X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Frss.rb;h=7313f252236d2507691fa075d64b834bf477b9a6;hb=b565bf81cbe456205b9f9fcf9e3960109c5b7de6;hp=af7901275b2e5a211552a63bd88c0efbc2ef881c;hpb=87e3751b3cbfc0de98a0b0b2dd44943e696b9eb9;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index af790127..7313f252 100644
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -1,18 +1,23 @@
+#-- vim:sw=2:et
+#++
+#
 # RSS feed plugin for RubyBot
 # (c) 2004 Stanislav Karchebny <berkus@madfire.net>
 # (c) 2005 Ian Monroe <ian@monroe.nu>
 # (c) 2005 Mark Kretschmann <markey@web.de>
+# (c) 2006 Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
+#
 # Licensed under MIT License.
 
 require 'rss/parser'
 require 'rss/1.0'
 require 'rss/2.0'
 require 'rss/dublincore'
-begin
-  # require 'rss/dublincore/2.0'
-rescue
-  warning "Unable to load RSS libraries, RSS plugin functionality crippled"
-end
+# begin
+#   require 'rss/dublincore/2.0'
+# rescue
+#   warning "Unable to load RSS libraries, RSS plugin functionality crippled"
+# end
 
 class ::String
   def shorten(limit)
@@ -54,19 +59,28 @@ class ::RssBlob
   end
 
   def watched_by?(who)
-    @watchers.include?(who)
+    # We need to check bot 'who' itself and the String form, because rss
+    # watches added before the new Irc framework represented watchers as
+    # Strings whereas they are now Channels.
+    #
+    @watchers.include?(who) || @watchers.include?(who.to_s) 
   end
 
   def add_watch(who)
     if watched_by?(who)
       return nil
     end
-    @watchers << who unless watched_by?(who)
+    # TODO FIXME? should we just store watchers as Strings instead?
+    # This should then be @watchers << who.downcase
+    @watchers << who
     return who
   end
 
   def rm_watch(who)
+    # See comment to watched_by?
+    #
     @watchers.delete(who)
+    @watchers.delete(who.to_s)
   end
 
   def to_a
@@ -92,6 +106,10 @@ class RSSFeedsPlugin < Plugin
     :default => 90, :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")
+
   @@watchThreads = Hash.new
   @@mutex = Mutex.new
 
@@ -106,6 +124,10 @@ class RSSFeedsPlugin < Plugin
     rewatch_rss
   end
 
+  def name
+    "rss"
+  end
+
   def watchlist
     @feeds.select { |h, f| f.watched? }
   end
@@ -132,7 +154,7 @@ class RSSFeedsPlugin < Plugin
   def help(plugin,topic="")
     case topic
     when "show"
-      "rss show #{Bold}handle#{Bold} [#{Bold}limit#{Bold}] : show #{Bold}limit#{Bold} (default: 5, max: 15) entries from rss #{Bold}handle#{Bold}"
+      "rss show #{Bold}handle#{Bold} [#{Bold}limit#{Bold}] : show #{Bold}limit#{Bold} (default: 5, max: 15) entries from rss #{Bold}handle#{Bold}; #{Bold}limit#{Bold} can also be in the form a..b, to display a specific range of items"
     when "list"
       "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
     when "watched"
@@ -156,37 +178,69 @@ class RSSFeedsPlugin < Plugin
     end
   end
 
-  def report_problem(report, m=nil)
-    if m
+  def report_problem(report, e=nil, m=nil)
+    if m && m.respond_to?(:reply)
       m.reply report
     else
       warning report
     end
+    if e
+      debug e.inspect
+      debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
+    end
   end
 
   def show_rss(m, params)
     handle = params[:handle]
-    limit = params[:limit].to_i
-    limit = 15 if limit > 15
-    limit = 1 if limit <= 0
+    lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
+    debug lims.to_a.inspect
+    if lims[2]
+      ll = [[lims[1].to_i-1,lims[2].to_i-1].min,  0].max
+      ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
+      rev = lims[1].to_i > lims[2].to_i
+    else
+      ll = 0
+      ul = [[lims[1].to_i-1, 1].max, 14].min
+      rev = false
+    end
+
     feed = @feeds.fetch(handle, nil)
     unless feed
       m.reply "I don't know any feeds named #{handle}"
       return
     end
-    m.reply("Please wait, querying...")
+
+    m.reply "lemme fetch it..."
     title = items = nil
     @@mutex.synchronize {
       title, items = fetchRss(feed, m)
     }
     return unless items
-    m.reply("Channel : #{title}")
-    # TODO: optional by-date sorting if dates present
-    items[0...limit].reverse.each do |item|
-      printRssItem(m.replyto,item)
+
+    # We sort the feeds in freshness order (newer ones first)
+    items = freshness_sort(items)
+    disp = items[ll..ul]
+    disp.reverse! if rev
+
+    m.reply "Channel : #{title}"
+    disp.each do |item|
+      printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
     end
   end
 
+  def itemDate(item,ex=nil)
+    return item.pubDate if item.respond_to?(:pubDate)
+    return item.date if item.respond_to?(:date)
+    return ex
+  end
+
+  def freshness_sort(items)
+    notime = Time.at(0)
+    items.sort { |a, b|
+      itemDate(b, notime) <=> itemDate(a, notime)
+    }
+  end
+
   def list_rss(m, params)
     wanted = params[:handle]
     reply = String.new
@@ -225,6 +279,10 @@ class RSSFeedsPlugin < Plugin
   def add_rss(m, params, force=false)
     handle = params[:handle]
     url = params[:url]
+    unless url.match(/https?/)
+      m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
+      return
+    end
     type = params[:type]
     if @feeds.fetch(handle, nil) && !force
       m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle].url})"
@@ -323,7 +381,7 @@ class RSSFeedsPlugin < Plugin
     return feed
   end
 
-  def rewatch_rss(m=nil)
+  def rewatch_rss(m=nil, params=nil)
     kill_threads
 
     # Read watches from list.
@@ -336,88 +394,109 @@ class RSSFeedsPlugin < Plugin
   private
   def watchRss(feed, m=nil)
     if @@watchThreads.has_key?(feed.handle)
-      report_problem("watcher thread for #{feed.handle} is already running", m)
+      report_problem("watcher thread for #{feed.handle} is already running", nil, m)
       return
     end
     @@watchThreads[feed.handle] = Thread.new do
-      debug 'watchRss thread started.'
+      debug "watcher for #{feed} started"
       oldItems = []
       firstRun = true
+      failures = 0
       loop do
         begin
-          debug 'Fetching rss feed...'
+          debug "fetching #{feed}"
           title = newItems = nil
           @@mutex.synchronize {
             title, newItems = fetchRss(feed)
           }
           unless newItems
-            m.reply "no items in feed"
-            break
-          end
-          debug "Checking if new items are available"
-          if firstRun
-            debug "First run, we'll see next time"
-            firstRun = false
+            debug "no items in feed #{feed}"
+            failures +=1
           else
-            otxt = oldItems.map { |item| item.to_s }
-            dispItems = newItems.reject { |item|
-              otxt.include?(item.to_s)
-            }
-            if dispItems.length > 0
-              debug "Found #{dispItems.length} new items"
-              dispItems.each { |item|
-                debug "showing #{item.title}"
-                @@mutex.synchronize {
-                  printFormattedRss(feed, item)
-                }
-              }
+            debug "Checking if new items are available for #{feed}"
+            if firstRun
+              debug "First run, we'll see next time"
+              firstRun = false
             else
-              debug "No new items found"
+              otxt = oldItems.map { |item| item.to_s }
+              dispItems = newItems.reject { |item|
+                otxt.include?(item.to_s)
+              }
+              if dispItems.length > 0
+                debug "Found #{dispItems.length} new items in #{feed}"
+                # When displaying watched feeds, publish them from older to newer
+                dispItems.reverse.each { |item|
+                  @@mutex.synchronize {
+                    printFormattedRss(feed, item)
+                  }
+                }
+              else
+                debug "No new items found in #{feed}"
+              end
             end
+            oldItems = newItems.dup
           end
-          oldItems = newItems.dup
         rescue Exception => e
-          error "IO failed: #{e.inspect}"
+          error "Error watching #{feed}: #{e.inspect}"
           debug e.backtrace.join("\n")
+          failures += 1
         end
 
-        seconds = 150 + rand(100)
-        debug "Thread going to sleep #{seconds} seconds.."
+        seconds = @bot.config['rss.thread_sleep'] * (failures + 1)
+        seconds += seconds * (rand(100)-50)/100
+        debug "watcher for #{feed} going to sleep #{seconds} seconds.."
         sleep seconds
       end
     end
   end
 
-  def printRssItem(loc,item)
-    if item.kind_of?(RSS::RDF::Item)
-      @bot.say loc, item.title.chomp.riphtml.shorten(@bot.config['rss.head_max']) + " @ " + item.link
-    else
-      desc = String.new
-      desc << item.pubDate.to_s.chomp + ": " if item.pubDate
-      desc << item.title.chomp.riphtml.shorten(@bot.config['rss.head_max']) + " :: " if item.title
-      desc << " @ " + item.link.chomp if item.link
-      @bot.say loc, desc
+  def printFormattedRss(feed, item, opts=nil)
+    places = feed.watchers
+    handle = "::#{feed.handle}:: "
+    date = String.new
+    if opts
+      places = opts[:places] if opts.key?(:places)
+      handle = opts[:handle].to_s if opts.key?(:handle)
+      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")
+          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")
+          else
+            date = item.date.to_s
+          end
+        else
+          date = "(no date)"
+        end
+        date += " :: "
+      end
     end
-  end
-
-  def printFormattedRss(feed, item)
-    debug "Printing formatted item #{item.inspect} for feed #{feed.to_s}"
-    feed.watchers.each { |loc|
+    title = "#{Bold}#{item.title.chomp.riphtml}#{Bold}" if item.title
+    desc = item.description.gsub(/\s+/,' ').strip.riphtml.shorten(@bot.config['rss.text_max']) if item.description
+    link = item.link.chomp if item.link
+    places.each { |loc|
       case feed.type
       when 'blog'
-        @bot.say loc, "::#{feed.handle}:: #{item.category.content} just blogged at #{item.link}::"
-        @bot.say loc, "::#{feed.handle}:: #{item.title.chomp.riphtml} - #{item.description.chomp.riphtml.shorten(@bot.config['rss.text_max'])}::"
+        @bot.say loc, "#{handle}#{date}#{item.category.content} blogged at #{link}"
+        @bot.say loc, "#{handle}#{title} - #{desc}"
       when 'forum'
-        @bot.say loc, "::#{feed.handle}:: #{item.pubDate.to_s.chomp+": " if item.pubDate}#{item.title.chomp.riphtml+" :: " if item.title}#{" @ "+item.link.chomp if item.link}"
+        @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
       when 'wiki'
-        @bot.say loc, "::#{feed.handle}:: #{item.title} has been edited by #{item.dc_creator}. #{item.description.split("\n")[0].chomp.riphtml.shorten(@bot.config['rss.text_max'])} #{item.link} ::"
+        @bot.say loc, "#{handle}#{date}#{item.title} has been edited by #{item.dc_creator}. #{desc} #{link}"
       when 'gmame'
-        @bot.say loc, "::#{feed.handle}:: Message #{item.title} sent by #{item.dc_creator}. #{item.description.split("\n")[0].chomp.riphtml.shorten(@bot.config['rss.text_max'])} ::"
+        @bot.say loc, "#{handle}#{date}Message #{title} sent by #{item.dc_creator}. #{desc}"
       when 'trac'
-        @bot.say loc, "::#{feed.handle}:: #{item.title} :: #{item.link}"
-        @bot.say loc, "::#{feed.handle}:: #{item.description.gsub(/\s+/,' ').strip.riphtml.shorten(@bot.config['rss.text_max'])}"
+        @bot.say loc, "#{handle}#{date}#{title} @ #{link}"
+        unless item.title =~ /^Changeset \[(\d+)\]/
+          @bot.say loc, "#{handle}#{date}#{desc}"
+        end
       else
-        printRssItem(loc,item)
+        @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
       end
     }
   end
@@ -425,46 +504,49 @@ class RSSFeedsPlugin < Plugin
   def fetchRss(feed, m=nil)
     begin
       # Use 60 sec timeout, cause the default is too low
-      xml = @bot.httputil.get_cached(feed.url,60,60)
+      # Do not use get_cached for RSS until we have proper cache handling
+      # xml = @bot.httputil.get_cached(feed.url,60,60)
+      xml = @bot.httputil.get(feed.url,60,60)
     rescue URI::InvalidURIError, URI::BadURIError => e
-      report_problem("invalid rss feed #{feed.url}", m)
+      report_problem("invalid rss feed #{feed.url}", e, m)
+      return
+    rescue => e
+      report_problem("error getting #{feed.url}", e, m)
       return
     end
-    debug 'fetched'
+    debug "fetched #{feed}"
     unless xml
-      report_problem("reading feed #{url} failed", m)
+      report_problem("reading feed #{feed} failed", nil, m)
       return
     end
 
     begin
       ## do validate parse
       rss = RSS::Parser.parse(xml)
-      debug 'parsed'
+      debug "parsed #{feed}"
     rescue RSS::InvalidRSSError
       ## do non validate parse for invalid RSS 1.0
       begin
         rss = RSS::Parser.parse(xml, false)
-      rescue RSS::Error
-        report_problem("parsing rss stream failed, whoops =(", m)
+      rescue RSS::Error => e
+        report_problem("parsing rss stream failed, whoops =(", e, m)
         return
       end
-    rescue RSS::Error
-      report_problem("parsing rss stream failed, oioi", m)
+    rescue RSS::Error => e
+      report_problem("parsing rss stream failed, oioi", e, m)
       return
     rescue => e
-      report_problem("processing error occured, sorry =(", m)
-      debug e.inspect
-      debug e.backtrace.join("\n")
+      report_problem("processing error occured, sorry =(", e, m)
       return
     end
     items = []
     if rss.nil?
-      report_problem("#{feed.url} does not include RSS 1.0 or 0.9x/2.0",m)
+      report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
     else
       begin
-        rss.output_encoding = "euc-jp"
-      rescue RSS::UnknownConvertMethod
-        report_problem("bah! something went wrong =(",m)
+        rss.output_encoding = 'UTF-8'
+      rescue RSS::UnknownConvertMethod => e
+        report_problem("bah! something went wrong =(", e, m)
         return
       end
       rss.channel.title ||= "Unknown"
@@ -476,7 +558,7 @@ class RSSFeedsPlugin < Plugin
     end
 
     if items.empty?
-      report_problem("no items found in the feed, maybe try weed?",m)
+      report_problem("no items found in the feed, maybe try weed?", e, m)
       return
     end
     return [title, items]
@@ -487,7 +569,7 @@ plugin = RSSFeedsPlugin.new
 
 plugin.map 'rss show :handle :limit',
   :action => 'show_rss',
-  :requirements => {:limit => /^\d+$/},
+  :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
   :defaults => {:limit => 5}
 plugin.map 'rss list :handle',
   :action => 'list_rss',
@@ -517,5 +599,6 @@ plugin.map 'rss unwatch :handle',
   :action => 'unwatch_rss'
 plugin.map 'rss rmwatch :handle',
   :action => 'unwatch_rss'
-plugin.map 'rss rewatch :handle',
+plugin.map 'rss rewatch',
   :action => 'rewatch_rss'
+