feedtools 0.2.5 → 0.2.6

data/CHANGELOG

@@ -1,3 +1,6 @@
+== FeedTools 0.2.6
+ * Added support for merging feeds
+ * Vastly improved handling of feed creation
 == FeedTools 0.2.5
  * fixed multiple rows being created in the cache after a 301 redirection
  * fixed broken table creation for postgresql and sqlite

data/lib/feed_tools.rb

@@ -25,7 +25,7 @@ FEED_TOOLS_ENV = ENV['FEED_TOOLS_ENV'] ||
                  ENV['RAILS_ENV'] ||
                  'production' # :nodoc:
 
-FEED_TOOLS_VERSION = "0.2.5"
+FEED_TOOLS_VERSION = "0.2.6"
 
 $:.unshift(File.dirname(__FILE__))
 $:.unshift(File.dirname(__FILE__) + "/../../activerecord/lib")
@@ -101,8 +101,9 @@ module FeedTools
         begin
           possible_config_files = [
             "./config/database.yml",
-            "../database.yml",
-            "./database.yml"
+            "../config/database.yml",
+            "./database.yml",
+            "../database.yml"
           ]
           database_config_file = nil
           for file in possible_config_files
@@ -560,6 +561,7 @@ module FeedTools
   
   # Escapes all html entities
   def FeedTools.escape_entities(html)
+    return nil if html.nil?
     escaped_html = CGI.escapeHTML(html)
     unescaped_html.gsub!(/'/, "'")
     unescaped_html.gsub!(/"/, """)
@@ -568,6 +570,7 @@ module FeedTools
   
   # Unescapes all html entities
   def FeedTools.unescape_entities(html)
+    return nil if html.nil?
     unescaped_html = html
     unescaped_html.gsub!(/&/, "&")
     unescaped_html.gsub!(/&/, "&")
@@ -579,6 +582,7 @@ module FeedTools
   
   # Removes all html tags from the html formatted text.
   def FeedTools.strip_html(html)
+    return nil if html.nil?
     # TODO: do this properly
     # ======================
     stripped_html = html.gsub(/<\/?[^>]+>/, "")
@@ -587,6 +591,7 @@ module FeedTools
 
   # Tidys up the html
   def FeedTools.tidy_html(html)
+    return nil if html.nil?
     if FeedTools.tidy_enabled?
       is_fragment = true
       html.gsub!(/&lt;!'/, "&amp;lt;!'")
@@ -628,6 +633,7 @@ module FeedTools
   # elements and all children will be removed entirely.
   # Dangerous or unknown attributes are always removed.
   def FeedTools.sanitize_html(html, mode=:strip)
+    return nil if html.nil?
     
     # Lists borrowed from Mark Pilgrim's feedparser
     acceptable_elements = ['a', 'abbr', 'acronym', 'address', 'area', 'b',
@@ -688,6 +694,21 @@ module FeedTools
     html = html_doc.root.inner_xml
     return html
   end
+  
+  # Creates a merged "planet" feed from a set of urls.
+  def FeedTools.build_merged_feed(url_array)
+    return nil if url_array.nil?
+    merged_feed = Feed.new
+    url_array.each do |feed_url|
+      feed = Feed.open(feed_url)
+      merged_feed.entries.concat(
+        feed.entries.collect do |entry|
+          entry.title = "#{feed.title}: #{entry.title}"
+          entry
+        end )
+    end
+    return merged_feed
+  end
 
   class Feed
     include REXML # :nodoc:
@@ -1089,7 +1110,7 @@ module FeedTools
     
     # Returns the channel node of the feed.
     def channel_node
-      if @channel_node.nil?
+      if @channel_node.nil? && root_node != nil
         @channel_node = XPath.first(root_node, "channel")
         if @channel_node == nil
           @channel_node = XPath.first(root_node, "CHANNEL")
@@ -1225,9 +1246,19 @@ module FeedTools
     # Returns the feed's unique id
     def id
       if @id.nil?
-        @id = XPath.first(root_node, "id/text()").to_s
-        if @id == ""
-          @id = XPath.first(root_node, "guid/text()").to_s
+        unless channel_node.nil?
+          @id = XPath.first(channel_node, "id/text()").to_s
+          if @id == ""
+            @id = XPath.first(channel_node, "guid/text()").to_s
+          end
+        end
+        unless root_node.nil?
+          if @id == "" || @id.nil?
+            @id = XPath.first(root_node, "id/text()").to_s
+          end
+          if @id == ""
+            @id = XPath.first(root_node, "guid/text()").to_s
+          end
         end
         @id = nil if @id == ""
       end
@@ -1257,13 +1288,15 @@ module FeedTools
     # Returns the feed title
     def title
       if @title.nil?
-        repair_entities = false
-        title_node = XPath.first(channel_node, "title")
-        if title_node.nil?
-          title_node = XPath.first(channel_node, "dc:title")
-        end
-        if title_node.nil?
-          title_node = XPath.first(channel_node, "TITLE")
+        unless channel_node.nil?
+          repair_entities = false
+          title_node = XPath.first(channel_node, "title")
+          if title_node.nil?
+            title_node = XPath.first(channel_node, "dc:title")
+          end
+          if title_node.nil?
+            title_node = XPath.first(channel_node, "TITLE")
+          end
         end
         if title_node.nil?
           return nil
@@ -1304,41 +1337,43 @@ module FeedTools
     # Returns the feed description
     def description
       if @description.nil?
-        repair_entities = false
-        description_node = XPath.first(channel_node, "description")
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "tagline")
-        end
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "subtitle")
-        end
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "summary")
-        end
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "abstract")
-        end
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "ABSTRACT")
-        end
-        if description_node.nil?
-          description_node = XPath.first(channel_node, "info")
-        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")
-          @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?
+        unless channel_node.nil?
+          repair_entities = false
+          description_node = XPath.first(channel_node, "description")
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "tagline")
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "subtitle")
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "summary")
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "abstract")
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "ABSTRACT")
+          end
+          if description_node.nil?
+            description_node = XPath.first(channel_node, "info")
+          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")
+            @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
         end
         if description_node.nil?
           return nil
@@ -1392,8 +1427,16 @@ module FeedTools
     # Returns the contents of the itunes:summary element
     def itunes_summary
       if @itunes_summary.nil?
-        @itunes_summary = FeedTools.unescape_entities(XPath.first(root_node,
-          "itunes:summary/text()").to_s)
+        unless channel_node.nil?
+          @itunes_summary = FeedTools.unescape_entities(XPath.first(channel_node,
+            "itunes:summary/text()").to_s)
+        end
+        unless root_node.nil?
+          if @itunes_summary == "" || @itunes_summary.nil?
+            @itunes_summary = FeedTools.unescape_entities(XPath.first(root_node,
+              "itunes:summary/text()").to_s)
+          end
+        end
         if @itunes_summary == ""
           @itunes_summary = nil
         end
@@ -1411,8 +1454,16 @@ module FeedTools
     # Returns the contents of the itunes:subtitle element
     def itunes_subtitle
       if @itunes_subtitle.nil?
-        @itunes_subtitle = FeedTools.unescape_entities(XPath.first(root_node,
-          "itunes:subtitle/text()").to_s)
+        unless channel_node.nil?
+          @itunes_subtitle = FeedTools.unescape_entities(XPath.first(channel_node,
+            "itunes:subtitle/text()").to_s)
+        end
+        unless root_node.nil?
+          if @itunes_subtitle == "" || @itunes_subtitle.nil?
+            @itunes_subtitle = FeedTools.unescape_entities(XPath.first(root_node,
+              "itunes:subtitle/text()").to_s)
+          end
+        end
         if @itunes_subtitle == ""
           @itunes_subtitle = nil
         end
@@ -1431,35 +1482,37 @@ module FeedTools
     # Returns the feed link
     def link
       if @link.nil?
-        # get the feed link from the xml document
-        @link = XPath.first(channel_node, "link[@rel='alternate' @type='text/html']/@href").to_s
-        if @link == ""
-          @link = XPath.first(channel_node, "link[@rel='alternate']/@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "link/@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "link/text()").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "@HREF").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "a/@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(channel_node, "A/@HREF").to_s
+        unless channel_node.nil?
+          # get the feed link from the xml document
+          @link = XPath.first(channel_node, "link[@rel='alternate' @type='text/html']/@href").to_s
+          if @link == ""
+            @link = XPath.first(channel_node, "link[@rel='alternate']/@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "link/@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "link/text()").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "@HREF").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "a/@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(channel_node, "A/@HREF").to_s
+          end
         end
-        if @link == ""
+        if @link == "" || @link.nil?
           if FeedTools.is_url? self.guid
             @link = self.guid
           end
         end
-        if @link == ""
+        if @link == "" && channel_node != nil
           # Technically, we shouldn't use the base attribute for this, but if the href attribute
           # is missing, it's already a given that we're looking at a messed up CDF file.  We can
           # always pray it's correct.
@@ -1755,53 +1808,55 @@ module FeedTools
     def images
       if @images.nil?
         @images = []
-        image_nodes = XPath.match(channel_node, "image")
-        if image_nodes.nil? || image_nodes.empty?
-          image_nodes = XPath.match(channel_node, "link")
-        end
-        if image_nodes.nil? || image_nodes.empty?
-          image_nodes = XPath.match(channel_node, "logo")
-        end
-        if image_nodes.nil? || image_nodes.empty?
-          image_nodes = XPath.match(channel_node, "LOGO")
-        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 == ""
-              image.url = XPath.first(image_node, "@rdf:resource").to_s
-            end
-            if image.url == "" && (image_node.name == "logo" ||
-                (image_node.attributes['type'] =~ /^image/) == 0)
-              image.url = XPath.first(image_node, "@href").to_s
-            end
-            if image.url == "" && image_node.name == "LOGO"
-              image.url = XPath.first(image_node, "@HREF").to_s
-            end
-            image.url.strip! unless image.url.nil?
-            image.url = nil if image.url == ""
-            image.title = XPath.first(image_node, "title/text()").to_s
-            image.title.strip! unless image.title.nil?
-            image.title = nil if image.title == ""
-            image.description =
-              XPath.first(image_node, "description/text()").to_s
-            image.description.strip! unless image.description.nil?
-            image.description = nil if image.description == ""
-            image.link = XPath.first(image_node, "link/text()").to_s
-            image.link.strip! unless image.link.nil?
-            image.link = nil if image.link == ""
-            image.height = XPath.first(image_node, "height/text()").to_s.to_i
-            image.height = nil if image.height <= 0
-            image.width = XPath.first(image_node, "width/text()").to_s.to_i
-            image.width = nil if image.width <= 0
-            image.style = XPath.first(image_node, "@style").to_s.downcase
-            if image.style == ""
-              image.style = XPath.first(image_node, "@STYLE").to_s.downcase
+        unless channel_node.nil?
+          image_nodes = XPath.match(channel_node, "image")
+          if image_nodes.nil? || image_nodes.empty?
+            image_nodes = XPath.match(channel_node, "link")
+          end
+          if image_nodes.nil? || image_nodes.empty?
+            image_nodes = XPath.match(channel_node, "logo")
+          end
+          if image_nodes.nil? || image_nodes.empty?
+            image_nodes = XPath.match(channel_node, "LOGO")
+          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 == ""
+                image.url = XPath.first(image_node, "@rdf:resource").to_s
+              end
+              if image.url == "" && (image_node.name == "logo" ||
+                  (image_node.attributes['type'] =~ /^image/) == 0)
+                image.url = XPath.first(image_node, "@href").to_s
+              end
+              if image.url == "" && image_node.name == "LOGO"
+                image.url = XPath.first(image_node, "@HREF").to_s
+              end
+              image.url.strip! unless image.url.nil?
+              image.url = nil if image.url == ""
+              image.title = XPath.first(image_node, "title/text()").to_s
+              image.title.strip! unless image.title.nil?
+              image.title = nil if image.title == ""
+              image.description =
+                XPath.first(image_node, "description/text()").to_s
+              image.description.strip! unless image.description.nil?
+              image.description = nil if image.description == ""
+              image.link = XPath.first(image_node, "link/text()").to_s
+              image.link.strip! unless image.link.nil?
+              image.link = nil if image.link == ""
+              image.height = XPath.first(image_node, "height/text()").to_s.to_i
+              image.height = nil if image.height <= 0
+              image.width = XPath.first(image_node, "width/text()").to_s.to_i
+              image.width = nil if image.width <= 0
+              image.style = XPath.first(image_node, "@style").to_s.downcase
+              if image.style == ""
+                image.style = XPath.first(image_node, "@STYLE").to_s.downcase
+              end
+              image.style.strip! unless image.style.nil?
+              image.style = nil if image.style == ""
+              @images << image
             end
-            image.style.strip! unless image.style.nil?
-            image.style = nil if image.style == ""
-            @images << image
           end
         end
       end
@@ -1855,86 +1910,88 @@ module FeedTools
     # Returns the number of seconds before the feed should expire
     def time_to_live
       if @time_to_live.nil?
-        # get the feed time to live from the xml document
-        update_frequency = XPath.first(channel_node, "syn:updateFrequency/text()").to_s
-        if update_frequency != ""
-          update_period = XPath.first(channel_node, "syn:updatePeriod/text()").to_s
-          if update_period == "daily"
-            @time_to_live = update_frequency.to_i.day
-          elsif update_period == "weekly"
-            @time_to_live = update_frequency.to_i.week
-          elsif update_period == "monthly"
-            @time_to_live = update_frequency.to_i.month
-          elsif update_period == "yearly"
-            @time_to_live = update_frequency.to_i.year
-          else
-            # hourly
-            @time_to_live = update_frequency.to_i.hour
+        unless channel_node.nil?
+          # get the feed time to live from the xml document
+          update_frequency = XPath.first(channel_node, "syn:updateFrequency/text()").to_s
+          if update_frequency != ""
+            update_period = XPath.first(channel_node, "syn:updatePeriod/text()").to_s
+            if update_period == "daily"
+              @time_to_live = update_frequency.to_i.day
+            elsif update_period == "weekly"
+              @time_to_live = update_frequency.to_i.week
+            elsif update_period == "monthly"
+              @time_to_live = update_frequency.to_i.month
+            elsif update_period == "yearly"
+              @time_to_live = update_frequency.to_i.year
+            else
+              # hourly
+              @time_to_live = update_frequency.to_i.hour
+            end
+          end
+          if @time_to_live.nil?
+            # usually expressed in minutes
+            update_frequency = XPath.first(channel_node, "ttl/text()").to_s
+            if update_frequency != ""
+              update_span = XPath.first(channel_node, "ttl/@span").to_s
+              if update_span == "seconds"
+                @time_to_live = update_frequency.to_i
+              elsif update_span == "minutes"
+                @time_to_live = update_frequency.to_i.minute
+              elsif update_span == "hours"
+                @time_to_live = update_frequency.to_i.hour
+              elsif update_span == "days"
+                @time_to_live = update_frequency.to_i.day
+              elsif update_span == "weeks"
+                @time_to_live = update_frequency.to_i.week
+              elsif update_span == "months"
+                @time_to_live = update_frequency.to_i.month
+              elsif update_span == "years"
+                @time_to_live = update_frequency.to_i.year
+              elsif update_frequency.to_i >= 3000
+                # Normally, this should default to minutes, but realistically,
+                # if they meant minutes, you're rarely going to see a value higher
+                # than 120.  If we see >= 3000, we're either dealing with a stupid
+                # pseudo-spec that decided to use seconds, or we're looking at
+                # someone who only has weekly updated content.  Worst case, we
+                # misreport the time, and we update too often.  Best case, we
+                # avoid accidentally updating the feed only once a year.  In the
+                # interests of being pragmatic, and since the problem we avoid
+                # is a far greater one than the one we cause, just run the check
+                # and hope no one actually gets hurt.
+                @time_to_live = update_frequency.to_i
+              else
+                @time_to_live = update_frequency.to_i.minute
+              end
+            end
           end
-        end
-      end
-      if @time_to_live.nil?
-        # usually expressed in minutes
-        update_frequency = XPath.first(channel_node, "ttl/text()").to_s
-        if update_frequency != ""
-          update_span = XPath.first(channel_node, "ttl/@span").to_s
-          if update_span == "seconds"
-            @time_to_live = update_frequency.to_i
-          elsif update_span == "minutes"
-            @time_to_live = update_frequency.to_i.minute
-          elsif update_span == "hours"
-            @time_to_live = update_frequency.to_i.hour
-          elsif update_span == "days"
-            @time_to_live = update_frequency.to_i.day
-          elsif update_span == "weeks"
-            @time_to_live = update_frequency.to_i.week
-          elsif update_span == "months"
-            @time_to_live = update_frequency.to_i.month
-          elsif update_span == "years"
-            @time_to_live = update_frequency.to_i.year
-          elsif update_frequency.to_i >= 3000
-            # Normally, this should default to minutes, but realistically,
-            # if they meant minutes, you're rarely going to see a value higher
-            # than 120.  If we see >= 3000, we're either dealing with a stupid
-            # pseudo-spec that decided to use seconds, or we're looking at
-            # someone who only has weekly updated content.  Worst case, we
-            # misreport the time, and we update too often.  Best case, we
-            # avoid accidentally updating the feed only once a year.  In the
-            # interests of being pragmatic, and since the problem we avoid
-            # is a far greater one than the one we cause, just run the check
-            # and hope no one actually gets hurt.
-            @time_to_live = update_frequency.to_i
-          else
-            @time_to_live = update_frequency.to_i.minute
+          if @time_to_live.nil?
+            @time_to_live = 0
+            update_frequency_days =
+              XPath.first(channel_node, "schedule/intervaltime/@days").to_s
+            update_frequency_hours =
+              XPath.first(channel_node, "schedule/intervaltime/@hour").to_s
+            update_frequency_minutes =
+              XPath.first(channel_node, "schedule/intervaltime/@min").to_s
+            update_frequency_seconds =
+              XPath.first(channel_node, "schedule/intervaltime/@sec").to_s
+            if update_frequency_days != ""
+              @time_to_live = @time_to_live + update_frequency_days.to_i.day
+            end
+            if update_frequency_hours != ""
+              @time_to_live = @time_to_live + update_frequency_hours.to_i.hour
+            end
+            if update_frequency_minutes != ""
+              @time_to_live = @time_to_live + update_frequency_minutes.to_i.minute
+            end
+            if update_frequency_seconds != ""
+              @time_to_live = @time_to_live + update_frequency_seconds.to_i
+            end
+            if @time_to_live == 0
+              @time_to_live = 1.hour
+            end
           end
         end
       end
-      if @time_to_live.nil?
-        @time_to_live = 0
-        update_frequency_days =
-          XPath.first(channel_node, "schedule/intervaltime/@days").to_s
-        update_frequency_hours =
-          XPath.first(channel_node, "schedule/intervaltime/@hour").to_s
-        update_frequency_minutes =
-          XPath.first(channel_node, "schedule/intervaltime/@min").to_s
-        update_frequency_seconds =
-          XPath.first(channel_node, "schedule/intervaltime/@sec").to_s
-        if update_frequency_days != ""
-          @time_to_live = @time_to_live + update_frequency_days.to_i.day
-        end
-        if update_frequency_hours != ""
-          @time_to_live = @time_to_live + update_frequency_hours.to_i.hour
-        end
-        if update_frequency_minutes != ""
-          @time_to_live = @time_to_live + update_frequency_minutes.to_i.minute
-        end
-        if update_frequency_seconds != ""
-          @time_to_live = @time_to_live + update_frequency_seconds.to_i
-        end
-        if @time_to_live == 0
-          @time_to_live = 1.hour
-        end
-      end
       if @time_to_live.nil? || @time_to_live == 0
         # Default to one hour
         @time_to_live = 1.hour
@@ -2009,17 +2066,19 @@ module FeedTools
     # Returns the feed language
     def language
       if @language.nil?
-        @language = XPath.first(channel_node, "language/text()").to_s
-        if @language == ""
-          @language = XPath.first(channel_node, "dc:language/text()").to_s
-        end
-        if @language == ""
-          @language = XPath.first(channel_node, "xml:lang/text()").to_s
-        end
-        if @language == ""
-          @language = XPath.first(root_node, "xml:lang/text()").to_s
+        unless channel_node.nil?
+          @language = XPath.first(channel_node, "language/text()").to_s
+          if @language == ""
+            @language = XPath.first(channel_node, "dc:language/text()").to_s
+          end
+          if @language == ""
+            @language = XPath.first(channel_node, "xml:lang/text()").to_s
+          end
+          if @language == ""
+            @language = XPath.first(root_node, "xml:lang/text()").to_s
+          end
         end
-        if @language == ""
+        if @language == "" || @language.nil?
           @language = "en-us"
         end
         @language = @language.downcase
@@ -2058,23 +2117,25 @@ module FeedTools
     # Returns the feed items
     def items
       if @items.nil?
-        raw_items = XPath.match(root_node, "item")
-        if raw_items == nil || raw_items == []
-          raw_items = XPath.match(channel_node, "item")
-        end
-        if raw_items == nil || raw_items == []
-          raw_items = XPath.match(channel_node, "ITEM")
-        end
-        if raw_items == nil || raw_items == []
-          raw_items = XPath.match(root_node, "ITEM")
-        end
-        if raw_items == nil || raw_items == []
-          raw_items = XPath.match(channel_node, "entry")
-        end
-        if raw_items == nil || raw_items == []
-          raw_items = XPath.match(root_node, "entry")
+        unless root_node.nil?
+          raw_items = XPath.match(root_node, "item")
+          if raw_items == nil || raw_items == []
+            raw_items = XPath.match(channel_node, "item")
+          end
+          if raw_items == nil || raw_items == []
+            raw_items = XPath.match(channel_node, "ITEM")
+          end
+          if raw_items == nil || raw_items == []
+            raw_items = XPath.match(root_node, "ITEM")
+          end
+          if raw_items == nil || raw_items == []
+            raw_items = XPath.match(channel_node, "entry")
+          end
+          if raw_items == nil || raw_items == []
+            raw_items = XPath.match(root_node, "entry")
+          end
         end
-        
+
         # create the individual feed items
         @items = []
         if raw_items != nil
@@ -2174,7 +2235,11 @@ module FeedTools
       if feed_type == "rss" && (version == 0.9 || version == 1.0 || version == 1.1)
         # RDF-based rss format
         return xml_builder.tag!("rdf:RDF") do
-          xml_builder.channel("rdf:about" => CGI.escapeHTML(link)) do
+          channel_attributes = {}
+          unless self.link.nil?
+            channel_attributes["rdf:about"] = CGI.escapeHTML(self.link)
+          end
+          xml_builder.channel(channel_attributes) do
             unless title.nil? || title == ""
               xml_builder.title(title)
             else
@@ -2534,9 +2599,11 @@ module FeedTools
     # Returns the feed items's unique id
     def id
       if @id.nil?
-        @id = XPath.first(root_node, "id/text()").to_s
-        if @id == ""
-          @id = XPath.first(root_node, "guid/text()").to_s
+        unless root_node.nil?
+          @id = XPath.first(root_node, "id/text()").to_s
+          if @id == ""
+            @id = XPath.first(root_node, "guid/text()").to_s
+          end
         end
         @id = nil if @id == ""
       end
@@ -2551,13 +2618,15 @@ module FeedTools
     # Returns the feed item title
     def title
       if @title.nil?
-        repair_entities = false
-        title_node = XPath.first(root_node, "title")
-        if title_node.nil?
-          title_node = XPath.first(root_node, "dc:title")
-        end
-        if title_node.nil?
-          title_node = XPath.first(root_node, "TITLE")
+        unless root_node.nil?
+          repair_entities = false
+          title_node = XPath.first(root_node, "title")
+          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
         end
         if title_node.nil?
           return nil
@@ -2606,41 +2675,43 @@ module FeedTools
     # Returns the feed item description
     def description
       if @description.nil?
-        repair_entities = false
-        description_node = XPath.first(root_node, "description")
-        if description_node.nil?
-          description_node = XPath.first(root_node, "xhtml:body")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "body")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "tagline")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "subtitle")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "summary")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "abstract")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "ABSTRACT")
-        end
-        if description_node.nil?
-          description_node = XPath.first(root_node, "content: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, "info")
-          @bozo = true unless description_node.nil?
+        unless root_node.nil?
+          repair_entities = false
+          description_node = XPath.first(root_node, "description")
+          if description_node.nil?
+            description_node = XPath.first(root_node, "xhtml:body")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "body")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "tagline")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "subtitle")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "summary")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "abstract")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "ABSTRACT")
+          end
+          if description_node.nil?
+            description_node = XPath.first(root_node, "content: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, "info")
+            @bozo = true unless description_node.nil?
+          end
         end
         if description_node.nil?
           return nil
@@ -2754,32 +2825,34 @@ module FeedTools
     # Returns the feed item link
     def link
       if @link.nil?
-        @link = XPath.first(root_node, "link[@rel='alternate']/@href").to_s
-        if @link == ""
-          @link = XPath.first(root_node, "link/@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "link/text()").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "@rdf:about").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "guid[@isPermaLink='true']/text()").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "a/@href").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "@HREF").to_s
-        end
-        if @link == ""
-          @link = XPath.first(root_node, "A/@HREF").to_s
+        unless root_node.nil?
+          @link = XPath.first(root_node, "link[@rel='alternate']/@href").to_s
+          if @link == ""
+            @link = XPath.first(root_node, "link/@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "link/text()").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "@rdf:about").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "guid[@isPermaLink='true']/text()").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "a/@href").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "@HREF").to_s
+          end
+          if @link == ""
+            @link = XPath.first(root_node, "A/@HREF").to_s
+          end
         end
-        if @link == ""
+        if @link == "" || @link.nil?
           if FeedTools.is_url? self.guid
             @link = self.guid
           end

data/rakefile

@@ -7,7 +7,7 @@ require 'rake/gempackagetask'
 require 'rake/contrib/rubyforgepublisher'
 
 PKG_NAME      = 'feedtools'
-PKG_VERSION   = '0.2.5'
+PKG_VERSION   = '0.2.6'
 PKG_FILE_NAME = "#{PKG_NAME}-#{PKG_VERSION}"
 
 RELEASE_NAME  = "REL #{PKG_VERSION}"

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.11
 specification_version: 1
 name: feedtools
 version: !ruby/object:Gem::Version 
-  version: 0.2.5
-date: 2005-08-19 00:00:00 -04:00
+  version: 0.2.6
+date: 2005-09-01 00:00:00 -04:00
 summary: "Parsing, generation, and caching system for xml news feeds."
 require_paths: 
   - lib