Config Finder

Finds a configuration file (e.g. pyproject.toml) and returns some sub-configuration with only python standard libraries.

Note

Supported formats:

Algorighm

  1. starts from the currently executed file (__main__.__file__)
  2. Checks if that folder contains the desired configuration file
  3. Goes to the parent directory and repeats at 2

Use Case

When defining machine learning projects and handling the project configuration by e.g. a pyproject.toml (e.g. with Astral UV or Poetry) you can utilize the configuration files to define and store important variables.

[tool.some_tool]
key1 = "some_value_1"
key2 = "some_value_2"

[tool.some_tool.default_config]
important_key = "some_value"

[tool.some_tool.special_config]
important_key = "another_value"

Tip

Instead of defining global variables in your code or using dotenv, a configuration file such as the pyproject.toml can be used to store configurations.

Access works via

find_configuration("pyproject.toml", ["tool", "some_tool", "default_config"])

{"important_key" : "some_value"}

This function can also be used to handle credentials.

Caution

Do not write your credentials into the pyproject.toml and ensure that you do not check your credentials into the source control.

Other Readers

Other readers can be used e.g. for YAML via PyYAML/PyPI: PyYAML works like this:

import yaml
from simpleconfigfinder import config_finder

config_finder(
        "some_file.yaml", additional_readers={"yaml": yaml.safe_load}
    )