25
Ruby Enumerator : the what, the why, the how
Article originally published here : https://www.bootrails.com/blog/ruby-enumerator-what-why-how/
This section will give you a general understanding of a few important terms and their distinctions. Before we start using them, we need to understand the concept of Enumeration and the tools Ruby gives us for Enumeration. Also, we will be assuming that you are comfortable with blocks in Ruby.
Enumeration: As a concept simply means to traverse a list of items according to some logic. In programming, we often come across lists and the need to traverse these lists is a common programming necessity.
Enumerable: When using lists we commonly use either the for
loop or the .each
method to iterate over the items in them. We know the for
loop is a form of flow-control but where does this .each
method come from ? You might think it comes from whichever Class the list belongs to but in fact, it is an inherited method from the Enumerable
module.
This module, when included inside a Class that contains a set of elements, allows any Class to inherit a number of methods that one would need to traverse that set of elements.
There are many built-in enumerables in Ruby such as Array, Hash, and Range and all of them receive their enumeration capabilities by using the Enumerable
module.
This module relies on a method called .each
, which needs to be implemented in any Class it's included in. Other important methods like .map
, .reduce
and .select
that rely on the implementation of the .each
method to function can then be used for free.
When called with a block on an array, the .each
method will execute the block for each of the array's elements:
nums = [1,2,3] # An Array is an Enumerable
nums.each { |i| puts "* #{i}" }
# => 1
# => 2
# => 3
Enumerator: A class that is instantiated by either defining a Enumerator.new
or by calling an instance method of an Enumerable
object. So if we call the .each
method on an array without passing a block to execute for each of its elements, we'll receive an instance of Enumerator
. Sounds a bit confusing I know. Lets simplify with a basic example:
nums = [1,2,3]
puts nums.each
# => <Enumerator: 0x00007fa3657f90f8>
We previously discussed how the for
loop can also be used to iterate over a list. So why do we need to use the methods provided by the Enumerable
module? Well, you see when we use a for
loop we risk introducing a bug into our code by inadvertently overwriting a previously declared variable's value.
Sample below:
fighter = 'Jackie'
fighters_list = ['Bruce', 'Rocky', 'Rambo']
for fighter in fighters_list
puts fighter
end
puts "Your Fighter has changed => " + fighter # Unintended change
Console output:
=> Bruce
=> Rocky
=> Rambo
=> Your Fighter has changed => Rambo
Clearly, this is not desirable. This is where Enumerable#each
can be pretty handy. If we were to use the following code, the value of the "fighter" variable would remain unchanged.
fighters_list.each { |fighter| puts fighter }
This is why it is almost always better to use the .each
method provided by an Enumerable to iterate over it. There are also many other common things we wish to do with lists such as reduce a list down to a single value like its sum, modify the whole list, and write our own custom logic to iterate over a list if we want to protect some data on the client-side. We can do all of that and more if we use an Enumerable
. We will discuss usage in detail in just a bit.
So how do we end up using Enumerable in our code? You have a lot of freedom in this and there are plenty of places where using enumerables is useful.
You can use the .each method to traverse over the list, but what if we wanted to modify the list and our mapping logic uses the index of each element. The Enumerable#map
method seems like a good first thought. However, this can modify the list but cannot track indexes of individual elements.
nums_enum = [1, 2, 3].map
nums_enum.each { |num| puts num }
# => 1
# => 2
# => 3
We also have the Enumerable#each_with_index
method. This won't modify the list but we can at least track indexes with it. You can already see where I am going with this. Yes indeed with the help of Enumerators
we can chain these two together to create our very own "Map with Index"-like function call ! Sample below:
nums_enum = [1, 2, 3].map # Called without a block so returns #Enumerator
# => #<Enumerator: ...>
p nums_enum.each_with_index{ |n, i| n * i } # Called with block so will iterate with index
# => [0, 2, 6]
p [1,2,3].map.each_with_index{ |n, i| n * i } # Shortened to a single line
# => [0, 2, 6]
You can play around with this and chain a large number of methods as long as it serves your goals.
10.times.reverse_each.cycle.first(11)
=> [9, 8, 7, 6, 5, 4, 3, 2, 1, 0, 9]
In some cases, you may want to have manual control over your iterations. The Enumerable module provides the .next
method and you can call this on an Enumerator as well:
nums_enum = [1, 2].each
nums_enum.next
# => 1
nums_enum.next
# => 2
nums_enum.next
# => `next': iteration reached an end (StopIteration)
The beauty of Enumeration in Ruby is the amount of customization it provides. What if you wanted to half custom traversal logic for some list or Class.
digits = (1..10).to_a # Range to Array
def odd_digits(digits)
index = -1
Enumerator.new do |yielder|
loop do
index += 1
yielder << digits[index] if digits[index] % 2 == 0
end
end
end
Console:
$> puts odd_digits(digits).first(4)
=> 2
=> 4
=> 6
=> 8
By far the coolest and most detailed implementation use case is when you have a custom class that contains a set of elements. As discussed previously we will need to add two things to the Class to grant it the powers of Enumeration. Firstly we need to include the Enumerable
module. Secondly, we need to implement the each
method in the class to iterate over the elements. Most cases allow falling back to the each
method of another object such as an array.
Let' implement a linked list Class with a custom structure so that it cannot rely on the Array#each
method to iterate over the nodes.
class LinkedList
def initialize(head, tail = nil)
@head, @tail = head, tail
end
def add(item)
LinkedList.new(item, self)
end
end
LinkedList.new(0).add(5).add(10)
# => <LinkedList:0x1 @head=10, @tail=<LinkedList:0x2 @head=5, @tail=<LinkedList:0x3 @head=0, @tail=nil>>>
We have created our linked list but have no way of iterating over the individual nodes/elements. To do this, we finally visit the last and possibly most important piece of code.
class LinkedList include Enumerable # We inherit enumerable methods that use .each
def initialize(head, tail = nil)
@head, @tail = head, tail
end
def add(item)
LinkedList.new(item, self)
end
def each(&block) # Implement our custom each
if block_given?
block.call(@head)
@tail.each(&block) if @tail
else
to_enum(:each) # Return enumerator if block not provided
end
end
end
So what does this code do? The inclusion of Enumerable
is self-explanatory by this point. More importantly, we implement the .each
method to let other enumerable methods know how to iterate our linked list.
In the .each
method if a block is given we simply call the block on the current node and recursively call .each on the rest of the list until the final node is nil
. If a block was not provided then return an Enumerator that uses our .each method. This last bit will allow the chaining of methods discussed in an earlier use case.
Console:
$> linked_list = LinkedList.new(1).add(5).add(10)
$> linked_list.each{ |node| puts node }
=> 0
=> 5
=> 10
$> linked_list.select{ |node| node % 5 == 0 } # Select if divisible by 5
=> [10, 5]
Congratulations! Our linked list implementation now has access to other useful methods like .map
and .select
as shown in the example above.
Hopefully, all of this made sense. In case some things are still hazy, then try practicing by creating your own enumerators and using them in your code.
Thanks for reading. That’s a wrap!
25