sitemap_checker 0.0.4 → 0.1.0

data/README.md

@@ -18,13 +18,19 @@ Or install it yourself as:
 
 ## Usage
 
-Get list of urls from xml or xml.gz sitemap url
+Get list of urls(locs) from xml or xml.gz sitemap url.
     
-    SitemapChecker::Checker.new(url)
+    @sitemap = SitemapChecker::Sitemap.new(url)
+    @sitemap.locs.size
 
-Get status of url
+Get status of url from Sitemap
 
-    SitemapChecker::Checker.get_status(url)
+    @sitemap = SitemapChecker::Sitemap.new(url)
+    @sitemap.locs.first.status
+
+or directly as a Path
+
+    SitemapChecker::Path.new(url).status
 
 ## Contributing
 

data/lib/sitemap_checker/path.rb

@@ -0,0 +1,22 @@
+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
+        @status ||= open(@url).status[0]
+      rescue OpenURI::HTTPError => e
+        e.io.status[0]
+      end
+    end
+  end
+end

data/lib/sitemap_checker/sitemap.rb

@@ -0,0 +1,62 @@
+module SitemapChecker
+  class Sitemap
+    attr_accessor :locs, :map
+
+    def initialize(map)
+      @map = map
+      @locs = process_map
+    end
+
+    private
+
+    def process_map
+      xml = get_xml_from_map(@map)
+      if is_siteindex?(xml)
+        process_siteindex(xml)
+      else
+        process_sitemap(xml)
+      end
+    end
+
+    def get_xml_from_map(map)
+      begin
+        Nokogiri::XML(Zlib::GzipReader.new(open(map)))
+      rescue
+        Nokogiri::XML(open(map))
+      end
+    end
+
+    def is_siteindex?(xml)
+      ixsd = Nokogiri::XML::Schema(open('http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd'))
+      ixsd.valid?(xml)
+    end
+
+    def is_sitemap?(xml)
+      mxsd = Nokogiri::XML::Schema(open('http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd'))
+      mxsd.valid?(xml)
+    end
+
+    def process_siteindex(xml)
+      @urls = []
+      maps = get_locs(xml)
+      maps.each do |map|
+        xml = get_xml_from_map(map.url)
+        @urls += process_sitemap(xml)
+      end
+      return @urls
+    end
+
+    def process_sitemap(xml)
+      if is_sitemap?(xml)
+        return get_locs(xml)
+      else raise 'Invalid Schema'
+        return false
+      end
+    end
+
+    def get_locs(xml)
+      xml.xpath("//xmlns:loc").map{|path| Path.new(path) }
+    end
+
+  end
+end

data/lib/sitemap_checker/version.rb

@@ -1,3 +1,3 @@
 module SitemapChecker
-  VERSION = "0.0.4"
+  VERSION = "0.1.0"
 end

data/lib/sitemap_checker.rb

@@ -3,72 +3,8 @@ require 'open-uri'
 require 'nokogiri'
 require 'zlib'
 
-module SitemapChecker
-  class Checker
-    attr_reader :url_list
-
-    def initialize(url,schema='')
-      @url = url
-      @url_list = Array.new
-      @status_list = Array.new
-      process_xml
-    end
-
-    def self.get_status_from_xml(url)
-      get_status(url.content)
-    end
-
-    def self.get_status(url)
-      begin
-        status = [url,open(url).status[0]]
-      rescue OpenURI::HTTPError => e
-        status = [url,e.io.status[0]]
-      end
-    end
-
-    private
-
-    def get_xml_from_url(url)
-      begin
-        Nokogiri::XML(Zlib::GzipReader.new(open(url)))
-      rescue
-        Nokogiri::XML(open(url))
-      end
-    end
-
-    def process_xml
-      mxsd = Nokogiri::XML::Schema(open('http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd'))
-      ixsd = Nokogiri::XML::Schema(open('http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd'))
-      xml = get_xml_from_url(@url)
-      if mxsd.valid?(xml)
-        @url_list = urls(xml)
-      elsif ixsd.valid?(xml)
-        maps = urls(xml)
-        maps.each do |map|
-          xml = get_xml_from_url(map)
-          @url_list = urls(xml)
-        end
-      else raise 'Invalid Schema'
-        false
-      end
-    end
-
-    def urls(xml)
-      xml.xpath("//xmlns:loc")
-    end
-
-    def get_status_list(urls)
-      statuses = []
-      urls.each do |url|
-        begin
-          status = [url.content,open(url).status[0]]
-        rescue OpenURI::HTTPError => e
-          status = [url.content,e.io.status[0]]
-        end
-        statuses << status
-      end
-      statuses
-    end
-  end
 
+module SitemapChecker
+  autoload :Sitemap, "./lib/sitemap_checker/sitemap"
+  autoload :Path, "./lib/sitemap_checker/path"
 end

data/spec/sitemap_checker_spec.rb

@@ -16,35 +16,39 @@ describe SitemapChecker do
     stub_request(:get, "http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd").to_return(:status => 200, :body => File.read(@dir + 'fixtures/siteindex.xsd'), :headers => {})
   end
 
-  it "accepts xml siteindexes" do
-    @xml_sitemap = SitemapChecker::Checker.new('http://www.github.com/siteindex.xml')
-    @xml_sitemap.url_list.size.should eq(2)
+  it "Sitemap accepts xml siteindexes" do
+    @list = SitemapChecker::Sitemap.new('http://www.github.com/siteindex.xml')
+    @list.locs.size.should eq(4)
   end
 
-  it "accepts gzipped siteindexes" do
-    @gz_sitemap = SitemapChecker::Checker.new('http://www.github.com/siteindex.xml.gz')
-    @gz_sitemap.url_list.size.should eq(2)
+  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 "accepts xml sitemaps" do
-    @xml_sitemap = SitemapChecker::Checker.new('http://www.github.com/sitemap.xml')
-    @xml_sitemap.url_list.size.should eq(2)
+  it "Sitemap accepts xml sitemaps" do
+    @list = SitemapChecker::Sitemap.new('http://www.github.com/sitemap.xml')
+    @list.locs.size.should eq(2)
   end
 
-  it "accepts xml and gzipped sitemaps" do
-    @xml_sitemap = SitemapChecker::Checker.new('http://www.github.com/sitemap.xml')
-    @gz_sitemap = SitemapChecker::Checker.new('http://www.github.com/sitemap.xml.gz')
-    @xml_sitemap.url_list.size.should eq(2)
-    @gz_sitemap.url_list.size.should eq(2)
+  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 "Errors if input doc does not match sitemap schema" do
-    lambda {SitemapChecker::Checker.new('http://www.github.com')}.should raise_error(RuntimeError, 'Invalid Schema')
+  it "Sitemap errors if input doc does not match sitemap schema" do
+    lambda {SitemapChecker::Sitemap.new('http://www.github.com')}.should raise_error(RuntimeError, 'Invalid Schema')
   end
 
-  it "returns status if given a url" do
-    @sitemap = SitemapChecker::Checker.new('http://www.github.com/sitemap.xml')
-    SitemapChecker::Checker.get_status_from_xml(@sitemap.url_list.first).should eq(['http://www.github.com','200'])
+  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

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: sitemap_checker
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-08-24 00:00:00.000000000 Z
+date: 2012-08-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
@@ -40,6 +40,8 @@ files:
 - README.md
 - Rakefile
 - lib/sitemap_checker.rb
+- lib/sitemap_checker/path.rb
+- lib/sitemap_checker/sitemap.rb
 - lib/sitemap_checker/version.rb
 - sitemap_checker.gemspec
 - spec/fixtures/siteindex.xml