Functional Programming in Ruby with Blocks and Procs

Object-Oriented Programming vs Functional Programming

Ruby is a true object-oriented programming language. Generally, we love object-oriented (OOP) programming! Object-oriented programming makes our code readable, reusable, modular, and is generally pretty simple paradigm to get our heads around.

In Ruby, everything is an object. We generally divide up our code into Classes and Instances, so this isn't an unfamiliar concept. Really, everything!

Let's try it out!

self
# => main

self.class
# => Object

nil.class
# => NilClass

true.class
# => TrueClass

Even nil? Yes, nil is the singleton instance of NilClass.

Okay, what is this main thing? Well, main is the default top level context provided to us in Ruby. Since everything is an object, we call this the global object. But even our global object is an instance of Object.

In fact, the class Object is an instance of the class Class!

self.class.class
# => Class

We can override the class of Class and confirm for ourselves!

class Class
  alias prev_new new

  def new(*args)
    print "Creating a new #{self.name} class. \n"
    prev_new(*args)
  end
end

class Text
end

t = Text.new
# => ...
# => Creating a new Text class.

What About Methods?

When we this is where things get a little tricky. A Ruby method is just a snippet of code and not an object.

In other languages that consider the functional paradigm (in this case TypeScript), we can assign functions to variables and call them.

function ourFunction():void {
    console.log("hello");
}

let variable = ourFunction
variable()
// "hello"
// => undefined

Let's try a similar approach in Ruby.

def method
    puts "hello"
end

variable = method

variable()
# NoMethodError (undefined method `variable' for main:Object)
variable
# => nil

In Ruby, variable is assigned the return value of method. As we call method, it is implicitly invoked. Calling method, the method is invoked, the same as method(). In our current paradigm, there is no way we can pass the function itself to another variable.

Higher-order functions

In some ways, this limitation impedes our ability to create DRY code.

We also lose the ability to create higher-order functions.

Higher-order function: a function that takes a function as a parameter or returns a function.

Let's take a look at a simple higher-order function in TypeScript.

function multiplier(number1: number): (number2: number) => number {
    return function(number2: number): number {
        return number1 * number2
    }
}

let doubler = multiplier(2)
doubler(6)
// => 12

In Ruby, we can echo some of the principles of functional programming using the powers of blocks and Procs! Let's give it a shot!

Proc 101

Before we get into the nitty-gritty code, let's pause and better understand procs. A Proc is a special type of Ruby object that allows us to store a block of code as an object. As we instantiate our Proc instance, we can pass a block to the instantiation immediately following any method parameters.

print_greetings = Proc.new() do
  puts "Welcome!"
  puts "Bonjour!"
  puts "¡Bienvenidas!"
end
print_greetings
# => #<Proc:0x00007fe0ff042a08>

We can call out proc in a few different ways:

  • Chaining the method .call
  • Chaining method .yield
  • Using bracket notation []
  • Invoking .call with syntactic sugar .() 
print_greetings.call
print_greetings.yield
print_greetings[]
print_greetings.()

# Welcome!
# Bonjour!
# Bienvenidas!
# => nil

Block Parameters

Similar to regular methods, a blocks can receive a parameter. In fact, they have a few things in common:

  • Block parameters can have default values
  • Block parameters can be set up to accept a keyword
  • We can use the splat operator * to allow an undetermined amount of arguments to be captured.

There are a few different rules when it comes to block parameters:

  • If an argument is not provided, nil will be assigned to the parameter
  • Arguments are placed inside the pipes following the opening of the block

Let's create a proc that receives an argument as a proc.

multiply_by_two = Proc.new do |item|
    item * 2
end

To transform our Proc back into a block, we can use the & operator inside of our method calls that expect a block. The two examples below are equivalent.

[1,2,3,4,5].map do |item|
     item * 2
end

[1,2,3,4,5].map(&multiply_by_two)
# => [2, 4, 6, 8, 10]

How cool is that! Let's review a few rules passing blocks to a method:

  • When passing a proc transformation, it must come as the final argument of the method e.g. .ourFunction(1, 2, &multiply_by_two)
  • We can only pass one block to a method. Trying to call [1,2,3,4,5].map(&multiply_by_two, &multiply_by_three) would produce a SyntaxError.

Re-creating Higher-order Functions with Procs

Let's revisit the higher-order function from our example above and re-create it with our newfound knowledge of blocks and Procs.

def multiplier(number1)
    Proc.new {|number2| number1 * number2}
end

doubler = multiplier(2)
# => #<Proc:0x00007fe7719b91b8>

doubler.call(6)
# => 12

Understanding yield

If you're familiar with Rails, you have likely seen the yield keyword. Allowing you to inject parts of your .erb inside of templates. Similarly, in plain old Ruby, the yield pauses execution of the current code and yields to the block that was passed.

def our_method
  puts "top of method"
  yield
  puts "bottom of method"
end

our_method {puts "we are inside the block"}
# top of method
# we are inside the block
# bottom of method
# => nil

We can pass parameters to the block by passing them following yield.

def our_method_w_parameters
  puts "top of method"
  yield("karson", "nyc")
  puts "bottom of method"
end

our_method_w_parameters do |name, loc|
  puts "my name is #{name}, and I live in #{loc}"
end

# top of method
# my name is karson, and I live in nyc
# bottom of method
# => nil

But how reusable is this code when it's hard coded? Let's pair this newfound power with principles of OOP to use instance attributes with yield.

Proc Scope

Procs exist in the scope where they are defined, not in the scope where they are called. This can lead to some misleading and confusing references to self. Let's take a look at the example below.

class Person
    attr_accessor :name, :loc

    def initialize(name, loc)
        @name = name
        @loc = loc
    end

    def ex_block
        yield
    end
end

k = Person.new("karson", "nyc")
# => #<Person:0x00007fded014edb0 @name="karson", @loc="nyc">
k.ex_block {puts self.name, self.loc}
# NoMethodError (undefined method `name' for main:Object)

self refers to the main object which tells us that we are in the global scope. If we want to bind self to the instance where the block is called, it must be defined upon instantiation.

class Person
    attr_accessor :name, :loc, :instance_proc

    def initialize(name, loc)
        @name = name
        @loc = loc
        @instance_proc = Proc.new() do
                puts self.name, self.loc
            end
    end

    def ex_proc(&proc)
        yield
    end

end
k = Person.new("karson", "nyc")
# => #<Person:0x00007ff6aa94c228 @name="karson", @loc="nyc", @instance_proc=#<Proc:0x00007ff6aa94c1d8>>
k.ex_proc(&k.instance_proc)
# karson
# nyc

Ending Challenge

Using what you now know about Procs and blocks, let's re-create the method .map on our custom class MArray.

class MArray < Array
  def initialize(*args)
    super(args)
  end

  def map(&block)
    newMArr = MArray.new()
    for element in self
      newMArr << yield(element)
    end
    newMArr
  end
end

m = MArray.new(1, 2, 3, 4, 5)
# =>[1, 2, 3, 4, 5]
m.map { |item| item * 2 }
# => [2, 4, 6, 8, 10]

Conclusion

Procs are a powerful Ruby concept that allow us to keep code DRY and implement features that play of OOP concepts and functional programming. Check out the official Ruby-Doc documentation on Procs and let me know your thoughts in the comments blow!

19

Categories:

Ruby

Tags:

RubyFunctionalBeginners