layered-settings
is a simple and configurable hierarchical settings library for Python, including Django, Flask, or any other
scripts that need settings from potentially a variety of sources. With it you can load
settings from the environment, Amazon's SSM, Amazon's Secrets Manager, local configparser .ini files, and more.
The package is available on pip as layered-settings
. Run:
pip install layered-settings
If you want to use the AWS SSM layer or Secrets Manager layer, include the [aws] extra:
pip install layered-settings[aws]
then import via:
from layered_settings import initialize_settings, loaders
import os
from layered_settings import initialize_settings
from layered_settings import loaders
get_setting = initialize_settings(
sources=[
# Lowest priority is setting-defaults.ini. All configuration values should be defaulted in here.
# The `loaders.ConfigParserLoader` is optional - if you pass a string ending in .ini, it wraps it in
# a `loaders.ConfigParserLoader` for you.
loaders.ConfigParserLoader(os.path.join(SCRIPT_DIR, "setting-defaults.ini")),
# Next priority is a "user" settings file.
os.path.expanduser("~/.app-settings.ini"),
# You can include a dictionary in the sources too...
{
"general": {"CLIENT_NAME": "client"},
"email": {"EMAIL_HOST": "smtp.example.com", "EMAIL_PORT": 25},
},
# Secrets Manager works similar to SSM. If the secret is a json object, the object's keys will be the layered setting "key"
# If the secret is just plain text, the secret's name will be used as the key
loaders.SecretsManagerLoader(f"/app/stage/", aws_region="us-east-1") if ALLOW_SSM_CONFIGURATION else None,
# If we are able/willing to reach out to AWS, do so. A `None` in the initialize_settings sources
# will simply be ignored.
loaders.SSMLoader(f"/app/stage/", aws_region="us-east-1") if ALLOW_SSM_CONFIGURATION else None,
# Top priority -- env var in the format APP__section__key.
loaders.EnvLoader("APP__{section}__{key}"),
]
)
DATABASE_HOST = get_setting("database", "DATABASE_HOST")
AWS_SECRET_KEY = get_setting("aws", "AWS_SECRET_KEY")
...
This library is tested using Python 3.x.