py-params

Python Versions Downloads

A type safe dict utility class in python.

LICENSE

Install

params is on the Python Package Index (PyPI):

pip install py-params

Usage

Params represents a set of parameters modeled as a dict with a fixed set of keys. Default values are provided as class level attributes in Params subclasses. Parameter values can then be specified when constructing a Params instance overriding the default values. The parameter values can then be accessed both as attributes and dict items, however the Params instance key set is closed for modification thus an exception is raised when a parameter name is misspelled.

Accessing parameters not defined as class level attributes would raise an AttributeError.

>>> import params as pp

>>> class TestParams(pp.Params):
...     param_a = 1
...     param_b = True

>>> params = TestParams()                     ## using the defaults
>>> params
{'param_a': 1, 'param_b': True}

>>> TestParams(param_a=2)                     ## setting a value for param_a
{'param_a': 2, 'param_b': True}

>>> params.param_a = 3                          ## access as attribute or key
>>> params["param_a"] = 4
>>> params.param_a == params["param_a"]
True

>>> params.param_c
AttributeError: 'TestParams' object has no attribute 'test_c'

>>> params.param_c = 3
AttributeError: Setting unexpected parameter 'param_c' in Params instance 'TestParams'

>>> params["param_d"] = 4
AttributeError: Setting unexpected parameter 'param_d' in Params instance 'TestParams'

Params instances can be used to generate CLI parser with argparse:

>>> import params as pp

>>> class TestParams(pp.Params):
...     number_of_things = pp.Param(None, doc="Specifies the number of things", dtype=int, required=True)
...     use_feature_x    = pp.Param(True, doc="whether to use feature X")
>>> parser = TestParams.to_argument_parser()
>>> parser.print_help()
usage: pydevconsole.py [-h] --number-of-things NUMBER_OF_THINGS
                       [--use-feature-x [USE_FEATURE_X]]

optional arguments:
  -h, --help            show this help message and exit
  --number-of-things NUMBER_OF_THINGS
                        Specifies the number of things
  --use-feature-x [USE_FEATURE_X]
                        whether to use feature X
>>> args = parser.parse_known_args(["--number-of-things", "7"])
>>> TestParams(args._get_kwargs())
{'number_of_things': 7, 'use_feature_x': True}

NEWS

05.Feb.2021 - v0.10.2 - passing any kwargs to argparser add_argument() from Param.__init__

10.Jan.2021 - v0.10.1 - YAML (de)serialization added; support for positional argument in argparse.

04.Apr.2020 - WithParams mixin added.

31.Mar.2020 - support for generating argparse CLI parser. Hierarchy aggregation refactored.

Resources

As an illustration of how Params could be used to reduce boilerplate code check:

kpe/params-flow - utilities for reducing keras boilerplate code in custom layers
kpe/bert-for-tf2 - BERT implementation using the TensorFlow 2 Keras API

Metadata

Release files for py-params 0.10.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for py-params 0.10.2
File	Size	Uploaded
py-params-0.10.2.tar.gz	7.4 kB	Feb 5, 2021	Details

Release files / py-params-0.10.2.tar.gz

Download URL	py-params-0.10.2.tar.gz
Size	7.4 kB
Tags	Source
SHA-256 checksum How to use checksums	`aaac44a8c5942656b6260ba4231f74ddf4eef98eef8a16676d7767cd147c5c84`
BLAKE2b-256 checksum How to use checksums	`aae04f663d8abf83c8084b75b995bd2ab3a9512ebc5b97206fde38cef906ab07`
Upload date	Feb 5, 2021
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/3.3.0 pkginfo/1.7.0 requests/2.25.1 setuptools/53.0.0 requests-toolbelt/0.9.1 tqdm/4.56.0 CPython/3.7.1`