rss plugin: strip whitespace from link, category and author

[user/henk/code/ruby/rbot.git] / data / rbot / plugins / rss.rb

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index bc4fd369da5b4bf7d0cc6c2652854cd3f3a0c363..77d897cab6b9d2d0edb28f1c9b975269aececca9 100644 (file)
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -25,21 +25,6 @@ end
 
 module ::RSS
 
-  # Make an  'unique' ID for a given item, based on appropriate bot options
-  # Currently only suppored is bot.config['rss.show_updated']: when true, the
-  # description is included in the uid hashing, otherwise it's not
-  #
-  def RSS.item_uid_for_bot(item, opts={})
-    options = { :show_updated => true}.merge(opts)
-    desc = nil
-    if options[:show_updated]
-      desc = item.content.content rescue item.description rescue nil
-    end
-    [(item.title.content rescue item.title rescue nil),
-     (item.link.href rescue item.link),
-     desc].hash
-  end
-
   # Add support for Slashdot namespace in RDF. The code is just an adaptation
   # of the DublinCore code.
   unless defined?(SLASH_PREFIX)
@@ -269,6 +254,38 @@ class RSSFeedsPlugin < Plugin
     :default => true,
     :desc => "Whether to display links from the text of a feed item.")
 
+  # Make an  'unique' ID for a given item, based on appropriate bot options
+  # Currently only suppored is bot.config['rss.show_updated']: when false,
+  # only the guid/link is accounted for.
+  
+  def block_rescue(df = nil, &block)
+    v = block.call rescue nil
+    (String === v && '' != v) ? v : nil
+  end
+
+  def make_uid(item)
+    uid = [
+      (block_rescue do item.guid.content end ||
+       block_rescue do item.guid end ||
+       block_rescue do item.link.href end ||
+       block_rescue do item.link end
+      )
+    ]
+    if @bot.config['rss.show_updated']
+      uid.push(
+        block_rescue do item.content.content end ||
+        block_rescue do item.description end
+      )
+      uid.unshift(
+        block_rescue do item.title.content end ||
+        block_rescue do item.title end
+      )
+    end
+    # debug "taking hash of #{uid.inspect}"
+    uid.hash
+  end
+
+
   # We used to save the Mutex with the RssBlob, which was idiotic. And
   # since Mutexes dumped in one version might not be resotrable in another,
   # we need a few tricks to be able to restore data from other versions of Ruby
@@ -803,7 +820,7 @@ class RSSFeedsPlugin < Plugin
     if params and handle = params[:handle]
       feed = @feeds.fetch(handle.downcase, nil)
       if feed
-        @bot.timer.reschedule(@watch[feed.handle], 0)
+        @bot.timer.reschedule(@watch[feed.handle], (params[:delay] || 0).to_f)
         m.okay if m
       else
         m.reply _("no such feed %{handle}") % { :handle => handle } if m
@@ -859,9 +876,8 @@ class RSSFeedsPlugin < Plugin
               otxt = []
 
               # These are used for checking new items vs old ones
-              uid_opts = { :show_updated => @bot.config['rss.show_updated'] }
               oids = Set.new feed.items.map { |item|
-                uid = RSS.item_uid_for_bot(item, uid_opts)
+                uid = make_uid item
                 otxt << item.to_s
                 debug [uid, item].inspect
                 debug [uid, otxt.last].inspect
@@ -880,7 +896,7 @@ class RSSFeedsPlugin < Plugin
                 # debug feed.xml
 
                 dispItems = feed.items.reject { |item|
-                  uid = RSS.item_uid_for_bot(item, uid_opts)
+                  uid = make_uid item
                   txt = item.to_s
                   if oids.include?(uid)
                     debug "rejecting old #{uid} #{item.inspect}"
@@ -937,13 +953,13 @@ class RSSFeedsPlugin < Plugin
   end
 
   def select_nonempty(*ar)
-    debug ar
-    ret = ar.map { |i| (i && i.empty?) ? nil : i }.compact.first
-    (ret && ret.empty?) ? nil : ret
+    # debug ar
+    ar.each { |i| return i unless i.nil_or_empty? }
+    return nil
   end
 
   def printFormattedRss(feed, item, opts=nil)
-    debug item
+    # debug item
     places = feed.watchers
     handle = feed.handle.empty? ? "" : "::#{feed.handle}:: "
     date = String.new
@@ -1019,10 +1035,13 @@ class RSSFeedsPlugin < Plugin
       desc = "(?)"
     end
 
-    link = item.link.href rescue item.link.chomp rescue nil
+    link = item.link.href rescue item.link rescue nil
+    link.strip! if link
 
     category = select_nonempty((item.category.content rescue nil), (item.dc_subject rescue nil))
+    category.strip! if category
     author = select_nonempty((item.author.name.content rescue nil), (item.dc_creator rescue nil), (item.author rescue nil))
+    author.strip! if author
 
     line1 = nil
     line2 = nil
@@ -1190,7 +1209,7 @@ plugin.map 'rss unwatch :handle [in :chan]',
   :action => 'unwatch_rss'
 plugin.map 'rss rmwatch :handle [in :chan]',
   :action => 'unwatch_rss'
-plugin.map 'rss rewatch [:handle]',
+plugin.map 'rss rewatch [:handle] [:delay]',
   :action => 'rewatch_rss'
 plugin.map 'rss types',
   :action => 'rss_types'