optify 0.9.0

Simplifies getting the right configuration options for a process using pre-loaded configurations from files to manage options for experiments or flights.

Optify Rust Bindings for Python

⚠️ Development in progress ⚠️
APIs are not final and may change, for example, interfaces may be used in the future or method names may change. This is just meant to be minimal to get started and help build a Python library.

See the homepage for details about how feature files are combined to build the options to process at runtime.

Usage

import json
from optify import OptionsProvider

provider = OptionsProvider.build('path/to/configs')
config = provider.get_options_json('key', ['feature_A', 'feature_B'])
print(config)

Outputs:

{
  "myArray": [
    "item 1",
    "item 2"
  ],
  "myObject": {
    "deeper": {
      "new": "new value",
      "num": 3333
    },
    "key": "val",
  },
  "rootString": "root string same"
}

See the tests directory for more examples.

Getting Typed Objects

The JSON output can be deserialized into typed objects using different libraries such as dacite (for dataclasses) or Pydantic.

Using dacite with dataclasses

Install with pip install dacite

import json
from dataclasses import dataclass

import dacite

from optify import OptionsProvider


@dataclass
class Deeper:
    newValue: str
    num: int


@dataclass
class MyObject:
    key: str
    deeper: Deeper


@dataclass
class MyConfig:
    rootString: str
    myArray: list[str]
    myObject: MyObject


provider = OptionsProvider.build('path/to/configs')
options_json = provider.get_options_json('myConfig', ['feature_A', 'feature_B'])
config = dacite.from_dict(data_class=MyConfig, data=json.loads(options_json))

print(config.rootString)
print(config.myObject.deeper.newValue)

Using Pydantic

Install with pip install pydantic

import json

from pydantic import BaseModel

from optify import OptionsProvider


class Deeper(BaseModel):
    newValue: str
    num: int


class MyObject(BaseModel):
    key: str
    deeper: Deeper


class MyConfig(BaseModel):
    rootString: str
    myArray: list[str]
    myObject: MyObject


provider = OptionsProvider.build('path/to/configs')
options_json = provider.get_options_json('myConfig', ['feature_A', 'feature_B'])
config = MyConfig.model_validate_json(options_json)

print(config.rootString)
print(config.myObject.deeper.newValue)

Watching for Changes

Use OptionsWatcher instead of OptionsProvider during development to have it automatically reload the options it stores when the files change:

from optify import OptionsWatcher

watcher = OptionsWatcher.build('path/to/configs')

Development

Setup

pyenv virtualenv optify-dev
pyenv local optify-dev
pyenv activate optify-dev

pip install -e '.[dev]'

Build

maturin develop

Tests

pytest

Formatting

To automatically change the Rust code, run:

cargo fmt && cargo clippy --fix --allow-dirty --allow-staged

Publishing

A GitHub Action will automatically publish new versions: https://github.com/juharris/optify/actions/workflows/python_publish.yml