Python Simple Settings

A simple way to manage your project settings.

simple-settings is inspired by Django's settings system but is generic for any python project.

With simple-settings you just need to specify your settings module using the --settings command line arg when invoking your python script (or SIMPLE_SETTINGS environment var) and all settings will be exposed as properties of the simple_settings.settings module.

>>> from simple_settings import settings
>>> print(settings.FOO)
'some value in foo'

Installation

simple-settings is available on Pypi.

$ pip install simple-settings

To install simple-settings with all dependencies use pip install simple-settings[all]

simple-settings is tested with Python 2.7, 3.4, 3.5, 3.6 and PyPy.

How this works

simple-settings reads and stores all variables (or constants if you prefer) of a python module that you specify. To store your settings you need at least one setting file (in any of supported formats).

To specify your settings module you have two approaches: with command line or environment.

For example, imagine that you have a python module for your project settings and this file is in "settings/development.py" (a common example). To load settings of this file you can run your project with command line arg --settings:

$ python app.py --settings=settings.development

simple-settings accepts --simple-settings command line arg also. Or set the environment variable SIMPLE_SETTINGS:

$ export SIMPLE_SETTINGS=settings.development
$ python app.py

the settings environment variable is deprecated

The simple_settings.settings object reads both the command line and environment in this order (but simple-settings takes the first value it encounters), to know which file to load.

Another option is use class LazySettings instead of singleton object settings. With LazySettings class is possible to determine settings files in object create:

from simple_settings import LazySettings


settings = LazySettings('settings.development')

If you don't pass any value in LazySettings init argument, this class follow the same behavior of settings object.

Example

This is a very dummy example, in real world you would use simple-settings in more complex cases.

project_settings.py

In this example we just store a simple string but any python type is accepted.

SIMPLE_CONF = 'simple'

app.py

You don't need specify which setting simple-settings must load, you can do this with command line or environment.

from simple_settings import settings

print(settings.SIMPLE_CONF)

Run

You can specify your settings module with command line:

$ python app.py --settings=project_settings
simple

Or environment:

$ export SIMPLE_SETTINGS=project_settings
$ python app.py
simple

Check examples, in project repository for more usage samples.

as_dict()

You can check the loaded settings through method settings.as_dict()

>>> settings.as_dict()
{'SIMPLE_CONF': 'simple'}

configure

You can change any settings (and add new settings) in runtime with method configure:

>>> settings.SOME_CONF
foo
>>> settings.configure(SOME_CONF='bar')
>>> settings.SOME_CONF
bar

Keep in mind that this only modifies the settings during runtime, and any modifications using configure will be lost on program exit.

If you're using dynamic settings the configure method will update the setting value in dynamic storage as well.

Types of settings

The simple-settings is prepared to play with the following files types:

• python modules.
• cfg files (simple key=value files).
• yaml files.
• json files.
• toml files.

To simple-settings load settings of yaml files is necessary to install with extra require yaml, e.g.: pip install simple-settings[yaml]

For toml files is necessary to install with extras require toml, e.g.: pip install simple-settings[toml]

Loading settings from environment variables

simple-settings can load all environment variables, e.g. python app.py --settings=.environ or only environment variables that start with a certain prefix, e.g. python app.py --settings=MYPREFIX_.environ.

Load multiple settings modules

simple-settings can load more than one setting module without use import approach, just specify yours settings modules separated by comma. For example:

$ python app.py --settings=production,amazon,new_relic,PREFIX_.environ

simple-setting will load all settings modules in order that was specified (production-> amazon -> new_relic -> PREFIX_.environ) overriding possibles conflicts.

This also works with LazySettings class:

from simple_settings import LazySettings


settings = LazySettings('production', 'amazon', 'new_relic', 'PREFIX_.environ')

You can combine any type of settings (python modules, yaml, etc.).

Ignored settings

• Python modules:
  • Variables starting with _.
• Cfg files:
  • Keys starting with #.

Special Settings

simple-settings has a list of special settings that change how simple-settings will load settings. This special settings are specified using a SIMPLE_SETTINGS dict in the settings module.

SIMPLE_SETTINGS = {
    'OVERRIDE_BY_ENV': True,
    'CONFIGURE_LOGGING': True,
    'REQUIRED_SETTINGS': ('API_TOKEN', 'DB_USER'),
    'DYNAMIC_SETTINGS': {
        'backend': 'redis',
        'pattern': 'DYNAMIC_*',
        'auto_casting': True,
        'prefix': 'MYAPP_'
    }
}

Configure logging

If you set the special setting CONFIGURE_LOGGING with True, simple-settings will configure the python logging to you. You just need to define your logging configuration with Python dictConfig format and place in LOGGING setting, e.g.

LOGGING = {
    'version': 1,
    'disable_existing_loggers': False,
    'formatters': {
        'default': {
            'format': '%(asctime)s %(levelname)s %(name)s %(message)s'
        },
    },
    'handlers': {
        'logfile': {
            'level': 'DEBUG',
            'class': 'logging.handlers.RotatingFileHandler',
            'filename': 'my_log.log',
            'maxBytes': 50 * 1024 * 1024,
            'backupCount': 10,
            'formatter': 'default'
        },
    },
    'loggers': {
        '': {
            'handlers': ['logfile'],
            'level': 'ERROR'
        },
        'my_project': {
            'level': 'INFO',
            'propagate': True,
        },
    }
}

To use just get logger with logging.getLogger(), e.g.

import logging
logger = logging.getLogger('my_project')


logger.info('Hello')

Don't forget, simple-settings is lazy and it only configures logging after runs setup() method or after reads some setting.

Override settings value

You can override the values of your settings module with environment variables. You just need set the special setting OVERRIDE_BY_ENV with True as value.

$ export SIMPLE_CONF="simple from env"
$ python app.py --settings=project_settings
simple from env

This is not a dynamic behavior, because settings are only overriden at "settings setup" time; see dynamic settings for a real dynamic behavior.

Required Settings

You can determine a list of mandatory settings, i.e. settings that require a valid value. For this, set the special setting REQUIRED_SETTINGS to a list (or any iterable) of your required settings. If any setting in this list has an invalid value (or is not present in setting file) then a ValueError is raised with a list of required settings not satified in the settings file.

Required Not None Settings

You can also determine a list of settings that must have a not none value, i.e. settings that cannot be set as none. For this, set the special setting REQUIRED_NOT_NONE_SETTINGS to a list (or any iterable) of the settings that you require to not be none. If any setting in this list has a value of none, then a ValueError is raised with a list of settings that must be set to not none.

Required Settings Types

You can enforce that settings must have a particular type. For this, set the special setting REQUIRED_SETTINGS_TYPES to a dictionary with the keys being the name of the setting and the value being the type of the setting (see list below for supported values).

If any of these settings has a value that is not of the type specified, or is a string that cannot be parsed to the type specified, a ValueError is raised with a list of settings that are of the wrong type. If there are no errors, the setting value will be converted into that type. If any of the values are none, their type is not evaluated.

The supported types are listed below. If you attempt to set a type that is not one of these types, then a ValueError will be raised with any unsupported types.

• "bool" - python's native boolean type, True values are y, yes, t, true, on and 1; false values are n, no, f, false, off and 0
• "int" - python's native integer type, parsed from a string using int(value)
• "float" - python's native float type, parsed from a string using float(value)
• "str" - python's native string type, not parsed from a string
• "json.loads" - Can be some types resulted of python's json.loads(value) function (e.g. dict: '{"foo": "bar"} -> {'foo': 'bar'}, int: '1' -> 1, bool: 'true' -> True, list: '[1, 2]' -> [1, 2], etc.)

Dynamic Settings

simple-settings has a list of dynamic settings mechanisms that change a value of setting dynamically. If dynamic setting is activate, for all setting the dynamic reader is called. The current dynamic mechanisms suported is:

Default Dynamic Settings Configuration

For all dynamic settings backends simple-settings accept this optional parameters:

• pattern: if you set some regex pattern the dynamic settings reader only get settings that match with this pattern. (Note that the pattern will be applied to key as entered, ignoring any configured prefix setting.)
• auto_casting: if you set this conf to True (default is False) simple settings use jsonpickle to encode settings value before save in dynamic storage and decode after read from dynamic storage. With this bahavior you can use complex types (like dict and list) in dynamic settings.
• prefix: if you set a prefix this value will be prepended to the keys when looked up on the backend. The value is prepended without any interpretation, so the key key="MYKEY" and prefix="my/namespace/" would resolve to key="my/namespace/MYKEY" and key="MYKEY" and prefix="MY_NAMESPACE_" would resolve to key="MY_NAMESPACE_MYKEY".

Redis

You can read your settings dynamically in redis if you activate the DYNAMIC_SETTINGS special setting with redis backend:

SIMPLE_SETTINGS = {
    'DYNAMIC_SETTINGS': {
        'backend': 'redis',
        'host': 'locahost',
        'port': 6379,
    }
}

for redis backend localhost is default value for host and 6379 is the default value for port.

In redis dynamic reader the binary types is automatically decoded.

To install with redis dependencies use: pip install simple-settings[redis]

Consul - Deprecated

You can read your settings dynamically from a consul server if you activate the DYNAMIC_SETTINGS special setting with the consul backend (uses consulate library):

SIMPLE_SETTINGS = {
    'DYNAMIC_SETTINGS': {
        'backend': 'consul',
        'host': 'locahost',
        'port': 8500,
        'prefix': 'mynamespace/'
    }
}

for consul backend localhost is default value for host and 8500 is the default value for port.

Additional attributes for consul backend: datacenter, token, scheme.

To install with consul dependencies use: pip install simple-settings[consul]

DATABASE

You can read your settings dynamically form a database if you activate the DYNAMIC_SETTINGS special setting with the database backend (uses sqlalchemy library)

SIMPLE_SETTINGS = {
    'DYNAMIC_SETTINGS': {
        'backend': 'database',
        'sqlalchemy.url': 'sqlite:///:memory:',
        ...
    }
}

To install with database dependencies use: pip install simple-settings[database]

AWS S3

You can read your settings dynamically form a AWS S3 bucket if you activate the DYNAMIC_SETTINGS special setting with the s3 backend (uses boto3 library)

SIMPLE_SETTINGS = {
    'DYNAMIC_SETTINGS': {
        'backend': 's3',
        'bucket_name': 'simple-settings',
        'region': 'us-east-1'
        ...
    }
}

To install with s3 dependencies use: pip install simple-settings[s3]

Memcached

You can read your settings dynamically with memcached if you activate the DYNAMIC_SETTINGS special setting with the memcached backend (uses pymemcached library)

SIMPLE_SETTINGS = {
    'DYNAMIC_SETTINGS': {
        'backend': 'memcached',
        'host': 'localhost',
        'port': 11211
        ...
    }
}

To install with memcached dependencies use: pip install simple-settings[memcached]

Utils

Settings Stub

A simple context manager (and decorator) class useful in tests which is necessary to change some setting in the safe way.

Context Manager example

from simple_settings import settings
from simple_settings.utils import settings_stub


with settings_stub(SOME_SETTING='foo'):
    assert settings.SOME_SETTING == 'foo'
assert settings.SOME_SETTING == 'bar'

Decorator example

from simple_settings import settings
from simple_settings.utils import settings_stub


@settings_stub(SOME_SETTING='foo')
def get_some_setting():
    return settings.SOME_SETTING

assert get_some_setting() == 'foo'
assert settings.SOME_SETTING == 'bar'

Advanced Usage

Custom Strategy

To implement a custom strategy:

from simple_settings import settings

class SettingsCustomStrategy:
    """
    See `/simple_settings/strategies` for sample strategies (e.g. python, json, cfg)
    """

settings.add_strategy(SettingsCustomStrategy)

Custom Dynamic Settings Reader

You can easily create your own dynamic settings reader. To do that you need to create a class than inherit from simple_settings.dynamic_settings.base.BaseReader and implement _get and _set methods, f.ex:

from simple_settings.dynamic_settings.base import BaseReader


class Reader(BaseReader):

    def __init__(self, conf):
        super(Reader, self).__init__(conf)
        self._dict = {}

    def _get(self, key):
        return self._dict.get(key)

    def _set(self, key, value):
        self._dict[key] = value

To use it, just configure SIMPLE_SETINGS special setting with the full path of the reader, f.ex:

'SIMPLE_SETTINGS': {
    'DYNAMIC_SETTINGS': {
      'backend': 'path.of.module.ClassName'
    }
}

Any other config of dynamic settings will be pass to reader backend on argument conf

Changelog

[1.2.0] - 2021-12-30

• Update dependencies
• Deprecated Consul Dynamic Settings Reader
• Implement strtobool function (and remove distutils dependency)

[1.1.0] - 2021-10-26

• Fix TypeError when load an empty config file.
• Update dependencies.
• Add log on Python Load Strategy to logging when an exception raises on import process.

[1.0.0] - 2020-09-29

• Drop support to python < 3.6.
• Update dependencies.

[0.19.1] - 2019-10-21

• Load dynamic settings reader with both ways, full class path and module path (assuming the reader class is called Reader)

[0.19.0] - 2019-10-18

• Change import dynamic reader mechanism to using full class path with dot notation
• Update several dependencies
• json.loads of REQUIRED_SETTINGS_TYPES now converts and validate lists

[0.18.0] - 2019-07-14

• Fix TypeError on jsonpickle.decode when auto_casting is True and dynamic backend returns None.
• Raise exception containing settings_file information when an error occurs in strategy.load_settings_file call from _load_settings_pipeline.
• If dynamic settings is enabled, query first the dynamic backend before raising an AttributeError.

[0.17.0] - 2019-07-10

• Allow settings to be loaded from environment variables via .environ or PREFIX_.environ
• Allow LazySettings.strategies to be easily overridden.
• Using strtobool from standard library on Required Settings Type feature.

[0.16.0] - 2019-02-23

• json.loads as new REQUIRED_SETTINGS_TYPES

[0.15.0] - 2019-02-23

• Enforce ordering of special settings being applied
• Dynamic settings behaviors with memcached.
• Fix TypeError on load settings by YAML file with a relative path

[0.14.0] - 2019-01-31

• Required not none special setting
• Required types special setting and type conversion

[0.13.0] - 2018-03-28

• Dynamic settings behaviors with AWS S3.

[0.12.1] - 2017-10-27

• Fix dynamic settings read behavior to ignore only None values and not zeros values ( #68)

[0.12.0] - 2017-03-07

• Load settings from toml files.

[0.11.0] - 2017-02-17

• Autoconfigure python logging with CONFIGURE_LOGGING special setting.

[0.10.0] - 2016-10-28

• Support configuring dynamic backends with an optional prefix.

[0.9.1] - 2016-09-15

• configure method now works even called before the LazySettings setup.

[0.9.0] - 2016-08-12

• configure method now update settings in dynamic settings.
• On get setting value in dynamic setting update local settings with this value.
• Auto casting value in dynamic storage to using complex types.

[0.8.1] - 2016-06-04

• Fix instalation with database extra requires.

[0.8.0] - 2016-06-04

• Better ImportError message if using a dynamic reader without your lib dependencies.
• Refactor in Settings Stub.
• Dynamic settings behaviors with SQLAlchemy (database backend).
• Load settings of json files.

[0.7.0] - 2016-06-02

• Nice python REPR for LazySettings objects.
• Dynamic settings behaviors with Redis.
• Dynamic settings behaviors with Consul.
• Generate package with python wheel.

[0.6.0] - 2016-05-17

• Some refactors.
• Determine settings files and modules directly in LazySettings object (to avoid use env or command line argument).
• configure method to update settings.
• Use safe_load instead load in yaml strategy.

[0.5.0] - 2016-02-03

• Some refactors.
• Load settings of yaml files.
• New SIMPLE_SETTINGS environment variable.
• New --simple-settings command line arg.

[0.4.0] - 2016-01-03

• Lazy settings load.

[0.3.1] - 2015-07-23

• Avoid to load python modules (as settings) in python files (with this, fix deepcopy bug in as_dict() method).

[0.3.0] - 2015-07-19

• Deepcopy in as_dict method to anticipate unexpected changes.
• Special Settings Behaviors.
  • Override settings values by environment.
  • Required settings validation.
• Remove default behavior of override settings values by environment (now it's a special settings).
• Settings Stub (useful for tests)
• Change bahavior of settings __getattr__ (before may raise KeyError if simple-settings do not locate the setting, now raise AttributeError)

[0.2.0] - 2015-06-19

• Load multiple settings separated by comma (like a pipeline).
• Load settings of cfg files.
• Filter python module attributes to read only user settings.

[0.1.1] - 2015-05-19

• Fix parser_args error if using simple-settings with others command line arguments.

[0.1.0] - 2015-05-14

• First release.