Pyproject TOML loader

If you use an unsupported file format or need to do extra processing of the file, you can make your own file loader and construct an appropriate callback.

Suppose you want to specify parameters in a section of pyproject.toml:

pyproject.toml

[tool.my_tool.parameters]
arg1 = "stuff"
opt1 = "things"
opt2 = "nothing"

Then, we can read the values in our typer CLI:

my_tool.py

from typing import Any, Dict

import typer
from typer_config import conf_callback_factory
from typer_config.loaders import toml_loader


def pyproject_loader(param_value: str) -> Dict[str, Any]:
    if not param_value:  # set a default path to read from
        param_value = "pyproject.toml"
    pyproject = toml_loader(param_value)
    conf = pyproject["tool"]["my_tool"]["parameters"]
    return conf


### You can define the same loader using some provided combinators:
#
# from typer_config.loaders import default_value_loader, subpath_loader, toml_loader
#
# pyproject_loader = subpath_loader(
#     default_value_loader(toml_loader, lambda: "pyproject.toml"),
#     ["tool", "my_tool", "parameters"],
# )

pyproject_callback = conf_callback_factory(pyproject_loader)

app = typer.Typer()


@app.command()
def main(
    arg1: str,
    config: str = typer.Option(
        "",
        callback=pyproject_callback,
        is_eager=True,  # THIS IS REALLY IMPORTANT (1)
    ),
    opt1: str = typer.Option(...),
    opt2: str = typer.Option("hello"),
):
    typer.echo(f"{opt1} {opt2} {arg1}")


if __name__ == "__main__":
    app()

1. You must use is_eager=True in the parameter definition because that will cause it to be processed first. If you don't use is_eager, then your parameter values will depend on the order in which they were processed (read: unpredictably).

And we get this behavior:

$ ls .
my_tool.py
other.toml
pyproject.toml

$ python my_tool.py
things nothing stuff

$ python my_tool.py others
things nothing others

$ python my_tool.py --config other.toml
something else entirely