application-settings 0.1.0

For providing a python application with configuration and/or settings

application_settings

What and why

Application_settings is a module for providing a python application with configuration and settings. It uses toml configuration files that are parsed into dataclasses. This brings some benefits:

• Configuration parameters are typed, which allows for improved static code analyses.
• IDEs will provide helpful hints and completion when using configuration parameters.
• More control over what happens when a config file contains mistakes (by leveraging the power of pydantic).
• Possibility to specify defaults when no config file is found or entries are missing.
• Configuration described in a human-usable, flexible, standardardized and not overly complex format.

Parsing is done once during first access and the resulting configuration is stored as a singleton.

How

Install the package

On Linux: python -m pip install -U application_settings. On Windows: py -m pip install -U application_settings

Define config section(s) and the container with application info

Example:

from application_settings import (
    ConfigBase,
    ConfigSectionBase,
)
from pydantic.dataclasses import dataclass


@dataclass(frozen=True)
class MyExample1ConfigSection(ConfigSectionBase):
    """Config section for an example"""

    field1: str = "field1"
    field2: int = 2


@dataclass(frozen=True)
class MyExampleConfig(ConfigBase):
    """Config for an example"""

    section1: MyExample1ConfigSection = MyExample1ConfigSection()

Note: a pydantic.dataclasses.dataclass is a drop-in replacement for the standard dataclasses.dataclass, with validation, see pydantic dataclasses.

Write a config file

For the example, the config file ~/.my_example/config.toml could be something like this:

[section1]
field1 = "my own version of field1"
field2 = 22

The section names in this file are equal to the fieldnames of your container class and the entries in a section consist of the fieldnames of your ConfigSection class(es). The order of sections and/or fields in the toml file does not have to adhere to the order in which fields have been specified in the Config(Section) classes.

Use config parameters in your code

# the first invocation of get() will create the singleton instance of MyExampleConfig
a_variable: str = MyExampleConfig.get().section1.field1  # a_variable == "my own version of field1"
another_variable: int = MyExampleConfig.get().section1.field2  # another_variable == 22

# you can reload a config and / or set a non-default path
another_config = MyExampleConfig.get(reload=True, configfile_path="./my_config.tml")

Location of the config file

The path for the config file can be specified via the optional argument configfile_path of the get method that creates the singleton. The path is not stored; if you reload then you again have to pass the configfile_path.

You can specify the path either as a string or as a pathlib Path. In case of a string spec, it is first validated for the platform that you are using; if the validation fails, a ValueError is raised, otherwise a Path is constructed from the string.

If you do not specify a configfile_path, then a default location is fetched via default_config_filepath(). Class ConfigBase provides a default implementation, being a filename config.toml located in a subfolder of your home directory. The default name of that subfolder is provided by default_config_foldername() and consists of a dot, followed by a name derived from your container class: the word Config is removed, underscores in front of capitals (except for the first letter) and all lower case. See also the example above. If you do not like the default implementation, you can override default_config_filepath() and/or default_config_foldername(). If you want to enforce that a configfile_path is specified in get(), then let default_config_filepath() return None.

Handling deviations in the config file

When your config file does not adhere to the specified types

When loading the config file, the values specified are coerced into the appropriate type where possible. If type coercion is not possible, then a pydantic.ValidationError is raised. Consider the case where you would use the following config file for the MyExampleConfig defined above:

[section1]
field1 = true
field2 = "22"

The bool specified for field1 will be coerced into a str value of "true". The str specified for field2 will be coerced into an int value of 22.

When your config file does not contain all specified attributes

If your Config has one of more sections with attributes that do not have a default value, then a config file must be loaded and these sections and attributes must be present in the loaded config file. If this is not the case, a TypeError is raised. Attributes that have default values can be omitted from the config file without problems.

Note that in the dataclass definitions, attributes without default value have to come before attributes with default values.

When your config file contains additional, unspecified attributes

Entries in a config file that are not defined in the Config(Section) classes will simply be ignored silently.

More advanced typing and validation with pydantic

• Non-standard types useful for configuration, such as network addresses, are offered, see pydantic types
• The value of numerous common types can be restricted using pydantic constrained types

License

This project is licensed under the terms of the MIT license.