サイトマップ生成

sitemap.rb

#!/usr/bin/env ruby
require "rubygems"
require "open-uri"
require "mechanize"

module Sitemap
  
  class UriList

    def initialize
      @url_list = []
      @agent = WWW::Mechanize.new
    end

    def gather(origin)
      begin
        page = @agent.get origin
      rescue StandardError => e
        puts e.message
        puts origin
        abort 'error...'
      end

      page.links.each do |link|
        abs_url = (page.uri + link.uri).to_s
        host = URI.parse(abs_url).host
        next if abs_url.match(/(#.+|jpg|png|gif|jpeg)$/i) || !abs_url.include?(host)

        if [email protected]?(link.uri)
          gather abs_url
        else
          @url_list << abs_url
        end
      end if page.class == WWW::Mechanize::Page
      @url_list.uniq
    end
  end
  
  class Writer
    include Nokogiri::XML
    
    def initialize
      @doc = Document.new
      @doc.root = Element.new('urlset', @doc)
      @doc.root.set_attribute('xmlns', "http://www.sitemaps.org/schemas/sitemap/0.9")
    end
    
    def to_sitemap(url_list)
      url_list.each do |url|
        elm = Element.new('url', @doc)
        loc = Element.new('loc', @doc)
        loc.content = url
        elm.add_child(loc)
        @doc.root.add_child(elm)
      end
      
      @doc.to_xml
    end
          
  end
  
end
if __FILE__ == $0
  uris = Sitemap::UriList.new
  writer = Sitemap::Writer.new
  puts writer.to_sitemap(uris.gather('http://www.example.com/'))
end