sitemap_spec.rb in w3clove-0.7.4

- old
+ new

@@ -9,10 +9,11 @@
     @sitemap_no_links                     = W3Clove::Sitemap.new('http://zigotica.com')
     @sitemap_with_trailing_slash          = W3Clove::Sitemap.new('http://eparreno.com')
     @sitemap_with_protocol_relative       = W3Clove::Sitemap.new('http://protocol-relative.com')
     @sitemap_with_protocol_relative_https = W3Clove::Sitemap.new('https://protocol-relative.com')
     @sitemap_for_exclusions               = W3Clove::Sitemap.new('http://example.com/exclusions')
+    @sitemap_for_absolute_urls            = W3Clove::Sitemap.new('http://w3clove.com/faqs')
 
     MarkupValidator.any_instance.stubs(:validate_uri).returns(stubbed_validator_results)
   end
 
   it "should have an URL" do
@@ -26,14 +27,14 @@
       @sitemap.pages[1].url.should == 'http://www.ryanair.com/es/careers/job'
       @sitemap.pages[2].url.should == 'http://www.ryanair.com/es/about'
     end
 
     it "should get pages from the sample guides.rubyonrails.org site" do
-      @sitemap_html.pages.length.should == 33
+      #@sitemap_html.pages.length.should == 34
       @sitemap_html.pages.map {|p| p.url}
-        .should == ["http://guides.rubyonrails.org/index.html",
-                    "http://guides.rubyonrails.org/",
+        .should == ["http://guides.rubyonrails.org",
+                    "http://guides.rubyonrails.org/index.html",
                     "http://guides.rubyonrails.org/getting_started.html",
                     "http://guides.rubyonrails.org/migrations.html",
                     "http://guides.rubyonrails.org/active_record_validations_callbacks.html",
                     "http://guides.rubyonrails.org/association_basics.html",
                     "http://guides.rubyonrails.org/active_record_querying.html",
@@ -63,30 +64,51 @@
                     "http://guides.rubyonrails.org/contribute.html",
                     "http://guides.rubyonrails.org/credits.html",
                     "http://guides.rubyonrails.org/v2.3.8/"]
     end
 
+    it "should get correct absolute links for internal pages" do
+      @sitemap_for_absolute_urls.pages.length.should == 9
+      @sitemap_for_absolute_urls.pages.map {|p| p.url}
+        .should == ["http://w3clove.com/faqs",
+                    "http://w3clove.com/",
+                    "http://w3clove.com/plans-and-pricing",
+                    "http://w3clove.com/contact",
+                    "http://w3clove.com/charts/errors",
+                    "http://w3clove.com/credits",
+                    "http://w3clove.com/signin",
+                    "http://w3clove.com/api_v1_reference",
+                    "http://w3clove.com/terms_of_service"]
+    end
+
     it "should include sitemap url at least, even if no links were found" do
       @sitemap_no_links.pages.length.should == 1
       @sitemap_no_links.pages[0].url.should == 'http://zigotica.com'
     end
 
-    it "should not repeat URLs with and without trailing slash" do
+    it "should not repeat sitemap URL with and without trailing slash" do
       urls = @sitemap_with_trailing_slash.pages.collect(&:url)
-      urls.should      include 'http://eparreno.com/'
-      urls.should_not  include 'http://eparreno.com'
+      urls.should      include 'http://eparreno.com'
+      urls.should_not  include 'http://eparreno.com/'
     end
 
+    it "should not repeat internal URLs with and without trailing slash" do
+      urls = @sitemap_for_absolute_urls.pages.collect(&:url)
+      urls.should      include 'http://w3clove.com/faqs'
+      urls.should_not  include 'http://w3clove.com/faqs/'
+    end
+
     it "should exclude non-html pages" do
-      @sitemap_for_exclusions.pages.length.should == 2
-      @sitemap_for_exclusions.pages.first.url.should == 'http://example.com/exclusions/'
-      @sitemap_for_exclusions.pages.last.url.should  == 'http://example.com/exclusions/faqs'
+      @sitemap_for_exclusions.pages.length.should == 3
+      @sitemap_for_exclusions.pages[0].url.should == 'http://example.com/exclusions'
+      @sitemap_for_exclusions.pages[1].url.should == 'http://example.com/'
+      @sitemap_for_exclusions.pages[2].url.should == 'http://example.com/faqs'
     end
 
     context "protocol-relative links" do
       it "should include only internal links" do
         @sitemap_with_protocol_relative.pages.size.should == 3
-        @sitemap_with_protocol_relative.pages.map {|p| p.url}.sort.should == ['http://protocol-relative.com/',
+        @sitemap_with_protocol_relative.pages.map {|p| p.url}.sort.should == ['http://protocol-relative.com',
                                                                               'http://protocol-relative.com/contact',
                                                                               'http://protocol-relative.com/faqs']
       end
 
       it "should include protocol-relative link to same domain" do
\ No newline at end of file