kfg 2.0.0

Setting and retrieving configuration values

Getting and setting configuration values is simple:

from kfg.config import Config

# construct a Config
c = Config()

# set the ('ui', 'font-color') value to 'blue'
c['ui', 'font-color'] = 'blue'

# retreive the value
font_color = c['ui', 'font-color']

Uniform exceptions

kfg provides a uniform set of exceptions that are used to signal configuration problems. This means that your program can catch these exceptions and know that they only ever indicate configuration problems. For example:

c = Config()

# kfg throws ConfigKeyError on missing keys.
try:
    x = c['foo']
except ConfigKeyError:
    print('foo not in the config')

Transforms and validation

kfg lets you associate a “transform” with a key. This transform is a 1-arity callable that will be passed the stored value of the configuration option, and kfg will pass the return value of the transform to the user when they access the value. If a transform fails, kfg will raise a ConfigValueError.

This lets you do two things. First, you can construct arbitrary values from stored configuration information in a centralized way, i.e. mediated by the configuration. Second, it let’s you validate configuration values. For example:

c = Config()
c['processing', 'timeout'] = "10 seconds"
c.set_transform(('processing', 'timeout'),
                float)

c['processing', 'timeout']  # Raises ConfigValueError because ``float``
                            # can't parse "10 seconds"

This system is intentionally low-powered and simple. For example, kfg allows you to set values which violate the transform. The goal is not to create completely water-tight configurations, but rather to create systems which are easy to get right and which warn you when you might be using an invalid value.