Brickset.com scraper

brickset.rb

require 'rubygems'
require 'nokogiri'
require 'open-uri'

class Brickset
	@@db = nil # database / cache
	class Set
		attr_accessor :id, :year, :pieces, :rrpp, :rrpd, :rating, :theme, :subtheme
		
		def initialize(id)
			begin
				# get data from web
				self.id = id
				# fetch brickset page
				html = Nokogiri::HTML(open("http://www.brickset.com/detail/?Set=#{id}-1"))

				# parse brickset details
				details = html.css("#menuPanel .menuPanel .setDetails li")
				
				details.each do |detail|
					set_property(detail)
				end
				
			rescue => e
				puts e.inspect
				return nil
			end
		end
		
		# set a brickset property from the nokogiri html node
		def set_property(li)
			attribute = li.children.first.content.strip
			value = li.children[li.children.size-1].content.strip
			case attribute
			when "Theme"
				self.theme = value
			when "Subtheme"
				self.subtheme = value
			when "Year released"
				self.year = value
			when "Pieces"
				self.pieces = value
			when "RRP"
				value =~ /(\d+).+\$(\d+)/
				self.rrpp = $1
				self.rrpd = $2
			end
		end
	end
end