A python package to build registries that can autodiscover values accross your project components
Persisting-theory is a small python utility designed to automate data discovering and access inside a list of packages. Use case: you are building an application that will have pluggable components. You want to allow these components to register data so it can be accessed by any other component of your app.
If you ever used Django framework, you may remember this:
from django.contrib import admin admin.autodiscover()
Basically, persisting-theory will do the same, except that it let you declare what you want to autodiscover.
Okay, I’m bad at explaining things, and english is not my mother tongue. Let’s build a simple example.
Install the package from PyPi. via pip (or any other tool):
pip install persisting-theory
Persisting-theory does not require any dependency but a python installation (it has been tested on python 2.7 and python 3.2).
A basic setup:
# registries.py from persiting_theory import Registry class CallbacksRegistry(Registry): """ Allow your apps to register callbacks """ # the package where the registry will try to find callbacks in each app look_into = "callbacks_registry" callbacks_registry = CallbacksRegistry() # app1/callbacks_registry.py from registries import callbacks_registry @callbacks_registry.register def dog(): print("Wouf") # app2/callbacks_registry.py from registries import callbacks_registry @callbacks_registry.register def cat(): print("Meow") # dosomething.py from registries import callbacks_registry APPS = ( 'app1', 'app2', ) # Trigger autodiscovering process callbacks_registry.autodiscover(APPS) for callback in callbacks_registry.values(): callback() # Wouf # Meow
Registry inherits from python built-in collections.OrderedDict, which means you can use regular dict methods to access registered data:
callbacks_registry.get("dog")() # will print Wouf assert callbacks_registry.get("chicken", None) is None
You can use this function as a decorator for registering functions and classes:
from persisting_theory import Registry class AwesomeRegistry(Registry): pass r = AwesomeRegistry() # register a class @r.register class AwesomeClass: pass # register a function @r.register def awesome_function(): pass # By default, the key in the registry for a given value is obtained from the function or class name, if possible assert r.get("AwesomeClass") == AwesomeClass assert r.get("awesome_function") == awesome_function # You can override this behaviour: @r.register(name="Chuck") class AwesomeClass: pass @r.register(name="Norris") def awesome_function(): pass assert r.get("Chuck") == AwesomeClass assert r.get("Norris") == awesome_function # You can also use the register method as is awesome_var = "Chuck Norris" r.register(awesome_var, name="Who am I ?") assert r.get("Who am I ?") == awesome_var # I f you are not registering a function or a class, you MUST provide a name argument
By default, a registry will accept any registered value. Sometimes, it’s not what you want, so you can restrict what kind of data your registry accepts:
from persisting_theory import Registry class StartsWithAwesomeRegistry(Registry): def validate(self, obj): if isinstance(obj, str): return obj.startswith("awesome") return False r = StartsWithAwesomeRegistry() # will pass registration r.register("awesome day", name="awesome_day") # will fail and raise ValueError r.register("not so awesome day", name="not_so_awesome_day")
If you have multiple registries, or want to allow your apps to declare their own registries, this is for you:
# registries.py from persisting_theory import meta_registry, Registry class RegistryA(Registry): look_into = "a" class RegistryB(Registry): look_into = "b" registry_a = RegistryA() meta_registry.register(registry_a, name="registry_a") registry_b = RegistryB() meta_registry.register(registry_b, name="registry_b") # dosomethingelse.py from persisting_theory import meta_registry # will import registries declared in `registries` packages, and trigger autodiscover() on each of them meta_registry.autodiscover(apps=("app1", "app2"))
What the hell is that name ?
It’s an anagram for “python registries”.
Contributions, bug reports, and “thank you” are welcomed. Feel free to contact me at <email@example.com>.
The project is licensed under BSD licence.