dlwh / simpleparsing

Simple, Elegant, Typed Argument Parsing with argparse

Simple, Elegant, Typed Argument Parsing

simple-parsing allows you to transform your ugly argparse scripts into beautifully structured, strongly typed little works of art. This isn't a fancy, complicated new command-line tool either, this simply adds new features to plain-old argparse! Using dataclasses, simple-parsing makes it easier to share and reuse command-line arguments - no more copy pasting!

Supports inheritance, nesting, easy serialization to json/yaml, automatic help strings from comments, and much more!

# examples/demo.py
from dataclasses import dataclass
from simple_parsing import ArgumentParser

parser = ArgumentParser()
parser.add_argument("--foo", type=int, default=123, help="foo help")

@dataclass
class Options:
    """ Help string for this group of command-line arguments """
    log_dir: str                # Help string for a required str argument
    learning_rate: float = 1e-4 # Help string for a float argument

parser.add_arguments(Options, dest="options")

args = parser.parse_args()
print("foo:", args.foo)
print("options:", args.options)

$ python examples/demo.py --log_dir logs --foo 123
foo: 123
options: Options(log_dir='logs', learning_rate=0.0001)

$ python examples/demo.py --help
usage: demo.py [-h] [--foo int] --log_dir str [--learning_rate float]

optional arguments:
  -h, --help            show this help message and exit
  --foo int             foo help (default: 123)

Options ['options']:
   Help string for this group of command-line arguments

  --log_dir str         Help string for a required str argument (default:
                        None)
  --learning_rate float
                        Help string for a float argument (default: 0.0001)

(new) Simplified API:

For a simple use-case, where you only want to parse a single dataclass, you can use the simple_parsing.parse or simple_parsing.parse_known_args functions:

options: Options = simple_parsing.parse(Options)
# or:
options, leftover_args = simple_parsing.parse_known_args(Options)

installation

pip install simple-parsing

Examples

API Documentation (Under construction)

Features

• Automatic "--help" strings

  As developers, we want to make it easy for people coming into our projects to understand how to run them. However, a user-friendly --help message is often hard to write and to maintain, especially as the number of arguments increases.

  With simple-parsing, your arguments and their descriptions are defined in the same place, making your code easier to read, write, and maintain.

• Modular, Reusable, Cleanly Grouped Arguments

  (no more copy-pasting)

  When you need to add a new group of command-line arguments similar to an existing one, instead of copy-pasting a block of argparse code and renaming variables, you can reuse your argument class, and let the ArgumentParser take care of adding relevant prefixes to the arguments for you:

  parser.add_arguments(Options, dest="train")
  parser.add_arguments(Options, dest="valid")
  args = parser.parse_args()
  train_options: Options = args.train
  valid_options: Options = args.valid
  print(train_options)
  print(valid_options)

  $ python examples/demo.py \
      --train.log_dir "training" \
      --valid.log_dir "validation"
  Options(log_dir='training', learning_rate=0.0001)
  Options(log_dir='validation', learning_rate=0.0001)

  These prefixes can also be set explicitly, or not be used at all. For more info, take a look at the Prefixing Guide

• Argument subgroups

  It's easy to choose between different argument groups of arguments, with the subgroups function!

• Setting defaults from Configuration files

  Default values for command-line arguments can easily be read from many different formats, including json/yaml!

• Easy serialization:

  Easily save/load configs to json or yaml!.

• Inheritance!

  You can easily customize an existing argument class by extending it and adding your own attributes, which helps promote code reuse across projects. For more info, take a look at the inheritance example

• Nesting!:

  Dataclasses can be nested within dataclasses, as deep as you need!

• Easier parsing of lists and tuples :

  This is sometimes tricky to do with regular argparse, but simple-parsing makes it a lot easier by using the python's builtin type annotations to automatically convert the values to the right type for you. As an added feature, by using these type annotations, simple-parsing allows you to parse nested lists or tuples, as can be seen in this example

• Enums support

• (More to come!)

Examples:

Additional examples for all the features mentioned above can be found in the examples folder