gistfile1.rb

require "nokogiri"

class SitemapBuilder
  attr_reader :root
  
  def initialize
    @urls = Array.new
    @root = File.expand_path("public", Rails.root)
    Dir[File.join(root, "sitemap*")].each{|f| FileUtils.rm f}
  end

  def add(url)
    @urls << url
  end

  def save(domain, name, defaults = {})
    defaults.reverse_merge!(:changefreq => 'monthly', :priority => 0.5, :lastmod => DateTime.now.utc.xmlschema)

    urls_groups = @urls.in_groups_of(50000)
    urls_groups.each_with_index do |urls, i|
      urls.compact!
      sitemap = Nokogiri::XML::Builder.new do |xml|
        xml.urlset(:xmlns => "http://www.sitemaps.org/schemas/sitemap/0.9") do
          urls.each do |url|
            url = url.reverse_merge(defaults)
            xml.url do
              xml.loc "http://#{domain}#{url[:loc]}"
              xml.lastmod url[:lastmod]
              xml.changefreq url[:changefreq]
              xml.priority url[:priority]
            end
          end
        end
      end
      Zlib::GzipWriter.open(File.expand_path("public/sitemap-#{name}-#{i}.xml.gz", Rails.root)) do |file|
        file << sitemap.to_xml
      end
    end

    sitemap_index = Nokogiri::XML::Builder.new do |xml|
      xml.sitemapindex(:xmlns => "http://www.sitemaps.org/schemas/sitemap/0.9") do
        urls_groups.length.times do |i|
          xml.sitemap do
            xml.loc "http://#{domain}/sitemap-#{name}-#{i}.xml.gz"
            xml.lastmod DateTime.now.utc.xmlschema
          end
        end
      end
    end
    File.open(File.expand_path("public/sitemap-#{name}.xml", Rails.root), "w") do |file|
      file << sitemap_index.to_xml
    end

  end
end