feed_item.rb in feedtools-0.2.15

- old
+ new

@@ -222,10 +222,13 @@
       if @title.nil?
         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
           if title_node.nil?
             title_node = XPath.first(root_node, "TITLE")
           end
@@ -248,11 +251,11 @@
           repair_entities = true
         end
         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
           # in the title... this is supremely ugly, and breaks any
           # applications which expect the title to be static, so we're
@@ -260,10 +263,11 @@
           #
           # If for some incredibly wierd reason you need the actual
           # 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 == ""
       end
       return @title
@@ -279,19 +283,30 @@
       if @description.nil?
         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
           if description_node.nil?
             description_node = XPath.first(root_node, "fullitem")
           end
           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
           if description_node.nil?
             description_node = XPath.first(root_node, "description")
           end          
@@ -309,21 +324,25 @@
           end
           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?
           return nil
         end
         unless description_node.nil?
           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" || 
               XPath.first(description_node, "@mode").to_s == "xml" ||
               XPath.first(description_node, "@type").to_s ==
@@ -348,11 +367,11 @@
         end
 
         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?
         @description = nil if @description == ""
       end
@@ -531,12 +550,9 @@
         end
         unless image_nodes.nil?
           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
             if image.url == "" && (image_node.name == "logo" ||
                 (image_node.attributes['type'] =~ /^image/) == 0)
\ No newline at end of file