lib/anemone/page.rb in anemone-0.0.3 vs lib/anemone/page.rb in anemone-0.0.4

@@ -1,27 +1,22 @@
 require 'anemone/http'
-require 'hpricot'
+require 'nokogiri'
+require 'facets/ostructable'
 
 module Anemone
   class Page
+    include OpenStructable
+
     # The URL of the page
     attr_reader :url
     # Array of distinct A tag HREFs from the page
     attr_reader :links
-    #Body of the HTTP response
-    attr_reader :body
     #Content-type of the  HTTP response
     attr_reader :content_type
-    #title of the page if it is an HTML document
-    attr_reader :title
-    #first h1 on the page, if present
-    attr_reader :h1
-    #first h2 on the page, if present
-    attr_reader :h2
-    #meta-description of the page, if present
-    attr_reader :description
     
+    #Nokogiri document for the HTML body
+    attr_accessor :doc
     # Integer response code of the page
     attr_accessor :code	
     # Array of redirect-aliases for the page
     attr_accessor :aliases
     # Boolean indicating whether or not this page has been visited in PageHash#shortest_paths!
@@ -52,40 +47,32 @@
     #
     # Create a new page
     #
     def initialize(url, body = nil, code = nil, content_type = nil, aka = nil)
       @url = url
-      @body = body unless Anemone.options.discard_page_bodies
       @code = code
       @content_type = content_type
       @links = []
       @aliases = []
-      
+	  
+	  #create empty storage for OpenStructable
+      update({})
+	  
       @aliases << aka if !aka.nil?
 
       if body
-        h = Hpricot(body)	  
+        begin
+          @doc = Nokogiri::HTML(body)
+        rescue
+          return
+        end
 
-        #save page title
-        title_elem = h.at('title')
-        @title = title_elem.inner_html if !title_elem.nil?
+        return if @doc.nil?
 
-        #save page h1
-        h1_elem = h.at('h1')
-        @h1 = h1_elem.inner_html if !h1_elem.nil?	  
-
-        #save page h2
-        h2_elem = h.at('h2')
-        @h2 = h2_elem.inner_html if !h2_elem.nil?
-
-        #save page meta-description
-        description_elem = h.at('meta[@name=description]')
-        @description = description_elem['content'] if !description_elem.nil?	  
-
         #get a list of distinct links on the page, in absolute url form
-        h.search('a').each do |a| 
-          u = a['href']
+        @doc.css('a').each do |a| 
+          u = a.attribute('href')
           next if u.nil?
           
           begin
             abs = to_absolute(URI(u))
           rescue
@@ -104,12 +91,12 @@
     # Return a new page with the same *response* and *url*, but
     # with a 200 response code
     #    
     def alias_clone(url)
       p = clone
-	    p.add_alias!(@aka) if !@aka.nil?
-	    p.code = 200
-	    p
+	  p.add_alias!(@aka) if !@aka.nil?
+	  p.code = 200
+	  p
     end
 
     #
     # Add a redirect-alias String *aka* to the list of the page's aliases
     #