dataparsers 1.0.0

A wrapper around argparse to get command line argument parsers from dataclasses

dataparsers

A simple module to wrap around argparse to get command line argument parsers from dataclasses.

Installation

pip install dataparsers

Basic usage

Create a dataclass describing your command line interface, and call parse() with the class:

# prog.py
from dataclasses import dataclass
from dataparsers import parse

@dataclass
class Args:
    foo: str
    bar: int = 42

args = parse(Args)
print("Printing `args`:")
print(args)

The dataclass fields that have a "default" value are turned into optional arguments, while the non default fields will be positional arguments.

The script can then be used in the same way as used with argparse:

$ python prog.py -h
usage: prog.py [-h] [--bar BAR] foo

positional arguments:
  foo

options:
  -h, --help  show this help message and exit
  --bar BAR

And the resulting type of args is Args (recognized by type checkers and autocompletes):

$ python prog.py test --bar 12
Printing `args`:
Args(foo='test', bar=12)

Argument specification

To specify detailed information about each argument, call the arg() function on the dataclass fields:

# prog.py
from dataclasses import dataclass
from dataparsers import parse, arg

@dataclass
class Args:
    foo: str = arg(help="foo help")
    bar: int = arg(default=42, help="bar help")

args = parse(Args)

It allows to customize the interface:

$ python prog.py -h
usage: prog.py [-h] [--bar BAR] foo

positional arguments:
  foo         foo help

options:
  -h, --help  show this help message and exit
  --bar BAR   bar help

In general, the arg() function accepts all parameters that are used in the original add_argument() method (with few exceptions) and some additional parameters. The default keyword argument used above makes the argument optional (i.e., passed with flags like --bar) except in some specific situations.

For more information, see the documentation.

Formalities, features, benefits and drawbacks

This project basically consists of a simple module dataparsers.py with a few functions that allows to define typed arguments parsers in a single place, based on dataclasses.

Formalities

The main strategy of the module is based on the same approach of the package datargs, which consists in using the metadata attribute of the dataclass fields to store argument parameters. Some additional features of this project have already been contributed back upstream.

There are a lot of alternative libraries out there that do similar things. The README file of the datargs repository provides a great list for existing solutions and differences. I could also add to that list the libraries Python fire and dargparser, just to give few examples.

Features and benefits

Use this project if you want particular added features, such as:

• More control over the interface display
• More control over the argument flag creation
• Support for argument groups and mutually exclusive argument groups
• More simplicity

The simplicity is mentioned because it is just a simple module dataparsers.py that doesn't have additional dependencies (it is pure Python) which can be downloaded directly and placed in your CLI scripts folder to import from.

In deed, the module consists of a 200 lines IPython code cell region (which starts and ends with a #%% line comment block), that can also be placed on top of your "single file" CLI script to directly distribute. The used names are just the few provided functions, the stdlib imports and Class (a TypeVar)

Additionally, this project also provides a stub file [dataparsers.pyi] that can be used by type checkers but, moreover, may be used by some code editors to give helper documentation including the related docs of argparse methods, which are also provided in this current documentation for convenience. The stub can be downloaded directly but it is installed with the module by default.

Drawbacks

Unlike the datargs package, dataparsers doesn't support:

• The library attrs (only works with pure python dataclasses)
• Enums
• Complex types (Sequences, Optionals, and Literals)
• Sub Commands (subparsers)

If you want any of these features, use the datargs package. If you need the added features of dataparsers, use this module instead.