Enable Starlette Config to load pyproject.toml metadata

[br3ndonland]

Description

Python now supports a pyproject.toml file for project metadata. PEP 621 defined how to store package metadata and dependencies in pyproject.toml.

Settings models like Starlette Config could benefit from the ability to read project metadata from pyproject.toml.

Suggestions

Add optional dependencies

Add Starlette and tomli as optional dependencies, with a marker to indicate that tomli is only needed on Python <3.11.

Python 3.11 added a new TOML library, tomllib. There doesn't appear to be a backport for earlier versions of Python, but the tomllib API is basically the same as tomli, so tomli can be used for Python <3.11.

Add a utility function for loading TOML files

To start, add a simple function to fastenv.utilities for loading TOML files.

The new code in fastenv.utilities could look something like this:

import sys
from typing import TYPE_CHECKING

if sys.version_info < (3, 11):
    import tomli as tomllib
else:
    import tomllib

if TYPE_CHECKING:
    from os import PathLike
    from typing import Any


def _read_toml_file(
    self, toml_file: PathLike | str = "pyproject.toml", table: str = "project"
) -> dict[str, Any]:
    """Read metadata from the pyproject.toml configuration file."""
    with open(pyproject_path, "rb") as pyproject_bytes:
        toml_data = tomllib.load(pyproject_bytes)
        project = toml_data[table]
    project_values = {key.upper(): project.get(key) for key in project}
    self.file_values |= project_values
    return project_values

Add the utility function to the Starlette Config model

Add a module fastenv.settings.starlette_config (if not already added by #19) that imports starlette.config and adds a toml_file argument.

Rough suggestion for fastenv.settings.starlette_config:

import sys
from typing import TYPE_CHECKING

import anyio
import starlette.config

from fastenv.utilities import _read_toml_file, logger

if sys.version_info < (3, 11):
    import tomllib
else:
    import tomli as tomllib

if TYPE_CHECKING:
    from os import PathLike
    from typing import Any


class Config:
    def __init__(
        self,
        env_file: PathLike[str] | str | None = None,
        toml_file: PathLike[str] | str | None = None,
        environ: dict[str, str] = environ,
        *,
        raise_exceptions: bool = True,
    ) -> None:
        self.environ = environ
        self.file_values: typing.Dict[str, str] = {}
        if toml_file and anyio.Path(toml_file).resolve(strict=raise_exceptions):
            self.file_values |= self._read_toml_file(toml_file)
        # logic for loading .env files could also be updated to use fastenv
        if env_file and anyio.Path(env_file).resolve(strict=raise_exceptions):
            self.file_values |= self._read_file(env_file)

And then a project might use it with something like:

import inboard

config = inboard.StarletteConfig(
  env_file=".env",
  toml_file="pyproject.toml",
)

DESCRIPTION = config("DESCRIPTION", cast=str)
NAME = config("NAME", cast=str)
URLS = config("URLS", cast=dict[str, str])

Add tests

Add some unit tests that load fastenv/pyproject.toml and make assertions on the resultant values, and potentially some tests on example TOML files (could be added to conftest.py as docstrings and made into pytest fixtures).

Related

• Enhance Starlette `Config` with fastenv #19
• Python 3 docs: Standard Library - tomllib