sitemap_generator 2.2.1 → 3.0.0

data/spec/sitemap_generator/news_sitemap_spec.rb

@@ -0,0 +1,42 @@
+require 'spec_helper'
+
+describe "SitemapGenerator" do
+
+  it "should add the news sitemap element" do
+    loc = 'http://www.example.com/my_article.html'
+
+    news_xml_fragment = SitemapGenerator::Builder::SitemapUrl.new('my_article.html', {
+      :host => 'http://www.example.com',
+
+      :news => {
+        :publication_name => "Example",
+        :publication_language => "en",
+        :title => "My Article",
+        :keywords => "my article, articles about myself",
+        :stock_tickers => "SAO:PETR3",
+        :publication_date => "2011-08-22",
+        :access => "Subscription",
+        :genres => "PressRelease"
+      }
+    }).to_xml
+
+    doc = Nokogiri::XML.parse("<root xmlns:news='http://www.google.com/schemas/sitemap-news/0.9'>#{news_xml_fragment}</root>")
+
+    url = doc.at_xpath("//url")
+    loc = url.at_xpath("loc")
+    loc.text.should == 'http://www.example.com/my_article.html'
+
+    news = doc.at_xpath("//news:news")
+
+    news.at_xpath('//news:title').text.should == "My Article"
+    news.at_xpath("//news:keywords").text.should == "my article, articles about myself"
+    news.at_xpath("//news:stock_tickers").text.should == "SAO:PETR3"
+    news.at_xpath("//news:publication_date").text.should == "2011-08-22"
+    news.at_xpath("//news:access").text.should == "Subscription"
+    news.at_xpath("//news:genres").text.should == "PressRelease"
+    news.at_xpath("//news:name").text.should == "Example"
+    news.at_xpath("//news:language").text.should == "en"
+
+    xml_fragment_should_validate_against_schema(news, 'http://www.google.com/schemas/sitemap-news/0.9', 'sitemap-news')
+  end
+end

data/spec/sitemap_generator/sitemap_generator_spec.rb

@@ -0,0 +1,232 @@
+require 'spec_helper'
+
+class Holder
+  class << self
+    attr_accessor :executed
+  end
+end
+
+def with_max_links(num)
+  original = SitemapGenerator::MAX_SITEMAP_LINKS
+  SitemapGenerator.const_set(:MAX_SITEMAP_LINKS, num)
+  yield
+  SitemapGenerator.const_set(:MAX_SITEMAP_LINKS, original)
+end
+
+describe "SitemapGenerator" do
+
+  describe "reset!" do
+    before :each do
+      SitemapGenerator::Sitemap.default_host # Force initialization of the LinkSet
+    end
+
+    it "should set a new LinkSet instance" do
+      first = SitemapGenerator::Sitemap.instance_variable_get(:@link_set)
+      first.should be_a(SitemapGenerator::LinkSet)
+      SitemapGenerator::Sitemap.reset!
+      second = SitemapGenerator::Sitemap.instance_variable_get(:@link_set)
+      second.should be_a(SitemapGenerator::LinkSet)
+      first.should_not be(second)
+    end
+  end
+
+  describe "root" do
+    it "should be set to the root of the gem" do
+      SitemapGenerator.root.should == File.expand_path('../../../' , __FILE__)
+    end
+  end
+
+  [:deprecated, :create].each do |extension|
+    describe "generate sitemap" do
+      before :all do
+        SitemapGenerator::Sitemap.reset!
+        clean_sitemap_files_from_rails_app
+        copy_sitemap_file_to_rails_app(extension)
+        with_max_links(10) { execute_sitemap_config }
+      end
+
+      it "should create sitemaps" do
+        file_should_exist(rails_path('public/sitemap_index.xml.gz'))
+        file_should_exist(rails_path('public/sitemap1.xml.gz'))
+        file_should_exist(rails_path('public/sitemap2.xml.gz'))
+        file_should_not_exist(rails_path('public/sitemap3.xml.gz'))
+      end
+
+      it "should have 14 links" do
+        SitemapGenerator::Sitemap.link_count.should == 14
+      end
+
+      it "index XML should validate" do
+        gzipped_xml_file_should_validate_against_schema rails_path('public/sitemap_index.xml.gz'), 'siteindex'
+      end
+
+      it "sitemap XML should validate" do
+        gzipped_xml_file_should_validate_against_schema rails_path('public/sitemap1.xml.gz'), 'sitemap'
+        gzipped_xml_file_should_validate_against_schema rails_path('public/sitemap2.xml.gz'), 'sitemap'
+      end
+
+      it "index XML should not have excess whitespace" do
+        gzipped_xml_file_should_have_minimal_whitespace rails_path('public/sitemap_index.xml.gz')
+      end
+
+      it "sitemap XML should not have excess whitespace" do
+        gzipped_xml_file_should_have_minimal_whitespace rails_path('public/sitemap1.xml.gz')
+      end
+    end
+  end
+
+  describe "sitemap with groups" do
+    before :all do
+      SitemapGenerator::Sitemap.reset!
+      clean_sitemap_files_from_rails_app
+      copy_sitemap_file_to_rails_app(:groups)
+      with_max_links(2) { execute_sitemap_config }
+      @expected = %w[
+        public/en/xxx1.xml.gz
+        public/fr/abc3.xml.gz
+        public/fr/abc4.xml.gz
+        public/fr/new_sitemaps_index.xml.gz
+        public/fr/new_sitemaps1.xml.gz
+        public/fr/new_sitemaps2.xml.gz
+        public/fr/new_sitemaps3.xml.gz
+        public/fr/new_sitemaps4.xml.gz]
+      @sitemaps = (@expected - %w[public/fr/new_sitemaps_index.xml.gz])
+    end
+
+    it "should create sitemaps" do
+      @expected.each { |file| file_should_exist(rails_path(file)) }
+      file_should_not_exist(rails_path('public/fr/new_sitemaps5.xml.gz'))
+      file_should_not_exist(rails_path('public/en/xxx2.xml.gz'))
+      file_should_not_exist(rails_path('public/fr/abc5.xml.gz'))
+    end
+
+    it "should have 14 links" do
+      SitemapGenerator::Sitemap.link_count.should == 13
+    end
+
+    it "index XML should validate" do
+      gzipped_xml_file_should_validate_against_schema rails_path('public/fr/new_sitemaps_index.xml.gz'), 'siteindex'
+    end
+
+    it "index XML should not have excess whitespace" do
+      gzipped_xml_file_should_have_minimal_whitespace rails_path('public/fr/new_sitemaps_index.xml.gz')
+    end
+
+    it "sitemaps XML should validate" do
+      @sitemaps.each { |file| gzipped_xml_file_should_validate_against_schema(rails_path(file), 'sitemap') }
+    end
+
+    it "sitemap XML should not have excess whitespace" do
+      @sitemaps.each { |file| gzipped_xml_file_should_have_minimal_whitespace(rails_path(file)) }
+    end
+  end
+
+  describe "sitemap path" do
+    before :each do
+      clean_sitemap_files_from_rails_app
+      ::SitemapGenerator::Sitemap.reset!
+      ::SitemapGenerator::Sitemap.default_host = 'http://test.local'
+      ::SitemapGenerator::Sitemap.filename = 'sitemap'
+    end
+
+    it "should allow changing of the filename" do
+      ::SitemapGenerator::Sitemap.create(:filename => :geo_sitemap) do
+        add '/goerss', :geo => { :format => 'georss' }
+        add '/kml', :geo => { :format => 'kml' }
+      end
+      file_should_exist(rails_path('public/geo_sitemap_index.xml.gz'))
+      file_should_exist(rails_path('public/geo_sitemap1.xml.gz'))
+    end
+
+    it "should support setting a sitemap path" do
+      directory_should_not_exist(rails_path('public/sitemaps/'))
+
+      sm = ::SitemapGenerator::Sitemap
+      sm.sitemaps_path = 'sitemaps/'
+      sm.create do
+        add '/'
+        add '/another'
+      end
+
+      file_should_exist(rails_path('public/sitemaps/sitemap_index.xml.gz'))
+      file_should_exist(rails_path('public/sitemaps/sitemap1.xml.gz'))
+    end
+
+    it "should support setting a deeply nested sitemap path" do
+      directory_should_not_exist(rails_path('public/sitemaps/deep/directory'))
+
+      sm = ::SitemapGenerator::Sitemap
+      sm.sitemaps_path = 'sitemaps/deep/directory/'
+      sm.create do
+        add '/'
+        add '/another'
+        add '/yet-another'
+      end
+
+      file_should_exist(rails_path('public/sitemaps/deep/directory/sitemap_index.xml.gz'))
+      file_should_exist(rails_path('public/sitemaps/deep/directory/sitemap1.xml.gz'))
+    end
+  end
+
+  describe "external dependencies" do
+    describe "rails" do
+      before :each do
+        @rails = Rails
+        Object.send(:remove_const, :Rails)
+      end
+
+      after :each do
+        Object::Rails = @rails
+      end
+
+      it "should work outside of Rails" do
+        defined?(Rails).should be_nil
+        lambda { ::SitemapGenerator::LinkSet.new }.should_not raise_exception
+      end
+    end
+  end
+
+  describe "verbose" do
+    it "should be set via ENV['VERBOSE']" do
+      original = SitemapGenerator.verbose
+      SitemapGenerator.verbose = nil
+      ENV['VERBOSE'] = 'true'
+      SitemapGenerator.verbose.should be_true
+      SitemapGenerator.verbose = nil
+      ENV['VERBOSE'] = 'false'
+      SitemapGenerator.verbose.should be_false
+      SitemapGenerator.verbose = original
+    end
+  end
+
+  protected
+
+  #
+  # Helpers
+  #
+
+  def rails_path(file)
+    SitemapGenerator.app.root + file
+  end
+
+  def copy_sitemap_file_to_rails_app(extension)
+    FileUtils.cp(File.join(SitemapGenerator.root, "spec/files/sitemap.#{extension}.rb"), SitemapGenerator.app.root + 'config/sitemap.rb')
+  end
+
+  def delete_sitemap_file_from_rails_app
+    FileUtils.remove(SitemapGenerator.app.root + 'config/sitemap.rb')
+  rescue
+    nil
+  end
+
+  def clean_sitemap_files_from_rails_app
+    FileUtils.rm_rf(rails_path('public/'))
+    FileUtils.mkdir_p(rails_path('public/'))
+  end
+
+  # Better would be to just invoke the environment task and use
+  # the interpreter.
+  def execute_sitemap_config
+   SitemapGenerator::Interpreter.run
+  end
+end

data/spec/sitemap_generator/sitemap_groups_spec.rb

@@ -0,0 +1,133 @@
+require "spec_helper"
+
+def with_max_links(num)
+  SitemapGenerator::Utilities.with_warnings(nil) do
+    original = SitemapGenerator::MAX_SITEMAP_LINKS
+    SitemapGenerator.const_set(:MAX_SITEMAP_LINKS, num)
+    yield
+    SitemapGenerator.const_set(:MAX_SITEMAP_LINKS, original)
+  end
+end
+
+describe "Sitemap Groups" do
+  before :each do
+    @sm = ::SitemapGenerator::LinkSet.new(:default_host => 'http://test.com')
+    FileUtils.rm_rf(SitemapGenerator.app.root + 'public/')
+  end
+
+  it "should not finalize the default sitemap if using groups" do
+    @sm.create do
+      group(:filename => :sitemap_en) do
+        add '/en'
+      end
+    end
+
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_en1.xml.gz')
+    file_should_not_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+  end
+
+  it "should add default links if no groups are created" do
+    @sm.create do
+    end
+    @sm.link_count.should == 2
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+  end
+
+  it "should add links to the default sitemap" do
+    @sm.create do
+      add '/before'
+      group(:filename => :sitemap_en) { }
+      add '/after'
+    end
+    @sm.link_count.should == 4
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_en1.xml.gz')
+  end
+
+  it "should rollover when sitemaps are full" do
+    with_max_links(1) {
+      @sm.include_index = false
+      @sm.include_root = false
+      @sm.create do
+        add '/before'
+        group(:filename => :sitemap_en, :sitemaps_path => 'en/') do
+          add '/one'
+          add '/two'
+        end
+        add '/after'
+      end
+    }
+    @sm.link_count.should == 4
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap2.xml.gz')
+    file_should_not_exist(SitemapGenerator.app.root + 'public/sitemap3.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/en/sitemap_en1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/en/sitemap_en2.xml.gz')
+    file_should_not_exist(SitemapGenerator.app.root + 'public/en/sitemap_en3.xml.gz')
+  end
+
+  it "should support multiple groups" do
+    @sm.create do
+      group(:filename => :sitemap_en, :sitemaps_path => 'en/') do
+        add '/one'
+      end
+      group(:filename => :sitemap_fr, :sitemaps_path => 'fr/') do
+        add '/one'
+      end
+    end
+    @sm.link_count.should == 2
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/en/sitemap_en1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/fr/sitemap_fr1.xml.gz')
+  end
+
+  it "the sitemap shouldn't be finalized if the groups don't conflict" do
+    @sm.create do
+      add 'one'
+      group(:filename => :first) { add '/two' }
+      add 'three'
+      group(:filename => :second) { add '/four' }
+      add 'five'
+    end
+    @sm.link_count.should == 7
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/first1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/second1.xml.gz')
+  end
+
+  it "groups should share the sitemap if the sitemap location is unchanged" do
+    @sm.create do
+      add 'one'
+      group(:default_host => 'http://newhost.com') { add '/two' }
+      add 'three'
+      group(:default_host => 'http://betterhost.com') { add '/four' }
+      add 'five'
+    end
+    @sm.link_count.should == 7
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+    file_should_not_exist(SitemapGenerator.app.root + 'public/sitemap2.xml.gz')
+  end
+
+  it "sitemaps should be finalized if virtual location settings are changed" do
+    @sm.create do
+      add 'one'
+      group(:sitemaps_path => :en) { add '/two' }
+      add 'three'
+      group(:sitemaps_host => 'http://newhost.com') { add '/four' }
+      add 'five'
+    end
+    @sm.link_count.should == 7
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap_index.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap1.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap2.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/sitemap3.xml.gz')
+    file_should_not_exist(SitemapGenerator.app.root + 'public/sitemap4.xml.gz')
+    file_should_exist(SitemapGenerator.app.root + 'public/en/sitemap1.xml.gz')
+  end
+end

data/spec/sitemap_generator/sitemap_location_spec.rb

@@ -0,0 +1,124 @@
+require 'spec_helper'
+
+describe SitemapGenerator::SitemapLocation do
+  before :all do
+    @default_host = 'http://example.com'
+  end
+
+  it "public_path should default to the public directory in the application root" do
+    @l = SitemapGenerator::SitemapLocation.new
+    @l.public_path.should == SitemapGenerator.app.root + 'public/'
+  end
+
+  it "should have a default namer" do
+    @l = SitemapGenerator::SitemapLocation.new
+    @l[:namer].should_not be_nil
+    @l[:filename].should be_nil
+    @l.filename.should == 'sitemap1.xml.gz'
+  end
+
+  it "should require a filename" do
+    @l = SitemapGenerator::SitemapLocation.new
+    @l[:filename] = nil
+    lambda {
+      @l.filename.should be_nil
+    }.should raise_error
+  end
+
+  it "should require a namer" do
+    @l = SitemapGenerator::SitemapLocation.new
+    @l[:namer] = nil
+    lambda {
+      @l.filename.should be_nil
+    }.should raise_error
+  end
+  
+  it "should require a host" do
+    @l = SitemapGenerator::SitemapLocation.new(:filename => nil, :namer => nil)
+    lambda {
+      @l.host.should be_nil
+    }.should raise_error
+  end
+
+  it "should accept a Namer option" do
+    @namer = SitemapGenerator::SitemapNamer.new(:xxx)
+    @l = SitemapGenerator::SitemapLocation.new(:namer => @namer)
+    @l.filename.should == @namer.to_s
+  end
+
+  it "should protect the filename from further changes in the Namer" do
+    @namer = SitemapGenerator::SitemapNamer.new(:xxx)
+    @l = SitemapGenerator::SitemapLocation.new(:namer => @namer)
+    @l.filename.should == @namer.to_s
+    @namer.next
+    @l.filename.should == @namer.previous.to_s
+  end
+
+  it "should allow changing the namer" do
+    @namer1 = SitemapGenerator::SitemapNamer.new(:xxx)
+    @l = SitemapGenerator::SitemapLocation.new(:namer => @namer1)
+    @l.filename.should == @namer1.to_s
+    @namer2 = SitemapGenerator::SitemapNamer.new(:yyy)
+    @l[:namer] = @namer2
+    @l.filename.should == @namer2.to_s
+  end
+
+  describe "testing options and #with" do
+    before :all do
+      @l = SitemapGenerator::SitemapLocation.new
+    end
+
+    # Array of tuples with instance options and expected method return values
+    tests = [
+      [{
+        :sitemaps_path => nil,
+        :public_path => '/public',
+        :filename => 'sitemap1.xml.gz',
+        :host => 'http://test.com' },
+      { :url => 'http://test.com/sitemap1.xml.gz',
+        :directory => '/public',
+        :path => '/public/sitemap1.xml.gz',
+        :path_in_public => 'sitemap1.xml.gz'
+      }],
+      [{
+        :sitemaps_path => 'sitemaps/en/',
+        :public_path => '/public/system/',
+        :filename => 'sitemap1.xml.gz',
+        :host => 'http://test.com/plus/extra/' },
+      { :url => 'http://test.com/plus/extra/sitemaps/en/sitemap1.xml.gz',
+        :directory => '/public/system/sitemaps/en',
+        :path => '/public/system/sitemaps/en/sitemap1.xml.gz',
+        :path_in_public => 'sitemaps/en/sitemap1.xml.gz'
+      }]
+    ]
+    tests.each do |opts, returns|
+      returns.each do |method, value|
+        it "#{method} should return #{value}" do
+          @l.with(opts).send(method).should == value
+        end
+      end
+    end
+  end
+
+  describe "when duplicated" do
+    it "should not inherit some objects" do
+      @l = SitemapGenerator::SitemapLocation.new(:filename => 'xxx', :host => @default_host, :public_path => 'public/')
+      @l.url.should == @default_host+'/xxx'
+      @l.public_path.to_s.should == 'public/'
+      dup = @l.dup
+      dup.url.should == @l.url
+      dup.url.should_not be(@l.url)
+      dup.public_path.to_s.should == @l.public_path.to_s
+      dup.public_path.should_not be(@l.public_path)
+    end
+  end
+end
+
+describe SitemapGenerator::SitemapIndexLocation do
+  it "should have a default namer" do
+    @l = SitemapGenerator::SitemapIndexLocation.new
+    @l[:namer].should_not be_nil
+    @l[:filename].should be_nil
+    @l.filename.should == 'sitemap_index.xml.gz'
+  end
+end