Skip to main content

Map a Python configuration from environment variables

Project description Codacy Badge


Ecological combines PEP526 and environment variables to make the configuration of 12 factor apps easy.

Getting Started

Ecological automatically gets and converts environment variables according to the configuration class definition.

For example, imagine your application has a configurable (integer) Port and (boolean) Debug flag and a (string) Log Level, that is INFO by default, you could simply declare your configuration as:

class Configuration(ecological.Config):
    port: int
    debug: bool
    log_level: str = "INFO"

And then set the environment variables PORT, DEBUG and LOG_LEVEL. Ecological will automatically set the class properties from the environment variables with the same (but upper cased) name.

By default the values are set at the class definition type and assigned to the class itself (i.e. the class doesn’t need to be instantiated). If needed this behavior can be changed (see the Autoloading section).


The tutorial can be used to get to know with the library’s basic features interactively.

Typing Support

Ecological also supports some of the types defined in PEP484, for example:

class Configuration(ecological.Config):
    list_of_values: List[str]

Will automatically parse the environment variable value as a list.

Prefixed Configuration

You can also decide to prefix your application configuration, for example, to avoid collisions:

class Configuration(ecological.Config, prefix='myapp'):
    home: str

In this case the home property will be fetched from the MYAPP_HOME environment property.

Nested Configuration

Ecological.Config also supports nested configurations, for example:

class Configuration(ecological.Config):
    integer: int

    class Nested(ecological.Config, prefix='nested'):
        boolean: bool

This way you can group related configuration properties hierarchically.


Fine-grained Control

You can control some behavior of how the configuration properties are set.

It can be achieved by providing a ecological.Variable instance as the default value for an attribute or by specifying global options on the class level:

my_source = {"KEY1": "VALUE1"}

class Configuration(ecological.Config, transform=lambda v, wt: v, wanted_type=int, ...):
    my_var1: WantedType = ecological.Variable(transform=lambda v, wt: wt(v), source=my_source, ...)
    my_var2: str
    # ...

All possible options and their meaning can be found in the table below:


Class level

Variable level







A prefix that is uppercased and prepended when a variable name is derived from an attribute name.




Derived from attribute name and prefixed with prefix if specified; uppercased.

When specified on the variable level it states the exact name of the source variable that will be used.

When specified on the class level it is treated as a function that returns a variable name from the attribute name with the following signature:

def func(attribute_name: str, prefix: Optional[str] = None)




(no default)

Default value for the property if it isn’t set.




A source value is casted to the wanted_type In case of non-scalar types (+ scalar bool) the value is Python-parsed first.

A function that converts a value from the source to the value and wanted_type you expect with the following signature:

def func(source_value: str, wanted_type: Union[Type, str])





Dictionary that the value will be loaded from.





Desired Python type of the attribute’s value.

On the variable level it is specified via a type annotation on the attribute: my_var_1: my_wanted_type.

However it can be also specified on the class level, then it acts as a default when the annotation is not provided:

class MyConfig(ecological.Config, wanted_type=int, ...)

The following rules apply when options are resolved:

  • when options are specified on both levels (variable and class), the variable ones take precedence over class ones,

  • when some options are missing on the variable level, their default values are taken from the class level,

  • it is not necessary to assign an ecological.Variable instance to change the behavior; it can still be changed on the class level (globally).


It is possible to defer/disable autoloading (setting) of variable values by specifying the autoload option on class definition.

On class creation (default)

When no option is provided values are loaded immediately on class creation and assigned to class attributes:

class Configuration(ecological.Config):
    port: int
# Values already read and set at this point.
# assert Configuration.port == <value-of-PORT-env-var>

When this option is chosen, no autoloading happens. In order to set variable values, the Config.load method needs to be called explicitly:

class Configuration(ecological.Config, autoload=ecological.Autoload.NEVER):
    port: int
# Values not set at this point.
# Accessing Configuration.port would throw AttributeError.

# Values read and set at this point.
# assert Configuration.port == <value-of-PORT-env-var>
On object instance initialization

If it is preferred to load and store attribute values on the object instance instead of the class itself, the Autoload.OBJECT strategy can be used:

class Configuration(ecological.Config, autoload=ecological.Autoload.OBJECT):
    port: int
# Values not set at this point.

config = Configuration()
# Values read and set at this point on ``config``.
# assert config.port == <value-of-PORT-env-var>
# Accessing ``Configuration.port`` would throw AttributeError.

Caveats and Known Limitations

  • Ecological doesn’t support (public) methods in Config classes

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ecological-2.0.0.tar.gz (11.2 kB view hashes)

Uploaded source

Built Distribution

ecological-2.0.0-py3-none-any.whl (9.2 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page