Argparse Pydantic

Config for argparse with pydantic model.

Simple wrapper for python argparse. Use pydantic model for you app config. It gives you typed config instead of default Namespace from argparse.

Tested on python 3.8 - 3.12

WIP

Install

Install from pypi:

pip install argparse_pydantic

Or install from github repo:

pip install git+https://github.com/ayasyrev/argparse_pydantic.git

Base use.

We use python argparse to parse arguments from command line. So, just create parser as usual:

import argparse

parser = argparse.ArgumentParser(prog="MyApp")

Than create config for you app with Pydantic.

from pydantic import  BaseModel


class AppCfg(BaseModel):
    echo: str

Now add argument to parser from you config.

from argparse_pydantic import add_args_from_model

parser = add_args_from_model(parser, AppCfg)

So we got parser with arguments from config.

It exactly like parser made classic way: parser.add_argument("echo")

Now we can use parser in you script usual way - parser.parse_args()

$ python my_app.py -h
usage: MyApp [-h] echo

positional arguments:
  echo

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

Parse command line as usual.

args = parser.parse_args(["argument from command line"])

When we parse command line, we got Namespace object. Bat we can convert it to config object.

from argparse_pydantic import create_model_obj

cfg = create_model_obj(AppCfg, args)

Now we got config with type checks / validation ont type hinting when use it at IDE.

cfg

AppCfg(echo='argument from command line')

cfg.echo

'argument from command line'

Optional if undefined.

We can use undefined arguments as positional or optional (but required).

class AppCfg2(BaseModel):
    arg_int: int
    arg_float: float = 0.1

parser = argparse.ArgumentParser(prog="MyApp")
parser = add_args_from_model(parser, AppCfg2, undefined_positional=False)

$ python my_app.py -h
usage: MyApp [-h] --arg_int ARG_INT [--arg_float ARG_FLOAT]

options:
  -h, --help            show this help message and exit
  --arg_int ARG_INT
  --arg_float ARG_FLOAT

Add types and defaults values.

And we can add type hints to help message from our config.

parser = argparse.ArgumentParser(prog="MyApp")
parser = add_args_from_model(
    parser,
    AppCfg2,
    undefined_positional=False,
    help_def_type=True,
)

$ python my_app.py -h
usage: MyApp [-h] --arg_int ARG_INT [--arg_float ARG_FLOAT]

options:
  -h, --help            show this help message and exit
  --arg_int ARG_INT     [int]
  --arg_float ARG_FLOAT
                        [float] default: 0.1

Examples

You can see examples at examples folder - Same examples as at python docs and tutorial for argparse.