Source code for wolframclient.utils.dispatch

from __future__ import absolute_import, print_function, unicode_literals

import inspect

from wolframclient.utils.functional import flatten

# original idea by Guido in person.
# https://www.artima.com/weblogs/viewpost.jsp?thread=101605


[docs]class Dispatch(object): """ A method dispatcher class allowing for multiple implementations of a function. Each implementation is associated to a specific input type. Implementations are registered with the annotation :meth:`~wolframclient.utils.dispatch.Dispatch.dispatch`. The Dispatch class is callable, it behaves as a function that uses the implementation corresponding to the input parameter. When a type is a subtype, the type and its parents are checked in the order given by :data:`__mro__` (method resolution order). *Example:* method :meth:`~wolframclient.utils.dispatch.Dispatch.resolve` applied to an instance of :class:`collections.OrderedDict`, check for the first implementation to match with :class:`collections.OrderedDict`, then with :class:`dict`, and ultimately to :data:`object`. Once the mapping is determined, it is cached for later use. """ def __init__(self): self.clear()
[docs] def dispatch(self, *args, **opts): """ Annotate a function and map it to a given set of type(s). Declare an implementation to use on :data:`bytearray` input:: @dispatcher.dispatch(bytearray) def my_func(...) The default implementation is associated with :data:`object`. Set a default:: @dispatcher.dispatch(object) def my_default_func(...) A tuple can be used as input to associate more than one type with a function. Declare a function used for both :data:`bytes` and :data:`bytearray`:: @dispatcher.dispatch((bytes, bytearray)) def my_func(...) Implementation must be unique. By default, registering the same combination of types will raise an error. Set `replace_existing` to :data:`True` to update the current mapping. Or, set `keep_existing` to :data:`True` to ignore duplicate registration and keep the existing mapping. """ def register(func): return self.register(func, *args, **opts) return register
[docs] def update(self, dispatch, **opts): """ Update current mapping with the one from `dispatch`. `dispatch` can be a Dispatch instance or a :class:`dict`. `**opts` are passed to :meth:`~wolframclient.utils.dispatch.Dispatch.register` """ if isinstance(dispatch, Dispatch): dispatchmapping = dispatch.dispatch_dict elif isinstance(dispatch, dict): dispatchmapping = dispatch else: raise ValueError("%s is not an instance of Dispatch" % dispatch) for t, function in dispatchmapping.items(): self.register(function, t, **opts)
[docs] def validate_types(self, types): for t in frozenset(flatten(types)): if not inspect.isclass(t): raise ValueError("%s is not a class" % t) yield t
[docs] def register(self, function, types=object, keep_existing=False, replace_existing=False): """ Equivalent to annotation :meth:`~wolframclient.utils.dispatch.Dispatch.dispatch` but as a function. """ if not callable(function): raise ValueError("Function %s is not callable" % function) if keep_existing and replace_existing: raise ValueError( "Option values keep_existing and replace_existing cannot be both True." ) self.clear_cache() for t in self.validate_types(types): if replace_existing: self.dispatch_dict[t] = function elif t in self.dispatch_dict: if not keep_existing: raise TypeError("Duplicated registration for input type(s): %s" % (t,)) else: self.dispatch_dict[t] = function return function
[docs] def unregister(self, types=object): """ Remove implementations associated with types. """ self.clear_cache() for t in self.validate_types(types): try: del self.dispatch_dict[t] except KeyError: pass
[docs] def clear(self): """ Reset the dispatcher to its initial state. """ self.dispatch_dict = dict() self.dispatch_dict_cache = dict()
[docs] def clear_cache(self): if self.dispatch_dict_cache: self.dispatch_dict_cache = dict()
[docs] def resolve(self, arg): """ Return the implementation better matching the type the argument type. """ for t in arg.__class__.__mro__: try: return self.dispatch_dict_cache[t] except KeyError: impl = self.dispatch_dict.get(t, None) if impl: self.dispatch_dict_cache[t] = impl return impl return self.default_function
[docs] def default_function(self, *args, **opts): """ Ultimately called when no type was found. """ raise ValueError("Unable to handle args")
def __call__(self, arg, *args, **opts): return self.resolve(arg)(arg, *args, **opts)
[docs] def as_method(self): """ Return the dispatch as a class method. Create a new dispatcher:: dispatch = Dispatcher() Use the dispatcher as a class method:: class MyClass(object): myMethod = dispatch.as_method() Call the class method:: o = MyClass() o.myMethod(arg, *args, **kwargs) """ def method(instance, arg, *args, **opts): return self.resolve(arg)(instance, arg, *args, **opts) return method