# encoding: utf-8 module Nanoc # Adds support for memoizing functions. # # @since 3.2.0 module Memoization # Memoizes the method with the given name. The modified method will cache # the results of the original method, so that calling a method twice with # the same arguments will short-circuit and return the cached results # immediately. # # Memoization assumes that the current object as well as the function # arguments are immutable. Mutating the object or the arguments will not # cause memoized methods to recalculate their results. There is no way to # un-memoize a result, and calculation results will remain in memory even # if they are no longer needed. # # @example A fast fib function due to memoization # # class FibFast # # extend Nanoc::Memoization # # def run(n) # if n == 0 # 0 # elsif n == 1 # 1 # else # run(n-1) + run(n-2) # end # end # memoize :run # # end # # @param [Symbol, String] method_name The name of the method to memoize # # @return [void] def memoize(method_name) # Alias original_method_name = '__nonmemoized_' + method_name.to_s alias_method original_method_name, method_name # Redefine define_method(method_name) do |*args| # Get cache @__memoization_cache ||= {} @__memoization_cache[method_name] ||= {} # Recalculate if necessary if !@__memoization_cache[method_name].key?(args) result = send(original_method_name, *args) @__memoization_cache[method_name][args] = result end # Done @__memoization_cache[method_name][args] end end end end