GitHub | PyPI | Docs | Examples | Installation | Changelog
Arguments parsing without boilerplate.
- arguments and type hints in IDE
- easy nested sub-commands
- sane defaults for arguments' params (ie if default of arg is 3 then type should be int, or when annotation/type/default is
bool
then generate 2 arguments: for true value--arg
and for false--no-arg
, ...) - 𝕡𝕣𝕖𝕥𝕥𝕪 𝕡𝕣𝕚𝕟𝕥𝕚𝕟𝕘
- support for argparse actions
- common options/arguments reusability
- auto shortcuts generation:
--verbose -> -v, --foo_bar -> --fb
- auto completion in shell (tnx to argcomplete)
pip install argser
pip install argser[tabulate] # for fancy tables support
pip install argser[argcomplete] # for shell auto completion
pip install argser[all]
If second parameter of parse_args
is string (as in almost all examples) then it will be parsed,
otherwise arguments to parse will be taken from command line.
from argser import parse_args
class Args:
a = 'a'
foo = 1
bar: bool
bar_baz = 42, "bar_baz help"
args = parse_args(Args, show=True)
argparse alternative
from argparse import ArgumentParser
parser = ArgumentParser()
parser.add_argument('-a', type=str, default='a', help="str, default: 'a'")
parser.add_argument('--foo', '-f', dest='foo', type=int, default=1, help="int, default: 1")
parser.add_argument('--bar', '-b', dest='bar', action='store_true', help="bool, default: None")
parser.add_argument('--no-bar', '--no-b', dest='bar', action='store_false')
parser.set_defaults(bar=None)
parser.add_argument('--bar-baz', dest='bar_baz', default=42, help="int, default: 42. bar_baz help")
args = parser.parse_args()
print(args)
❯ python playground.py -a "aaa bbb" -f 100500 --no-b
>>> Args(bar=False, a='aaa bbb', foo=100500, bar_baz=42)
❯ python playground.py -h
usage: playground.py [-h] [--bar] [--no-bar] [-a A] [--foo F] [--bar-baz B]
optional arguments:
-h, --help show this help message and exit
--bar, -b bool, default: None
--no-bar, --no-b
-a A str, default: 'a'
--foo F, -f F int, default: 1
--bar-baz B, --bb B int, default: 42. bar_baz help
import argser
def foo(a, b: int, c=1.2):
return [a, b, c]
assert argser.call(foo, '1 2 -c 3.4') == ['1', 2, 3.4]
from argser import parse_args, sub_command
class Args:
a: bool
b = []
c = 5
class SubArgs:
d = 1
e = '2'
sub = sub_command(SubArgs, help='help message for sub-command')
args = parse_args(Args, '-a -b a b -c 10', parser_help='help message for root parser')
assert args.a is True
assert args.b == ['a', 'b']
assert args.c == 10
assert args.sub is None
args = parse_args(Args, '--no-a -c 10 sub -d 5 -e "foo bar"')
assert args.a is False
assert args.sub.d == 5
assert args.sub.e == 'foo bar'
❯ python playground.py -h
usage: playground.py [-h] [-a] [--no-a] [-b [B [B ...]]] [-c C] {sub} ...
positional arguments:
{sub}
optional arguments:
-h, --help show this help message and exit
-a bool, default: None
--no-a
-b [B [B ...]] List[str], default: []
-c C int, default: 5
❯ python playground.py sub1 -h
usage: playground.py sub [-h] [-d D] [-e E]
help message for sub-command
optional arguments:
-h, --help show this help message and exit
-d D int, default: 1
-e E str, default: '2'
Can be deep nested:
from argser import parse_args, sub_command
class Args:
a = 1
class Sub1:
b = 1
class Sub2:
c = 1
class Sub3:
d = 1
sub3 = sub_command(Sub3)
sub2 = sub_command(Sub2)
sub1 = sub_command(Sub1)
args = parse_args(Args, '-a 1 sub1 -b 2 sub2 -c 3 sub3 -d 4')
import argser
subs = argser.SubCommands()
@subs.add
def foo():
return 'foo'
@subs.add(description="foo bar") # with additional arguments for sub-parser
def bar(a, b=1):
return [a, b]
assert subs.parse('foo') == 'foo'
assert subs.parse('bar 1 -b 2') == ['1', 2]