cookpad_scraper.rb

require 'open-uri'
require 'nokogiri'

class CookpadInspector
  def initialize(recipe_id)
      @doc = Nokogiri::HTML(open("http://cookpad.com/recipe/#{recipe_id.to_i}"))
  end

  def get_title
    @doc.css("h1")[0].inner_html.strip
  end

  def get_ingredients
    ingredient_names = @doc.css(".ingredient_name span").map do |elem|
      next elem.inner_html if elem.child.is_a? Nokogiri::XML::Text
      elem.child.inner_html if elem.child.is_a? Nokogiri::XML::Element
    end

    ingredient_quantities = @doc.css(".ingredient_quantity").map do |elem|
      elem.inner_html
    end

    ary = [ingredient_names, ingredient_quantities].transpose
    Hash[*ary.flatten]
  end

  def get_process
    @doc.css(".step_text").map {|elem| elem.inner_html }
  end

end

begin
  c=CookpadInspector.new(ARGV[0].to_i)
rescue OpenURI::HTTPError
  p 'not found!'
  exit
end

p '=' * 20

p 'TITLE'
p c.get_title
p '=' * 20

p 'INGREDIENTS'
c.get_ingredients.each{|i| p i}
p '=' * 20

p 'PROCESSES'
c.get_process.each{|pr| p pr}

p '=' * 20