From e9ce55afeabf34e4e689b10b71be481279b7df8d Mon Sep 17 00:00:00 2001
From: Jessica Lynn Suttles <jlsuttles@gmail.com>
Date: Mon, 4 Feb 2013 14:55:06 -0800
Subject: [PATCH] walks dom tree looking for root microformat classes

---
 lib/microformats2.rb           | 34 +++++++++++++++++++++++++++++++++-
 spec/lib/microformats2_spec.rb |  6 ++++++
 2 files changed, 39 insertions(+), 1 deletion(-)

diff --git a/lib/microformats2.rb b/lib/microformats2.rb
index 03e1ac8..0c60890 100644
--- a/lib/microformats2.rb
+++ b/lib/microformats2.rb
@@ -5,7 +5,9 @@ require "microformats2/version"
 module Microformats2
   def self.parse(html)
     html = read_html(html)
-    Nokogiri::HTML(html)
+    document = Nokogiri::HTML(html)
+    parsed_document = parse_nodeset(document.children)
+    parsed_document.flatten.compact
   end
 
   def self.read_html(html)
@@ -13,4 +15,34 @@ module Microformats2
   rescue Errno::ENOENT => e
     html
   end
+
+  def self.parse_nodeset(nodeset)
+    nodeset.map do |node| parse_node(node) end
+  end
+
+  def self.parse_node(node)
+    case
+    when node.is_a?(Nokogiri::XML::NodeSet) then parse_nodeset(node)
+    when node.is_a?(Nokogiri::XML::Element) then parse_element(node)
+    end
+  end
+
+  def self.parse_element(element)
+    # innocent until proven guilty
+    microformat = false
+
+    # look for root microformat class
+    element.attribute("class").to_s.split.each do |html_class|
+      microformat = microformat || html_class =~ /^h-/
+    end
+
+    # if found root microformat, yay
+    if microformat
+      "YAY MICROFORMAT"
+
+    # if no root microformat found, look at children
+    else
+      parse_nodeset(element.children)
+    end
+  end
 end
diff --git a/spec/lib/microformats2_spec.rb b/spec/lib/microformats2_spec.rb
index 8e54335..903b6b6 100644
--- a/spec/lib/microformats2_spec.rb
+++ b/spec/lib/microformats2_spec.rb
@@ -2,6 +2,12 @@ require "spec_helper"
 require "microformats2"
 
 describe Microformats2 do
+  describe "::parse" do
+    it "returns an array of found microformats" do
+      html = "spec/support/simple.html"
+      Microformats2.parse(html).should == ["YAY MICROFORMAT"]
+    end
+  end
   describe "::read_html" do
     before do
       @html = <<-HTML.strip