Blargparse extends ``argparse.ArgumentParser``,
adding an ``add_aggregate`` method that can help you separate
your user interface logic from everything else.

How to use
-----------

Install from PyPI. ::

    pip install blargparse

Here's the simplest way to use it. ::

    b = BlargParser()
    b.add_argument('--left', '-l', type = int, default = 0)
    b.add_argument('--right', 'r', type = int, default = 100)
    b.add_aggregate('numbers', lambda args: range(args.left, args.right))

    b.parse_args([]).numbers == range(0, 100)

You can get a bit fancier. ::

    b = BlargParser()
    b.add_argument('--left', '-l', type = int, default = 0)
    b.add_argument('--right', 'r', type = int, default = 100)
    def f(args):
        output = range(args.left, args.right)
        del(args.left, args.right)
    b.add_aggregate(f)

    b.parse_args([]).numbers == range(0, 100)

Why
--------
I wrote Blargparse because I found myself doing stuff like this. ::

    # Construct the parser.
    a = argparse.ArgumentParser()
    a.add_argument('--left', '-l', type = int, default = 0)
    a.add_argument('--right', 'r', type = int, default = 100)
    
    # Parse the arguments
    import sys
    ns = a.parse_args(sys.argv)

    # Convert them into the form I want for my application.
    numbers = range(ns.left, ns.right)

The input argument that made most sense to me while programming the
thing was the ``numbers`` range, but I thought it was easiest if the
end user could specify this range as a combination of ``--left`` and
``--right`` flags. And I might want to change that based on what
users seem to prefer.

The above approach is fine because it's small, but I had several
arguments like this spread across several subparsers in different places,
and the construction, parsing, and conversion of the arguments thus
wound up spread across functions and even files.

With vanilla ``argparse`` I wound up doing stuff like this. ::

    def _build_subparsers(argparser):
        'This adds subparsers to the parser.'
        subparsers = argparser.add_subparsers()
        # ...

    def _apply_aggregates(args):
        args.range = range(args.left, args.right)
        del(args.left, args.right)

    argparser = argparser.ArgumentParser()
    _build_subparsers(argparser)
    args = argparser.parse_args()
    _apply_aggregates(args)

This is a bit better except that it groups subparsers together and
aggregates together. I wanted subparsers to be grouped with relevant
aggregates, and I did not want to group by whether something was a
subparser or an aggregate; I think of the aggregates as components of
the subparsers.

A more confusing alternative would have been to come up with my own
non-argparse interface on top of this for specifying groupings of
subparsers and aggregates.