sitemap_checker 0.1.8 → 0.2.1

data/spec/sitemap_spec.rb

@@ -0,0 +1,49 @@
+require 'rubygems'
+require 'webmock/rspec'
+require './lib/sitemap_checker/sitemap'
+require './lib/sitemap_checker/uri'
+WebMock.disable_net_connect!(:allow_localhost => true)
+
+describe 'SitemapChecker::Sitemap' do
+  before(:each) do
+    @dir  = Pathname.new(File.dirname(__FILE__))
+    stub_request(:any, "http://www.github.com").to_return(:status => 200, :body => 'foo', headers: {'Content-type' => 'text/html'})
+    stub_request(:any, "http://www.github.com/404").to_return(:status => 404, :body => 'foo', headers: {'Content-type' => 'text/html'})
+    stub_request(:any, "http://www.github.com/sitemap.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml'), headers: {'Content-type' => 'application/xml'})
+    stub_request(:any, "http://www.github.com/sitemap.xml.gz").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml.gz'), headers: {'Content-type' => 'application/octet-stream'})
+    stub_request(:any, "http://www.github.com/siteindex.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/siteindex.xml'), headers: {'Content-type' => 'application/xml'})
+    stub_request(:any, "http://www.github.com/siteindex.xml.gz").to_return(:status => 200, :body => File.read(@dir + 'fixtures/siteindex.xml.gz'), headers: {'Content-type' => 'application/octet-stream'})
+    stub_request(:any, "http://store.apple.com/apple-robots.txt").to_return(:status => 200, :body => File.read(@dir + 'fixtures/apple-robots.txt'), headers: {'Content-type' => 'text/plain'})
+    stub_request(:any, "http://store.apple.com/apple-index.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/apple-index.xml'), headers: {'Content-type' => 'application/xml'})
+    stub_request(:any, "http://store.apple.com/apple-sitemap.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/apple-sitemap.xml'), headers: {'Content-type' => 'application/xml'})
+    stub_request(:any, "http://store.apple.com/apple-sitemap-new.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/apple-sitemap-new.xml'), headers: {'Content-type' => 'application/xml'})
+  end
+
+  it "Sitemap gracefully handles 404s" do
+    lambda { SitemapChecker::Sitemap.new('http://www.github.com/404') }.should_not raise_error
+  end
+
+  it "Sitemap accepts xml siteindexes" do
+    SitemapChecker::Sitemap.new('http://www.github.com/siteindex.xml').locs.size.should eq(4)
+    SitemapChecker::Sitemap.new('http://store.apple.com/apple-index.xml').locs.size.should eq(419)
+    SitemapChecker::Sitemap.new('http://store.apple.com/apple-sitemap.xml').locs.size.should eq(214)
+    SitemapChecker::Sitemap.new('http://store.apple.com/apple-sitemap-new.xml').locs.size.should eq(205)
+  end
+
+  it "Sitemap accepts gzipped siteindexes" do
+    SitemapChecker::Sitemap.new('http://www.github.com/siteindex.xml.gz').locs.size.should eq(4)
+  end
+
+  it "Sitemap accepts xml sitemaps" do
+    SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml').locs.size.should eq(2)
+  end
+
+  it "Sitemap accepts gzipped sitemaps" do
+    SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml.gz').locs.size.should eq(2)
+  end
+
+  it "Sitemap locs are String objects" do
+    SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml').locs.first.class.should eq(String)
+  end
+
+end

data/spec/uri_spec.rb

@@ -0,0 +1,27 @@
+require 'rubygems'
+require 'webmock/rspec'
+require './lib/sitemap_checker/uri'
+WebMock.disable_net_connect!(:allow_localhost => true)
+
+describe 'SitemapChecker::Uri' do
+
+  before(:each) do
+    @dir  = Pathname.new(File.dirname(__FILE__))
+    stub_request(:any, "http://www.github.com/404").to_return(:status => 404)
+    stub_request(:any, "http://www.github.com/sitemap.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml'), headers: {'Content-type' => 'application/xml'})
+    stub_request(:any, "http://www.github.com/sitemap.xml.gz").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml.gz'), headers: {'Content-type' => 'application/octet-stream'})
+  end
+
+  it "Accepts XML" do
+    SitemapChecker::Uri.new('http://www.github.com/sitemap.xml').xml.class.should eq(Nokogiri::XML::Document)
+  end
+
+  it "Accepts Gzipped XML" do
+    SitemapChecker::Uri.new('http://www.github.com/sitemap.xml.gz').xml.class.should eq(Nokogiri::XML::Document)
+  end
+
+  it "does not contain IO object if not xml or gz" do
+    SitemapChecker::Uri.new('http://www.github.com/404').xml.class.should eq(NilClass)
+  end
+
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: sitemap_checker
 version: !ruby/object:Gem::Version
-  version: 0.1.8
+  version: 0.2.1
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-12-23 00:00:00.000000000 Z
+date: 2012-12-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
@@ -41,18 +41,23 @@ files:
 - Rakefile
 - lib/sitemap_checker.rb
 - lib/sitemap_checker/open_uri.rb
-- lib/sitemap_checker/path.rb
 - lib/sitemap_checker/sitemap.rb
+- lib/sitemap_checker/uri.rb
 - lib/sitemap_checker/version.rb
 - sitemap_checker.gemspec
+- spec/fixtures/apple-index.xml
+- spec/fixtures/apple-robots.txt
+- spec/fixtures/apple-sitemap-new.xml
+- spec/fixtures/apple-sitemap.xml
 - spec/fixtures/siteindex.xml
 - spec/fixtures/siteindex.xml.gz
 - spec/fixtures/siteindex.xsd
 - spec/fixtures/sitemap.xml
 - spec/fixtures/sitemap.xml.gz
 - spec/fixtures/sitemap.xsd
-- spec/sitemap_checker_spec.rb
+- spec/sitemap_spec.rb
 - spec/spec_helper.rb
+- spec/uri_spec.rb
 homepage: https://github.com/gerlandop/sitemap_checker
 licenses: []
 post_install_message: 
@@ -78,11 +83,17 @@ signing_key:
 specification_version: 3
 summary: Gets status of Urls in SiteMap
 test_files:
+- spec/fixtures/apple-index.xml
+- spec/fixtures/apple-robots.txt
+- spec/fixtures/apple-sitemap-new.xml
+- spec/fixtures/apple-sitemap.xml
 - spec/fixtures/siteindex.xml
 - spec/fixtures/siteindex.xml.gz
 - spec/fixtures/siteindex.xsd
 - spec/fixtures/sitemap.xml
 - spec/fixtures/sitemap.xml.gz
 - spec/fixtures/sitemap.xsd
-- spec/sitemap_checker_spec.rb
+- spec/sitemap_spec.rb
 - spec/spec_helper.rb
+- spec/uri_spec.rb
+has_rdoc: 

data/lib/sitemap_checker/path.rb

@@ -1,24 +0,0 @@
-module SitemapChecker
-  class Path
-    attr_accessor :url, :status
-
-    def initialize(url)
-      @url = url
-      @status = nil
-    end
-
-    def get_status_from_xml(url)
-      status(url.content)
-    end
-
-    def status
-      begin
-        open(@url, "Accept" => @url[/\.xml$/] ? 'application/xml' : 'text/html').status[0]
-      rescue RuntimeError => e
-        e
-      rescue OpenURI::HTTPError => e
-        e.io.status[0]
-      end
-    end
-  end
-end

data/spec/sitemap_checker_spec.rb

@@ -1,48 +0,0 @@
-require 'rubygems'
-require 'webmock/rspec'
-require './lib/sitemap_checker'
-WebMock.disable_net_connect!(:allow_localhost => true)
-
-describe SitemapChecker do
-  before(:each) do
-    @dir  = Pathname.new(File.dirname(__FILE__))
-    stub_request(:any, "http://www.github.com").to_return(:status => 200, :body => 'foo')
-    stub_request(:any, "http://www.github.com/404").to_return(:status => 404, :body => 'foo')
-    stub_request(:any, "http://www.github.com/sitemap.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml'))
-    stub_request(:any, "http://www.github.com/sitemap.xml.gz").to_return(:status => 200, :body => File.read(@dir + 'fixtures/sitemap.xml.gz'))
-    stub_request(:any, "http://www.github.com/siteindex.xml").to_return(:status => 200, :body => File.read(@dir + 'fixtures/siteindex.xml'))
-    stub_request(:any, "http://www.github.com/siteindex.xml.gz").to_return(:status => 200, :body => File.read(@dir + 'fixtures/siteindex.xml.gz'))
-  end
-
-  it "Sitemap accepts xml siteindexes" do
-    @list = SitemapChecker::Sitemap.new('http://www.github.com/siteindex.xml')
-    @list.locs.size.should eq(4)
-  end
-
-  it "Sitemap accepts gzipped siteindexes" do
-    @list = SitemapChecker::Sitemap.new('http://www.github.com/siteindex.xml.gz')
-    @list.locs.size.should eq(4)
-  end
-
-  it "Sitemap accepts xml sitemaps" do
-    @list = SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml')
-    @list.locs.size.should eq(2)
-  end
-
-  it "Sitemap accepts xml and gzipped sitemaps" do
-    @xml_sitemap = SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml')
-    @gz_sitemap = SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml.gz')
-    @xml_sitemap.locs.size.should eq(2)
-    @gz_sitemap.locs.size.should eq(2)
-  end
-
-  it "Sitemap locs are Path objects" do
-    @xml_sitemap = SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml')
-    @xml_sitemap.locs.first.class.should eq(SitemapChecker::Path)
-  end
-
-  it "Path#status returns status code" do
-    SitemapChecker::Path.new('http://www.github.com').status.should eq('200')
-  end
-
-end