sitemap_generator 1.4.0 → 1.5.2

data/Gemfile

@@ -5,12 +5,12 @@ gem 'sitemap_generator', :path => './'
 gem 'rails', '2.3.8'
 gem 'mocha', '0.9.10'
 gem 'rake', '0.8.7'
-gem 'rspec', '1.3.0', :require => 'spec'
+gem 'rspec', '1.3.1', :require => 'spec'
 gem 'rspec-rails', '~>1.3.2', :require => 'spec/rails'
 
 # Simplify these dependencies
 gem 'git', '1.2.5'
-gem 'jeweler', '1.4.0'
+gem 'jeweler', '=1.4.0'
 gem 'github', '0.4.5'
 gem 'gemcutter', '0.6.1'
 
@@ -21,4 +21,4 @@ gem 'sqlite3-ruby', '1.3.1', :require => 'sqlite3'
 group :test do
   gem 'ruby-debug', '0.10.3'
   gem 'ruby-debug-base', '0.10.3'
-end
+end

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: ./
   specs:
-    sitemap_generator (1.4.0)
+    sitemap_generator (1.5.0)
 
 GEM
   remote: http://rubygems.org/
@@ -16,7 +16,7 @@ GEM
     activeresource (2.3.8)
       activesupport (= 2.3.8)
     activesupport (2.3.8)
-    columnize (0.3.2)
+    columnize (0.3.1)
     gemcutter (0.6.1)
     git (1.2.5)
     github (0.4.5)
@@ -29,12 +29,12 @@ GEM
       git (>= 1.2.5)
       rubyforge (>= 2.0.0)
     json (1.5.1)
-    json_pure (1.5.1)
+    json_pure (1.4.6)
     linecache (0.43)
     mocha (0.9.10)
       rake
     nokogiri (1.4.4)
-    rack (1.1.1)
+    rack (1.1.2)
     rails (2.3.8)
       actionmailer (= 2.3.8)
       actionpack (= 2.3.8)
@@ -43,7 +43,7 @@ GEM
       activesupport (= 2.3.8)
       rake (>= 0.8.3)
     rake (0.8.7)
-    rspec (1.3.0)
+    rspec (1.3.1)
     rspec-rails (1.3.2)
       rack (>= 1.0.0)
       rspec (>= 1.3.0)
@@ -71,7 +71,7 @@ DEPENDENCIES
   nokogiri (= 1.4.4)
   rails (= 2.3.8)
   rake (= 0.8.7)
-  rspec (= 1.3.0)
+  rspec (= 1.3.1)
   rspec-rails (~> 1.3.2)
   ruby-debug (= 0.10.3)
   ruby-debug-base (= 0.10.3)

data/README.md

@@ -18,7 +18,8 @@ Features
 Changelog
 -------
 
-- v1.4.0: [Geo sitemap][geo_tags] support, support for generate multiple sitemap sets with different filenames
+- v1.5.0: Major refactoring & testing in preparation for new API & features
+- v1.4.0: [Geo sitemap][geo_tags] support, multiple sitemap support via CONFIG_FILE rake option
 - v1.3.0: Support setting the sitemaps path
 - v1.2.0: Verified working with Rails 3 stable release
 - v1.1.0: [Video sitemap][sitemap_video] support

data/Rakefile

@@ -98,7 +98,6 @@ Spec::Rake::SpecTask.new(:spec) do |spec|
   spec.libs << 'lib' << 'spec'
   spec.spec_files = FileList['spec/sitemap_generator/**/*_spec.rb']
 end
-task :spec => :check_dependencies
 
 Spec::Rake::SpecTask.new(:rcov) do |spec|
   spec.libs << 'lib' << 'spec'

data/VERSION

@@ -1 +1 @@
-1.4.0
+1.5.2

data/lib/sitemap_generator.rb

@@ -1,16 +1,17 @@
 require 'sitemap_generator/builder'
+require 'sitemap_generator/sitemap_namer'
 require 'sitemap_generator/link_set'
 require 'sitemap_generator/templates'
 require 'sitemap_generator/utilities'
-require 'sitemap_generator/railtie' if SitemapGenerator::Utilities.rails3?
-
+require 'sitemap_generator/application'
+require 'sitemap_generator/sitemap_location'
 require 'active_support/core_ext/numeric'
 
 module SitemapGenerator
-  class SitemapError <StandardError; end
-  class SitemapFullError < SitemapError; end
-  class SitemapFinalizedError < SitemapError; end
-  
+  SitemapError = Class.new(StandardError)
+  SitemapFullError = Class.new(SitemapError)
+  SitemapFinalizedError = Class.new(SitemapError)
+
   silence_warnings do
     VERSION = File.read(File.dirname(__FILE__) + "/../VERSION").strip
     MAX_SITEMAP_FILES    = 50_000        # max sitemap links per index file
@@ -18,14 +19,21 @@ module SitemapGenerator
     MAX_SITEMAP_IMAGES   = 1_000         # max images per url
     MAX_SITEMAP_FILESIZE = 10.megabytes  # bytes
 
-    Sitemap = LinkSet.new
+    # Lazy-initialize the LinkSet instance
+    Sitemap = (Class.new do
+      def method_missing(*args, &block)
+        (@link_set ||= LinkSet.new).send(*args, &block)
+      end
+    end).new
   end
 
   class << self
-    attr_accessor :root, :templates
+    attr_accessor :root, :app, :templates
   end
 
   self.root = File.expand_path(File.join(File.dirname(__FILE__), '../'))
   self.templates = SitemapGenerator::Templates.new(self.root)
+  self.app = SitemapGenerator::Application.new
+end
 
-end
+require 'sitemap_generator/railtie' if SitemapGenerator.app.rails3?

data/lib/sitemap_generator/application.rb

@@ -0,0 +1,22 @@
+require 'pathname'
+
+module SitemapGenerator
+  class Application
+    def rails?
+      defined?(Rails)
+    end
+
+    # Returns a boolean indicating whether this environment is Rails 3
+    #
+    # @return [Boolean]    
+    def rails3?
+      rails? && Rails.version.to_f >= 3
+    rescue
+      false # Rails.version defined in 2.1.0
+    end
+   
+    def root
+      Pathname.new(rails? && Rails.root || Dir.getwd)
+    end
+  end
+end

data/lib/sitemap_generator/builder/sitemap_file.rb

@@ -7,31 +7,30 @@ module SitemapGenerator
     #
     # General Usage:
     #
-    #   sitemap = SitemapFile.new('public/', 'sitemap.xml', 'http://example.com')
+    #   sitemap = SitemapFile.new(:location => SitemapLocation.new(...))
     #   sitemap.add('/', { ... })    <- add a link to the sitemap
-    #   sitemap.finalize!            <- creates a new sitemap file in directory public/
-    #       and freezes the object to protect it from further modification
+    #   sitemap.finalize!            <- write the sitemap file and freeze the object to protect it from further modification
     #
     class SitemapFile
       include ActionView::Helpers::NumberHelper
       include ActionView::Helpers::TextHelper   # Rails 2.2.2 fails with missing 'pluralize' otherwise
-      attr_accessor :sitemap_path, :public_path, :filesize, :link_count, :hostname
+      attr_reader :link_count, :filesize, :filename, :location
 
-      # <tt>public_path</tt> full path of the directory to write sitemaps in.
-      #   Usually your Rails <tt>public/</tt> directory.
+      # Options:
       #
-      # <tt>sitemap_path</tt> relative path including filename of the sitemap
-      #   file relative to <tt>public_path</tt>
+      # <tt>location</tt> a SitemapGenerator::SitemapLocation instance
       #
-      # <tt>hostname</tt> hostname including protocol to use in all links
-      #   e.g. http://en.google.ca
-      def initialize(public_path, sitemap_path, hostname='http://example.com')
-        self.sitemap_path = sitemap_path
-        self.public_path = public_path
-        self.hostname = hostname
-        self.link_count = 0
-
-        @xml_content       = ''     # XML urlset content
+      # <tt>filename</tt> a symbol giving the base of the sitemap fileaname.  Default: :sitemap
+      #
+      # <tt>namer</tt> (optional) if provided is used to get the next sitemap filename, overriding :filename
+      def initialize(opts={})
+        SitemapGenerator::Utilities.assert_valid_keys(opts, [:location, :filename, :namer])
+
+        @location = opts.delete(:location) || SitemapGenerator::SitemapLocation.new
+        @namer = opts.delete(:namer) || new_namer(opts.delete(:filename))
+        @filename = @location[:filename] = @namer.next
+        @link_count = 0
+        @xml_content = '' # XML urlset content
         @xml_wrapper_start = <<-HTML
           <?xml version="1.0" encoding="UTF-8"?>
             <urlset
@@ -46,61 +45,49 @@ module SitemapGenerator
         HTML
         @xml_wrapper_start.gsub!(/\s+/, ' ').gsub!(/ *> */, '>').strip!
         @xml_wrapper_end   = %q[</urlset>]
-        self.filesize = bytesize(@xml_wrapper_start) + bytesize(@xml_wrapper_end)
+        @filesize = bytesize(@xml_wrapper_start) + bytesize(@xml_wrapper_end)
       end
 
       def lastmod
-        File.mtime(self.full_path) rescue nil
+        File.mtime(path) rescue nil
       end
 
       def empty?
-        self.link_count == 0
-      end
-
-      def full_url
-        URI.join(self.hostname, self.sitemap_path).to_s
-      end
-
-      def full_path
-        @full_path ||= File.join(self.public_path, self.sitemap_path)
+        @link_count == 0
       end
 
       # Return a boolean indicating whether the sitemap file can fit another link
-      # of <tt>bytes</tt> bytes in size.
+      # of <tt>bytes</tt> bytes in size.  You can also pass a string and the
+      # bytesize will be calculated for you.
       def file_can_fit?(bytes)
-        (self.filesize + bytes) < SitemapGenerator::MAX_SITEMAP_FILESIZE && self.link_count < SitemapGenerator::MAX_SITEMAP_LINKS
+        bytes = bytes.is_a?(String) ? bytesize(bytes) : bytes
+        (@filesize + bytes) < SitemapGenerator::MAX_SITEMAP_FILESIZE && @link_count < SitemapGenerator::MAX_SITEMAP_LINKS
       end
 
       # Add a link to the sitemap file.
       #
       # If a link cannot be added, for example if the file is too large or the link
-      # limit has been reached, a SitemapGenerator::SitemapFullError exception is raised.
+      # limit has been reached, a SitemapGenerator::SitemapFullError exception is raised
+      # and the sitemap is finalized.
       #
       # If the Sitemap has already been finalized a SitemapGenerator::SitemapFinalizedError
       # exception is raised.
       #
+      # Return the new link count.
+      #
       # Call with:
       #   sitemap_url - a SitemapUrl instance
       #   sitemap, options - a Sitemap instance and options hash
       #   path, options - a path for the URL and options hash
       def add(link, options={})
-        xml = if link.is_a?(SitemapGenerator::Builder::SitemapUrl)
-          link.to_xml
-        else
-          SitemapGenerator::Builder::SitemapUrl.new(link, options).to_xml
-        end
+        raise SitemapGenerator::SitemapFinalizedError if finalized?
+        xml = (link.is_a?(SitemapUrl) ? link : SitemapUrl.new(link, options)).to_xml
+        raise SitemapGenerator::SitemapFullError if !file_can_fit?(xml)
 
-        if self.finalized?
-          raise SitemapGenerator::SitemapFinalizedError
-        elsif !file_can_fit?(bytesize(xml))
-          raise SitemapGenerator::SitemapFullError
-        end
-
-        # Add the XML
+        # Add the XML to the sitemap
         @xml_content << xml
-        self.filesize += bytesize(xml)
-        self.link_count += 1
-        true
+        @filesize += bytesize(xml)
+        @link_count += 1
       end
 
       # Write out the Sitemap file and freeze this object.
@@ -111,17 +98,18 @@ module SitemapGenerator
       # A SitemapGenerator::SitemapFinalizedError exception is raised if the Sitemap
       # has already been finalized
       def finalize!
-        raise SitemapGenerator::SitemapFinalizedError if self.finalized?
+        raise SitemapGenerator::SitemapFinalizedError if finalized?
 
         # Ensure that the directory exists
-        dir = File.dirname(self.full_path)
+        dir = @location.directory
+        path = @location.path
         if !File.exists?(dir)
           FileUtils.mkdir_p(dir)
         elsif !File.directory?(dir)
           raise SitemapError.new("#{dir} should be a directory!")
         end
 
-        open(self.full_path, 'wb') do |file|
+        open(path, 'wb') do |file|
           gz = Zlib::GzipWriter.new(file)
           gz.write @xml_wrapper_start
           gz.write @xml_content
@@ -129,22 +117,41 @@ module SitemapGenerator
           gz.close
         end
         @xml_content = @xml_wrapper_start = @xml_wrapper_end = ''
-        self.freeze
+        freeze
       end
 
       def finalized?
-        return self.frozen?
+        frozen?
+      end
+
+      # Return a new instance of the sitemap file with the same options, and the next name in the sequence.
+      def next
+        self.class.new(:location => @location.dup, :namer => @namer)
       end
 
       # Return a summary string
-      def summary
-        uncompressed_size = number_to_human_size(filesize) rescue "#{filesize / 8} KB"
-        compressed_size =   number_to_human_size(File.size?(full_path)) rescue "#{File.size?(full_path) / 8} KB"
-        "+ #{'%-21s' % self.sitemap_path} #{'%13s' % self.link_count} links / #{'%10s' % uncompressed_size} / #{'%10s' % compressed_size} gzipped"
+      def summary(opts={})
+        uncompressed_size = number_to_human_size(@filesize) rescue "#{@filesize / 8} KB"
+        compressed_size =   number_to_human_size(@location.filesize) rescue "#{@location.filesize / 8} KB"
+        "+ #{'%-21s' % @location.path_in_public} #{'%13s' % @link_count} links / #{'%10s' % uncompressed_size} / #{'%10s' % compressed_size} gzipped"
+      end
+
+      # Create a new namer given a filename base and set the filename of this sitemap from it.
+      # It is a bit confusing because the setter takes a filename base whereas the getter
+      # returns a full filename including extension.
+      def filename=(base)
+        @namer = new_namer(base)
+        @filename = @location[:filename] = @namer.next
       end
 
       protected
 
+      # Return a new namer given a filename base and set the filename of this sitemap from it.
+      # Default filename base is 'sitemap'.
+      def new_namer(base=nil)
+        SitemapGenerator::SitemapNamer.new(base ||= :sitemap)
+      end
+
       # Return the bytesize length of the string.  Ruby 1.8.6 compatible.
       def bytesize(string)
         string.respond_to?(:bytesize) ? string.bytesize : string.length

data/lib/sitemap_generator/builder/sitemap_index_file.rb

@@ -3,10 +3,16 @@ module SitemapGenerator
     class SitemapIndexFile < SitemapFile
       attr_accessor :sitemaps
 
-      def initialize(*args)
-        super(*args)
+      def initialize(opts={})
+        @options = [:location, :filename]
+        SitemapGenerator::Utilities.assert_valid_keys(opts, @options)
 
-        self.sitemaps = []
+        @location = opts.delete(:location) || SitemapGenerator::SitemapLocation.new
+        @filename = "#{opts.fetch(:filename, :sitemap_index)}.xml.gz"
+        @location[:filename] = @filename
+        
+        @link_count = 0
+        @sitemaps_link_count = 0
         @xml_content = '' # XML urlset content
         @xml_wrapper_start = <<-HTML
           <?xml version="1.0" encoding="UTF-8"?>
@@ -19,28 +25,38 @@ module SitemapGenerator
         HTML
         @xml_wrapper_start.gsub!(/\s+/, ' ').gsub!(/ *> */, '>').strip!
         @xml_wrapper_end   = %q[</sitemapindex>]
-        self.filesize = bytesize(@xml_wrapper_start) + bytesize(@xml_wrapper_end)
+        @filesize = bytesize(@xml_wrapper_start) + bytesize(@xml_wrapper_end)
       end
 
       # Finalize sitemaps as they are added to the index
       def add(link, options={})
-        if link.is_a?(SitemapFile)
-          self.sitemaps << link
-          link.finalize!
+        if file = link.is_a?(SitemapFile) && link
+          @sitemaps_link_count += file.link_count
+          file.finalize!
         end
         super(SitemapGenerator::Builder::SitemapIndexUrl.new(link, options))
       end
 
       # Return the total number of links in all sitemaps reference by this index file
       def total_link_count
-        self.sitemaps.inject(0) { |link_count_sum, sitemap| link_count_sum + sitemap.link_count }
+        @sitemaps_link_count
+      end
+
+      # Set a new filename on the instance.  Should not include any extensions e.g. :sitemap_index
+      def filename=(filename)
+        @filename = @location[:filename] = "#{filename}_index.xml.gz"
       end
 
       # Return a summary string
-      def summary
-        uncompressed_size = number_to_human_size(filesize) rescue "#{filesize / 8} KB"
-        compressed_size =   number_to_human_size(File.size?(full_path)) rescue "#{File.size?(full_path) / 8} KB"
-        "+ #{'%-21s' % self.sitemap_path} #{'%10s' % self.link_count} sitemaps / #{'%10s' % uncompressed_size} / #{'%10s' % compressed_size} gzipped"
+      def summary(opts={})
+        uncompressed_size = number_to_human_size(@filesize) rescue "#{@filesize / 8} KB"
+        compressed_size =   number_to_human_size(@location.filesize) rescue "#{@location.filesize / 8} KB"
+        "+ #{'%-21s' % @location.path_in_public} #{'%10s' % @link_count} sitemaps / #{'%10s' % uncompressed_size} / #{'%10s' % compressed_size} gzipped"
+      end
+
+      def stats_summary(opts={})
+        str = "Sitemap stats: #{number_with_delimiter(@sitemaps_link_count)} links / #{@link_count} sitemaps"
+        str += " / %dm%02ds" % opts[:time_taken].divmod(60) if opts[:time_taken]
       end
     end
   end

data/lib/sitemap_generator/builder/sitemap_index_url.rb

@@ -5,9 +5,9 @@ module SitemapGenerator
     class SitemapIndexUrl < SitemapUrl
 
       def initialize(path, options={})
-        if path.is_a?(SitemapGenerator::Builder::SitemapIndexFile)
-          options.reverse_merge!(:host => path.hostname, :lastmod => Time.now, :changefreq => 'always', :priority => 1.0)
-          path = path.sitemap_path
+        if index = path.is_a?(SitemapGenerator::Builder::SitemapIndexFile) && path
+          options.reverse_merge!(:host => index.location.host, :lastmod => Time.now, :changefreq => 'always', :priority => 1.0)
+          path = index.location.path_in_public
           super(path, options)
         else
           super

data/lib/sitemap_generator/builder/sitemap_url.rb

@@ -9,13 +9,13 @@ module SitemapGenerator
       #   sitemap - a Sitemap instance, or
       #   path, options - a path for the URL and options hash
       def initialize(path, options={})
-        if path.is_a?(SitemapGenerator::Builder::SitemapFile)
-          options.reverse_merge!(:host => path.hostname, :lastmod => path.lastmod)
-          path = path.sitemap_path
+        if sitemap = path.is_a?(SitemapGenerator::Builder::SitemapFile) && path
+          options.reverse_merge!(:host => sitemap.location.host, :lastmod => sitemap.lastmod)
+          path = sitemap.location.path_in_public
         end
 
-        options.assert_valid_keys(:priority, :changefreq, :lastmod, :host, :images, :video, :geo)
-        options.reverse_merge!(:priority => 0.5, :changefreq => 'weekly', :lastmod => Time.now, :host => Sitemap.default_host, :images => [])
+        SitemapGenerator::Utilities.assert_valid_keys(options, :priority, :changefreq, :lastmod, :host, :images, :video, :geo)
+        options.reverse_merge!(:priority => 0.5, :changefreq => 'weekly', :lastmod => Time.now, :images => [])
         self.merge!(
           :path => path,
           :priority => options[:priority],
@@ -94,7 +94,7 @@ module SitemapGenerator
       def prepare_images(images, host)
         images.delete_if { |key,value| key[:loc] == nil }
         images.each do |r|
-          r.assert_valid_keys(:loc, :caption, :geo_location, :title, :license)
+          SitemapGenerator::Utilities.assert_valid_keys(r, :loc, :caption, :geo_location, :title, :license)
           r[:loc] = URI.join(host, r[:loc]).to_s
         end
         images[0..(SitemapGenerator::MAX_SITEMAP_IMAGES-1)]

data/lib/sitemap_generator/interpreter.rb

@@ -6,9 +6,9 @@ module SitemapGenerator
   # Rails URL helpers.
   class Interpreter
 
-    if SitemapGenerator::Utilities.rails3?
+    if SitemapGenerator.app.rails3?
       include ::Rails.application.routes.url_helpers
-    else
+    elsif SitemapGenerator.app.rails?
       require 'action_controller'
       include ActionController::UrlWriter
     end
@@ -23,7 +23,7 @@ module SitemapGenerator
       if block_given?
         instance_eval(&block)
       else
-        sitemap_config_file ||= File.join(::Rails.root, 'config/sitemap.rb')
+        sitemap_config_file ||= SitemapGenerator.app.root + 'config/sitemap.rb'
         eval(File.read(sitemap_config_file), nil, sitemap_config_file.to_s)
       end
     end

data/lib/sitemap_generator/link_set.rb

@@ -1,15 +1,12 @@
 require 'builder'
-require 'action_view'
 
 # A LinkSet provisions a bunch of links to sitemap files.  It also writes the index file
 # which lists all the sitemap files written.
 module SitemapGenerator
   class LinkSet
-    include ActionView::Helpers::NumberHelper  # for number_with_delimiter
 
-    attr_reader :default_host, :public_path, :sitemaps_path, :filename
-    attr_accessor :sitemap, :sitemap_index
-    attr_accessor :verbose, :yahoo_app_id
+    attr_reader :default_host, :public_path, :sitemaps_path, :filename, :sitemap, :location
+    attr_accessor :verbose, :yahoo_app_id, :include_root, :include_index
 
     # Evaluate the sitemap config file and write all sitemaps.
     #
@@ -21,85 +18,102 @@ module SitemapGenerator
     def create(config_file = 'config/sitemap.rb', &block)
       require 'sitemap_generator/interpreter'
 
-      start_time = Time.now
-      if self.sitemap_index.finalized?
-        self.sitemap_index = SitemapGenerator::Builder::SitemapIndexFile.new(@public_path, sitemap_index_path)
-        self.sitemap = SitemapGenerator::Builder::SitemapFile.new(@public_path, new_sitemap_path)
-      end
+      # Clear out the current objects.  New objects will be lazy-initialized.
+      @sitemap_index = @sitemap = nil
 
+      start_time = Time.now
       SitemapGenerator::Interpreter.new(self, config_file, &block)
-      unless self.sitemap.finalized?
-        self.sitemap_index.add(self.sitemap)
-        puts self.sitemap.summary if verbose
-      end
-      self.sitemap_index.finalize!
+      finalize_sitemap
+      finalize_sitemap_index
       end_time = Time.now
 
-      if verbose
-        puts self.sitemap_index.summary
-        puts "\nSitemap stats: #{number_with_delimiter(self.sitemap_index.total_link_count)} links / #{self.sitemap_index.sitemaps.size} sitemaps / " +
-              ("%dm%02ds" % (end_time - start_time).divmod(60))
-      end
+      puts sitemap_index.stats_summary(:time_taken => end_time - start_time) if verbose
     end
 
     # Constructor
     #
-    # <tt>public_path</tt> (optional) full path to the directory to write sitemaps in.
+    # Call with a hash of options.  Options:
+    #
+    # <tt>public_path</tt> full path to the directory to write sitemaps in.
     #   Defaults to your Rails <tt>public/</tt> directory.
     #
-    # <tt>sitemaps_path</tt> (optional) path fragment within public to write sitemaps
+    # <tt>sitemaps_path</tt> path fragment within public to write sitemaps
     #   to e.g. 'en/'.  Sitemaps are written to <tt>public_path</tt> + <tt>sitemaps_path</tt>
     #
-    # <tt>default_host</tt> hostname including protocol to use in all sitemap links
+    # <tt>default_host</tt> host including protocol to use in all sitemap links
     #   e.g. http://en.google.ca
     #
-    # <tt>filename</tt> used in the name of the file like "#{@filename}1.xml.gzip" and "#{@filename}_index.xml.gzip"
-    #   Defaults to <tt>sitemap</tt>
-    def initialize(public_path = nil, sitemaps_path = nil, default_host = nil, filename = 'sitemap')
-      @default_host = default_host
-      @public_path = public_path
-      @sitemaps_path = sitemaps_path
-      @filename = filename
-
-      if @public_path.nil?
-        @public_path = File.join(::Rails.root, 'public/') rescue 'public/'
+    # <tt>filename</tt> symbol giving the base name for files (default <tt>:sitemap</tt>).
+    #   The sitemap names are generated like "#{@filename}1.xml.gzip", "#{@filename}2.xml.gzip"
+    #   and the index name is like "#{@filename}_index.xml.gzip".
+    #
+    # <tt>include_root</tt> whether to include the root url i.e. '/' in each group of sitemaps.
+    #   Default is false.
+    #
+    # <tt>include_index</tt> whether to include the sitemap index URL in each group of sitemaps.
+    #   Default is false.
+    def initialize(*args)
+
+      # Extract options
+      options = if (!args.first.nil? && !args.first.is_a?(Hash)) || args.size > 1
+        warn "Deprecated. Please call with an options hash instead."
+        [:public_path, :sitemaps_path, :default_host, :filename].each_with_index.inject({}) do |hash, arg|
+          hash[arg[0]] = args[arg[1]]
+          hash
+        end
+      else
+        args.first || {}
       end
 
-      # Default host is not set yet.  Set it on these objects when `add_links` is called
-      self.sitemap_index = SitemapGenerator::Builder::SitemapIndexFile.new(@public_path, sitemap_index_path)
-      self.sitemap = SitemapGenerator::Builder::SitemapFile.new(@public_path, new_sitemap_path)
+      # Option defaults
+      options.reverse_merge!({
+        :include_root => true,
+        :include_index => true,
+        :filename => :sitemap,
+        :public_path => SitemapGenerator.app.root + 'public/',
+        :sitemaps_path => nil
+      })
+      options.each_pair { |k, v| instance_variable_set("@#{k}".to_sym, v) }
+
+
+      # Create a location object to store all the location options
+      @location = SitemapGenerator::SitemapLocation.new(
+        :sitemaps_path => @sitemaps_path,
+        :public_path => @public_path,
+        :host => @default_host
+      )
     end
 
     # Entry point for users.
     #
     # Called within the user's eval'ed sitemap config file.  Add links to sitemap files
-    # passing a block.
+    # passing a block.  This instance is passed in as an argument.  You can call
+    # `add` on it to add links.
     #
-    # TODO: Refactor.  The call chain is confusing and convoluted here.
+    # Example:
+    #   add_links do |sitemap|
+    #     sitemap.add '/'
+    #   end
     def add_links
-      raise ArgumentError, "Default hostname not set" if default_host.blank?
-
-      # Set default host on the sitemap objects and seed the sitemap with the default links
-      self.sitemap.hostname = self.sitemap_index.hostname = default_host
-      self.sitemap.add('/', :lastmod => Time.now, :changefreq => 'always', :priority => 1.0)
-      self.sitemap.add(self.sitemap_index, :lastmod => Time.now, :changefreq => 'always', :priority => 1.0)
-
+      sitemap.add('/', :lastmod => Time.now, :changefreq => 'always', :priority => 1.0, :host => @location.host) if include_root
+      sitemap.add(sitemap_index, :lastmod => Time.now, :changefreq => 'always', :priority => 1.0) if include_index
       yield self
     end
 
     # Add a link to a Sitemap.  If a new Sitemap is required, one will be created for
     # you.
+    #
+    # link - string link e.g. '/merchant', '/article/1' or whatever.
+    # options - see README.
+    #   host - host for the link, defaults to your <tt>default_host</tt>.
     def add(link, options={})
-      begin
-        self.sitemap.add(link, options)
-      rescue SitemapGenerator::SitemapError => e
-        if e.is_a?(SitemapGenerator::SitemapFullError)
-          self.sitemap_index.add(self.sitemap)
-          puts self.sitemap.summary if verbose
-        end
-        self.sitemap = SitemapGenerator::Builder::SitemapFile.new(public_path, new_sitemap_path, default_host)
-        retry
-      end
+      sitemap.add(link, options.reverse_merge!(:host => @location.host))
+    rescue SitemapGenerator::SitemapFullError
+      finalize_sitemap
+      retry
+    rescue SitemapGenerator::SitemapFinalizedError
+      @sitemap = sitemap.next
+      retry
     end
 
     # Ping search engines.
@@ -108,7 +122,7 @@ module SitemapGenerator
     def ping_search_engines
       require 'open-uri'
 
-      sitemap_index_url = CGI.escape(self.sitemap_index.full_url)
+      sitemap_index_url = CGI.escape(sitemap_index.location.url)
       search_engines = {
         :google         => "http://www.google.com/webmasters/sitemaps/ping?sitemap=#{sitemap_index_url}",
         :yahoo          => "http://search.yahooapis.com/SiteExplorerService/V1/ping?sitemap=#{sitemap_index_url}&appid=#{yahoo_app_id}",
@@ -140,49 +154,89 @@ module SitemapGenerator
       end
     end
 
+    # Return a count of the total number of links in all sitemaps
     def link_count
-      self.sitemap_index.total_link_count
+      sitemap_index.total_link_count
     end
 
+    # Set the host name, including protocol, that will be used by default on each
+    # of your sitemap links.  You can pass a different host in your options to `add`
+    # if you need to change it on a per-link basis.
     def default_host=(value)
-      @default_host = value
-      self.sitemap_index.hostname = value unless self.sitemap_index.finalized?
-      self.sitemap.hostname = value unless self.sitemap.finalized?
+      update_location_info(:host, value)
     end
 
+    # Set the public_path.  This path gives the location of your public directory.
+    # The default is the public/ directory in your Rails root.  Or if Rails is not
+    # found, it defaults to public/ in the current directory (of the process).
+    #
+    # Example: 'tmp/' if you don't want to generate in public for some reason.
+    #
+    # Set to nil to use the current directory.
     def public_path=(value)
-      @public_path = value
-      self.sitemap_index.public_path = value unless self.sitemap_index.finalized?
-      self.sitemap.public_path = value unless self.sitemap.finalized?
+      update_location_info(:public_path, value)
     end
 
+    # Set the sitemaps_path.  This path gives the location to write sitemaps to
+    # relative to your public_path.
+    # Example: 'sitemaps/' to generate your sitemaps in 'public/sitemaps/'.
     def sitemaps_path=(value)
-      @sitemaps_path = value
-      self.sitemap_index.sitemap_path = sitemap_index_path unless self.sitemap_index.finalized?
-      self.sitemap.sitemap_path = new_sitemap_path unless self.sitemap.finalized?
+      update_location_info(:sitemaps_path, value)
     end
 
+    # Set the filename base to use when generating sitemaps and sitemap indexes.
     def filename=(value)
       @filename = value
-      self.sitemap_index.sitemap_path = sitemap_index_path unless self.sitemap_index.finalized?
-      self.sitemap.sitemap_path = new_sitemap_path unless self.sitemap.finalized?
+      update_sitemap_info(:filename, value)
+    end
+
+    # Lazy-initialize a sitemap instance when it's accessed
+    def sitemap
+      @sitemap ||= SitemapGenerator::Builder::SitemapFile.new(
+        :location => @location.dup,
+        :filename => @filename
+      )
+    end
+
+    # Lazy-initialize a sitemap index instance when it's accessed
+    def sitemap_index
+      @sitemap_index ||= SitemapGenerator::Builder::SitemapIndexFile.new(
+        :location => @location.dup,
+        :filename => "#{@filename}_index"
+      )
     end
 
     protected
 
-    # Return the current sitemap filename with index.
-    #
-    # The index depends on the length of the <tt>sitemaps</tt> array.
-    def new_sitemap_path
-      File.join(self.sitemaps_path || '', "#{@filename}#{self.sitemap_index.sitemaps.length + 1}.xml.gz")
+    # Finalize a sitemap by including it in the index and outputting a summary line.
+    # Do nothing if it has already been finalized.
+    def finalize_sitemap
+      return if sitemap.finalized?
+      sitemap_index.add(sitemap)
+      puts sitemap.summary if verbose
     end
 
-    # Return the current sitemap index filename.
-    #
-    # At the moment we only support one index file which can link to
-    # up to 50,000 sitemap files.
-    def sitemap_index_path
-      File.join(self.sitemaps_path || '', "#{@filename}_index.xml.gz")
+    # Finalize a sitemap index and output a summary line.  Do nothing if it has already
+    # been finalized.
+    def finalize_sitemap_index
+      return if sitemap_index.finalized?
+      sitemap_index.finalize!
+      puts sitemap_index.summary if verbose
+    end
+
+    # Update the given attribute on the current sitemap index and sitemap files.  But
+    # don't create the index or sitemap files yet if they are not already created.
+    def update_sitemap_info(attribute, value)
+      sitemap_index.send("#{attribute}=", value) if @sitemap_index && !@sitemap_index.finalized?
+      sitemap.send("#{attribute}=", value) if @sitemap && !@sitemap.finalized?
+    end
+
+    # Update the given attribute on the current sitemap index and sitemap file location objects.
+    # But don't create the index or sitemap files yet if they are not already created.
+    def update_location_info(attribute, value)
+      @location.merge!(attribute => value)
+      sitemap_index.location.merge!(attribute => value) if @sitemap_index && !@sitemap_index.finalized?
+      sitemap.location.merge!(attribute => value) if @sitemap && !@sitemap.finalized?
     end
   end
 end

data/lib/sitemap_generator/sitemap_location.rb

@@ -0,0 +1,71 @@
+module SitemapGenerator
+  class SitemapLocation < Hash
+
+    [:filename, :host].each do |method|
+      define_method(method) do
+        raise SitemapGenerator::SitemapError, "No value set for #{method}" unless self[method]
+        self[method]
+      end
+    end
+
+    [:public_path, :sitemaps_path].each do |method|
+      define_method(method) do
+        Pathname.new(self[method].nil? ? '' : self[method])
+      end
+    end
+
+    # The filename is not required at initialization but must be set when calling
+    # methods that depend on it like <tt>path</tt>.
+    #
+    # All options are optional.  Supported options are:
+    #   public_path   - path to the "public" directory, or the directory you want to
+    #                   write sitemaps in.  Default is a directory <tt>public/</tt>
+    #                   in the current working directory, or relative to the Rails root
+    #                   directory if running under Rails.
+    #   sitemaps_path - gives the path relative to the <tt>public_path</tt> in which to
+    #                 write sitemaps e.g. <tt>sitemaps/</tt>.
+    #   host          - host name for URLs.  The full URL to the file is then constructed from
+    #                   the <tt>host</tt>, <tt>sitemaps_path</tt> and <tt>filename</tt>
+    #   filename      - name of the file
+    def initialize(opts={})
+      SitemapGenerator::Utilities.assert_valid_keys(opts, [:public_path, :sitemaps_path, :host, :filename])
+      opts.reverse_merge!(
+        :sitemaps_path => nil,
+        :public_path => SitemapGenerator.app.root + 'public/',
+        :host => nil,
+        :filename => nil
+      )
+      self.merge!(opts)
+    end
+
+    # Return a new Location instance with the given options merged in
+    def with(opts={})
+      self.merge(opts)
+    end
+
+    # Full path to the directory of the file.
+    def directory
+      (public_path + sitemaps_path).to_s
+    end
+
+    # Full path of the file including the filename.
+    def path
+      (public_path + sitemaps_path + filename).to_s
+    end
+
+    # Relative path of the file (including the filename) relative to <tt>public_path</tt>
+    def path_in_public
+      (sitemaps_path + filename).to_s
+    end
+    
+    # Full URL of the file.
+    def url
+      URI.join(host, sitemaps_path.to_s, filename.to_s).to_s
+    end
+    
+    # Return the size of the file at <tt>path</tt>
+    def filesize
+      File.size?(path)
+    end
+  end
+end

data/lib/sitemap_generator/sitemap_namer.rb

@@ -0,0 +1,30 @@
+module SitemapGenerator
+  # A poor excuse for an enumerator, but it will have to do.
+  # Return an object with a method `next` that generates sitemaps with the given name
+  # and an index appended.
+  #
+  # For example:
+  #   SitemapNamer.new(:sitemap) generates 'sitemap1.xml.gz', 'sitemap2.xml.gz' etc
+  class SitemapNamer
+    # Params:
+    #   name - string or symbol name that is the base of the generated filename
+    #
+    # Options include:
+    #   :extension - Default: '.xml.gz'. File extension to append.
+    #   :start     - Default: 1. Index at which to start counting.
+    def initialize(name, options={});
+      @options = options.reverse_merge(
+        :extension => '.xml.gz',
+        :start => 1
+      )
+      @name = name
+      @count = @options[:start]
+    end
+
+    def next
+      "#{@name}#{@count}#{@options[:extension]}"
+    ensure
+      @count += 1
+    end
+  end
+end

data/lib/sitemap_generator/utilities.rb

@@ -4,35 +4,34 @@ module SitemapGenerator
 
     # Copy templates/sitemap.rb to config if not there yet.
     def install_sitemap_rb(verbose=false)
-      if File.exist?(File.join(Rails.root, 'config/sitemap.rb'))
+      if File.exist?(SitemapGenerator.app.root + 'config/sitemap.rb')
         puts "already exists: config/sitemap.rb, file not copied" if verbose
       else
         FileUtils.cp(
           SitemapGenerator.templates.template_path(:sitemap_sample),
-          File.join(Rails.root, 'config/sitemap.rb'))
+          SitemapGenerator.app.root + 'config/sitemap.rb')
         puts "created: config/sitemap.rb" if verbose
       end
     end
 
     # Remove config/sitemap.rb if exists.
     def uninstall_sitemap_rb
-      if File.exist?(File.join(Rails.root, 'config/sitemap.rb'))
-        File.rm(File.join(Rails.root, 'config/sitemap.rb'))
+      if File.exist?(SitemapGenerator.app.root + 'config/sitemap.rb')
+        File.rm(SitemapGenerator.app.root + 'config/sitemap.rb')
       end
     end
 
     # Clean sitemap files in output directory.
     def clean_files
-      FileUtils.rm(Dir[File.join(Rails.root, 'public/sitemap*.xml.gz')])
+      FileUtils.rm(Dir[SitemapGenerator.app.root + 'public/sitemap*.xml.gz'])
     end
 
-    # Returns a boolean indicating whether this environment is Rails 3
-    #
-    # @return [Boolean]
-    def self.rails3?
-      Rails.version.to_f >= 3
-    rescue
-      false  # Rails.version defined in 2.1.0
+    # Validate all keys in a hash match *valid keys, raising ArgumentError on a
+    # mismatch. Note that keys are NOT treated indifferently, meaning if you use
+    # strings for keys but assert symbols as keys, this will fail.
+    def assert_valid_keys(hash, *valid_keys)
+      unknown_keys = hash.keys - [valid_keys].flatten
+      raise(ArgumentError, "Unknown key(s): #{unknown_keys.join(", ")}") unless unknown_keys.empty?
     end
   end
-end
+end

metadata

@@ -2,12 +2,12 @@
 name: sitemap_generator
 version: !ruby/object:Gem::Version 
   hash: 7
-  prerelease: false
+  prerelease: 
   segments: 
   - 1
-  - 4
-  - 0
-  version: 1.4.0
+  - 5
+  - 2
+  version: 1.5.2
 platform: ruby
 authors: 
 - Karl Varga
@@ -16,14 +16,12 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-03-05 00:00:00 -08:00
+date: 2011-03-31 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
-  prerelease: false
   type: :development
-  name: rspec
-  version_requirements: &id001 !ruby/object:Gem::Requirement 
+  requirement: &id001 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -32,12 +30,12 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id001
-- !ruby/object:Gem::Dependency 
+  name: rspec
+  version_requirements: *id001
   prerelease: false
+- !ruby/object:Gem::Dependency 
   type: :development
-  name: nokogiri
-  version_requirements: &id002 !ruby/object:Gem::Requirement 
+  requirement: &id002 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -46,12 +44,12 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id002
-- !ruby/object:Gem::Dependency 
+  name: nokogiri
+  version_requirements: *id002
   prerelease: false
+- !ruby/object:Gem::Dependency 
   type: :development
-  name: sqlite3-ruby
-  version_requirements: &id003 !ruby/object:Gem::Requirement 
+  requirement: &id003 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -60,12 +58,12 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id003
-- !ruby/object:Gem::Dependency 
+  name: sqlite3-ruby
+  version_requirements: *id003
   prerelease: false
+- !ruby/object:Gem::Dependency 
   type: :development
-  name: jeweler
-  version_requirements: &id004 !ruby/object:Gem::Requirement 
+  requirement: &id004 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -74,12 +72,12 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id004
-- !ruby/object:Gem::Dependency 
+  name: jeweler
+  version_requirements: *id004
   prerelease: false
+- !ruby/object:Gem::Dependency 
   type: :development
-  name: github
-  version_requirements: &id005 !ruby/object:Gem::Requirement 
+  requirement: &id005 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -88,12 +86,12 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id005
-- !ruby/object:Gem::Dependency 
+  name: github
+  version_requirements: *id005
   prerelease: false
+- !ruby/object:Gem::Dependency 
   type: :development
-  name: git
-  version_requirements: &id006 !ruby/object:Gem::Requirement 
+  requirement: &id006 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -102,7 +100,9 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  requirement: *id006
+  name: git
+  version_requirements: *id006
+  prerelease: false
 description: SitemapGenerator is a Rails gem that makes it easy to generate enterprise-class Sitemaps readable by all search engines.  Generated Sitemaps adhere to the Sitemap protocol specification.  When you generate new Sitemaps, SitemapGenerator can automatically ping the major search engines (including Google, Yahoo and Bing) to notify them.  SitemapGenerator includes rake tasks to easily manage your sitemaps.
 email: kjvarga@gmail.com
 executables: []
@@ -121,6 +121,7 @@ files:
 - Rakefile
 - VERSION
 - lib/sitemap_generator.rb
+- lib/sitemap_generator/application.rb
 - lib/sitemap_generator/builder.rb
 - lib/sitemap_generator/builder/sitemap_file.rb
 - lib/sitemap_generator/builder/sitemap_index_file.rb
@@ -129,6 +130,8 @@ files:
 - lib/sitemap_generator/interpreter.rb
 - lib/sitemap_generator/link_set.rb
 - lib/sitemap_generator/railtie.rb
+- lib/sitemap_generator/sitemap_location.rb
+- lib/sitemap_generator/sitemap_namer.rb
 - lib/sitemap_generator/tasks.rb
 - lib/sitemap_generator/templates.rb
 - lib/sitemap_generator/utilities.rb
@@ -166,7 +169,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.3.7
+rubygems_version: 1.6.2
 signing_key: 
 specification_version: 3
 summary: Easily generate enterprise class Sitemaps for your Rails site using a familiar Rails Routes-like DSL