gem install inverse_methods
require 'inverse_methods'
There is a single module InverseMethods which can be included where it's needed or patched on Object for global access.
To load it globally, use Object.include InverseMethods. Or use a more specific scope such as MyClass.include InverseMethods or MyClass.extend InverseMethods.
It can be used as a refinement: using InverseMethods.
Two methods which can be added on Object:
Both use debug_inspector to do some stuff with Ruby that wouldn't otherwise be possible.
pass_to is kind of like a reverse tap. It still works as a "tap" (it returns the caller) but takes arguments differently:
foo = []
1.pass_to %i{ foo.push foo.push }
# => 1
puts foo # => [1,1]
The caller is being passed as an argument (serialized using Marshal.dump), and the symbols in the list are methods which it is passed to. They are evaluated in the caller context, which is why the local variable foo can be referenced.
The method names passed to pass_to are not chained. This is why [1,1] was returned instead of [1, [...]] (the three dots being a recursive reference). The foo.push 1 calls happen independently of one another.
The above example, if modified like so, wouldn't work:
foo = []
1.pass_to [:foo.push, :foo.push]
puts foo
The reason is that :foo.push is a SyntaxError. It needs to be written as :"foo.push", which the %i{ array.of symbols } shorthand would do automatically.
The second method is chain_to, which is perhaps the 'functional programming' alternative to pass_to (and tap, which it's based on). It works similarly:
foo = []
[1].chain_to %i{[2].concat [3].concat}
puts foo # => [3,2,1]
Only the first symbol (:"[2].concat") gets the original argument, [1], passed when evaluated. The next symbol :"[3].concat" gets the first evaluation's result ([2, 1]) passed.