configya 0.3.0

A generic YAML config file generator

status

configya

What is this?

I often make YAML configuration files for codes to input default parameters. I kept repeating myself and wanting more robustness to the structure required in the file... so I decided to make a generic one.

What it does?

• Inherit the YAMLConfig class and pass to the super class a dict with the structure of your config and default values as well as the location and name of the config file.
• Upon creating a config object, it will check if there is an existing config, and if not create a default one
• If there is an existing config, it makes sure the structure and value types conform to default file structure and types
• If it finds errors, it yells at you, backs yours up and tries to correct the corrupted one.
• Only simple values like strings and numbers are checked... but maybe more in the future

What next?

fork it and add more... I'm sure this could be better

Check it out

from yaml_config.yaml_config import YAMLConfig

Create a dictionary with the (nested) structure you desire and give it some default values

my_structure = {}
my_structure['n_workers'] = 4
my_structure['server'] = dict(name='supercomputer', address='10.10.1.1')
my_structure['my_iq'] = 80
my_structure['hours_spent_this'] = 'too many'

Create your class somewhere in your package

class MyConfig(YAMLConfig):  
    def __init__(self):

        super(MyConfig, self).__init__(my_structure,
                                       '~/.my_cool_program',
                                       'config.yml')
    

The first time you make it, the default config will be written. It will yell at you.

config = MyConfig()

!ls /Users/jburgess/.my_cool_program/

config.yml

Yes, it is there.

config

/Users/jburgess/.my_cool_program/config.yml





hours_spent_this: too many
my_iq: 80
n_workers: 4
server:
  address: 10.10.1.1
  name: supercomputer

config['server']['name']

'supercomputer'

config['my_iq']

80

Well...

config['my_iq'] = 200
config['my_iq']

200

Safety first

It will not allow you to clobber nested dicts

config['server'] = 5

---------------------------------------------------------------------------

AssertionError                            Traceback (most recent call last)

<ipython-input-7-3d0b32c9a17b> in <module>
----> 1 config['server'] = 5


~/coding/projects/yaml_config/yaml_config/yaml_config.py in __setitem__(self, key, item)
    294         if key in self._configuration:
    295 
--> 296             assert not isinstance(self._configuration[key], dict), f"Woah, you are going to overwrite the structure"
    297 
    298             self._configuration[key] = item


AssertionError: Woah, you are going to overwrite the structure

Cannot add things that are not there

config['ooops'] = 10

---------------------------------------------------------------------------

ValueError                                Traceback (most recent call last)

<ipython-input-8-ebff4c5248b8> in <module>
----> 1 config['ooops'] = 10


~/coding/projects/yaml_config/yaml_config/yaml_config.py in __setitem__(self, key, item)
    301 
    302             raise ValueError(
--> 303                 f"Configuration key {key} does not exist"
    304             )
    305 


ValueError: Configuration key ooops does not exist

Or find them

config['what?'] 

---------------------------------------------------------------------------

ValueError                                Traceback (most recent call last)

<ipython-input-9-37ec546ab0de> in <module>
----> 1 config['what?']


~/coding/projects/yaml_config/yaml_config/yaml_config.py in __getitem__(self, key)
    287 
    288             raise ValueError(
--> 289                 f"Configuration key {key} does not exist"
    290             )
    291 


ValueError: Configuration key what? does not exist

What if I manually change the config in my github repo, and the one on my local computer is not up to date?

my_structure = {}
my_structure['n_workers'] = 4
my_structure['server'] = dict(name='supercomputer', address='10.10.1.1')
my_structure['my_iq'] = 80
my_structure['hours_spent_this'] = 'too many'
# Added a new number!
my_structure['total_number_of_postdocs'] = 1e5

class MyConfig(YAMLConfig):  
    def __init__(self):

        super(MyConfig, self).__init__(my_structure,
                                       '~/.my_cool_program',
                                       'config.yml')
    

config = MyConfig()

/Users/jburgess/coding/projects/yaml_config/yaml_config/yaml_config.py:109: BadStructureWarning: The user config file /Users/jburgess/.my_cool_program/config.yml was corrupt
  BadStructureWarning,
/Users/jburgess/coding/projects/yaml_config/yaml_config/yaml_config.py:113: BadStructureWarning: and has been backed up to /Users/jburgess/.my_cool_program/config.yml.bak and replaced
  BadStructureWarning,
/Users/jburgess/coding/projects/yaml_config/yaml_config/yaml_config.py:117: BadStructureWarning: with the default config. The CURRENT config is now default
  BadStructureWarning,

config

/Users/jburgess/.my_cool_program/config.yml





hours_spent_this: too many
my_iq: 80
n_workers: 4
server:
  address: 10.10.1.1
  name: supercomputer
total_number_of_postdocs: 100000.0

!ls /Users/jburgess/.my_cool_program/

config.yml     config.yml.bak

Also, if you manually edit the config in an editor, it will check if the types are correct. If not, it will replace that value with the default value and backup your config