Dynamically Naming a Class Variable with a Param

[bivouac]

I swear this is like throwing a football or baseball left handed! I'm stumped on something that I think should be easy - but since I'm still new to Ruby, I'm not sure how to pull it off.

Basically, I want to be able to create a class var based on the value I pass in the constructor. Something like this:

# BEGIN CODE
class Foo

def initialize(var)

@#{var} = []

end

end

b = new.Foo("baz")
p b
# END CODE

I was hoping that I would get a new 'Foo' that has a class array called 'baz' - instead I get a syntax error. I just need a short way of constructing empty arrays when I instantiate an object. Otherwise, I guess I could statically name all of the arrays I want, but that doesn't seem really elegant.

And apologies about the code block - I couldn't get the Ruby to format correctly.

[KTottE]

What are you even trying to do?

Correct Ruby syntax for instantiating an object is: bar = Foo.new.

Why do you want to name the variable yourself from the outside of the object, an object's internal data is just that: An object's internal data.

Could you please elaborate on what you are trying to achieve, and maybe we could tell you how you're supposed to be doing it.

[bivouac]

Sorry about the syntax gaffe.

I'm trying to create a class called 'Address.' And this class should have two public attributes - 'location' and 'type.' It will have one public method called 'getAddress.' Based on the 'provider' and 'type' attributes, 'getAddress' will return a 'street' and a 'zip code.' The mechanics of the 'getAddress' method are to:

1. Load a CSV file that has the following elements ('location', 'street', 'zip')
2. Parse the file and load 'street' and 'zip' into a 'array' that should be named 'location.'

So let's say the CSV file has the following data:

  Home,123 Foo,90210
    Home,345 Wombat,00921
    Work,456 Baz,11210
    Work,991 Bazbat,92128

Now I have my class:

   class Address
 	def initialize(location, type="default")
 		@location = location
 		@type = type
 		# i want to initialize an empty array here that's named
 		# whatever the value that is given for the 'location' parameter
 		# lets just call it @array but ideally these would be class
 		# vars that could be called '@home' or '@work' based on 
 		# 'location' parameter when the object is instantiated
 	end
 
 	def getAddress()
 		CSV.open('csv_filename_here.csv', 'r') do | row |
 			# push the contents of the row into the initialized array
 		end
 
 		case @type
 			when "random"
 			    # return a random element from the array
 			when "default"
 			    # return the first element from the array
 		end
 	end
    end

I'm sure I'm doing this a backasswards - but when I pseudocoded it out, it seemed like a good way of handling the problem.

Maybe not.

[Luke Redpath]

It sounds to me like you've got candidates for two child classes, HomeAddress and WorkAddress. Perhaps some sort of factory class, AddressFactory that parses a CSV file and returns an array of Address objects. Your Address classes shouldn't know anything about a CSV file (what if you wanted to use a different source?).

Of course, you only need the sub-classes if there is a need for different behaviour depending on whether its a home or work address, otherwise, a single Address class will do. It still shouldn't know anything about where the data comes from though.

What I am confused about is why the array needs to be named according to the location. What difference does it make?

So, as an example:

class Address
  attr_reader :city, :zip, :type

  def initialize(type, city, zip)
    @city = city
    @zip = zip
    @type = type
  end
end

class AddressCsvReader
  attr_reader :addresses

  def initialize(path)
    @addresses = Array.new
    CSV.open(path, 'r') do |row|
      address = Address.new( row[0], row[1], row[2] )
      @addresses << address
    end
  end
end

reader = AddressCsvReader.new("my_csv_file.csv")
reader.addresses.each do |addr|
  puts "This is a #{addr.type} address in the city #{addr.city}"
end

Of course, because in the above example "reader.addresses" returns a normal array of Address objects, you can use any of the built-in Ruby address functions, such as "first". You might have to write your own "random" method on the AddressCsvReader class though.

[DougBTX]

I swear this is like throwing a football or baseball left handed!

Probably more like trying to throw one with your appendix, but just so you know how to do it, here's an example from the documentation:

   class Fred
     def initialize(p1, p2)
       @a, @b = p1, p2
     end
   end
   fred = Fred.new('cat', 99)
   fred.instance_variable_set(:@a, 'dog')   #=> "dog"
   fred.instance_variable_set(:@c, 'cat')   #=> "cat"
   fred.inspect                             #=> "#<Fred:0x401b3da8 @a=\"dog\", @b=99, @c=\"cat\">"

You should be able to call instance_variable_set from inside Fred, like this:

class Foo
  def initialize(var)
     instance_variable_set(var, [])
  end
end

You say you think it should be easy - what does the code look like in a language you are more familiar with?

What you are doing is setting the [b]name[b] of a variable dynamically. That's quite unusual. Normally you change the values of variables, not what the variables are themselves. If you want to set a key (a name) and a value at the same time, a hash is much more common. For example:

def add_a_key_and_a_value(key, value)
  @params[key] = value
end

That way you still know what the name of the variable is: @params. If you set the name of the variable dynamically, it is hard to find the variable again, because you don't know what it is called!

As an aside, getAddress is a bit foreign in Ruby. CamelCase is only really used in class and module names, where the case of the first letter is important. So if you wanted to use two words, it would be get_address.

Using "get" at the start of a method name is a little unusual too. Ruby supports this syntax for getters and setters, so an explicit "get" or "set" isn't needed:

class Fred

  def foo=(val)
    @foo = val
  end

  def foo
    @foo
  end

end

So normally you would use something more descriptive than "get" if you needed something special.

Douglas

[bivouac]

You know the more I think about it - the more I think you guys are right. And I apologize for the real lack of clarity with my request, this is a work problem and of course I'm translating the problem space from work to example poorly.

I think I'm going about this the totally wrong way - never have I ever wanted to dynamically set the name of a var inside of function, let along a class method. The factory class that Luke mentions is definitely worth investigating and Douglas' mention of the hash also has some merit (too bad data structures scare me).

In any case, everything that you guys have posted has been immeasurably helpful. I really appreciate you guys putting the time and effort forward to helping me out. Going to try and re-attack this problem tonight with some of the tips given here and some of the tips I got @ work and see if I can't come up with something. I'll repost if anyone's interesting.

Thanks again.

[stereofrog]

def add_a_key_and_a_value(key, value)
  @params[key] = value
end

Perhaps more common is to override [] and []=

class A
	def initialize(params)
		@params = params
	end
	def [](key)
		@params[key]
	end
	def []=(key, value)
		@params[key] = value
	end
end

a = A.new('name' => 'Fred', 'age' => 20)

puts a['name']
a['age'] += 10
puts a['age']

Delegation looks even more "rubysh":

require "delegate"

class B < SimpleDelegator
	def initialize(params)
		super(params)
	end
end

b = B.new('name' => 'Sally', 'age' => 17)

puts b['name']
puts b['age']