Middleman xml sitemap (to aid Google and other search engines)

sitemap.xml.builder

# encoding: utf-8
xml.instruct!

# Drop this file in your source directory
#
# Uses the builder gem.
# Add
# gem 'builder', '~> 2.0' 
# to the Gemfile, and run builder install
#
# Add this line to config.rb: 
# page "/sitemap.xml", :layout => false  
# 
# To tell Google and other search services about the sitemap, add
# Sitemap: http://www.your-domain.com/sitemap.xml   
# to the robots.txt file. Or see
# http://support.google.com/webmasters/bin/answer.py?hl=en&answer=183669&topic=8476&ctx=topic for other options.
#
# For details on how to construct a site map, see
# http://www.sitemaps.org/protocol.html
# http://support.google.com/webmasters/bin/answer.py?hl=en&answer=156184&topic=8476&ctx=topic
#
# Remember to exchange www.your-domain.com for your domain (line 64)
# and to set an apropriate update frequency for your site (line 84)

xml.urlset "xmlns" => "http://www.sitemaps.org/schemas/sitemap/0.9" do
  page_count = 0
  sitemap.resources.each do |page|
    catch :next_page do

      # If all you want the sitemap to cover is normal html pages, use this:
      # throw :next_page unless page.url.end_with?('.html')
      #
      # If you need more detailed control, this is a starting point:
      throw :next_page if page.url.start_with?('/images/')
      throw :next_page if page.url.start_with?('/javascripts/')
      throw :next_page if page.url.start_with?('/stylesheets/')
      throw :next_page if page.url.start_with?('/layouts/')
      throw :next_page if page.url.start_with?('/apple-touch-icon')
      throw :next_page if page.url.end_with?('.xml')
      throw :next_page if page.url.start_with?('/.')   # .htaccess, .DS_Store, .git etc.
      throw :next_page if page.url == '/robots.txt'
      throw :next_page if page.url == '/favicon.ico'
      throw :next_page if page.url == '/humans.txt'

      # Exclude drafts
      throw :next_page if defined?(page.data) && page.data['published'] == false
      throw :next_page if defined?(page.data) && page.data['date'] && page.data['date'].to_time > Time.now

      # Exclude archive indexes
      #     I prefer not to have the calendar archive index pages in the sitemap.
      #     Remove this if you want them.
      (2000..2030).each do |year|
        # Exclude year index, directories and subdirectories
        throw :next_page if page.url == "/#{year}.html" || page.url == "/#{year}" || page.url.start_with?("/#{year}/")
      end

      # Count pages
      page_count += 1
      raise "Over 50 000 pages, to much for a single sitemap" if page_count > 50000

      # Build xml of sitemap
      xml.url do
        xml.loc 'http://www.your-domain.com' + page.url   # TODO: possibly remove .html

        # As usual, it's hard to get any solid information on how the search engines are
        # using the information in the site map. lastmod (last modified) and changefreq
        # (change frequency) seams to overlap each other. My guess is that it's best to
        # use lastmod for pages where you know when the file changed, and changefreq for
        # dynamic pages, like the first page of a blog and archive pages, where it's
        # hard to know when the dynamic information was changed.
        #
        # Possible values for changefreq
        # always
        # hourly
        # daily
        # weekly
        # monthly
        # yearly
        # never

        case page.url
          when '/index.html'
            xml.changefreq 'daily'    # Set this to a suitable value for your site
          else
            # Setting last modified (lastmod)
            #
            # Checks date in frontmatter and modification date of file, and uses the most recent.
            # (Note: Will not be able to get the correct date for pages with dynamic content)
            frontmatter_date = nil
            frontmatter_date = page.data['date'].to_time.getlocal if defined?(page.data) && page.data['date']
            file_mtime = nil
            file_mtime = File.mtime(page.source_file).to_time.getlocal if defined?(page.source_file)
            most_recent = [frontmatter_date, file_mtime].compact.max if  frontmatter_date || file_mtime

            xml.lastmod most_recent.strftime("%Y-%m-%d")  # Using the more recent of the two dates.
                                                          # Using only date, stripping clock
        end

        # Priority (priority)
        #
        # To adjust the priority of a page, set sitemap_priority: in the frontmatter.
        frontmatter_sitemap_priority = page.data['sitemap_priority'].to_f if (defined?(page.data) && page.data['sitemap_priority'])
        if frontmatter_sitemap_priority && frontmatter_sitemap_priority < 0 && frontmatter_sitemap_priority > 1
          raise "Incorrect sitemap_priority #{sitemap_priority} for #{page.destination_path}"
        end
        if frontmatter_sitemap_priority && frontmatter_sitemap_priority >= 0 && frontmatter_sitemap_priority <= 1
          xml.priority frontmatter_sitemap_priority unless frontmatter_sitemap_priority == 0.5 # (No use setting the default priority)
        end

      end
    end
  end
end

Thank you. Instead of "page.url.end_with?('.html')" I'd rather use "page.destination_path.end_with?('.html')" or it won't add index.html files.

For those that get confused like I did setting this up, @iparr is correct that line should be bundle install as builder is run the next time middleman site is built.

Also, if you're using dynamic pages, you'll want to do as @pointum suggests.

I found it useful to add @ line 49:

# Exclude priority zero 
throw :next_page if defined?(page.data) && page.data['sitemap_priority'].to_f == -1

Which allows for the setting of Priority = -1 and therefore allowing content editors the option of skipping individual pages in the XML Sitemap.