feedtools 0.2.14 → 0.2.15

data/CHANGELOG

@@ -1,3 +1,9 @@
+== FeedTools 0.2.15
+ * changed so that tidied elements no longer handle entities differently
+ * tidy now assumes utf-8 encoding instead of ascii
+ * began the first bit of work on namespace support
+ * better support for cdatas
+ * removed "bozo bit" - no need for a parser to moonlight as a validator
 == FeedTools 0.2.14
  * fixed bug with feeds not expiring properly
 == FeedTools 0.2.13

data/lib/feed_tools.rb

@@ -32,7 +32,54 @@ FEED_TOOLS_ENV = ENV['FEED_TOOLS_ENV'] ||
                  ENV['RAILS_ENV'] ||
                  'production' # :nodoc:
 
-FEED_TOOLS_VERSION = "0.2.14"
+FEED_TOOLS_VERSION = "0.2.15"
+
+FEED_TOOLS_NAMESPACES = {
+  "admin" => "http://webns.net/mvcb/",
+  "ag" => "http://purl.org/rss/1.0/modules/aggregation/",
+  "annotate" => "http://purl.org/rss/1.0/modules/annotate/",
+  "atom" => "http://www.w3.org/2005/Atom",
+  "audio" => "http://media.tangent.org/rss/1.0/",
+  "blogChannel" => "http://backend.userland.com/blogChannelModule",
+  "cc" => "http://web.resource.org/cc/",
+  "creativeCommons" => "http://backend.userland.com/creativeCommonsRssModule",
+  "co" => "http://purl.org/rss/1.0/modules/company",
+  "content" => "http://purl.org/rss/1.0/modules/content/",
+  "cp" => "http://my.theinfo.org/changed/1.0/rss/",
+  "dc" => "http://purl.org/dc/elements/1.1/",
+  "dcterms" => "http://purl.org/dc/terms/",
+  "email" => "http://purl.org/rss/1.0/modules/email/",
+  "ev" => "http://purl.org/rss/1.0/modules/event/",
+  "icbm" => "http://postneo.com/icbm/",
+  "image" => "http://purl.org/rss/1.0/modules/image/",
+  "feedburner" => "http://rssnamespace.org/feedburner/ext/1.0",
+  "foaf" => "http://xmlns.com/foaf/0.1/",
+  "fm" => "http://freshmeat.net/rss/fm/",
+  "itunes" => "http://www.itunes.com/DTDs/Podcast-1.0.dtd",
+  "l" => "http://purl.org/rss/1.0/modules/link/",
+  "media" => "http://search.yahoo.com/mrss",
+  "pingback" => "http://madskills.com/public/xml/rss/module/pingback/",
+  "prism" => "http://prismstandard.org/namespaces/1.2/basic/",
+  "rdf" => "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
+  "rdfs" => "http://www.w3.org/2000/01/rdf-schema#",
+  "ref" => "http://purl.org/rss/1.0/modules/reference/",
+  "reqv" => "http://purl.org/rss/1.0/modules/richequiv/",
+  "search" => "http://purl.org/rss/1.0/modules/search/",
+  "slash" => "http://purl.org/rss/1.0/modules/slash/",
+  "soap" => "http://schemas.xmlsoap.org/soap/envelope/",
+  "ss" => "http://purl.org/rss/1.0/modules/servicestatus/",
+  "str" => "http://hacks.benhammersley.com/rss/streaming/",
+  "sub" => "http://purl.org/rss/1.0/modules/subscription/",
+  "sy" => "http://purl.org/rss/1.0/modules/syndication/",
+  "taxo" => "http://purl.org/rss/1.0/modules/taxonomy/",
+  "thr" => "http://purl.org/rss/1.0/modules/threading/",
+  "ti" => "http://purl.org/rss/1.0/modules/textinput/",
+  "trackback" => "http://madskills.com/public/xml/rss/module/trackback/",
+  "wfw" => "http://wellformedweb.org/CommentAPI/",
+  "wiki" => "http://purl.org/rss/1.0/modules/wiki/",
+  "xhtml" => "http://www.w3.org/1999/xhtml",
+  "xml" => "http://www.w3.org/XML/1998/namespace"
+}
 
 $:.unshift(File.dirname(__FILE__))
 $:.unshift(File.dirname(__FILE__) + "/feed_tools/vendor")
@@ -424,7 +471,7 @@ module FeedTools
   end
 
   # Tidys up the html
-  def FeedTools.tidy_html(html)
+  def FeedTools.tidy_html(html, options = {})
     return nil if html.nil?
     if FeedTools.tidy_enabled?
       is_fragment = true
@@ -438,9 +485,15 @@ module FeedTools
       end
       tidy_html = Tidy.open(:show_warnings=>false) do |tidy|
         tidy.options.output_xml = true
+        tidy.options.numeric_entities = true
+        tidy.options.markup = true
         tidy.options.indent = false
         tidy.options.wrap_attributes = true
         tidy.options.logical_emphasis = true
+        # TODO: Make this match the actual encoding of the feed
+        # =====================================================
+        tidy.options.char_encoding = "utf8"
+        tidy.options.ascii_chars = true
         tidy.options.doctype = "omit"        
         xml = tidy.clean(html)
         xml

data/lib/feed_tools/feed.rb

@@ -750,8 +750,9 @@ module FeedTools
         unless @title.nil?
           @title = FeedTools.sanitize_html(@title, :strip)
           @title = FeedTools.unescape_entities(@title) if repair_entities
-          @title = FeedTools.tidy_html(@title)
+          @title = FeedTools.tidy_html(@title) unless repair_entities
         end
+        @title.gsub!(/>\n</, "><")
         @title.gsub!(/\n/, " ")
         @title.strip!
         @title = nil if @title == ""
@@ -792,19 +793,25 @@ module FeedTools
           end
           if description_node.nil?
             description_node = XPath.first(channel_node, "content:encoded")
-            @bozo = true unless description_node.nil?
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "content:encoded",
+              FEED_TOOLS_NAMESPACES)
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "encoded")
           end
           if description_node.nil?
             description_node = XPath.first(channel_node, "content")
-            @bozo = true unless description_node.nil?
           end
           if description_node.nil?
             description_node = XPath.first(channel_node, "xhtml:body")
-            @bozo = true unless description_node.nil?
           end
           if description_node.nil?
             description_node = XPath.first(channel_node, "body")
-            @bozo = true unless description_node.nil?
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "blurb")
           end
         end
         if description_node.nil?
@@ -814,6 +821,8 @@ module FeedTools
           if XPath.first(description_node, "@encoding").to_s != ""
             @description =
               "[Embedded data objects are not currently supported.]"
+          elsif description_node.cdatas.size > 0
+            @description = description_node.cdatas.first.value
           elsif XPath.first(description_node, "@type").to_s == "xhtml" || 
               XPath.first(description_node, "@mode").to_s == "xhtml" ||
               XPath.first(description_node, "@type").to_s == "xml" || 
@@ -842,7 +851,7 @@ module FeedTools
         unless @description.nil?
           @description = FeedTools.sanitize_html(@description, :strip)
           @description = FeedTools.unescape_entities(@description) if repair_entities
-          @description = FeedTools.tidy_html(@description)
+          @description = FeedTools.tidy_html(@description) unless repair_entities
         end
 
         @description = @description.strip unless @description.nil?
@@ -1193,13 +1202,13 @@ module FeedTools
     #
     # Returns any incorrectly placed channel-level itunes:author
     # elements.  They're actually amazingly common.  People don't read specs.
-    # There is no setter for this, since this is a "bozo" attribute.
+    # There is no setter for this, since this is an incorrectly placed
+    # attribute.
     def itunes_author
       if @itunes_author.nil?
         @itunes_author = FeedTools.unescape_entities(XPath.first(channel_node,
           "itunes:author/text()").to_s)
         @itunes_author = nil if @itunes_author == ""
-        @bozo = true unless @itunes_author.nil? 
       end
       return @itunes_author
     end
@@ -1776,14 +1785,6 @@ module FeedTools
       return vidlog
     end
   
-    # True if this feed is malformed somehow
-    def bozo?
-      if @bozo.nil?
-        @bozo = false
-      end
-      return @bozo
-    end
-  
     # True if the feed was not last retrieved from the cache.
     def live?
       return @live

data/lib/feed_tools/feed_item.rb

@@ -223,6 +223,9 @@ module FeedTools
         unless root_node.nil?
           repair_entities = false
           title_node = XPath.first(root_node, "title")
+          if title_node.nil?
+            title_node = XPath.first(root_node, "atom:title")
+          end
           if title_node.nil?
             title_node = XPath.first(root_node, "dc:title")
           end
@@ -250,7 +253,7 @@ module FeedTools
         unless @title.nil?
           @title = FeedTools.sanitize_html(@title, :strip)
           @title = FeedTools.unescape_entities(@title) if repair_entities
-          @title = FeedTools.tidy_html(@title)
+          @title = FeedTools.tidy_html(@title) unless repair_entities
         end
         if @title != ""
           # Some blogging tools include the number of comments in a post
@@ -262,6 +265,7 @@ module FeedTools
           # unstripped title, just use find_node("title/text()").to_s
           @title = @title.strip.gsub(/\[\d*\]$/, "").strip
         end
+        @title.gsub!(/>\n</, "><")
         @title.gsub!(/\n/, " ")
         @title.strip!
         @title = nil if @title == ""
@@ -280,6 +284,13 @@ module FeedTools
         unless root_node.nil?
           repair_entities = false
           description_node = XPath.first(root_node, "content:encoded")
+          if description_node.nil?
+            description_node = XPath.first(root_node, "content:encoded",
+              FEED_TOOLS_NAMESPACES)
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "encoded")
+          end
           if description_node.nil?
             description_node = XPath.first(root_node, "content")
           end
@@ -289,6 +300,10 @@ module FeedTools
           if description_node.nil?
             description_node = XPath.first(root_node, "xhtml:body")
           end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "xhtml:body",
+              FEED_TOOLS_NAMESPACES)
+          end
           if description_node.nil?
             description_node = XPath.first(root_node, "body")
           end
@@ -310,9 +325,11 @@ module FeedTools
           if description_node.nil?
             description_node = XPath.first(root_node, "ABSTRACT")
           end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "blurb")
+          end
           if description_node.nil?
             description_node = XPath.first(root_node, "info")
-            @bozo = true unless description_node.nil?
           end
         end
         if description_node.nil?
@@ -322,6 +339,8 @@ module FeedTools
           if XPath.first(description_node, "@encoding").to_s != ""
             @description =
               "[Embedded data objects are not currently supported.]"
+          elsif description_node.cdatas.size > 0
+            @description = description_node.cdatas.first.value
           elsif XPath.first(description_node, "@type").to_s == "xhtml" || 
               XPath.first(description_node, "@mode").to_s == "xhtml" ||
               XPath.first(description_node, "@type").to_s == "xml" || 
@@ -350,7 +369,7 @@ module FeedTools
         unless @description.nil?
           @description = FeedTools.sanitize_html(@description, :strip)
           @description = FeedTools.unescape_entities(@description) if repair_entities
-          @description = FeedTools.tidy_html(@description)
+          @description = FeedTools.tidy_html(@description) unless repair_entities
         end
 
         @description = @description.strip unless @description.nil?
@@ -533,9 +552,6 @@ module FeedTools
           for image_node in image_nodes
             image = FeedTools::Feed::Image.new
             image.url = XPath.first(image_node, "url/text()").to_s
-            if image.url != ""
-              self.feed.bozo = true
-            end
             if image.url == ""
               image.url = XPath.first(image_node, "@rdf:resource").to_s
             end

data/rakefile

@@ -7,7 +7,7 @@ require 'rake/gempackagetask'
 require 'rake/contrib/rubyforgepublisher'
 
 PKG_NAME      = 'feedtools'
-PKG_VERSION   = '0.2.14'
+PKG_VERSION   = '0.2.15'
 PKG_FILE_NAME = "#{PKG_NAME}-#{PKG_VERSION}"
 
 RELEASE_NAME  = "REL #{PKG_VERSION}"

data/test/amp_test.rb

@@ -10,448 +10,384 @@ class AmpTest < Test::Unit::TestCase
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp01.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
   
   def test_amp_02
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp02.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
   
   def test_amp_03
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp03.xml')
     assert_equal("&", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
   
   def test_amp_04
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp04.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_05
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp05.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_06
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp06.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_07
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp07.xml')
     assert_equal("&", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_08
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp08.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_09
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp09.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_10
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp10.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_11
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp11.xml')
     assert_equal("&", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_12
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp12.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_13
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp13.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_14
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp14.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_15
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp15.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_16
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp16.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_17
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp17.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_18
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp18.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_19
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp19.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_20
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp20.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_21
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp21.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_22
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp22.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_23
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp23.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_24
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp24.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_25
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp25.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_26
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp26.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_27
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp27.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_28
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp28.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_29
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp29.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_30
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp30.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_31
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp31.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_32
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp32.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_33
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp33.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_34
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp34.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_35
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp35.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_36
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp36.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_37
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp37.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_38
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp38.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_39
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp39.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_40
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp40.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_41
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp41.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_42
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp42.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_43
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp43.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_44
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp44.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_45
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp45.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_46
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp46.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_47
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp47.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_48
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp48.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_49
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp49.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_50
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp50.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_51
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp51.xml')
     assert_equal("&", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_52
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp52.xml')
     assert_equal("&amp;", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_53
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp53.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_54
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp54.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_55
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp55.xml')
     assert_equal("<b>&amp;</b>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_56
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp56.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_57
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp57.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_58
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp58.xml')
     assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_59
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp59.xml')
     assert_equal("<div><b>&amp;</b></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_60
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp60.xml')
     assert_equal("<div><b>&amp;</b></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_61
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp61.xml')
     assert_equal("<div><b>&amp;</b></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_62
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp62.xml')
     assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_63
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp63.xml')
     assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
 
   def test_amp_64
     feed = FeedTools::Feed.open(
       'http://feedparser.org/tests/wellformed/amp/amp64.xml')
     assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
-    assert_equal(false, feed.bozo?)
   end
   
   def test_amp_65
@@ -470,6 +406,472 @@ class AmpTest < Test::Unit::TestCase
     assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.tagline)
     assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.entries.first.title)
     assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.entries.first.content)
-    assert_equal(false, feed.bozo?)
+  end
+
+  def test_amp_tidy_01
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp01.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+  
+  def test_amp_tidy_02
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp02.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+  
+  def test_amp_tidy_03
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp03.xml')
+    assert_equal("&", feed.entries.first.title)
+  end
+  
+  def test_amp_tidy_04
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp04.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_05
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp05.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_06
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp06.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_07
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp07.xml')
+    assert_equal("&", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_08
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp08.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_09
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp09.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_10
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp10.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_11
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp11.xml')
+    assert_equal("&", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_12
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp12.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_13
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp13.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_14
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp14.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_15
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp15.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_16
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp16.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_17
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp17.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_18
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp18.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_19
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp19.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_20
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp20.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_21
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp21.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_22
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp22.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_23
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp23.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_24
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp24.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_25
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp25.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_26
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp26.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_27
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp27.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_28
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp28.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_29
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp29.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_30
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp30.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_31
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp31.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_32
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp32.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_33
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp33.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_34
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp34.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_35
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp35.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_36
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp36.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_37
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp37.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_38
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp38.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_39
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp39.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_40
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp40.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_41
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp41.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_42
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp42.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_43
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp43.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_44
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp44.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_45
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp45.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_46
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp46.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_47
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp47.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_48
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp48.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_49
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp49.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_50
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp50.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_51
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp51.xml')
+    assert_equal("&", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_52
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp52.xml')
+    assert_equal("&amp;", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_53
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp53.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_54
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp54.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_55
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp55.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_56
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp56.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_57
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp57.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_58
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp58.xml')
+    assert_equal("<strong>&amp;</strong>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_59
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp59.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_60
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp60.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_61
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp61.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_62
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp62.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_63
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp63.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+
+  def test_amp_tidy_64
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.open(
+      'http://feedparser.org/tests/wellformed/amp/amp64.xml')
+    assert_equal("<div><strong>&amp;</strong></div>", feed.entries.first.title)
+  end
+  
+  def test_amp_tidy_65
+    FeedTools.tidy_enabled = true
+    feed = FeedTools::Feed.new
+    feed.feed_data = <<-FEED
+      <feed version="0.3">
+        <title>&lt;strong>1 &amp;amp; 2 &amp; 3&lt;/strong></title>
+        <tagline>&lt;strong>1 &amp;amp; 2 &amp; 3&lt;/strong></tagline>
+    	  <entry>
+          <title>&lt;strong>1 &amp;amp; 2 &amp; 3&lt;/strong></title>
+          <content>&lt;strong>1 &amp;amp; 2 &amp; 3&lt;/strong></content>
+        </entry>
+      </feed>
+    FEED
+    assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.title)
+    assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.tagline)
+    assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.entries.first.title)
+    assert_equal("<strong>1 &amp; 2 & 3</strong>", feed.entries.first.content)
   end
 end