argdantic

Typed command line interfaces with argparse and pydantic.

Features

argdantic provides a thin boilerplate layer to provide a modern CLI experience, including:

Typed arguments: arguments require full typing by default, enforcing clarity and help your editor provide better support (linting, hinting).
Nested models: exploit pydantic models to scale from simple primitives to complex nested configurations with little effort.
Nested commands: combine commands and build complex hierarchies to build complex interfaces.
Validation by default: thanks to pydantic, field validation is provided by default, with the desired complexity.
Multiple sources: arguments can be provided from multiple sources, including environment variables, JSON, TOML and YAML files.

Quickstart

Installation

Installing argdantic can be done from source, or simply using pip. The only required dependency is, of course, pydantic, while the remaining can be selected depending on your needs:

recommended choice: install everything
this includes orjson, pyyaml, tomli, python-dotenv
user@pc:~$ pip install argdantic[all]

env, json, toml or yaml dependencies
user@pc:~$ pip install argdantic[env|json|toml|yaml]

minimum requirement, only pydantic included
user@pc:~$ pip install argdantic

A Simple Example

Creating a CLI with argdantic can be as simple as:

from argdantic import ArgParser

# 1. create a CLI instance
parser = ArgParser()


# 2. decorate the function to be called
@parser.command()
def buy(name: str, quantity: int, price: float):
    print(f"Bought {quantity} {name} at ${price:.2f}.")

# 3. Use your CLI by simply calling it
if __name__ == "__main__":
    parser()

Then, in a terminal, the help command can provide the usual information:

$ python cli.py --help
> usage: buy [-h] --name TEXT --quantity INT --price FLOAT
>
> optional arguments:
>   -h, --help      show this help message and exit
>   --name TEXT
>   --quantity INT
>   --price FLOAT

This gives us the required arguments for the execution:

$ python cli.py --name apples --quantity 10 --price 3.4
> Bought 10 apples at $3.40.

Using Models

Plain arguments and pydantic models can be mixed together:

from argdantic import ArgParser
from pydantic import BaseModel

parser = ArgParser()


class Item(BaseModel):
    name: str
    price: float


@parser.command()
def buy(item: Item, quantity: int):
    print(f"Bought {quantity} X {item.name} at ${item.price:.2f}.")

if __name__ == "__main__":
    parser()

This will produce the following help:

usage: cli.py [-h] --item.name TEXT --item.price FLOAT --quantity INT

optional arguments:
  -h, --help          show this help message and exit
  --item.name TEXT
  --item.price FLOAT
  --quantity INT

Arguments From Different Sources

argdantic supports several inputs:

.env files, environment variables, and secrets thanks to pydantic.
JSON files, using either the standard json library, or orjson if available.
YAML files, using the pyyaml library.
TOML files, using the lightweight tomli library.

Sources can be imported and added to each command independently, as such:

from argdantic import ArgParser
from argdantic.sources import EnvSettingsSource, JsonSettingsSource

parser = ArgParser()


@parser.command(
    sources=[
        EnvSettingsSource(env_file=".env", env_file_encoding="utf-8"),
        JsonSettingsSource(path="settings.json"),
    ]
)
def sell(item: str, quantity: int, value: float):
    print(f"Selling: {item} x {quantity}, {value:.2f}$")


if __name__ == "__main__":
    parser()

This is just a brief introduction to the library, more examples and details can be found in the documentation.

Contributing

Contributions are welcome! You can open a new issue to report bugs, or suggest new features. If you're brave enough, pull requests are also welcome.

edornd/argdantic