Use a pydantic model to define the command's options

[choucavalier]

In the case where I have something like this

class MyParams(BaseModel):
  foo: int
  bar: str


def main(params: MyParams):
    typer.echo(f"{params.foo}")

if __name__ == "__main__":
    pydantic_typer.run(main)

I would like to generate a CLI with arguments --foo and --bar directly instead of --params.foo and --params.bar.

In other words, I would like to define the set of parameters externally, using a pydantic model.

Is this possible?

The reason why I want to do that is that I have a bunch of BaseModel that define the arguments of multiple commands. And I would like to automatically generate the CLI based an these pydantic models.

A second question then is: how can I programmatically generate CLI commands with arguments defined by pydantic models?

Thanks a lot for your work on this extension!

[synvima]

For the first question: you can flatten the argument list with this helper decorator:

from collections.abc import Callable
from typing import get_args
from typer.models import OptionInfo

def flatten_args(func: Callable) -> Callable:
    """Flatten command line arguments to the underling pydantic class field names."""
    if hasattr(func, "__annotations__"):
        for value in func.__annotations__.values():
            args = get_args(value)
            option_info = next((arg for arg in args if isinstance(arg, OptionInfo)), None)
            qualifiers = next((arg for arg in args if isinstance(arg, list)), None)
            if option_info is not None and qualifiers is not None:
                option_info.default = f"--{qualifiers[-1]}"

    def wrapper(*args, **kwargs):  # noqa: ANN002, ANN003, ANN202
        return func(*args, **kwargs)

    return wrapper

Usage:

@flatten_args
@app.command()
def set_config(
    config: models.Configuration,
) -> None:
  ...

Second question:
the best way is probably to traverse your models programmatically and generate appropriate commands using jinja.