zycelium.dataconfig 0.0.3

Create dataclasses backed by configuration files.

zycelium.dataconfig

Create dataclasses backed by configuration files.

Usage

Use defaults:

Create a new python script and name it example.py

from zycelium.dataconfig import dataconfig


@dataconfig
class Config:
    name: str = "World"

config = Config().load()

print(f"Hello, {config.name}!")

Create a config.ini file in the same directory as example.py

name = "DataConfig"

Finally, from the same directory, run python example.py , your console session should look something like this:

$ python example.py
Hello, DataConfig!

The defaults here are:

• Config file name: "config.ini"
• Paths to look for the config file (current working directory): ["."]

Specify file-name for configuration:

from zycelium.dataconfig import dataconfig


@dataconfig(file="custom_config.ini")
class Config:
    name: str = "World"

config = Config().load()

print(f"Hello, {config.name}!")

In this example, we specify the file-name on this line: @dataconfig(file="custom_config.ini") with keyword arguments file="custom_config.ini" passed to @dataconfig().

Specify file-lookup-paths:

from zycelium.dataconfig import dataconfig


@dataconfig(paths=[".", "examples", "/usr/local/etc"])
class Config:
    name: str = "World"

config = Config().load()

print(f"Hello, {config.name}!")

Here, we pass paths=[".", "examples"] to @dataconfig() to specify the paths on filesystem where dataconfig should look for the default "config.ini" file. We can also specify the filename along with the paths. Paths can be relative to current working directory or absolute.

Save configuration to file:

from zycelium.dataconfig import dataconfig

FILE_NAME = "newconfig.ini"

@dataconfig(file=FILE_NAME)
class Config:
    name: str = "World"

config = Config()
config.save()

print(f"Saved config to file: {FILE_NAME}.")

Here, we set the config-file-name while creating the class, when save() is called, it will create the file and save contents of Config.

If we try running the same example again, we will get an error:

FileExistsError: File newconfig.ini exists, refusing to overwrite.

This is to protect us from accidentally overwriting an existing config file. To overwrite it, pass overwrite=True to save() like this:

config.save(overwrite=True)

Frozen configuration:

from zycelium.dataconfig import dataconfig


@dataconfig(frozen=True)
class Config:
    name: str = "World"

config = Config().load(replace=True)

print(f"Hello, {config.name}!")

To load a frozen config, we need to pass replace=True to load(), if we forget, we get the error:

dataclasses.FrozenInstanceError: cannot assign to field 'name'

Once loaded, we cannot overwrite the configuration.

Use with Click Integration for CLI apps:

Here, dataconfig will generate options for click CLI framework, one to add defaults to all options with names that exist in the dataconfig class, overridden by values found in the configuration file. These options can be overridden by passing values as usual to the command line.

There's also a new option added to the command: "--conf", which can be used to specify a different configuration file to load defaults.

And finally, any changes made in the command line are applied to the dataconfig object, but not saved to the configuration file unless the save() method is called later.

Frozen dataconfig does not work with commandline integration.

import click
from zycelium.dataconfig import dataconfig


@dataconfig
class Config:
    name: str = "World"


config = Config()
# No need to load() config when using click_option()


@click.command()
@click.option("--name")
@config.click_option()
def main(name):
    print(f"Hello, {name}!")
    print(f"Hello, {config.name}!")


main()

For more examples:

Read through the tests/ directory, where you will find the expected usage and how and why dataconfig can fail.

Install

From PyPI

pip install zycelium.dataconfig

From source:

git clone https://github.com/zycelium/dataconfig.git
cd dataconfig
pip install -e .