Skip to main content

Wrapper around OmegaConf for loading configuration from various types of files.

Project description

VariConf - Load Configurations from Various Types of Files

VariConf provides a wrapper around OmegaConf for loading configuration from various types of files.

Supported file types are JSON, YAML and TOML. Support for more file types can easily be added by registering a custom loader function.

Thanks to the power of OmegaConf, you can provide a configuration schema which the defines expected parameters, default values and optionally expected types.

Design Goals

This package is developed with the following goals in mind:

  • Load configuration from files with expected parameters and default values provided in an easy way.
  • Provide config as a simple Namespace-like object (with option to convert to dictionary).
  • Do not commit to a specific file format. All formats that can easily be loaded into a dictionary should be supported (json, toml, yaml, ...).
  • Optionally check types.
  • Optionally check for unknown parameters in the file (to raise an error).
  • Keep it simple. Prefer fewer features over too complicated API.

Installation

Basic (does not include dependencies for YAML and TOML):

pip install variconf

With optional dependencies:

# with TOML support
pip install "variconf[toml]"

# with YAML support
pip install "variconf[yaml]"

# to include everything:
pip install "variconf[all]"

Usage

The package provides a class WConf for loading and merging configurations from different sources. When creating an instance of it, a configuration "schema" needs to be given, i.e. a structure (dictionary or dataclass) that defines what sections and parameters the configuration has and that provides default values.

A number of "load"-methods is provided to load configurations from different sources (e.g. JSON files, YAML files, dictionaries from other sources, ...). When called, the corresponding parameters are merged into the existing configuration, overwriting existing values. This means that an input does not need to provide all parameters, in this case the default values are kept. Further, if calling multiple "load"-methods after another, the later calls will overwrite values set by previous ones.

All "load"-methods return self, so they can be chained:

import variconf

schema = {"sec1": {"foo": 42, "bar": 13}, "sec2": {"bla": ""}}
wconf = variconf.WConf(schema)
config = (
    wconf.load_file("~/global_config.toml")
    .load_file("./local_config.yml", fail_if_not_found=False)
    .load_dotlist(sys.argv[1:])  # easily allow overwriting parameters via
                                 # command-line arguments
    .get()  # return the final config object
)

Allow Unknown Parameters

By default an error is raised if the loaded configuration contains parameters that are not declared in the schema. If you want to allow these unknown parameters, initialise WConf with strict=False:

wconf = variconf.WConf(schema, strict=False)

This will result in the unknown parameters being merged into the config object.

With this you can even omit the schema altogether by simply passing an empty dictionary:

wconf = variconf.WConf({}, strict=False)

Search File

Assuming an application where the config file can be located in one of several places (e.g. ~, ~/.config or /etc/myapp). This situation is supported by the optional search_paths argument of load_file():

wconf.load_file(
    "config.yml",
    search_paths=[os.expanduser("~"), os.expanduser("~/.config"), "/etc/myapp"],
    fail_if_not_found=False,
)

This will search for a file "config.yml" in the listed directories (in the given order) and use the first match. By setting fail_if_not_found=False, we specify that it's okay if the file is not found in any of these directories. In this case, we simply keep the default values of all parameters.

Using XDG Base Directory Specification

If your application follows the XDG Base Directory Specification you can use load_xdg_config() (currently not supported on Windows!):

wconf.load_xdg_config("myapp/config.toml")

Will search for the file in the directories specified in the environment variables XDG_CONFIG_HOME and XDG_CONFIG_DIRS (defaulting to ~/.config).

Like for load_file() there is an argument fail_if_not_found but here it defaults to False as providing a config in XDG_CONFIG_HOME is typically optional.

Supported File Types

Supported file types are JSON, YAML and TOML. Support for custom file types can be added by providing a loader function. Example for adding XML support:

import xml.etree.ElementTree as ET

def xml_loader(fp: typing.IO) -> dict:
    xml_str = fp.read()
    xml_tree = ET.fromstring(xml_str)
    # do some magic to convert XML tree to dictionary
    xml_dict = tree_to_dict(xml_tree)
    return xml_dict

wconf.add_file_loader("xml", [".xml"], xml_loader)

# now, XML files can be read by WConf.load and WConf.load_file
wconf.load_file("config.xml")

Type Checking

OmegaConf supports type-checking by providing a schema as dataclass with type hints:

@dataclasses.dataclass
class ConfigSchema:
    foo: int = 42
    bar: str = 13

wconf = variconf.WConf(ConfigSchema)

# raises ValidationError: Value 'hi' of type 'str' could not be converted to Integer
wconf.load_dict({"foo": "hi"})

Required Values

Required parameters without default value are supported through OmegaConf's concept of missing values.

When using a dictionary schema:

schema = {
    "optional_param": "default value",
    "required_param": "???",
}

When using a dataclass schema:

@dataclasses.dataclass
class Schema:
    required_param1: float  # not providing a default makes it required
    optional_param: str = "default value"
    required_param2: int = omegaconf.MISSING  # alternative for required parameters

If there is a required parameter for which no value has been provided by any of the load*-methods, calling get() will raise an error.

You can avoid that error by using get(allow_missing=True). However, the error is still raised when trying to access the actual value of the missing parameter.

Variable Interpolation

OmegaConf has a feature called variable interpolation that allows to refer to other fields within the config file:

server:
  host: localhost
  port: 80

client:
  url: http://${server.host}:${server.port}/
  server_port: ${server.port}
  # relative interpolation
  description: Client of ${.url}

See the documentation of OmegaConf for more information.

Copyright and License

VariConf is written and maintained by Felix Kloss at the Max Planck Institute for Intelligent Systems, Tübingen.

Copyright (c) 2022, Max Planck Gesellschaft. All rights reserved.

License: BSD 3-clause

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

variconf-1.0.1.tar.gz (14.6 kB view details)

Uploaded Source

Built Distribution

variconf-1.0.1-py3-none-any.whl (10.0 kB view details)

Uploaded Python 3

File details

Details for the file variconf-1.0.1.tar.gz.

File metadata

  • Download URL: variconf-1.0.1.tar.gz
  • Upload date:
  • Size: 14.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.6

File hashes

Hashes for variconf-1.0.1.tar.gz
Algorithm Hash digest
SHA256 4848da3f5335dd870b51221b68db42eb260216c1433d83a2e82ceb1401f69c45
MD5 de7e44068ddf8cc39a3ef2b66b342b5b
BLAKE2b-256 154ffcdeaa5e6d24c0256d5c4a941f521c36999c359d75ebe14e828e01e59c3b

See more details on using hashes here.

File details

Details for the file variconf-1.0.1-py3-none-any.whl.

File metadata

  • Download URL: variconf-1.0.1-py3-none-any.whl
  • Upload date:
  • Size: 10.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.6

File hashes

Hashes for variconf-1.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 4ae7b484b256e2eccf72d8c8b526e61d2e714779ab3de1bec740ea8e1a503fcd
MD5 9f7369b5b2d89573ee8a2520a2b60d94
BLAKE2b-256 e3015716521b719ae5b72aa9f94de3602c0cab2744f31477faafdf2d5195728b

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page