conffu 2.2.1

A simple, but powerful JSON, XML and command line configuration package

Conf Fu

A configuration package that allows you to configure your Python scripts, through a combination of JSON configuration files and command line options, with a minimum of code.

Install

Install the package:

pip install conffu

Note: on Linux, you may need to install libffi-dev and re-install Python if installation causes a 'No module named _ctypes error'. This is not specific to conffu, but any package requiring _ctypes in one of its dependencies.

If you want to be able to read/write XML configurations as well, there is a dependency on lxml, install using:

pip install conffu[xml]

Example

With the package installed, try running this script:

from conffu import Config

cfg = Config({
    '_globals': {
        'temp': 'C:/Temp'
    },
    'temp_file': '{temp}/text.txt',
    'number': 3
})

print(f'The number is {cfg.number}')

cfg.save('example_config.json')

After running that, this also works:

from conffu import Config

cfg = Config.from_file('example_config.json')
print(f'The number is {cfg.number}')

Make a change and save this script as example.py:

from conffu import Config

cfg = Config.from_file('example_config.json').update_from_arguments()
print(f'The number is {cfg.number}')

Then try running it like this:

python example.py -number 7

There's many more options, check the documentation for more examples.

Caveat

Note that Config allows you to do this:

from conffu import Config

cfg = Config()
cfg['test'] = 1
print(cfg.test)  # prints 1
cfg.test = 2
print(cfg.test)  # prints 2

That is, you're allowed to access configuration keys as if they were attributes on the configuration. However, if you try to access a key that happens to also be an attribute on the object, you get the attribute. This has the advantage that the feature doesn't break how objects work in Python, but the disadvantage that you'll need to access keys that have the same name as object attributes using the dictionary syntax. For example:

from conffu import Config

cfg = Config()
# create a new config value a
cfg['a'] = 1
# change the value of a, accessing it as you would a property / attribute
cfg.a = 2
# create a new object attribute b
cfg.b = 3
# this does not change the attribute b, because you're directly accessing a new config value b
cfg.['b'] = 4
print(cfg.a, cfg.b)  # prints 2 3
print(cfg['a'], cfg['b'])  # prints 2 4

If you don't like this behaviour, consider using the DictConfig class instead of Config - they are identical, except that the DictConfig does not have this behaviour and you must always access keys like cfg['test'], or cfg['key.subkey'].

If you do like the ability to access configuration values like attributes, you should take care that your code only assigns to existing values (initialise them as 'null' when creating the config from json for example), or that you use the dictionary syntax when assigning values and you only use the attribute syntax for reading values.

License

This project is licensed under the MIT license. See LICENSE.txt.

Changelog

See CHANGELOG.md.