kedro.config.ConfigLoader¶
-
class
kedro.config.
ConfigLoader
(conf_paths)[source]¶ Recursively scan the directories specified in
conf_paths
for configuration files with ayaml
,yml
,json
,ini
,pickle
,xml
,properties
orshellvars
extension, load them, and return them in the form of a config dictionary.When the same top-level key appears in any 2 config files located in the same
conf_path
(sub)directory, aValueError
is raised.When the same key appears in any 2 config files located in different
conf_path
directories, the last processed config path takes precedence and overrides this key.For example, if your
conf_path
looks like this:. └── conf ├── README.md ├── base │ ├── catalog.yml │ ├── logging.yml │ └── experiment1 │ └── parameters.yml └── local ├── catalog.yml ├── db.ini ├── experiment1 │ ├── parameters.yml │ └── model_parameters.yml └── experiment2 └── parameters.yml
You can access the different configurations as follows:
import logging.config from kedro.config import ConfigLoader conf_paths = ['conf/base', 'conf/local'] conf_loader = ConfigLoader(conf_paths) conf_logging = conf_loader.get('logging*') logging.config.dictConfig(conf_logging) # set logging conf conf_catalog = conf_loader.get('catalog*', 'catalog*/**') conf_params = conf_loader.get('**/parameters.yml')
-
__init__
(conf_paths)[source]¶ Instantiate a ConfigLoader.
Parameters: conf_paths ( Union
[str
,List
[str
]]) – Non-empty path or list of paths to configuration directories.Raises: ValueError
– Ifconf_paths
is empty.
Methods
__init__
(conf_paths)Instantiate a ConfigLoader. get
(*patterns)Recursively scan for configuration files, load and merge them, and return them in the form of a config dictionary. -