remove whitespace

[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 87f79ec57c240c947d4fd7a95ad789d7aa56eacf..bfc700d3e90a6738b3282e3f662c89fc1dd5c0f6 100644 (file)
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -16,7 +16,7 @@
 
 require 'rss'
 
-# Try to load rss/content/2.0 so we can access the data in <content:encoded> 
+# Try to load rss/content/2.0 so we can access the data in <content:encoded>
 # tags.
 begin
   require 'rss/content/2.0'
@@ -185,7 +185,7 @@ end
 
 class ::RssBlob
   attr_accessor :url, :handle, :type, :refresh_rate, :xml, :title, :items,
-    :mutex, :watchers, :last_fetched, :http_cache
+    :mutex, :watchers, :last_fetched, :http_cache, :last_success
 
   def initialize(url,handle=nil,type=nil,watchers=[], xml=nil, lf = nil)
     @url = url
@@ -203,6 +203,7 @@ class ::RssBlob
     @items = nil
     @mutex = Mutex.new
     @last_fetched = lf
+    @last_success = nil
     sanitize_watchers(watchers)
   end
 
@@ -277,6 +278,10 @@ class RSSFeedsPlugin < Plugin
     :default => 300, :validate => Proc.new{|v| v > 30},
     :desc => "How many seconds to sleep before checking RSS feeds again")
 
+  Config.register Config::IntegerValue.new('rss.announce_timeout',
+    :default => 0,
+    :desc => "Don't announce watched feed if these many seconds elapsed since the last successful update")
+
   Config.register Config::BooleanValue.new('rss.show_updated',
     :default => true,
     :desc => "Whether feed items for which the description was changed should be shown as new")
@@ -288,7 +293,7 @@ class RSSFeedsPlugin < Plugin
   # 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! || item.link!]
     if @bot.config['rss.show_updated']
@@ -888,7 +893,13 @@ class RSSFeedsPlugin < Plugin
       failures = status[:failures]
       begin
         debug "fetching #{feed}"
-        first_run = !feed.last_fetched
+
+        first_run = !feed.last_success
+        if (@bot.config['rss.announce_timeout'] > 0 &&
+           (Time.now - feed.last_success > @bot.config['rss.announce_timeout']))
+          debug "#{feed} wasn't polled for too long, supressing output"
+          first_run = true
+        end
         oldxml = feed.xml ? feed.xml.dup : nil
         unless fetchRss(feed, nil, feed.http_cache)
           failures += 1
@@ -1007,7 +1018,7 @@ class RSSFeedsPlugin < Plugin
           else
             date = item.source.updated.content.to_s
           end
-        elsif item.respond_to?(:pubDate) 
+        elsif item.respond_to?(:pubDate)
           if item.pubDate.class <= Time
             date = item.pubDate.strftime("%Y/%m/%d %H:%M")
           else
@@ -1044,7 +1055,7 @@ class RSSFeedsPlugin < Plugin
     desc_opt[:limit] = @bot.config['rss.text_max']
     desc_opt[:a_href] = :link_out if @bot.config['rss.show_links']
 
-    # We prefer content_encoded here as it tends to provide more html formatting 
+    # We prefer content_encoded here as it tends to provide more html formatting
     # for use with ircify_html.
     if item.respond_to?(:content_encoded) && item.content_encoded
       desc = item.content_encoded.ircify_html(desc_opt)
@@ -1126,6 +1137,7 @@ class RSSFeedsPlugin < Plugin
     end
     feed.mutex.synchronize do
       feed.xml = xml
+      feed.last_success = Time.now
     end
     return true
   end
@@ -1134,25 +1146,33 @@ class RSSFeedsPlugin < Plugin
     return nil unless feed.xml
     feed.mutex.synchronize do
       xml = feed.xml
-      begin
-        ## do validate parse
-        rss = RSS::Parser.parse(xml)
-        debug "parsed and validated #{feed}"
-      rescue RSS::InvalidRSSError
-        ## do non validate parse for invalid RSS 1.0
+      rss = nil
+      errors = []
+      RSS::AVAILABLE_PARSERS.each do |parser|
         begin
-          rss = RSS::Parser.parse(xml, false)
-          debug "parsed but not validated #{feed}"
+          ## do validate parse
+          rss = RSS::Parser.parse(xml, true, true, parser)
+          debug "parsed and validated #{feed} with #{parser}"
+          break
+        rescue RSS::InvalidRSSError
+          begin
+            ## do non validate parse for invalid RSS 1.0
+            rss = RSS::Parser.parse(xml, false, true, parser)
+            debug "parsed but not validated #{feed} with #{parser}"
+            break
+          rescue RSS::Error => e
+            errors << [parser, e, "parsing rss stream failed, whoops =("]
+          end
         rescue RSS::Error => e
-          report_problem("parsing rss stream failed, whoops =(", e, m)
-          return nil
+          errors << [parser, e, "parsing rss stream failed, oioi"]
+        rescue => e
+          errors << [parser, e, "processing error occured, sorry =("]
         end
-      rescue RSS::Error => e
-        report_problem("parsing rss stream failed, oioi", e, m)
-        return nil
-      rescue => e
-        report_problem("processing error occured, sorry =(", e, m)
-        return nil
+      end
+      unless errors.empty?
+        debug errors
+        self.send(:report_problem, errors.last[2], errors.last[1], m)
+        return nil unless rss
       end
       items = []
       if rss.nil?