sitemap_generator 4.0.alpha → 4.0

data/lib/sitemap_generator/sitemap_location.rb

@@ -14,15 +14,15 @@ module SitemapGenerator
       end
     end
 
-    # If no +filename+ or +namer+ is provided, the default namer is used.  For sitemap
-    # files this generates names like <tt>sitemap1.xml.gz</tt>, <tt>sitemap2.xml.gz</tt> and so on,
+    # If no +filename+ or +namer+ is provided, the default namer is used, which
+    # generates names like <tt>sitemap.xml.gz</tt>, <tt>sitemap1.xml.gz</tt>, <tt>sitemap2.xml.gz</tt> and so on.
     #
     # === Options
     # * <tt>adapter</tt> - SitemapGenerator::Adapter subclass
     # * <tt>filename</tt> - full name of the file e.g. <tt>'sitemap1.xml.gz'<tt>
     # * <tt>host</tt> - 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>
-    # * <tt>namer</tt> - a SitemapGenerator::SitemapNamer instance.  Can be passed instead of +filename+.
+    # * <tt>namer</tt> - a SitemapGenerator::SimpleNamer instance.  Can be passed instead of +filename+.
     # * <tt>public_path</tt> - 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
@@ -30,7 +30,7 @@ module SitemapGenerator
     # * <tt>sitemaps_path</tt> - gives the path relative to the <tt>public_path</tt> in which to
     #   write sitemaps e.g. <tt>sitemaps/</tt>.
     # * <tt>verbose</tt> - whether to output summary into to STDOUT.  Default +false+.
-    # * <tt>create_index</tt> - whether to create a sitemap index.  Default +true+.  See LinkSet.
+    # * <tt>create_index</tt> - whether to create a sitemap index.  Default `:auto`.  See LinkSet.
     #                           Only applies to the SitemapIndexLocation object.
     def initialize(opts={})
       SitemapGenerator::Utilities.assert_valid_keys(opts, [:adapter, :public_path, :sitemaps_path, :host, :filename, :namer, :verbose, :create_index])
@@ -127,7 +127,7 @@ module SitemapGenerator
   class SitemapIndexLocation < SitemapLocation
     def initialize(opts={})
       if !opts[:filename] && !opts[:namer]
-        opts[:namer] = SitemapGenerator::SitemapIndexNamer.new(:sitemap_index)
+        opts[:namer] = SitemapGenerator::SitemapIndexNamer.new(:sitemap)
       end
       super(opts)
     end

data/lib/sitemap_generator/sitemap_namer.rb

@@ -1,5 +1,7 @@
 module SitemapGenerator
   # A class for generating sitemap names given the base for the filename.
+  # Deprecated.  Rather use the <tt>SitemapGenerator::SimpleNamer</tt> class and the
+  # +namer+ option on your sitemap object.
   #
   # === Example
   # namer = SitemapNamer.new(:sitemap)
@@ -50,7 +52,9 @@ module SitemapGenerator
     end
   end
 
-  # A Namer for Sitemap Indexes.  The name never changes.
+  # A Namer for Sitemap Indexes.
+  # Deprecated.  Rather use the <tt>SitemapGenerator::SimpleNamer</tt> class and the
+  # +namer+ option on your sitemap object.
   class SitemapIndexNamer < SitemapNamer
     def to_s
       "#{@base}#{@options[:extension]}"
@@ -68,14 +72,19 @@ module SitemapGenerator
   #   * sitemap3.xml.gz
   #   * ...
   #
+  # Arguments:
+  #   base - string or symbol that forms the base of the generated filename e.g.
+  #          if `:geo` files are generated like `geo.xml.gz`, `geo1.xml.gz`, `geo2.xml.gz` etc.
+  #
   # Options:
   #   :extension - Default: '.xml.gz'. File extension to append.
   #   :start     - Default: 1. Numerical index at which to start counting.
-  #   :zero      - Default: nil.  Could be a string or number that gives part
-  #                of the first name in the sequence.  So in the old naming scheme
-  #                setting this to '_index' would produce 'sitemap_index.xml.gz' as
+  #   :zero      - Default: nil.  A string or number that is appended to +base+
+  #                to create the first name in the sequence.  So setting this
+  #                to '_index' would produce 'sitemap_index.xml.gz' as
   #                the first name.  Thereafter, the numerical index defined by +start+
-  #                is used.
+  #                is used, and subsequent names would be 'sitemap1.xml.gz', 'sitemap2.xml.gz', etc.
+  #                In these examples the `base` string is assumed to be 'sitemap'.
   class SimpleNamer < SitemapNamer
     def initialize(base, options={})
       super_options = SitemapGenerator::Utilities.reverse_merge(options,

data/spec/files/sitemap.deprecated.rb

@@ -1,5 +1,7 @@
 SitemapGenerator::Sitemap.default_host = "http://www.example.com"
 SitemapGenerator::Sitemap.yahoo_app_id = false
+SitemapGenerator::Sitemap.create_index = true
+SitemapGenerator::Sitemap.namer = SitemapGenerator::SimpleNamer.new(:sitemap, :zero => '_index')
 
 SitemapGenerator::Sitemap.add_links do |sitemap|
   sitemap.add '/contents', :priority => 0.7, :changefreq => 'daily'

data/spec/files/sitemap.groups.rb

@@ -1,4 +1,5 @@
 SitemapGenerator::Sitemap.default_host = "http://www.example.com"
+
 SitemapGenerator::Sitemap.create(
     :include_root => true, :include_index => true,
     :filename => :new_sitemaps, :sitemaps_path => 'fr/') do
@@ -13,16 +14,25 @@ SitemapGenerator::Sitemap.create(
     add '/three'
   end
 
-  # Test a namer
+  # Test a deprecated namer
   group(:sitemaps_namer => SitemapGenerator::SitemapNamer.new(:abc, :start => 3)) do
     add '/four'
     add '/five'
     add '/six'
   end
 
+  # Test a simple namer
+  group(:sitemaps_namer => SitemapGenerator::SimpleNamer.new(:def)) do
+    add '/four'
+    add '/five'
+    add '/six'
+  end
+
   add '/seven'
 
-  # This should be in a file of its own
+  # This should be in a file of its own.
+  # Not technically valid to have a link with a different host, but people like
+  # to do strange things sometimes.
   group(:sitemaps_host => "http://exceptional.com") do
     add '/eight'
     add '/nine'
@@ -33,5 +43,7 @@ SitemapGenerator::Sitemap.create(
   # This should have no effect.  Already added default links.
   group(:include_root => true, :include_index => true) {}
 
+  # Not technically valid to have a link with a different host, but people like
+  # to do strange things sometimes
   add "/merchant_path", :host => "https://www.merchanthost.com"
 end

data/spec/sitemap_generator/alternate_sitemap_spec.rb

@@ -25,4 +25,29 @@ describe "SitemapGenerator" do
     alternate.attribute('href').value.should == 'http://www.example.de/link_with_alternate.html'
   end
 
+  it "should add alternate links to sitemap with rel nofollow" do
+    xml_fragment = SitemapGenerator::Builder::SitemapUrl.new('link_with_alternates.html',
+      :host => 'http://www.example.com',
+      :alternates => [
+        {
+          :lang => 'de',
+          :href => 'http://www.example.de/link_with_alternate.html',
+          :nofollow => true
+        }
+      ]
+    ).to_xml
+
+    doc = Nokogiri::XML.parse("<root xmlns='http://www.sitemaps.org/schemas/sitemap/0.9' xmlns:xhtml='http://www.w3.org/1999/xhtml'>#{xml_fragment}</root>")
+    url = doc.css('url')
+    url.should_not be_nil
+    url.css('loc').text.should == 'http://www.example.com/link_with_alternates.html'
+
+    alternate = url.at_xpath('xhtml:link')
+    alternate.should_not be_nil
+    alternate.attribute('rel').value.should == 'alternate nofollow'
+    alternate.attribute('hreflang').value.should == 'de'
+    alternate.attribute('href').value.should == 'http://www.example.de/link_with_alternate.html'
+  end
+
 end
+

data/spec/sitemap_generator/builder/sitemap_file_spec.rb

@@ -1,35 +1,33 @@
 require 'spec_helper'
 
 describe 'SitemapGenerator::Builder::SitemapFile' do
-  before :each do
-    @loc = SitemapGenerator::SitemapLocation.new(:namer => SitemapGenerator::SitemapNamer.new(:sitemap), :public_path => 'tmp/', :sitemaps_path => 'test/', :host => 'http://example.com/')
-    @s = SitemapGenerator::Builder::SitemapFile.new(@loc)
-  end
+  let(:location) { SitemapGenerator::SitemapLocation.new(:namer => SitemapGenerator::SitemapNamer.new(:sitemap), :public_path => 'tmp/', :sitemaps_path => 'test/', :host => 'http://example.com/') }
+  let(:sitemap)  { SitemapGenerator::Builder::SitemapFile.new(location) }
 
   it "should have a default namer" do
-    @s = SitemapGenerator::Builder::SitemapFile.new
-    @s.location.filename.should == 'sitemap1.xml.gz'
+    sitemap = SitemapGenerator::Builder::SitemapFile.new
+    sitemap.location.filename.should == 'sitemap1.xml.gz'
   end
 
   it "should return the name of the sitemap file" do
-    @s.location.filename.should == 'sitemap1.xml.gz'
+    sitemap.location.filename.should == 'sitemap1.xml.gz'
   end
 
   it "should return the URL" do
-    @s.location.url.should == 'http://example.com/test/sitemap1.xml.gz'
+    sitemap.location.url.should == 'http://example.com/test/sitemap1.xml.gz'
   end
 
   it "should return the path" do
-    @s.location.path.should == File.expand_path('tmp/test/sitemap1.xml.gz')
+    sitemap.location.path.should == File.expand_path('tmp/test/sitemap1.xml.gz')
   end
 
   it "should be empty" do
-    @s.empty?.should be_true
-    @s.link_count.should == 0
+    sitemap.empty?.should be_true
+    sitemap.link_count.should == 0
   end
 
   it "should not be finalized" do
-    @s.finalized?.should be_false
+    sitemap.finalized?.should be_false
   end
 
   it "should raise if no default host is set" do
@@ -39,41 +37,96 @@ describe 'SitemapGenerator::Builder::SitemapFile' do
   describe "lastmod" do
     it "should be the file last modified time" do
       lastmod = (Time.now - 1209600)
-      @s.location.reserve_name
-      File.expects(:mtime).with(@s.location.path).returns(lastmod)
-      @s.lastmod.should == lastmod
+      sitemap.location.reserve_name
+      File.expects(:mtime).with(sitemap.location.path).returns(lastmod)
+      sitemap.lastmod.should == lastmod
     end
 
     it "should be nil if the location has not reserved a name" do
       File.expects(:mtime).never
-      @s.lastmod.should be_nil
+      sitemap.lastmod.should be_nil
     end
-    
+
     it "should be nil if location has reserved a name and the file DNE" do
-      @s.location.reserve_name
+      sitemap.location.reserve_name
       File.expects(:mtime).raises(Errno::ENOENT)
-      @s.lastmod.should be_nil
+      sitemap.lastmod.should be_nil
     end
   end
 
   describe "new" do
+    let(:original_sitemap) { sitemap }
+    let(:new_sitemap)      { sitemap.new }
+
     before :each do
-      @orig_s = @s
-      @s = @s.new
+      original_sitemap
+      new_sitemap
     end
 
     it "should inherit the same options" do
       # The name is the same because the original sitemap was not finalized
-      @s.location.url.should == 'http://example.com/test/sitemap1.xml.gz'
-      @s.location.path.should == File.expand_path('tmp/test/sitemap1.xml.gz')
+      new_sitemap.location.url.should == 'http://example.com/test/sitemap1.xml.gz'
+      new_sitemap.location.path.should == File.expand_path('tmp/test/sitemap1.xml.gz')
     end
 
     it "should not share the same location instance" do
-      @s.location.should_not be(@orig_s.location)
+      new_sitemap.location.should_not be(original_sitemap.location)
     end
 
     it "should inherit the same namer instance" do
-      @s.location.namer.should == @orig_s.location.namer
+      new_sitemap.location.namer.should == original_sitemap.location.namer
+    end
+  end
+
+  describe "reserve_name" do
+    it "should reserve the name from the location" do
+      sitemap.reserved_name?.should be_false
+      sitemap.location.expects(:reserve_name).returns('name')
+      sitemap.reserve_name
+      sitemap.reserved_name?.should be_true
+      sitemap.instance_variable_get(:@reserved_name).should == 'name'
+    end
+
+    it "should be safe to call multiple times" do
+      sitemap.location.expects(:reserve_name).returns('name').once
+      sitemap.reserve_name
+      sitemap.reserve_name
+    end
+  end
+
+  describe "add" do
+    it "should use the host provided" do
+      url = SitemapGenerator::Builder::SitemapUrl.new('/one', :host => 'http://newhost.com/')
+      SitemapGenerator::Builder::SitemapUrl.expects(:new).with('/one', :host => 'http://newhost.com').returns(url)
+      sitemap.add '/one', :host => 'http://newhost.com'
+    end
+
+    it "should use the host from the location" do
+      url = SitemapGenerator::Builder::SitemapUrl.new('/one', :host => 'http://example.com/')
+      SitemapGenerator::Builder::SitemapUrl.expects(:new).with('/one', :host => 'http://example.com/').returns(url)
+      sitemap.add '/one'
+    end
+  end
+
+  describe "ellipsis" do
+    it "should not modify when less than or equal to max" do
+      (1..10).each do |i|
+        string = 'a'*i
+        sitemap.send(:ellipsis, string, 10).should == string
+      end
+    end
+
+    it "should replace last 3 characters with ellipsis when greater than max" do
+      (1..5).each do |i|
+        string = 'aaaaa' + 'a'*i
+        sitemap.send(:ellipsis, string, 5).should == 'aa...'
+      end
+    end
+
+    it "should not freak out when string too small" do
+      sitemap.send(:ellipsis, 'a', 1).should == 'a'
+      sitemap.send(:ellipsis, 'aa', 1).should == '...'
+      sitemap.send(:ellipsis, 'aaa', 1).should == '...'
     end
   end
 end

data/spec/sitemap_generator/builder/sitemap_index_file_spec.rb

@@ -1,38 +1,101 @@
 require 'spec_helper'
 
 describe 'SitemapGenerator::Builder::SitemapIndexFile' do
+  let(:location) { SitemapGenerator::SitemapLocation.new(:filename => 'sitemap.xml.gz', :public_path => '/public/', :host => 'http://example.com/') }
+  let(:index)    { SitemapGenerator::Builder::SitemapIndexFile.new(location) }
+
   before :each do
-    @loc = SitemapGenerator::SitemapLocation.new(:filename => 'sitemap_index.xml.gz', :public_path => '/public/', :sitemaps_path => 'test/', :host => 'http://example.com/')
-    @s = SitemapGenerator::Builder::SitemapIndexFile.new(@loc)
+    index.location[:sitemaps_path] = 'test/'
   end
 
   it "should return the URL" do
-    @s.location.url.should == 'http://example.com/test/sitemap_index.xml.gz'
+    index.location.url.should == 'http://example.com/test/sitemap.xml.gz'
   end
 
   it "should return the path" do
-    @s.location.path.should == '/public/test/sitemap_index.xml.gz'
+    index.location.path.should == '/public/test/sitemap.xml.gz'
   end
 
   it "should be empty" do
-    @s.empty?.should be_true
-    @s.link_count.should == 0
+    index.empty?.should be_true
+    index.link_count.should == 0
   end
 
   it "should not have a last modification data" do
-    @s.lastmod.should be_nil
+    index.lastmod.should be_nil
   end
 
   it "should not be finalized" do
-    @s.finalized?.should be_false
+    index.finalized?.should be_false
   end
 
-  it "filename should default to sitemap_index" do
-    @s.location.filename.should == 'sitemap_index.xml.gz'
+  it "filename should be set" do
+    index.location.filename.should == 'sitemap.xml.gz'
   end
 
   it "should have a default namer" do
-    @s = SitemapGenerator::Builder::SitemapIndexFile.new
-    @s.location.filename.should == 'sitemap_index.xml.gz'
+    index = SitemapGenerator::Builder::SitemapIndexFile.new
+    index.location.filename.should == 'sitemap.xml.gz'
+  end
+
+  describe "link_count" do
+    it "should return the link count" do
+      index.instance_variable_set(:@link_count, 10)
+      index.link_count.should == 10
+    end
+  end
+
+  describe "create_index?" do
+    it "should return false" do
+      index.location[:create_index] = false
+      index.create_index?.should be_false
+
+      index.instance_variable_set(:@link_count, 10)
+      index.create_index?.should be_false
+    end
+
+    it "should return true" do
+      index.location[:create_index] = true
+      index.create_index?.should be_true
+
+      index.instance_variable_set(:@link_count, 1)
+      index.create_index?.should be_true
+    end
+
+    it "when :auto, should be true if more than one link" do
+      index.instance_variable_set(:@link_count, 1)
+      index.location[:create_index] = :auto
+      index.create_index?.should be_false
+
+      index.instance_variable_set(:@link_count, 2)
+      index.create_index?.should be_true
+    end
+  end
+
+  describe "add" do
+    it "should use the host provided" do
+      url = SitemapGenerator::Builder::SitemapIndexUrl.new('/one', :host => 'http://newhost.com/')
+      SitemapGenerator::Builder::SitemapIndexUrl.expects(:new).with('/one', :host => 'http://newhost.com').returns(url)
+      index.add '/one', :host => 'http://newhost.com'
+    end
+
+    it "should use the host from the location" do
+      url = SitemapGenerator::Builder::SitemapIndexUrl.new('/one', :host => 'http://example.com/')
+      SitemapGenerator::Builder::SitemapIndexUrl.expects(:new).with('/one', :host => 'http://example.com/').returns(url)
+      index.add '/one'
+    end
+
+    describe "when adding manually" do
+      it "should reserve a name" do
+        index.expects(:reserve_name)
+        index.add '/link'
+      end
+
+      it "should create index" do
+        index.create_index?.should be_false
+        index.add '/one'
+        index.create_index?.should be_true
+      end
+    end
   end
 end

data/spec/sitemap_generator/builder/sitemap_index_url_spec.rb

@@ -1,16 +1,28 @@
 require 'spec_helper'
 
 describe SitemapGenerator::Builder::SitemapIndexUrl do
-  before :all do
-    @s = SitemapGenerator::Builder::SitemapIndexFile.new(
+  let(:index) {
+    SitemapGenerator::Builder::SitemapIndexFile.new(
       :sitemaps_path => 'sitemaps/',
       :host => 'http://test.com',
       :filename => 'sitemap_index.xml.gz'
     )
-  end
+  }
+  let(:url)  { SitemapGenerator::Builder::SitemapUrl.new(index) }
 
   it "should return the correct url" do
-    @u = SitemapGenerator::Builder::SitemapUrl.new(@s)
-    @u[:loc].should == 'http://test.com/sitemaps/sitemap_index.xml.gz'
+    url[:loc].should == 'http://test.com/sitemaps/sitemap_index.xml.gz'
+  end
+
+  it "should use the host from the index" do
+    host = 'http://myexample.com'
+    index.location.expects(:host).returns(host)
+    url[:host].should == host
+  end
+
+  it "should use the public path for the link" do
+    path = '/path'
+    index.location.expects(:path_in_public).returns(path)
+    url[:loc].should == 'http://test.com/path'
   end
 end