moebrowne / bash-argument-parser Goto Github PK

When including in a project which has set -e the library fails to run

Add the ability to define what the type/format of an argument is, for example you could define an argument to only be valid if its only numbers or at least 4 characters long or even better match a regex pattern

Add the ability to turn an argument into an array when its repeatedly passed.
For example --set 1 --set 2 --set 3 would cause argValue 'set' to return (1,2,3) as an array

Currently if you leave the description of an argument blank it will still be shown in the usage, it should be skipped instead

The README doesn't mention about the argList function.

Currently the order of the arguments in the usage output is somewhat arbitrary need to order A-Z.

Best way to approach it may be to get all the keys of the argument associative array order them using sort then loop over each one

Currently there is no test to check the usage output is generated correctly

Add the functionality to give arguments a default value

If you leave the description of an argument blank the default value is ignored

A common pattern is to define the -h and --help arguments to show the argument usage, it would be useful if this was included by default maybe with an opt out.

Implementation would look something like this:

argExpected['h|help']="showHelp; Show this help message"

argPassed 'showHelp' && { echo "$(argList)"; exit 0; }

It would be good to support the -- argument to mark the end of the argument list, and perhaps assign $1, $2, etc.. to any words that follow?

Add the functionality to set parameters as required

argument parser doesn't work when executed from a function.

_arguments() {
  # Define the expected arguments
  declare -A argExpected
  argExpected['h|help']="help - print this message"
  argExpected['debug']="debug - enable debugging"
  argExpected['o|option']="option - some option"

  source argument-parser.sh
}
_main() {
  _arguments "$@"
  argPassed 'debug' && _debug
  argPassed 'help' && _print_help
  _script "$@"
}

Currently when the tests run there is no programmatic way to tell if they passed or not, they should return a 0 exit code if all is okay and a non-zero otherwise