N.B. This fork is Ruby 1.8.6-compatible (and probably therefore not compatible with 1.8.7+).

SitemapGenerator

A Rails 3-compatible gem/plugin to generate 'enterprise-class' Sitemaps using a familiar Rails Routes-like DSL. Sitemaps are readable by all search engines and adhere to the 'Sitemap protocol specification'. Automatically pings search engines to notify them of new sitemaps (including Google, Yahoo and Bing). Provides rake tasks to easily manage your sitemaps. Supports image sitemaps and handles millions of links.

Features

v0.2.6: Support 'image sitemaps'!
v0.2.5: Support Rails 3!
Adheres to the 'Sitemap protocol specification'
Handles millions of links
Automatic Gzip of Sitemap files
Automatic ping of search engines to notify them of new sitemaps: Google, Yahoo, Bing, Ask, SitemapWriter
Won't clobber your old sitemaps if the new one fails to generate
Set the priority of links, change frequency etc
You control which links are included
You set the host name, so it doesn't matter if your application is in a subdirectory

Foreword

Unfortunately, Adam Salter passed away in 2009. Those who knew him know what an amazing guy he was, and what an excellent Rails programmer he was. His passing is a great loss to the Rails community.

Karl Varga has taken over development of SitemapGenerator. The canonical repository is http://github.com/kjvarga/sitemap_generator

Installation

Rails 3:

Add the gem to your Gemspec

gem 'sitemap_generator'
$ rake sitemap:install

Rails 2.x: As a gem

Add the gem as a dependency in your config/environment.rb

config.gem 'sitemap_generator', :lib => false
$ rake gems:install

Add the following to your RAILS_ROOT/Rakefile

begin
  require 'sitemap_generator/tasks'
rescue Exception => e
  puts "Warning, couldn't load gem tasks: #{e.message}! Skipping..."
end

$ rake sitemap:install

Rails 2.x: As a plugin

$ ./script/plugin install git://github.com/kjvarga/sitemap_generator.git

Installation creates a config/sitemap.rb file which will contain your logic for generating the Sitemap files. If you want to create this file manually run rake sitemap:install.

You can run rake sitemap:refresh as needed to create Sitemap files. This will also ping these 'major search engines': Google, Yahoo, Bing, Ask, SitemapWriter. If you want to disable all non-essential output run the rake task with rake -s sitemap:refresh.

To keep your Sitemaps up-to-date, setup a cron job. Pass the -s option to the rake task to silence all but the most important output. If you're using Whenever, then your schedule would look something like:

# config/schedule.rb
every 1.day, :at => '5:00 am' do
  rake "-s sitemap:refresh"
end

Optionally, you can add the following to your public/robots.txt file, so that robots can find the sitemap file:

Sitemap: <hostname>/sitemap_index.xml.gz

The Sitemap URL in the robots file should be the complete URL to the Sitemap Index, such as http://www.example.org/sitemap_index.xml.gz

Example 'config/sitemap.rb'

# Set the host name for URL creation
SitemapGenerator::Sitemap.default_host = "http://www.example.com"

SitemapGenerator::Sitemap.add_links do |sitemap|
  # Put links creation logic here.
  #
  # The Root Path ('/') and Sitemap Index file are added automatically.
  # Links are added to the Sitemap output in the order they are specified.
  #
  # Usage: sitemap.add path, options
  #        (default options are used if you don't specify them)
  #
  # Defaults: :priority => 0.5, :changefreq => 'weekly',
  #           :lastmod => Time.now, :host => default_host


  # Examples:

  # add '/articles'
  sitemap.add articles_path, :priority => 0.7, :changefreq => 'daily'

  # add all individual articles
  Article.find(:all).each do |a|
    sitemap.add article_path(a), :lastmod => a.updated_at
  end

  # add merchant path
  sitemap.add '/purchase', :priority => 0.7, :host => "https://www.example.com"

  # add all individual news with images
  News.all.each do |n|
    sitemap.add news_path(n), :lastmod => n.updated_at, :images=>n.images.collect{ |r| :loc=>r.image.url, :title=>r.image.name }
  end

end

# Including Sitemaps from Rails Engines.
#
# These Sitemaps should be almost identical to a regular Sitemap file except
# they needn't define their own SitemapGenerator::Sitemap.default_host since
# they will undoubtedly share the host name of the application they belong to.
#
# As an example, say we have a Rails Engine in vendor/plugins/cadability_client
# We can include its Sitemap here as follows:
#
file = File.join(Rails.root, 'vendor/plugins/cadability_client/config/sitemap.rb')
eval(open(file).read, binding, file)

Raison d'être

Most of the Sitemap plugins out there seem to try to recreate the Sitemap links by iterating the Rails routes. In some cases this is possible, but for a great deal of cases it isn't.

a) There are probably quite a few routes in your routes file that don't need inclusion in the Sitemap. (AJAX routes I'm looking at you.)

and

b) How would you infer the correct series of links for the following route?

map.zipcode 'location/:state/:city/:zipcode', :controller => 'zipcode', :action => 'index'

Don't tell me it's trivial, because it isn't. It just looks trivial.

So my idea is to have another file similar to 'routes.rb' called 'sitemap.rb', where you can define what goes into the Sitemap.

Here's my solution:

Zipcode.find(:all, :include => :city).each do |z|
  sitemap.add zipcode_path(:state => z.city.state, :city => z.city, :zipcode => z)
end

Easy hey?

Other Sitemap settings for the link, like lastmod, priority, changefreq and host are entered automatically, although you can override them if you need to.

Compatibility

Tested and working on:

Rails 3.0.0, sitemap_generator version >= 0.2.5
Rails 1.x - 2.3.5
Ruby 1.8.7, 1.9.1

Notes

1) For large sitemaps it may be useful to split your generation into batches to avoid running out of memory. E.g.:

# add movies
Movie.find_in_batches(:batch_size => 1000) do |movies|
  movies.each do |movie|
    sitemap.add "/movies/show/#{movie.to_param}", :lastmod => movie.updated_at, :changefreq => 'weekly'
  end
end

2) New Capistrano deploys will remove your Sitemap files, unless you run rake sitemap:refresh. The way around this is to create a cap task:

after "deploy:update_code", "deploy:copy_old_sitemap"

namespace :deploy do
  task :copy_old_sitemap do
      run "if [ -e #{previous_release}/public/sitemap_index.xml.gz ]; then cp #{previous_release}/public/sitemap* #{current_release}/public/; fi"
  end
end

3) If generation of your sitemap fails for some reason, the old sitemap will remain in public/. This ensures that robots will always find a valid sitemap. Running silently (rake -s sitemap:refresh) and with email forwarding setup you'll only get an email if your sitemap fails to build, and no notification when everything is fine - which will be most of the time.

Known Bugs

There's no check on the size of a URL which isn't supposed to exceed 2,048 bytes.
Currently only supports one Sitemap Index file, which can contain 50,000 Sitemap files which can each contain 50,000 urls, so it only supports up to 2,500,000,000 (2.5 billion) urls. I personally have no need of support for more urls, but plugin could be improved to support this.

Wishlist & Coming Soon

Support for generating sitemaps for sites with multiple domains. Sitemaps are generated into subdirectories and we use a Rack middleware to rewrite requests for sitemaps to the correct subdirectory based on the request host.
I want to refactor the code because it has grown a lot. Part of this refactoring will include implementing some more checks to make sure we adhere to standards as well as making sure that the sitemaps are being generated as efficiently as possible.

I'd like to simplify adding links to a sitemap. Right now it's all or nothing. I'd like to break it up so you can add batches.

Auto coverage testing. Generate a report of broken URLs by checking the status codes of each page in the sitemap.

Thanks (in no particular order)