Envier

Python application configuration from the environment

Synopsis

Envier is a Python library for extracting configuration from environment variables in a declarative and (eventually) 12-factor-app-compliant way.

Usage

The following example shows how to declare the configuration for an application that uses the MYAPP_DEBUG, MYAPP_SERVICE_HOST and MYAPP_SERVICE_PORT variables from the environment.

>>> from envier import Env
>>> 
>>> class GlobalConfig(Env):
>>>     __prefix__ = "myapp"
>>>     
>>>     debug_mode = Env.var(bool, "debug", default=False)
>>> 
>>>     service_host = Env.var(str, "service.host", default="localhost")
>>>     service_port = Env.var(int, "service.port", default=3000)
>>> 
>>>     _is_default_port = Env.der(bool, lambda c: c.service_port == c.spec.service_port.default)
>>> 
>>> config = GlobalConfig()
>>> config.service_port
3000
>>> config._is_default_port
True

Configurations can also be nested to create namespaces:

>>> from envier import Env
>>> 
>>> class ServiceConfig(Env):
>>>     __prefix__ = "service"
>>> 
>>>     host = Env.var(str, "host", default="localhost")
>>>     port = Env.var(int, "port", default=3000)
>>> 
>>> class GlobalConfig(Env):
>>>     __prefix__ = "myapp"
>>>     
>>>     debug_mode = Env.var(bool, "debug", default=False)
>>> 
>>>     service = ServiceConfig
>>> 
>>> config = GlobalConfig()
>>> config.service.port
3000

The same configuration can be obtained with implicit nesting by declaring the ServiceConfig subclass inside GlobalConfig, and setting the class attribute __item__ to the name of the item the sub-configuration should be assigned to, viz.

>>> from envier import Env
>>> 
>>> class GlobalConfig(Env):
>>>     __prefix__ = "myapp"
>>>     
>>>     debug_mode = Env.var(bool, "debug", default=False)
>>> 
>>>     class ServiceConfig(Env):
>>>         __item__ = __prefix__ = "service"
>>>         
>>>         host = Env.var(str, "host", default="localhost")
>>>         port = Env.var(int, "port", default=3000)
>>> 
>>> config = GlobalConfig()
>>> config.service.port
3000

Type Checking

The library ships with a mypy plugin to allow for type checking. If you want to use it, either install the library with the mypy extra or ensure that mypy is installed, and then add envier.mypy to the list of extra plugins in the mypy configuration.

Sphinx Plugin

The library comes with a Sphinx plugin at envier.sphinx to allow generating documentation from the configuration spec class directly. It exposes the envier directive that takes a mandatory argument, the configuration spec class in the form module:class; additionally, the options heading and recursive can be used to control whether to add heading and whether to recursively get help information from nested configuration spec classes respectively. By default, the plugin will display the table heading and will not recurse over nested configuration spec classes.

Here is an example for a configuration class GlobalConfig located in the myapp.config module. We omit the table header and recurse over nested configuration.

.. envier:: myapp.config:GlobalConfig
   :heading: false
   :recursive: true

Roadmap

• Add support for environment files.
• Rely on type hints as support for older versions of Python is dropped.
• Derivations might require an evaluation order.