X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Frss.rb;h=30d09bd3a3280b7f875e5efa33a84c62e52bb1bd;hb=16336b4a240a4265d1f2df1e30d7b68d3a924287;hp=f80236f08eef79d8cca4911df5e381f465f2a622;hpb=8797fa565f88d1843011150d95ff6df6e1fefb41;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index f80236f0..30d09bd3 100644
--- 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
+    :mutex, :watchers, :last_fetched, :http_cache, :last_success
 
   def initialize(url,handle=nil,type=nil,watchers=[], xml=nil, lf = nil)
     @url = url
@@ -197,11 +197,13 @@ class ::RssBlob
     @type = type
     @watchers=[]
     @refresh_rate = nil
+    @http_cache = false
     @xml = xml
     @title = nil
     @items = nil
     @mutex = Mutex.new
     @last_fetched = lf
+    @last_success = nil
     sanitize_watchers(watchers)
   end
 
@@ -276,6 +278,14 @@ 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::IntegerValue.new('rss.announce_max',
+    :default => 3,
+    :desc => "Maximum number of new items to announce when a watched feed is updated")
+
   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")
@@ -287,7 +297,7 @@ 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']
@@ -643,19 +653,34 @@ class RSSFeedsPlugin < Plugin
 
   def list_rss(m, params)
     wanted = params[:handle]
-    reply = String.new
-    @feeds.each { |handle, feed|
-      next if wanted and !handle.match(/#{wanted}/i)
-      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
-      (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
-      (reply << " (watched)") if feed.watched_by?(m.replyto)
-      reply << "\n"
-    }
-    if reply.empty?
+    listed = @feeds.keys
+    if wanted
+      wanted_rx = Regexp.new(wanted, true)
+      listed.reject! { |handle| !handle.match(wanted_rx) }
+    end
+    listed.sort!
+    debug listed
+    if @bot.config['send.max_lines'] > 0 and listed.size > @bot.config['send.max_lines']
+      reply = listed.inject([]) do |ar, handle|
+        feed = @feeds[handle]
+        string = handle.dup
+        (string << " (#{feed.type})") if feed.type
+        (string << " (watched)") if feed.watched_by?(m.replyto)
+        ar << string
+      end.join(', ')
+    elsif listed.size > 0
+      reply = listed.inject([]) do |ar, handle|
+        feed = @feeds[handle]
+        string = "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
+        (string << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
+        (string << " (watched)") if feed.watched_by?(m.replyto)
+        ar << string
+      end.join("\n")
+    else
       reply = "no feeds found"
       reply << " matching #{wanted}" if wanted
     end
-    m.reply reply, :max_lines => reply.length
+    m.reply reply, :max_lines => 0
   end
 
   def watched_rss(m, params)
@@ -836,6 +861,7 @@ class RSSFeedsPlugin < Plugin
     if params and handle = params[:handle]
       feed = @feeds.fetch(handle.downcase, nil)
       if feed
+        feed.http_cache = false
         @bot.timer.reschedule(@watch[feed.handle], (params[:delay] || 0).to_f)
         m.okay if m
       else
@@ -855,7 +881,7 @@ class RSSFeedsPlugin < Plugin
   private
   def watchRss(feed, m=nil)
     if @watch.has_key?(feed.handle)
-      report_problem("watcher thread for #{feed.handle} is already running", nil, m)
+      # report_problem("watcher thread for #{feed.handle} is already running", nil, m)
       return
     end
     status = Hash.new
@@ -871,11 +897,18 @@ 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)
+        unless fetchRss(feed, nil, feed.http_cache)
           failures += 1
         else
+          feed.http_cache = true
           if first_run
             debug "first run for #{feed}, getting items"
             parseRss(feed)
@@ -928,7 +961,19 @@ class RSSFeedsPlugin < Plugin
                 }
 
                 if dispItems.length > 0
+                  max = @bot.config['rss.announce_max']
                   debug "Found #{dispItems.length} new items in #{feed}"
+                  if max > 0 and dispItems.length > max
+                    debug "showing only the latest #{dispItems.length}"
+                    feed.watchers.each do |loc|
+                      @bot.say loc, (_("feed %{feed} had %{num} updates, showing the latest %{max}") % {
+                        :feed => feed.handle,
+                        :num => dispItems.length,
+                        :max => max
+                      })
+                    end
+                    dispItems.slice!(max..-1)
+                  end
                   # When displaying watched feeds, publish them from older to newer
                   dispItems.reverse.each { |item|
                     printFormattedRss(feed, item)
@@ -989,7 +1034,7 @@ 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
@@ -1026,7 +1071,7 @@ 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)
@@ -1108,6 +1153,7 @@ class RSSFeedsPlugin < Plugin
     end
     feed.mutex.synchronize do
       feed.xml = xml
+      feed.last_success = Time.now
     end
     return true
   end
@@ -1116,25 +1162,33 @@ 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?