An open API service indexing awesome lists of open source software.

https://github.com/zipfile/fiicha

A minimalistic feature flag implementation with type hints
https://github.com/zipfile/fiicha

Last synced: 7 months ago
JSON representation

A minimalistic feature flag implementation with type hints

Awesome Lists containing this project

README

          

======
Fiicha
======

Fiicha is minimalistic mypy-friendly feature flag implementation with no
external dependencies.

Installation
============

.. code-block:: sh

pip install fiicha

For Development
---------------

Clone the repo, change current working dir to the root of the repo and execute:

.. code-block:: sh

python -m venv env
. env/bin/activate
pip install -e .[test,lint]

Usage
=====

Basic
-----

Basic usage is simple:

1. Sublcass from ``FeatureFlags``
2. Add ``FeatureFlag`` attributes representing individual feature flags.

.. code-block:: python

from fiicha import FeatureFlag, FeatureFlags

class MyProjectFeatureFlags(FeatureFlags):
release_x = FeatureFlag("Enable features from upcoming release X")
use_new_algorithm = FeatureFlag("Use new algorithm for ...")

ff = MyProjectFeatureFlags({"release_x": True})

print(ff.release_x) # True
print(ff.use_new_algorithm) # False

In addition, you kan specify a key defining default state of the feature flags:

.. code-block:: python

ff = MyProjectFeatureFlags({"all": True, "release_x": False}, default_key="all")

print(ff.release_x) # False
print(ff.use_new_algorithm) # True

Classes subclassed from ``FeatureFlags`` has full type hint support, so
when you are accessing removed or otherwise non-existent feature flags, you
can detect it before running tests:

.. code-block:: python

ff.xxx # will trigger mypy's attr-defined error

Utils
-----

The package comes with few util functions to avoid building your own framework
for simple use cases.

Single-line Configuration
~~~~~~~~~~~~~~~~~~~~~~~~~

.. code-block:: python

from fiicha import parse_feature_flags_string

flags_string = "release_x !use_new_algorithm" # stored in config
ff = MyProjectFeatureFlags(parse_feature_flags_string(flags_string))

print(ff.release_x) # True
print(ff.use_new_algorithm) # False

Load Flags From Environment Variables
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. code-block:: python

import os
from fiicha import feature_flags_from_environ

os.environ["MYPROJ_FEATURE_RELEASE_X"] = "True" # can also be 1, t, yes, on
os.environ["MYPROJ_FEATURE_USE_NEW_ALGORITHM"] = "False" # can also be 0, f, no, off or empty string

ff = MyProjectFeatureFlags(feature_flags_from_environ("MYPROJ_FEATURE_"))

print(ff.release_x) # True
print(ff.use_new_algorithm) # False

Automatically Document Your Feature Flags
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. code-block:: python

from fiicha import make_napoleon_doc, make_sphinx_doc

class GoogleStyleDocFeatureFlags(FeatureFlags, make_doc=make_napoleon_doc):
a = FeatureFlag("Enable feature A")
b = FeatureFlag("Enable feature B")

class SphinxStyleDocFeatureFlags(FeatureFlags, make_doc=make_sphinx_doc):
x = FeatureFlag("Enable feature X")
y = FeatureFlag("Enable feature Y")

print(GoogleStyleDocFeatureFlags.__doc__)
print(SphinxStyleDocFeatureFlags.__doc__)

Context Variable Support
~~~~~~~~~~~~~~~~~~~~~~~~

You might want to do A/B testing or otherwise toggle feature flags for
individual users without affecting the rest of the system. For such cases,
there is ``FeatureFlagsContext`` - a wrapper around ``ContextVar``. It copies
current feature flags and sets them as new context variable on enter and
resets them back on exit. This way you will be able to achieve global feature
flags protection from changes made within context of a request or a task.

.. code-block:: python

from contextvars import ContextVar
from fiicha import FeatureFlag, FeatureFlags, FeatureFlagsContext

class MyFeatureFlags(FeatureFlags):
a = FeatureFlag("Enable feature A")
b = FeatureFlag("Enable feature B")

root = MyFeatureFlags()
var: ContextVar[MyFeatureFlags] = ContextVar("test", default=root)
ff_ctx = FeatureFlagsContext(var)

with ff_ctx:
ff = var.get()
ff.a = True

assert ff is not root # is a copy

assert not root.a # not changed

Advanced
--------

See ``examples`` subfolder within the repo.