How does defining [square bracket] method in Ruby work?

Question

I am going through Programming Ruby - a pragmatic programmers guide and have stumbled on this piece of code:

class SongList
  def [](key)
    if key.kind_of?(Integer)
      return @songs[key]
    else
      for i in [email protected]
        return @songs[i] if key == @songs[i].name
      end
    end
    return nil
  end
end

I do not understand how defining [ ] method works?

Why is the key outside the [ ], but when the method is called, it is inside [ ]?

Can key be without parenthesis?

I realize there are far better ways to write this, and know how to write my own method that works, but this [ ] method just baffles me... Any help is greatly appreciated, thanks

Galloromance

Answer 1

Methods in ruby, unlike many languages can contain some special characters. One of which is the array lookup syntax.

If you were to implement your own hash class where when retrieving an item in your hash, you wanted to reverse it, you could do the following:

class SillyHash < Hash

  def [](key)
    super.reverse
  end

end

You can prove this by calling a hash with the following:

a = {:foo => "bar"}
 => {:foo=>"bar"} 
a.[](:foo)
 => "bar" 
a.send(:[], :foo)
 => "bar" 

So the def [] defined the method that is used when you do my_array["key"] Other methods that may look strange to you are:

class SillyHash < Hash

  def [](key)
    super.reverse
  end

  def []=(key, value)
    #do something
  end

  def some_value=(value)
    #do something
  end

  def is_valid?(value)
    #some boolean expression
  end

end

Just to clarify, the definition of a [] method is unrelated to arrays or hashes. Take the following (contrived) example:

class B
  def []
    "foo"
  end
end

 B.new[]
 => "foo" 

Control

I think the OP was asking why we wouldn't call it: my_array.[]("key") instead and how my_array["key"] could possibly work... – Ilex

so by definition, whenever I create [] method for some class in ruby, it knows that it is being used on some kind of array and expects (key) parameter wich it later puts in []? – Galloromance

@Galloromance If you define a method called [], then it will allow you to call ["key"] on an instance of that class. An Array or Hash are two examples of where ruby uses this internally. – Control

@Control one more question... can i specify key without (). like this: def [] key ...? – Galloromance

@Galloromance Yeah, you are allowed to omit the braces in the definition of a ruby method. – Control

@Control just noticed... it isn't a problem that SongList class from the example isn't inheriting from Array class? I can still create [] method? – Galloromance

@Galloromance Not at all. This was purely for my demonstration. The Array implementation just happens to be the one that utilizes the [] method well that people are most familiar with. Try and disassociate [] from an Array and just treat it as another method. Although this can be difficult (as my answer proves!) – Control

There's an infinite recursion in your code. I think you meant to pass the call to super. – Gae

@Control Fix your code. super[key] inside a method won't work. super(key).reverse would. super[key] = blah won't work. super(key, value) would. You should refresh yourself on how to use super in a ruby method! – Own

@Ben You are quite correct. I have updated the answer with super calls. – Control

Answer 2

It's just syntactic sugar. There are certain syntax patterns that get translated into message sends. In particular

a + b

is the same as

a.+(b)

and the same applies to ==, !=, <, >, <=, >=, <=>, ===, &, |, *, /, -, %, **, >>, <<, !==, =~ and !~ as well.

Also,

!a

is the same as

a.!

and the same applies to ~.

Then,

+a

is the same as

a.+@

and the same applies to -.

Plus,

a.(b)

is the same as

a.call(b)

There is also special syntax for setters:

a.foo = b

is the same as

a.foo=(b)

And last but not least, there is special syntax for indexing:

a[b]

is the same as

a.[](b)

and

a[b] = c

is the same as

a.[]=(b, c)

Arapaima

Your list got me thinking about Ruby's sugar. It's interesting that the string interpolator #{} (e.g. "say hi, #{my_name}") isn't sugar for a method call . The first page from the index of the pick axe book has a nice list. – Iveson

@SooDesuNe: It does however call to_s. – Bjorn

Answer 3

the square brackets are the method name like Array#size you have Array#[] as a method and you can even use it like any other method:

array = [ 'a', 'b', 'c']
array.[](0) #=> 'a'
array.[] 1  #=> 'b'
array[2]    #=> 'c'

the last one is something like syntactic sugar and does exactly the same as the first one. The Array#+ work similar:

array1 = [ 'a', 'b' ]
array2 = [ 'c', 'd' ]
array1.+(array2) #=> [ 'a', 'b', 'c', 'd' ]
array1.+ array2  #=> [ 'a', 'b', 'c', 'd' ]
array1 + array2  #=> [ 'a', 'b', 'c', 'd' ]

You can even add numbers like this:

1.+(1) #=> 2
1.+ 1  #=> 2
1 + 1  #=> 2

the same works with /, *, - and many more.

Grounds

Answer 4

It's an operator overloader, it overrides or supplements the behavior of a method inside a class you have defined, or a class the behavior of which you are modifying. You can do it to other operators different from []. In this case you are modifying the behavior of [] when it is called on any instances of class SongList.

If you have songlist = SongList.new and then you do songlist["foobar"] then your custom def will come into operation and will assume that "foobar" is to be passed as the parameter (key) and it will do to "foobar" whatever the method says should be done to key.

Try

class Overruler
    def [] (input)
          if input.instance_of?(String)
            puts "string"
          else
            puts "not string"
          end
     end
end
foo = Overruler.new
foo["bar"].inspect
foo[1].inspect

Latrena