* rss.rb: use guids to check for seen items

[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 97640838c3dd56d4bbbcad6f3c116d917bf0ae00..cbdb2a82ec5cf12bab04ba8d6597656c7d94fee4 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 +254,26 @@ 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 make_uid(item)
+    uid = [
+      (item.guid.content rescue \
+       item.guid rescue \
+       item.link.href rescue \
+       item.link rescue ''
+      )
+    ]
+    if @bot.config['rss.show_updated']
+      uid.push((item.content.content rescue item.description rescue nil))
+      uid.unshift((item.title.content rescue item.title rescue nil))
+    end
+    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
@@ -370,11 +375,12 @@ class RSSFeedsPlugin < Plugin
   def htmlinfo_filter(s)
     return nil unless s[:headers] and s[:headers]['x-rbot-location']
     return nil unless s[:headers]['content-type'].first.match(/xml|rss|atom|rdf/i) or
-      s[:text].include?("<rdf:RDF") or s[:text].include?("<rss") or s[:text].include?("<feed") or
+      (s[:text].include?("<rdf:RDF") and s[:text].include?("<channel")) or
+      s[:text].include?("<rss") or s[:text].include?("<feed") or
       s[:text].match(FEED_NS)
     blob = RssBlob.new(s[:headers]['x-rbot-location'],"", :htmlinfo)
-    unless fetchRss(blob, nil) and parseRss(blob, nil)
-      debug "tried to filter #{s.inspect} which is not an RSS feed"
+    unless (fetchRss(blob, nil) and parseRss(blob, nil) rescue nil)
+      debug "#{s.pretty_inspect} is not an RSS feed, despite the appearances"
       return nil
     end
     output = []
@@ -802,7 +808,7 @@ 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
@@ -858,9 +864,8 @@ 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
@@ -879,7 +884,7 @@ 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}"
@@ -937,8 +942,8 @@ class RSSFeedsPlugin < Plugin
 
   def select_nonempty(*ar)
     debug ar
-    ret = ar.map { |i| (i && i.empty?) ? nil : i }.compact.first
-    (ret && ret.empty?) ? nil : ret
+    ar.each { |i| return i unless i.nil_or_empty? }
+    return nil
   end
 
   def printFormattedRss(feed, item, opts=nil)
@@ -1189,7 +1194,7 @@ 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'