X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;ds=sidebyside;f=data%2Frbot%2Fplugins%2Frss.rb;h=905daad81ffcfa49c36fc88ea484441c8a5bdbda;hb=ad4317c01095675546e5d907a917168d82a0580e;hp=a067cbba1aad5a258a2281637b622eb111d682b7;hpb=b14aff9aee87df83c2aab47c92b58cc7bf74432e;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index a067cbba..905daad8 100644
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -16,10 +16,20 @@
 
 require 'rss'
 
-# Add support for Slashdot namespace in RDF. The code is just an adaptation of
-# the DublinCore code.
 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 = options[:show_updated] ? item.description : nil
+    [item.title, 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)
     SLASH_PREFIX = 'slash'
     SLASH_URI = "http://purl.org/rss/1.0/modules/slash/"
@@ -220,17 +230,21 @@ end
 
 class RSSFeedsPlugin < Plugin
   BotConfig.register BotConfigIntegerValue.new('rss.head_max',
-    :default => 30, :validate => Proc.new{|v| v > 0 && v < 200},
+    :default => 100, :validate => Proc.new{|v| v > 0 && v < 200},
     :desc => "How many characters to use of a RSS item header")
 
   BotConfig.register BotConfigIntegerValue.new('rss.text_max',
-    :default => 90, :validate => Proc.new{|v| v > 0 && v < 400},
+    :default => 200, :validate => Proc.new{|v| v > 0 && v < 400},
     :desc => "How many characters to use of a RSS item text")
 
   BotConfig.register BotConfigIntegerValue.new('rss.thread_sleep',
     :default => 300, :validate => Proc.new{|v| v > 30},
     :desc => "How many seconds to sleep before checking RSS feeds again")
 
+  BotConfig.register BotConfigBooleanValue.new('rss.show_updated',
+    :default => true,
+    :desc => "Whether feed items for which the description was changed should be shown as new")
+
   # 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
@@ -246,6 +260,8 @@ class RSSFeedsPlugin < Plugin
     end
   end
 
+  attr_reader :feeds
+
   def initialize
     super
     if @registry.has_key?(:feeds)
@@ -273,6 +289,7 @@ class RSSFeedsPlugin < Plugin
       }
 
       @feeds = @registry[:feeds]
+      raise unless @feeds
 
       @registry.recovery = nil
 
@@ -302,6 +319,7 @@ class RSSFeedsPlugin < Plugin
 
   def cleanup
     stop_watches
+    super
   end
 
   def save
@@ -320,7 +338,7 @@ class RSSFeedsPlugin < Plugin
         debug "Stopping watch #{handle}"
         @bot.timer.remove(@watch[handle])
         @watch.delete(handle)
-      rescue => e
+      rescue Exception => e
         report_problem("Failed to stop watch for #{handle}", e, nil)
       end
     end
@@ -397,7 +415,25 @@ class RSSFeedsPlugin < Plugin
 
     m.reply "lemme fetch it..."
     title = items = nil
-    fetched = fetchRss(feed, m, false)
+    we_were_watching = false
+
+    if @watch.key?(feed.handle)
+      # If a feed is being watched, we run the watcher thread
+      # so that all watchers can be informed of changes to
+      # the feed. Before we do that, though, we remove the
+      # show requester from the watchlist, if present, lest
+      # he gets the update twice.
+      if feed.watched_by?(m.replyto)
+        we_were_watching = true
+        feed.rm_watch(m.replyto)
+      end
+      @bot.timer.reschedule(@watch[feed.handle], 0)
+      if we_were_watching
+        feed.add_watch(m.replyto)
+      end
+    else
+      fetched = fetchRss(feed, m, false)
+    end
     return unless fetched or feed.xml
     if not fetched and feed.items
       m.reply "using old data"
@@ -447,7 +483,7 @@ class RSSFeedsPlugin < Plugin
       reply = "no feeds found"
       reply << " matching #{wanted}" if wanted
     end
-    m.reply reply
+    m.reply reply, :max_lines => reply.length
   end
 
   def watched_rss(m, params)
@@ -558,6 +594,7 @@ class RSSFeedsPlugin < Plugin
 
   def del_rss(m, params, pass=false)
     feed = unwatch_rss(m, params, true)
+    return unless feed
     if feed.watched?
       m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
       return
@@ -641,16 +678,21 @@ class RSSFeedsPlugin < Plugin
     end
     status = Hash.new
     status[:failures] = 0
+    status[:first_run] = true
     @watch[feed.handle] = @bot.timer.add(0, status) {
       debug "watcher for #{feed} started"
       failures = status[:failures]
+      first_run = status.delete(:first_run)
       begin
         debug "fetching #{feed}"
         oldxml = feed.xml ? feed.xml.dup : nil
         unless fetchRss(feed)
           failures += 1
         else
-          if oldxml and oldxml == feed.xml
+          if first_run
+            debug "first run for #{feed}, getting items"
+            parseRss(feed)
+          elsif oldxml and oldxml == feed.xml
             debug "xml for #{feed} didn't change"
             failures -= 1 if failures > 0
           else
@@ -659,16 +701,45 @@ class RSSFeedsPlugin < Plugin
               parseRss(feed)
               failures -= 1 if failures > 0
             else
-              otxt = feed.items.map { |item| item.to_s }
+              # This one is used for debugging
+              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)
+                otxt << item.to_s
+                debug [uid, item].inspect
+                debug [uid, otxt.last].inspect
+                uid
+              }
+
               unless parseRss(feed)
                 debug "no items in feed #{feed}"
                 failures += 1
               else
                 debug "Checking if new items are available for #{feed}"
                 failures -= 1 if failures > 0
+                # debug "Old:"
+                # debug oldxml
+                # debug "New:"
+                # debug feed.xml
+
                 dispItems = feed.items.reject { |item|
-                  otxt.include?(item.to_s)
+                  uid = RSS.item_uid_for_bot(item, uid_opts)
+                  txt = item.to_s
+                  if oids.include?(uid)
+                    debug "rejecting old #{uid} #{item.inspect}"
+                    debug [uid, txt].inspect
+                    true
+                  else
+                    debug "accepting new #{uid} #{item.inspect}"
+                    debug [uid, txt].inspect
+                    warning "same text! #{txt}" if otxt.include?(txt)
+                    false
+                  end
                 }
+
                 if dispItems.length > 0
                   debug "Found #{dispItems.length} new items in #{feed}"
                   # When displaying watched feeds, publish them from older to newer
@@ -710,13 +781,13 @@ class RSSFeedsPlugin < Plugin
       if opts.key?(:date) && opts[:date]
         if item.respond_to?(:pubDate) 
           if item.pubDate.class <= Time
-            date = item.pubDate.strftime("%Y/%m/%d %H.%M.%S")
+            date = item.pubDate.strftime("%Y/%m/%d %H:%M")
           else
             date = item.pubDate.to_s
           end
         elsif  item.respond_to?(:date)
           if item.date.class <= Time
-            date = item.date.strftime("%Y/%m/%d %H.%M.%S")
+            date = item.date.strftime("%Y/%m/%d %H:%M")
           else
             date = item.date.to_s
           end
@@ -727,15 +798,17 @@ class RSSFeedsPlugin < Plugin
       end
     end
 
-    title = "#{Bold}#{item.title.chomp.riphtml}#{Bold}" if item.title
+    title = "#{Bold}#{item.title.ircify_html :limit => @bot.config['rss.head_max']}#{Bold}" if item.title
 
-    desc = item.description.gsub(/\s+/,' ').strip.riphtml if item.description
+    desc = item.description.ircify_html(:limit => @bot.config['rss.text_max'], :a_href => :link_out) if item.description
 
     link = item.link.chomp if item.link
 
     debug item.inspect
-    category = item.dc_subject rescue item.category rescue nil
+    category = item.dc_subject rescue item.category.content rescue nil
+    category = nil if category and category.empty?
     author = item.dc_creator rescue item.author rescue nil
+    author = nil if author and author.empty?
 
     line1 = nil
     line2 = nil
@@ -743,8 +816,9 @@ class RSSFeedsPlugin < Plugin
     at = ((item.title && item.link) ? ' @ ' : '')
     case feed.type
     when 'blog'
+      author << " " if author
       abt = category ? "about #{category} " : ""
-      line1 = "#{handle}#{date}#{author} blogged #{abt}at #{link}"
+      line1 = "#{handle}#{date}#{author}blogged #{abt}at #{link}"
       line2 = "#{handle}#{title} - #{desc}"
     when 'forum'
       line1 = "#{handle}#{date}#{title}#{at}#{link}"
@@ -891,6 +965,9 @@ plugin.map 'rss replace :handle :url :type',
 plugin.map 'rss forcereplace :handle :url :type',
   :action => 'forcereplace_rss',
   :defaults => {:type => nil}
+plugin.map 'rss watch :handle [in :chan]',
+  :action => 'watch_rss',
+  :defaults => {:url => nil, :type => nil}
 plugin.map 'rss watch :handle :url :type [in :chan]',
   :action => 'watch_rss',
   :defaults => {:url => nil, :type => nil}