views:

621

answers:

1

Hi,

I'm trying to to define a class called "HTML" to extend Nokogiri - which uses modules.

I tried the following:

require 'nokogiri'

class HTML
    include Nokogiri
end

and

require 'nokogiri'

class HTML
    extend Nokogiri
end

but so far it's been impossible that the class HTML inherit of all the functions in nokogiri. So stuff like are not working:

doc = HTML.new
doc.HTML(open('http://www.google.com/search?q=tenderlove'))

undefined method `HTML' for # (NoMethodError)

Does anyone know how I could manage to program a class that inherits of all the methods of a module?

+3  A: 

Nokogiri doesn't have any instance methods to inherit:

irb> Nokogiri.instance_methods
#=>  []

But normally, you would use extend

% ri extend
---------------------------------------------------------- Object#extend
     obj.extend(module, ...)    => obj
------------------------------------------------------------------------
     Adds to obj the instance methods from each module given as a
     parameter.

        module Mod
          def hello
            "Hello from Mod.\n"
          end
        end

        class Klass
          def hello
            "Hello from Klass.\n"
          end        end

        k = Klass.new
        k.hello         #=> "Hello from Klass.\n"
        k.extend(Mod)   #=> #<Klass:0x401b3bc8>
        k.hello         #=> "Hello from Mod.\n"

%

What you want to do is use all the class methods of the Nokogiri module as instance methods of your class. Which is a bit nonstandard, which is why the syntax doesn't support it. Most programmers use ruby modules for the Singleton pattern - there only needs to be one Nokogiri, so other things shouldn't be able to use its methods.

You could do some hacking with UndefinedMethods to get around this, but considering that Nokogiri has some compiled code in the backend, this may produce undefined bugs.

Which isn't to say you can't forward calls to Nokogiri:

# nokogiri_wrapper.rb
require 'rubygems'
require 'nokogiri'

class NokogiriWrapper
  def method_missing(meth, *args, &blk)
    puts "call for #{meth.inspect}, #{args}, #{blk ? "with block" : "and no block"}"
    if Nokogiri.methods.include? meth.to_s
      puts "forwarding to Nokogiri"
      Nokogiri.send(meth, *args, &blk)
    else
      puts "falling back to default behaviour"
      super
    end
  end
end

html = "<html></html>"

puts "calling Nokogiri directly"
p Nokogiri.HTML(html)

wrapper = NokogiriWrapper.new

puts "calling Nokogiri through wrapper"
p wrapper.HTML(html)

puts "calling non-Nokogiri method with wrapper"
p(begin
    wrapper.scooby_dooby_doo!
  rescue NoMethodError => e
    [e.message, e.backtrace]
  end)
% ruby nokogiri_wrapper.rb
calling Nokogiri directly
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd"&gt;
<html></html>

calling Nokogiri through wrapper
call for :HTML, <html></html>, and no block
forwarding to Nokogiri
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd"&gt;
<html></html>

calling non-Nokogiri method with wrapper
call for :scooby_dooby_doo!, , and no block
falling back to default behaviour
["undefined method `scooby_dooby_doo!' for #<NokogiriWrapper:0x581f74>", ["nokogiri_wrapper.rb:12:in `method_missing'", "nokogiri_wrapper.rb:29"]]

This is one way to implement the delegator pattern in ruby (another way is to use one of the Delegator classes).

rampion
Thanks for that.
Benjamin