glorpen-config 2.0.0

Loads, validates and normalizes configuration.

Config framework for Your projects - with validation, interpolation and value normalization!

Full documentation: https://glorpen-config.readthedocs.io/

Official repositories

GitHub: https://github.com/glorpen/glorpen-config

BitBucket: https://bitbucket.org/glorpen/glorpen-config

Features

You can:

• create custom fields for custom data

• define configuration schema inside Python app

• convert configuration values to Python objects

• validate configuration

• use interpolation to fill config values

• set default values

Loading data

glorpen.config.Config class allows loading data from three sources:

• path, filepath constructor argument

• file-like object, fileobj constructor argument

• dict object passed to glorpen.config.Config.load_data or glorpen.config.Config.finalize.

Interpolation

You can reuse values from config with {{ path.to.value }} notation, eg:

project:
   path: "/tmp"
   cache_path: "{{ project.path }}/cache"

String interpolation currently can be used only with glorpen.config.fields.String fields.

Normalization and validation

Each field type has own normalization rules, eg. for glorpen.config.fields.LogLevel:

logging: DEBUG

config.get("logging") would yield value 10 as is logging.DEBUG.

Additionally it will raise glorpen.config.exceptions.ValidationError if invalid level name is given.

Default values

Each field can have default value. If no value is given in config but default one is set, it will be used instead.

Default values adhere to same interpolation and normalization rules - each default value is denormalized and then passed to normalizers. That way complex object can still profit from config interpolation. There should not be any real impact on performance as it is done only once.

Example usage

Your first step should be defining configuration schema:

from glorpen.config import Config
from glorpen.config.fields import Dict, String, Path, LogLevel

project_path = "/tmp/project"

spec = Dict({
   "project_path": Path(default=project_path),
   "project_cache_path": Path(default="{{ project_path }}/cache"),
   "logging": Dict({
       "level": LogLevel(default=logging.INFO)
   }),
   "database": String(),
   "sources": Dict(
       "some_param": String(),
       "some_path": Path(),
   ),
   "maybe_string": Variant([
       String(),
       Number()
   ])
})

Example yaml config:

logging: "DEBUG"
database: "mysql://...."
sources:
   some_param: "some param"
   some_path: "/tmp"
maybe_string: 12

Then you can create glorpen.config.Config instance:

cfg = Config(filepath=config_path, spec=spec).finalize()

cfg.get("sources.some_param") #=> "some param"
cfg.get("project_path") #=> "/tmp/project"
cfg.get("project_cache_path") #=> "/tmp/project/cache"
cfg.get("logging") #=> 10
cfg.get("maybe_string") #=> 12

Creating custom fields

Custom field class should extend glorpen.config.fields.Field or glorpen.config.fields.FieldWithDefault.

glorpen.config.fields.Field.make_resolvable method should register normalizer functions which later will be called in registration order. Each value returned by normalizer is passed to next one. After chain end value is returned as config value.

Returned glorpen.config.fields.ResolvableObject instance is resolved before passing it to next normalizer.

If value passed to normalizator is invalid it should raise glorpen.config.exceptions.ValidationError. Sometimes value can be lazy loaded - it is represented as glorpen.config.fields.ResolvableObject. You can get real value by using glorpen.config.fields.resolve(value, config).

class MyValue(object):
   def __init__(self, value):
      super(MyValue, self).__init__()
      self.value = value

class MyField(Field):

    def to_my_value(self, value, config):
        return MyValue(value)

    def is_value_supported(self, value):
        return True

    def make_resolvable(self, r):
        r.on_resolve(self.to_my_value)

The last thing is to use prepared custom field in configuration spec.