First pass at hatom support.

* reltag extract not working yet (placeholdered) * Only simple examples, not following spec 100% yet * Also includes refactoring microformat map some to avoid class loading issues * Also had to add yaml require for tests to run
mwunsch · Apr 27, 2013 · a1084ba · a1084ba
1 parent 2830c5d
commit a1084ba
Show file tree

Hide file tree

Showing 6 changed files with 247 additions and 12 deletions.
diff --git a/lib/prism/microformat.rb b/lib/prism/microformat.rb
@@ -3,21 +3,27 @@ module Microformat
 
     def self.map(name)
       map = microformats[Prism.normalize(name)]
+      map = format_to_class(map) if map
       raise "#{name} is not a recognized microformat." unless map
       map
     end
 
     def self.microformats
-      { :hcard      => Prism::Microformat::HCard,
-        :hcalendar  => Prism::Microformat::HCalendar,
-        :geo        => Prism::Microformat::Geo,
-        :adr        => Prism::Microformat::Adr,
-        :rellicense => Prism::Microformat::RelLicense,
-        :reltag     => Prism::Microformat::RelTag,
-        :votelinks  => Prism::Microformat::VoteLinks,
-        :xfn        => Prism::Microformat::XFN,
-        :xmdp       => Prism::Microformat::XMDP,
-        :xoxo       => Prism::Microformat::XOXO }
+      { :hatom      => "HAtom",
+        :hcard      => "HCard",
+        :hcalendar  => "HCalendar",
+        :geo        => "Geo",
+        :adr        => "Adr",
+        :rellicense => "RelLicense",
+        :reltag     => "RelTag",
+        :votelinks  => "VoteLinks",
+        :xfn        => "XFN",
+        :xmdp       => "XMDP",
+        :xoxo       => "XOXO" }
+    end
+
+    def self.format_to_class(format)
+      Prism::Microformat.const_get(format)
     end
 
     def self.find(html, uformat = nil)
@@ -31,7 +37,8 @@ def self.find(html, uformat = nil)
     def self.find_all(html)
       doc = Prism.get(html)
       uformats = microformats.values.collect do |uf|
-        uf.parse(doc)
+        klass = format_to_class(uf)
+        klass.parse(doc)
       end
       uformats.compact.flatten
     end
@@ -49,3 +56,4 @@ def self.find_all(html)
 require 'prism/microformat/adr'
 require 'prism/microformat/hcalendar'
 require 'prism/microformat/hcard'
+require 'prism/microformat/hatom'
diff --git a/lib/prism/microformat/hatom.rb b/lib/prism/microformat/hatom.rb
@@ -0,0 +1,70 @@
+module Prism
+  module Microformat
+    class HAtom < POSH
+      FRIENDLY_NAME = "hAtom"
+      WIKI_URL = "http://microformats.org/wiki/hatom"
+      XMDP = 'http://microformats.org/profile/hatom'
+
+      name :hfeed
+
+      search do |doc|
+        hfeed = doc.css("#{name}")
+        if hfeed.empty? && !doc.css('.hfeed').empty?
+          doc
+        else
+          hfeed
+        end
+      end
+
+      #TODO move to reltags
+      has_many :category do
+        search do |node|
+          node.css('a[rel~="tag"]')
+        end
+        extract do |node|
+          node['href'].split('/').last
+        end
+      end
+
+      has_many :hentry do
+        has_one :entry_title, :entry_summary
+
+        has_one :updated, :published do
+          extract :typevalue
+        end
+
+        has_one :author do
+          extract :hcard
+        end
+
+        has_one :entry_content do
+          extract do |node|
+            node.inner_html.strip
+          end
+        end
+
+        #TODO move to reltags
+        has_many :tags do
+          search do |node|
+            node.css('a[rel~="tag"]')
+          end
+          extract do |node|
+            node['href'].split('/').last
+          end
+        end
+
+        has_one :bookmark do
+          search do |node|
+            node.css('a[rel~="bookmark"]')
+          end
+          extract do |node|
+            node['href']
+          end
+        end
+
+        required!
+      end
+
+    end
+  end
+end
diff --git a/test/fixtures/test-fixture/hatom/example1.html b/test/fixtures/test-fixture/hatom/example1.html
@@ -0,0 +1,40 @@
+<!-- from http://microformats.org/wiki/hatom-examples -->
+<body>
+ <div id="uf">
+  <div class="hfeed" id="content">
+   <div class="hentry entry" id="post-60">
+    <h3 class="entry-title">
+     <a href="http://www.microformats.org/blog/..." rel="bookmark" title="...">Wiki Attack</a>
+    </h3>
+    <div class="entry-content">
+     <p>We had a bit of trouble with ...</p>
+     <p>We've restored the wiki and ...</p>
+     <p>If anyone is working to combat said spammers ...</p>
+    </div>
+
+    <h4 class="tags">Technorati Tags:</h4>
+    <ul class="tags">
+     <li><a href="http://technorati.com/tag/mediawiki" rel="tag">MediaWiki</a></li>
+     <li><a href="http://technorati.com/tag/microformats" rel="tag">MicroFormats</a></li>
+     <li><a href="http://technorati.com/tag/spam" rel="tag">Spam</a></li>
+    </ul>
+
+    <ul class="post-info">
+     <li>
+      <a href="http://www.microformats.org/blog/..." rel="bookmark" 
+        title="..."><abbr class="published" title="2005-10-10T14:07:00-07:00">October 10th, 2005</abbr></a>
+     </li>
+     <li>
+      <address class="vcard author"><a class="url fn" href="http://theryanking.com">Ryan King</a></address>
+     </li>
+     <li>
+      <a href="http://www.microformats.org/blog/...">4 Comments</a>
+     </li>
+    </ul>
+   </div>
+
+   <div class="hentry entry" id="post-59">
+   </div>
+  </div>
+ </div>
+</body>
diff --git a/test/fixtures/test-fixture/hatom/example2.html b/test/fixtures/test-fixture/hatom/example2.html
@@ -0,0 +1,35 @@
+<!-- http://microformats.org/wiki/hatom-examples -->
+<body>
+ <div id="uf">
+ <div class="hfeed">
+  <div class="hentry posts" id="112993192128302715">
+   <strong class="entry-title entry-content">
+    Nelson's final prayer
+   </strong> 
+   <span class="entry-content">
+    written on the night before Trafalgar:<blockquote>May the Great God, ... heart.
+   </span>
+   <div>
+    <span class="byline">posted by <span class="author vcard"><span class="fn">Natalie</span></span> at 
+     <a rel="bookmark" href="2005_10_16_nataliesolent_archive.html#112993192128302715">
+     <abbr class="published" title="2005-10-24t09:49:00-00:00">9:49 PM</abbr></a>
+    </span>
+   </div>
+  </div>
+
+  <div class="hentry posts" id="112993022840118939">
+   <strong class="heading content>"I really, truly </strong>
+   <span class="entry-content">
+    didn't go ... view.
+   </span>
+   <div>
+    <span class="byline">
+     posted by <span class="author vcard"><span class="fn">Natalie</span></span> at 
+     <a rel="bookmark" href="2005_10_16_nataliesolent_archive.html#112993022840118939">
+     <abbr class="published" title="2005-10-24t09:49:00-00:00">9:28 pm</abbr></a>
+    </span>
+   </div>
+  </div>
+ </div>
+</div>
+</body>
diff --git a/test/microformat/hatom_test.rb b/test/microformat/hatom_test.rb
@@ -0,0 +1,81 @@
+require File.join(File.dirname(File.absolute_path(__FILE__)),'..','test_helper')
+
+class HAtomTest < Test::Unit::TestCase
+  @@klass = Prism::Microformat::HAtom
+
+  describe 'first example test' do
+    def self.before_all
+      @doc ||= test_fixture('hatom/example1.html')
+      @hatom ||= @@klass.parse(@doc)
+    end
+
+    setup do
+      @hatom ||= self.class.before_all
+    end
+
+    test 'The title is a singular value' do
+      hentry = @hatom.hentry
+      assert_equal "Wiki Attack", hentry[0].entry_title
+    end
+
+    test 'The content contains html' do
+      hentry = @hatom.hentry
+      content = <<-EOS
+     <p>We had a bit of trouble with ...</p>
+     <p>We've restored the wiki and ...</p>
+     <p>If anyone is working to combat said spammers ...</p>
+      EOS
+      assert_equal content.strip, hentry[0].entry_content
+    end
+
+    test 'The published is a time' do
+      hentry = @hatom.hentry
+      assert_equal Time.parse('2005-10-10 14:07:00 -0700'), hentry[0].published
+    end
+
+    test 'The author is an hcard' do
+      hentry = @hatom.hentry
+      hcard = hentry[0].author
+      assert_equal Prism::Microformat::HCard, hcard.class
+      assert_equal 'Ryan King', hcard.fn
+      assert_equal 'King', hcard.n[:family_name]
+      assert_equal 'Ryan', hcard.n[:given_name]
+      assert_equal 'http://theryanking.com/', hcard.url[0]
+    end
+
+    test 'The tags are a list of values' do
+      hentry = @hatom.hentry
+      assert_equal ['mediawiki', 'microformats', 'spam'], hentry[0].tags
+    end
+
+  end
+
+  describe 'second example test' do
+    def self.before_all
+      @doc ||= test_fixture('hatom/example2.html')
+      @hatom ||= @@klass.parse(@doc)
+    end
+
+    setup do
+      @hatom ||= self.class.before_all
+    end
+
+    test 'The title is a singular value' do
+      hentry = @hatom.hentry
+      assert_equal "Nelson's final prayer", hentry[0].entry_title
+    end
+
+    test 'The content is an html value' do
+      hentry = @hatom.hentry
+      assert_equal "Nelson's final prayer", hentry[0].entry_content
+    end
+
+    test 'The bookmark is a rel-bookmark' do
+      hentry = @hatom.hentry
+      assert_equal "2005_10_16_nataliesolent_archive.html#112993192128302715", hentry[0].bookmark
+    end
+
+  end
+
+end
+
diff --git a/test/test_helper.rb b/test/test_helper.rb
@@ -8,6 +8,7 @@
   Bundler.setup
 end
 
+require 'yaml'
 require 'test/unit'
 require 'contest'
 require 'fakeweb'
@@ -34,4 +35,4 @@ def get_fixture(filename)
 def test_fixture(filename)
   html = get_fixture('test-fixture/' + filename)
   doc = Nokogiri::HTML.parse(html).css('#uf').first
-end
+end