by a legal flag or switch" "${argstr}" "$usage" 1>&2
            exit 1
         fi

      fi


      if [[ "$1" = "--" ]]; then shift; break; fi     # -- marks end of options

      if [[ -z "$optstr" ]]; then break; fi     # no further flags or switches to process


      # flag followed by space (example: -f file)
      if [[ $optid -eq 1 ]]; then 

         if [[ -z "$2" ]]; then
            printf "%s\n%s\n" "no argument given to flag: ${1}" "$usage" 1>&2
            exit 1
         fi 

         flagvar="${1#"${1%%[!-]*}"}"     # remove leading - or --
         argvar="$2"
         eval $flagvar='"${argvar}"'
         shift 2     # shift positional parameters $1 & $2 (that is, a flag plus its argument)
         continue

      # single switch (example: -a)
      elif [[ $optid -eq 2 ]]; then
         flagvar="${1#"${1%%[!-]*}"}"
         eval $flagvar='"1"'
         shift
         continue
  
      # combined switch (example: -abcc)
      elif [[ $optid -eq 3 ]]; then
         flagvar="${1#"${1%%[!-]*}"}"
         while [[ -n "$flagvar" ]]; do
            char="$(printf "%s" "$flagvar" | /usr/bin/sed -E "s/^${switches}.*$/\1/")"
            eval $char='"1"'
            flagvar="$(printf "%s" "$flagvar" | /usr/bin/sed -E "s/^${switches}//")"
         done
         shift
         continue

      # flag without following space (example: -ffile)
      elif [[ $optid -eq 4 ]]; then 

: <<-'COMMENT'

         argvar="$(printf "%s" "$1" | /usr/bin/sed -E "s/^\-\-?${flags}\=?//")"

         argvar2="${argvar//\\\\/\\\\}"       # escape \  (for Bash version 2.05b.0(1)-release)
         #argvar2="${argvar//\\/\\\\}"          # escape \  

         flagvar="${1%${argvar2}}"          # remove escaped $argvar string
         flagvar="${flagvar%=}"             # remove trailing =
         flagvar="${flagvar#"${flagvar%%[!-]*}"}"           # remove leading - or --
         eval $flagvar='"${argvar}"'
         shift
         continue

COMMENT

         # alternative: no string escaping necessary
         #argvar="$(printf "%s" "$1" | /usr/bin/sed -E "s/^\-\-?${flags}\=?//")"
         #flagvar="$(printf "%s" "$1" | /usr/bin/sed -E -n -e "s/^\-\-?${flags}\=?.*$/\\1/p")"

         argvar="$(printf "%s" "${1#"${1%%[!-]*}"}" | /usr/bin/sed -E "s/^${flags}\=?//")"
         flagvar="$(printf "%s" "${1#"${1%%[!-]*}"}" | /usr/bin/sed -E -n -e "s/^${flags}\=?.*$/\\1/p")"

         eval $flagvar='"${argvar}"'
         shift
         continue


      # the special read-from-stdin switch "-"
      elif [[ $optid -eq 5 ]]; then 
         pipedvar=1
         eval $piped='"${pipedstr}"'
         shift
         break

      fi

      # remove positional parameter $1 from "$@"
      shift

   done

fi   # if [[$pipedvar -eq 0 ]]; then ...


echo 

printf "%s\t%s\n" "a:" "${a}"
printf "%s\t%s\n" "b:" "${b}"
printf "%s\t%s\n" "c:" "${c}"
printf "%s\t%s\n" "cc:" "${cc}"
printf "%s\t%s\n" "zz:" "${zz}"
printf "%s\t%s\n" "flag1:" "${flag1}"
printf "%s\t%s\n" "flag2:" "${flag2}"
printf "%s\t%s\n" "flag3:" "${flag3}"
printf "%s\t%s\n" "flag4:" "${flag4}"
printf "%s\t%s\n" "flag5:" "${flag5}"
printf "%s\t%s\n" "flag6:" "${flag6}"
printf "%s\t%s\n" "flag7:" "${flag7}"
printf "%s\t%s\n" "flag8:" "${flag8}"
printf "%s\t%s\n" "flag9:" "${flag9}"

echo


if [[ $pipedvar -eq 1 ]] && [[ -z "$@" ]]; then 
   echo "remaining string-piped: ${piped}"
else 
   echo "remaining string: ${@}"
fi

echo

if [[ $flag9 == '~/Desktop/*.txt' ]]; then printf "%s\n" ~/Desktop/*.txt | nl; fi

echo

exit 0

Version 2: does not modify (the number of) command line arguments ($# and $@).


#!/bin/bash

# create a fake command line
#set -- -abcc -c -zz -flag1="" -flag2=arg$'\n'plus_newline -flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' -flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ !' -flag9 ~/Desktop/*.txt filename1 filename2 filename3

#set -- -abcc -c -zz -flag1="" -flag2=arg$'\n'plus_newline -flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' -flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ !' -flag9 '~/Desktop/*.txt' filename1 filename2 filename3


printf "%s\n" "$@" | nl
#printf "%s" "$@"$'\n' | nl
#printf "%s" "${@/%/ }" | nl


: <<-'COMMENT'

# copy & paste examples

echo "filename1" "filename2" "filename3" | ~/Downloads/Mac-OS-X-bash-scripts/bash-cmdparser/cmdparser-non-destructive-1.txt -abcc -c -zz -flag1 arg -flag2=arg$'\n'plus_newline --flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' --flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ !' -flag9 ~/Desktop/*.txt -

echo "filename1" "filename2" "filename3" | ~/Downloads/Mac-OS-X-bash-scripts/bash-cmdparser/cmdparser-non-destructive-1.txt -abcc -c -zz -flag1 arg -flag2=arg$'\n'plus_newline --flag3="arg" -flag4='arg1=*,arg2=?,arg3=!' -flag5 '(arg1|arg2|arg3)' -flag6 'arg1=ag,arg2=bg,arg3=cg' --flag7 An\ argument\ with\ spaces\! -flag8='Yet another argument with spaces / * + ` \ !' -flag9 '~/Desktop/*.txt' -

COMMENT


echo

echo "Number of positional parameters: ${#}"

echo


# cmdparser

export PATH=/usr/bin:/bin:/usr/sbin:/sbin
export IFS=$' \t\n'


# non-builtin commands used
# cf. man builtin
declare basename=/usr/bin/basename egrep=/usr/bin/egrep sed=/usr/bin/sed


# Note: names of flags and switches may not contain the characters "-", "." or "=".

# define the names of flags as a regular expression
# flags are command line options that require arguments

flags="(flag1|flag2|flag3|flag4|flag5|flag6|flag7|flag8|flag9)"


# define the names of switches as a regular expression
# Switches are command line options that do not take arguments.
# Make sure multi-char switches precede single-char switches in the regular expression.
# Note that the regular expression contains neither the special read-from-stdin switch "-" 
# nor the special end-of-options switch "--".

switches="(cc|zz|a|b|c)"  


usage="usage: $(${basename} "$0") [-a] [-b] [-c] [-cc] [-zz] [-flag1 arg] [-flag2 'arg1 arg2 ...'] [-flag3=arg] [-flag4=\"arg1 arg2\"] ..."

declare flag1 flag2 flag3 flag4 flag5 flag6 flag7 flag8 flag9                # flags
declare -i a=0 b=0 c=0 cc=0 zz=0                                            # switches
                         
declare argn argstr argvar argvar_escaped char flagvar optstr piped pipedstr       # script variables
declare -i optid pipedvar
declare -a argarslice

# piped="piped" will be used for variable creation 
# example: piped="piped"; pipedstr="piped arg"; eval $piped='"$(echo "$pipedstr")"'; echo "$piped"

piped="piped"

# default value is set to "no pipe"
pipedvar=0
pipedstr=""

# if /dev/stdin has a size greater than zero ...
if [[ -s /dev/stdin ]]; then pipedstr="$(</dev/stdin)"; fi 

if [[ $# -eq 0 ]] && [[ -z "$pipedstr" ]]; then
  printf "\n%s\n\n%s\n\n" 'No arguments specified!' "$usage" 1>&2
  exit 1
fi 

if [[ $# -eq 0 ]] && [[ -n "$pipedstr" ]]; then
  eval $piped='"${pipedstr}"'  
  pipedvar=1
fi 

# if there are command line arguments ...
# Note that $pipedvar may still be set to 1 below if the special read-from-stdin switch "-" is given

if [[ $pipedvar -eq 0 ]]; then

   optstr=" "  
   optid=0


   # processing one positional parameter at a time without modifying $# or $@
   # Process positional parameters non-destructively in Bash, http://codesnippets.joyent.com/posts/show/1706

   for (( i=1; i <= $#; i++ )); do 


      argn="${@:${i}:1}"     # current positional parameter
                             # "${@:(${i}+1):1}": the positional parameter following the current one
                             # "${@:${i}}": all positional parameters starting with the current one


      ### if [[ ${argn:0:1} != '-' ]]; then break; fi   # every flag or switch has to have a leading -

      optstr="$(printf "%s" "${argn}" | ${egrep} -e "^--?${flags}$")"

      if [[ -n "$optstr" ]]; then optid=1; fi
      if [[ -z "$optstr" ]]; then optid=2; optstr="$(printf "%s" "${argn}" | ${egrep} -e "^--?${switches}$")"; fi
      if [[ -z "$optstr" ]]; then optid=3; optstr="$(printf "%s" "${argn}" | ${egrep} -e "^--?${switches}+$")"; fi
      if [[ -z "$optstr" ]]; then optid=4; optstr="$(printf "%s" "${argn}" | ${egrep} -e "^--?(${flags}=.*|${flags}[^[:space:]]+)$")"; fi


      if [[ -z "$optstr" ]]; then  

         if [[ "${argn}" = "-" ]] && [[ "${@:${i}}" = "-" ]]; then  
            optid=5
            optstr="-"
         fi

      fi


      if [[ -z "$optstr" ]]; then  

         # create argstr by appending a space to each command line argument
         argarslice=( "${@:${i}}" )
         argstr="$(printf "%s" "${argarslice[@]/%/ }")"
         argstr="$(printf "%s" "${argstr}" | ${egrep} -o -m 1 -e "[[:space:]]--?(${flags}|${switches})" | /usr/bin/head -n 1  )"

         if [[ -n "${argstr}" ]]; then 
            printf "\n%s\x21\n\n%s\n\n%s\n\n" "Undefined non-option string: ${argn:0:1000} is followed by a legal flag or switch" "${argstr}" "$usage" 1>&2
            exit 1
         fi

      fi


      if [[ "${argn}" = "--" ]]; then break; fi     # -- marks end of options

      if [[ -z "$optstr" ]]; then break; fi     # no further flags or switches to process


      # flag followed by space (example: -f file)
      if [[ $optid -eq 1 ]]; then 

         if [[ -z "${@:(${i}+1):1}" ]]; then
            printf "%s\n%s\n" "no argument given to flag: ${argn}" "$usage" 1>&2
            exit 1
         fi 

         flagvar="${argn#"${argn%%[!-]*}"}"     # remove leading dashes
         argvar="${@:(${i}+1):1}"
         eval $flagvar='"${argvar}"'
         let "i += 1"     # skip argument of current flag in next for loop
         continue

      # single switch (example: -a)
      elif [[ $optid -eq 2 ]]; then
         flagvar="${argn#"${argn%%[!-]*}"}"
         eval $flagvar='"1"'
         continue
  
      # combined switch (example: -abcc)
      elif [[ $optid -eq 3 ]]; then
         flagvar="${argn#"${argn%%[!-]*}"}"
         while [[ -n "$flagvar" ]]; do
            char="$(printf "%s" "$flagvar" | ${sed} -E "s/^${switches}.*$/\1/")"
            eval $char='"1"'
            flagvar="$(printf "%s" "$flagvar" | ${sed} -E "s/^${switches}//")"
         done
         continue

      # flag without following space (example: -ffile)
      elif [[ $optid -eq 4 ]]; then 

: <<-'COMMENT'

         argvar="$(printf "%s" "${argn}" | ${sed} -E "s/^\-\-?${flags}\=?//")"

         argvar2="${argvar//\\\\/\\\\}"       # escape \  (for Bash version 2.05b.0(1)-release)
         #argvar2="${argvar//\\/\\\\}"          # escape \  

         flagvar="${argn%${argvar2}}"                 # remove escaped $argvar string
         flagvar="${flagvar%=}"                            # remove trailing =
         flagvar="${argn#"${argn%%[!-]*}"}"             # remove leading - or --
         eval $flagvar='"${argvar}"'
         continue

COMMENT

        # alternative: no string escaping required
         #argvar="$(printf "%s" "${argn}" | ${sed} -E "s/^\-\-?${flags}\=?//")"
         #flagvar="$(printf "%s" "${argn}" | ${sed} -E -n -e "s/^\-\-?${flags}\=?.*$/\\1/p")"

         argvar="$(printf "%s" "${argn#"${argn%%[!-]*}"}" | ${sed} -E "s/^${flags}\=?//")"
         flagvar="$(printf "%s" "${argn#"${argn%%[!-]*}"}" | ${sed} -E -n -e "s/^${flags}\=?.*$/\\1/p")"

         eval $flagvar='"${argvar}"'
         continue


      # the special read-from-stdin switch "-"
      elif [[ $optid -eq 5 ]]; then 
         pipedvar=1
         eval $piped='"${pipedstr}"'
         break

      fi

   done   # for loop

fi   # if [[$pipedvar -eq 0 ]]; then ...


echo 

printf "%s\t%s\n" "a:" "${a}"
printf "%s\t%s\n" "b:" "${b}"
printf "%s\t%s\n" "c:" "${c}"
printf "%s\t%s\n" "cc:" "${cc}"
printf "%s\t%s\n" "zz:" "${zz}"
printf "%s\t%s\n" "flag1:" "${flag1}"
printf "%s\t%s\n" "flag2:" "${flag2}"
printf "%s\t%s\n" "flag3:" "${flag3}"
printf "%s\t%s\n" "flag4:" "${flag4}"
printf "%s\t%s\n" "flag5:" "${flag5}"
printf "%s\t%s\n" "flag6:" "${flag6}"
printf "%s\t%s\n" "flag7:" "${flag7}"
printf "%s\t%s\n" "flag8:" "${flag8}"
printf "%s\t%s\n" "flag9:" "${flag9}"

echo


if [[ $pipedvar -eq 1 ]] && [[ -z "$@" ]]; then 
   echo "remaining string-piped: ${piped}"
else 
   echo "remaining string: ${@}"
fi

echo

echo "Number of positional parameters: ${#}"

echo

if [[ $flag9 == '~/Desktop/*.txt' ]]; then printf "%s\n" ~/Desktop/*.txt | nl; fi

echo

exit 0

Further information:

- Command-line argument
- In the Beginning... was the Command Line
- Handling Command Line Arguments
- Utility Conventions - Utility Argument Syntax (POSIX)
- Utility Conventions - Utility Syntax Guidelines (POSIX)
- GNU coding standards: 4.7 Standards for Command Line Interfaces
- bash-getopts
- getoptx.bash (getopt function that parses long-named options)
- Processing Command-Line Options in the Shell
- Shell Corner: Processing Command-line Arguments with my_getopts
- my_getopts (Bash version)
- Parsing arguments for your shell script
- Bash: parsing arguments with 'getopts'
- More Power with Bash Getopts
- Getopt and getopts
- Option-ize your shell scripts
- Emulating getopt
- Positional Parameters
- Parsing Command Line Options in Shell Scripts
- Command Line Processing in Cocoa
- ddcli: An Objective-C Command Line Helper
- Command-line processing with getopt() (in C)
- Arg_parser

CodeSnippets

Never been to CodeSnippets before?

cmdparser - parse command line arguments (See related posts)

Related Posts

Joyent

CodeSnippets

Never been to CodeSnippets before?

cmdparser - parse command line arguments (See related posts)

Related Posts