[user/henk/code/ruby/rbot.git] / data / rbot / plugins / rss.rb

#-- vim:sw=2:et\r
#++\r
#\r
# :title: RSS feed plugin for rbot\r
#\r
# Author:: Stanislav Karchebny <berkus@madfire.net>\r
# Author:: Ian Monroe <ian@monroe.nu>\r
# Author:: Mark Kretschmann <markey@web.de>\r
# Author:: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>\r
#\r
# Copyright:: (C) 2004 Stanislav Karchebny\r
# Copyright:: (C) 2005 Ian Monroe, Mark Kretschmann\r
# Copyright:: (C) 2006-2007 Giuseppe Bilotta\r
#\r
# License:: MIT license\r
\r
require 'rss'\r
\r
# Try to load rss/content/2.0 so we can access the data in <content:encoded> \r
# tags.\r
begin\r
  require 'rss/content/2.0'\r
rescue LoadError\r
end\r
\r
module ::RSS\r
\r
  # Make an  'unique' ID for a given item, based on appropriate bot options\r
  # Currently only suppored is bot.config['rss.show_updated']: when true, the\r
  # description is included in the uid hashing, otherwise it's not\r
  #\r
  def RSS.item_uid_for_bot(item, opts={})\r
    options = { :show_updated => true}.merge(opts)\r
    desc = nil\r
    if options[:show_updated]\r
      desc = item.content.content rescue item.description rescue nil\r
    end\r
    [(item.title.content rescue item.title rescue nil),\r
     (item.link.href rescue item.link),\r
     desc].hash\r
  end\r
\r
  # Add support for Slashdot namespace in RDF. The code is just an adaptation\r
  # of the DublinCore code.\r
  unless defined?(SLASH_PREFIX)\r
    SLASH_PREFIX = 'slash'\r
    SLASH_URI = "http://purl.org/rss/1.0/modules/slash/"\r
\r
    RDF.install_ns(SLASH_PREFIX, SLASH_URI)\r
\r
    module BaseSlashModel\r
      def append_features(klass)\r
        super\r
\r
        return if klass.instance_of?(Module)\r
        SlashModel::ELEMENT_NAME_INFOS.each do |name, plural_name|\r
          plural = plural_name || "#{name}s"\r
          full_name = "#{SLASH_PREFIX}_#{name}"\r
          full_plural_name = "#{SLASH_PREFIX}_#{plural}"\r
          klass_name = "Slash#{Utils.to_class_name(name)}"\r
\r
          # This will fail with older version of the Ruby RSS module\r
          begin\r
            klass.install_have_children_element(name, SLASH_URI, "*",\r
                                                full_name, full_plural_name)\r
            klass.install_must_call_validator(SLASH_PREFIX, SLASH_URI)\r
          rescue ArgumentError\r
            klass.module_eval("install_have_children_element(#{full_name.dump}, #{full_plural_name.dump})")\r
          end\r
\r
          klass.module_eval(<<-EOC, *get_file_and_line_from_caller(0))\r
          remove_method :#{full_name}     if method_defined? :#{full_name}\r
          remove_method :#{full_name}=    if method_defined? :#{full_name}=\r
          remove_method :set_#{full_name} if method_defined? :set_#{full_name}\r
\r
          def #{full_name}\r
            @#{full_name}.first and @#{full_name}.first.value\r
          end\r
\r
          def #{full_name}=(new_value)\r
            @#{full_name}[0] = Utils.new_with_value_if_need(#{klass_name}, new_value)\r
          end\r
          alias set_#{full_name} #{full_name}=\r
        EOC\r
        end\r
      end\r
    end\r
\r
    module SlashModel\r
      extend BaseModel\r
      extend BaseSlashModel\r
\r
      TEXT_ELEMENTS = {\r
      "department" => nil,\r
      "section" => nil,\r
      "comments" =>  nil,\r
      "hit_parade" => nil\r
      }\r
\r
      ELEMENT_NAME_INFOS = SlashModel::TEXT_ELEMENTS.to_a\r
\r
      ELEMENTS = TEXT_ELEMENTS.keys\r
\r
      ELEMENTS.each do |name, plural_name|\r
        module_eval(<<-EOC, *get_file_and_line_from_caller(0))\r
        class Slash#{Utils.to_class_name(name)} < Element\r
          include RSS10\r
\r
          content_setup\r
\r
          class << self\r
            def required_prefix\r
              SLASH_PREFIX\r
            end\r
\r
            def required_uri\r
              SLASH_URI\r
            end\r
          end\r
\r
          @tag_name = #{name.dump}\r
\r
          alias_method(:value, :content)\r
          alias_method(:value=, :content=)\r
\r
          def initialize(*args)\r
            begin\r
              if Utils.element_initialize_arguments?(args)\r
                super\r
              else\r
                super()\r
                self.content = args[0]\r
              end\r
            # Older Ruby RSS module\r
            rescue NoMethodError\r
              super()\r
              self.content = args[0]\r
            end\r
          end\r
\r
          def full_name\r
            tag_name_with_prefix(SLASH_PREFIX)\r
          end\r
\r
          def maker_target(target)\r
            target.new_#{name}\r
          end\r
\r
          def setup_maker_attributes(#{name})\r
            #{name}.content = content\r
          end\r
        end\r
      EOC\r
      end\r
    end\r
\r
    class RDF\r
      class Item; include SlashModel; end\r
    end\r
\r
    SlashModel::ELEMENTS.each do |name|\r
      class_name = Utils.to_class_name(name)\r
      BaseListener.install_class_name(SLASH_URI, name, "Slash#{class_name}")\r
    end\r
\r
    SlashModel::ELEMENTS.collect! {|name| "#{SLASH_PREFIX}_#{name}"}\r
  end\r
end\r
\r
\r
class ::RssBlob\r
  attr_accessor :url, :handle, :type, :refresh_rate, :xml, :title, :items,\r
    :mutex, :watchers, :last_fetched\r
\r
  def initialize(url,handle=nil,type=nil,watchers=[], xml=nil, lf = nil)\r
    @url = url\r
    if handle\r
      @handle = handle\r
    else\r
      @handle = url\r
    end\r
    @type = type\r
    @watchers=[]\r
    @refresh_rate = nil\r
    @xml = xml\r
    @title = nil\r
    @items = nil\r
    @mutex = Mutex.new\r
    @last_fetched = lf\r
    sanitize_watchers(watchers)\r
  end\r
\r
  def dup\r
    @mutex.synchronize do\r
      self.class.new(@url,\r
                     @handle,\r
                     @type ? @type.dup : nil,\r
                     @watchers.dup,\r
                     @xml ? @xml.dup : nil,\r
                     @last_fetched)\r
    end\r
  end\r
\r
  # Downcase all watchers, possibly turning them into Strings if they weren't\r
  def sanitize_watchers(list=@watchers)\r
    ls = list.dup\r
    @watchers.clear\r
    ls.each { |w|\r
      add_watch(w)\r
    }\r
  end\r
\r
  def watched?\r
    !@watchers.empty?\r
  end\r
\r
  def watched_by?(who)\r
    @watchers.include?(who.downcase)\r
  end\r
\r
  def add_watch(who)\r
    if watched_by?(who)\r
      return nil\r
    end\r
    @mutex.synchronize do\r
      @watchers << who.downcase\r
    end\r
    return who\r
  end\r
\r
  def rm_watch(who)\r
    @mutex.synchronize do\r
      @watchers.delete(who.downcase)\r
    end\r
  end\r
\r
  def to_a\r
    [@handle,@url,@type,@refresh_rate,@watchers]\r
  end\r
\r
  def to_s(watchers=false)\r
    if watchers\r
      a = self.to_a.flatten\r
    else\r
      a = self.to_a[0,3]\r
    end\r
    a.compact.join(" | ")\r
  end\r
end\r
\r
class RSSFeedsPlugin < Plugin\r
  Config.register Config::IntegerValue.new('rss.head_max',\r
    :default => 100, :validate => Proc.new{|v| v > 0 && v < 200},\r
    :desc => "How many characters to use of a RSS item header")\r
\r
  Config.register Config::IntegerValue.new('rss.text_max',\r
    :default => 200, :validate => Proc.new{|v| v > 0 && v < 400},\r
    :desc => "How many characters to use of a RSS item text")\r
\r
  Config.register Config::IntegerValue.new('rss.thread_sleep',\r
    :default => 300, :validate => Proc.new{|v| v > 30},\r
    :desc => "How many seconds to sleep before checking RSS feeds again")\r
\r
  Config.register Config::BooleanValue.new('rss.show_updated',\r
    :default => true,\r
    :desc => "Whether feed items for which the description was changed should be shown as new")\r
\r
  Config.register Config::BooleanValue.new('rss.show_links',\r
    :default => true,\r
    :desc => "Whether to display links from the text of a feed item.")\r
\r
  # We used to save the Mutex with the RssBlob, which was idiotic. And\r
  # since Mutexes dumped in one version might not be resotrable in another,\r
  # we need a few tricks to be able to restore data from other versions of Ruby\r
  #\r
  # When migrating 1.8.6 => 1.8.5, all we need to do is define an empty\r
  # #marshal_load() method for Mutex. For 1.8.5 => 1.8.6 we need something\r
  # dirtier, as seen later on in the initialization code.\r
  unless Mutex.new.respond_to?(:marshal_load)\r
    class ::Mutex\r
      def marshal_load(str)\r
        return\r
      end\r
    end\r
  end\r
\r
  attr_reader :feeds\r
\r
  def initialize\r
    super\r
    if @registry.has_key?(:feeds)\r
      # When migrating from Ruby 1.8.5 to 1.8.6, dumped Mutexes may render the\r
      # data unrestorable. If this happens, we patch the data, thus allowing\r
      # the restore to work.\r
      #\r
      # This is actually pretty safe for a number of reasons:\r
      # * the code is only called if standard marshalling fails\r
      # * the string we look for is quite unlikely to appear randomly\r
      # * if the string appears somewhere and the patched string isn't recoverable\r
      #   either, we'll get another (unrecoverable) error, which makes the rss\r
      #   plugin unsable, just like it was if no recovery was attempted\r
      # * if the string appears somewhere and the patched string is recoverable,\r
      #   we may get a b0rked feed, which is eventually overwritten by a clean\r
      #   one, so the worst thing that can happen is that a feed update spams\r
      #   the watchers once\r
      @registry.recovery = Proc.new { |val|\r
        patched = val.sub(":\v@mutexo:\nMutex", ":\v@mutexo:\vObject")\r
        ret = Marshal.restore(patched)\r
        ret.each_value { |blob|\r
          blob.mutex = nil\r
          blob\r
        }\r
      }\r
\r
      @feeds = @registry[:feeds]\r
      raise unless @feeds\r
\r
      @registry.recovery = nil\r
\r
      @feeds.keys.grep(/[A-Z]/) { |k|\r
        @feeds[k.downcase] = @feeds[k]\r
        @feeds.delete(k)\r
      }\r
      @feeds.each { |k, f|\r
        f.mutex = Mutex.new\r
        f.sanitize_watchers\r
        parseRss(f) if f.xml\r
      }\r
    else\r
      @feeds = Hash.new\r
    end\r
    @watch = Hash.new\r
    rewatch_rss\r
  end\r
\r
  def name\r
    "rss"\r
  end\r
\r
  def watchlist\r
    @feeds.select { |h, f| f.watched? }\r
  end\r
\r
  def cleanup\r
    stop_watches\r
    super\r
  end\r
\r
  def save\r
    unparsed = Hash.new()\r
    @feeds.each { |k, f|\r
      unparsed[k] = f.dup\r
      # we don't want to save the mutex\r
      unparsed[k].mutex = nil\r
    }\r
    @registry[:feeds] = unparsed\r
  end\r
\r
  def stop_watch(handle)\r
    if @watch.has_key?(handle)\r
      begin\r
        debug "Stopping watch #{handle}"\r
        @bot.timer.remove(@watch[handle])\r
        @watch.delete(handle)\r
      rescue Exception => e\r
        report_problem("Failed to stop watch for #{handle}", e, nil)\r
      end\r
    end\r
  end\r
\r
  def stop_watches\r
    @watch.each_key { |k|\r
      stop_watch(k)\r
    }\r
  end\r
\r
  def help(plugin,topic="")\r
    case topic\r
    when "show"\r
      "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"\r
    when "list"\r
      "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"\r
    when "watched"\r
      "rss watched [#{Bold}handle#{Bold}] [in #{Bold}chan#{Bold}]: list all watched rss feeds (matching #{Bold}handle#{Bold}) (in channel #{Bold}chan#{Bold})"\r
    when "who", "watches", "who watches"\r
      "rss who watches [#{Bold}handle#{Bold}]]: list all watchers for rss feeds (matching #{Bold}handle#{Bold})"\r
    when "add"\r
      "rss add #{Bold}handle#{Bold} #{Bold}url#{Bold} [#{Bold}type#{Bold}] : add a new rss called #{Bold}handle#{Bold} from url #{Bold}url#{Bold} (of type #{Bold}type#{Bold})"\r
    when "change"\r
      "rss change #{Bold}what#{Bold} of #{Bold}handle#{Bold} to #{Bold}new#{Bold} : change the #{Underline}handle#{Underline}, #{Underline}url#{Underline}, #{Underline}type#{Underline} or #{Underline}refresh#{Underline} rate of rss called #{Bold}handle#{Bold} to value #{Bold}new#{Bold}"\r
    when /^(del(ete)?|rm)$/\r
      "rss del(ete)|rm #{Bold}handle#{Bold} : delete rss feed #{Bold}handle#{Bold}"\r
    when "replace"\r
      "rss replace #{Bold}handle#{Bold} #{Bold}url#{Bold} [#{Bold}type#{Bold}] : try to replace the url of rss called #{Bold}handle#{Bold} with #{Bold}url#{Bold} (of type #{Bold}type#{Bold}); only works if nobody else is watching it"\r
    when "forcereplace"\r
      "rss forcereplace #{Bold}handle#{Bold} #{Bold}url#{Bold} [#{Bold}type#{Bold}] : replace the url of rss called #{Bold}handle#{Bold} with #{Bold}url#{Bold} (of type #{Bold}type#{Bold})"\r
    when "watch"\r
      "rss watch #{Bold}handle#{Bold} [#{Bold}url#{Bold} [#{Bold}type#{Bold}]]  [in #{Bold}chan#{Bold}]: watch rss #{Bold}handle#{Bold} for changes (in channel #{Bold}chan#{Bold}); when the other parameters are present, the feed will be created if it doesn't exist yet"\r
    when /(un|rm)watch/\r
      "rss unwatch|rmwatch #{Bold}handle#{Bold} [in #{Bold}chan#{Bold}]: stop watching rss #{Bold}handle#{Bold} (in channel #{Bold}chan#{Bold}) for changes"\r
    when  /who(?: watche?s?)?/\r
      "rss who watches #{Bold}handle#{Bold}: lists watches for rss #{Bold}handle#{Bold}"\r
    when "rewatch"\r
      "rss rewatch : restart threads that watch for changes in watched rss"\r
    else\r
      "manage RSS feeds: rss show|list|watched|add|change|del(ete)|rm|(force)replace|watch|unwatch|rmwatch|rewatch|who watches"\r
    end\r
  end\r
\r
  def report_problem(report, e=nil, m=nil)\r
    if m && m.respond_to?(:reply)\r
      m.reply report\r
    else\r
      warning report\r
    end\r
    if e\r
      debug e.inspect\r
      debug e.backtrace.join("\n") if e.respond_to?(:backtrace)\r
    end\r
  end\r
\r
  def show_rss(m, params)\r
    handle = params[:handle]\r
    lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)\r
    debug lims.to_a.inspect\r
    if lims[2]\r
      ll = [[lims[1].to_i-1,lims[2].to_i-1].min,  0].max\r
      ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min\r
      rev = lims[1].to_i > lims[2].to_i\r
    else\r
      ll = 0\r
      ul = [[lims[1].to_i-1, 0].max, 14].min\r
      rev = false\r
    end\r
\r
    feed = @feeds.fetch(handle.downcase, nil)\r
    unless feed\r
      m.reply "I don't know any feeds named #{handle}"\r
      return\r
    end\r
\r
    m.reply "lemme fetch it..."\r
    title = items = nil\r
    we_were_watching = false\r
\r
    if @watch.key?(feed.handle)\r
      # If a feed is being watched, we run the watcher thread\r
      # so that all watchers can be informed of changes to\r
      # the feed. Before we do that, though, we remove the\r
      # show requester from the watchlist, if present, lest\r
      # he gets the update twice.\r
      if feed.watched_by?(m.replyto)\r
        we_were_watching = true\r
        feed.rm_watch(m.replyto)\r
      end\r
      @bot.timer.reschedule(@watch[feed.handle], 0)\r
      if we_were_watching\r
        feed.add_watch(m.replyto)\r
      end\r
    else\r
      fetched = fetchRss(feed, m, false)\r
    end\r
    return unless fetched or feed.xml\r
    if not fetched and feed.items\r
      m.reply "using old data"\r
    else\r
      parsed = parseRss(feed, m)\r
      m.reply "using old data" unless parsed\r
    end\r
    return unless feed.items\r
    title = feed.title\r
    items = feed.items\r
\r
    # We sort the feeds in freshness order (newer ones first)\r
    items = freshness_sort(items)\r
    disp = items[ll..ul]\r
    disp.reverse! if rev\r
\r
    m.reply "Channel : #{title}"\r
    disp.each do |item|\r
      printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})\r
    end\r
  end\r
\r
  def itemDate(item,ex=nil)\r
    return item.pubDate if item.respond_to?(:pubDate) and item.pubDate\r
    return item.date if item.respond_to?(:date) and item.date\r
    return ex\r
  end\r
\r
  def freshness_sort(items)\r
    notime = Time.at(0)\r
    items.sort { |a, b|\r
      itemDate(b, notime) <=> itemDate(a, notime)\r
    }\r
  end\r
\r
  def list_rss(m, params)\r
    wanted = params[:handle]\r
    reply = String.new\r
    @feeds.each { |handle, feed|\r
      next if wanted and !handle.match(/#{wanted}/i)\r
      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"\r
      (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate\r
      (reply << " (watched)") if feed.watched_by?(m.replyto)\r
      reply << "\n"\r
    }\r
    if reply.empty?\r
      reply = "no feeds found"\r
      reply << " matching #{wanted}" if wanted\r
    end\r
    m.reply reply, :max_lines => reply.length\r
  end\r
\r
  def watched_rss(m, params)\r
    wanted = params[:handle]\r
    chan = params[:chan] || m.replyto\r
    reply = String.new\r
    watchlist.each { |handle, feed|\r
      next if wanted and !handle.match(/#{wanted}/i)\r
      next unless feed.watched_by?(chan)\r
      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"\r
      (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate\r
      reply << "\n"\r
    }\r
    if reply.empty?\r
      reply = "no watched feeds"\r
      reply << " matching #{wanted}" if wanted\r
    end\r
    m.reply reply\r
  end\r
\r
  def who_watches(m, params)\r
    wanted = params[:handle]\r
    reply = String.new\r
    watchlist.each { |handle, feed|\r
      next if wanted and !handle.match(/#{wanted}/i)\r
      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"\r
      (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate\r
      reply << ": watched by #{feed.watchers.join(', ')}"\r
      reply << "\n"\r
    }\r
    if reply.empty?\r
      reply = "no watched feeds"\r
      reply << " matching #{wanted}" if wanted\r
    end\r
    m.reply reply\r
  end\r
\r
  def add_rss(m, params, force=false)\r
    handle = params[:handle]\r
    url = params[:url]\r
    unless url.match(/https?/)\r
      m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"\r
      return\r
    end\r
    type = params[:type]\r
    if @feeds.fetch(handle.downcase, nil) && !force\r
      m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"\r
      return\r
    end\r
    unless url\r
      m.reply "You must specify both a handle and an url to add an RSS feed"\r
      return\r
    end\r
    @feeds[handle.downcase] = RssBlob.new(url,handle,type)\r
    reply = "Added RSS #{url} named #{handle}"\r
    if type\r
      reply << " (format: #{type})"\r
    end\r
    m.reply reply\r
    return handle\r
  end\r
\r
  def change_rss(m, params)\r
    handle = params[:handle].downcase\r
    feed = @feeds.fetch(handle, nil)\r
    unless feed\r
      m.reply "No such feed with handle #{handle}"\r
      return\r
    end\r
    case params[:what].intern\r
    when :handle\r
      new = params[:new].downcase\r
      if @feeds.key?(new) and @feeds[new]\r
        m.reply "There already is a feed with handle #{new}"\r
        return\r
      else\r
        feed.mutex.synchronize do\r
          @feeds[new] = feed\r
          @feeds.delete(handle)\r
          feed.handle = new\r
        end\r
        handle = new\r
      end\r
    when :url\r
      new = params[:new]\r
      feed.mutex.synchronize do\r
        feed.url = new\r
      end\r
    when :format, :type\r
      new = params[:new]\r
      new = nil if new == 'default'\r
      feed.mutex.synchronize do\r
        feed.type = new\r
      end\r
    when :refresh\r
      new = params[:new].to_i\r
      new = nil if new == 0\r
      feed.mutex.synchronize do\r
        feed.refresh_rate = new\r
      end\r
    else\r
      m.reply "Don't know how to change #{params[:what]} for feeds"\r
      return\r
    end\r
    m.reply "Feed changed:"\r
    list_rss(m, {:handle => handle})\r
  end\r
\r
  def del_rss(m, params, pass=false)\r
    feed = unwatch_rss(m, params, true)\r
    return unless feed\r
    if feed.watched?\r
      m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"\r
      return\r
    end\r
    @feeds.delete(feed.handle.downcase)\r
    m.okay unless pass\r
    return\r
  end\r
\r
  def replace_rss(m, params)\r
    handle = params[:handle]\r
    if @feeds.key?(handle.downcase)\r
      del_rss(m, {:handle => handle}, true)\r
    end\r
    if @feeds.key?(handle.downcase)\r
      m.reply "can't replace #{feed.handle}"\r
    else\r
      add_rss(m, params, true)\r
    end\r
  end\r
\r
  def forcereplace_rss(m, params)\r
    add_rss(m, params, true)\r
  end\r
\r
  def watch_rss(m, params)\r
    handle = params[:handle]\r
    chan = params[:chan] || m.replyto\r
    url = params[:url]\r
    type = params[:type]\r
    if url\r
      add_rss(m, params)\r
    end\r
    feed = @feeds.fetch(handle.downcase, nil)\r
    if feed\r
      if feed.add_watch(chan)\r
        watchRss(feed, m)\r
        m.okay\r
      else\r
        m.reply "Already watching #{feed.handle} in #{chan}"\r
      end\r
    else\r
      m.reply "Couldn't watch feed #{handle} (no such feed found)"\r
    end\r
  end\r
\r
  def unwatch_rss(m, params, pass=false)\r
    handle = params[:handle].downcase\r
    chan = params[:chan] || m.replyto\r
    unless @feeds.has_key?(handle)\r
      m.reply("dunno that feed")\r
      return\r
    end\r
    feed = @feeds[handle]\r
    if feed.rm_watch(chan)\r
      m.reply "#{chan} has been removed from the watchlist for #{feed.handle}"\r
    else\r
      m.reply("#{chan} wasn't watching #{feed.handle}") unless pass\r
    end\r
    if !feed.watched?\r
      stop_watch(handle)\r
    end\r
    return feed\r
  end\r
\r
  def rewatch_rss(m=nil, params=nil)\r
    if params and handle = params[:handle]\r
      feed = @feeds.fetch(handle.downcase, nil)\r
      if feed\r
        @bot.timer.reschedule(@watch[feed.handle], 0)\r
        m.okay if m\r
      else\r
        m.reply _("no such feed %{handle}") % { :handle => handle } if m\r
      end\r
    else\r
      stop_watches\r
\r
      # Read watches from list.\r
      watchlist.each{ |handle, feed|\r
        watchRss(feed, m)\r
      }\r
      m.okay if m\r
    end\r
  end\r
\r
  private\r
  def watchRss(feed, m=nil)\r
    if @watch.has_key?(feed.handle)\r
      report_problem("watcher thread for #{feed.handle} is already running", nil, m)\r
      return\r
    end\r
    status = Hash.new\r
    status[:failures] = 0\r
    tmout = 0\r
    if feed.last_fetched\r
      tmout = feed.last_fetched + calculate_timeout(feed) - Time.now\r
      tmout = 0 if tmout < 0\r
    end\r
    debug "scheduling a watcher for #{feed} in #{tmout} seconds"\r
    @watch[feed.handle] = @bot.timer.add(tmout) {\r
      debug "watcher for #{feed} wakes up"\r
      failures = status[:failures]\r
      begin\r
        debug "fetching #{feed}"\r
        first_run = !feed.last_fetched\r
        oldxml = feed.xml ? feed.xml.dup : nil\r
        unless fetchRss(feed)\r
          failures += 1\r
        else\r
          if first_run\r
            debug "first run for #{feed}, getting items"\r
            parseRss(feed)\r
          elsif oldxml and oldxml == feed.xml\r
            debug "xml for #{feed} didn't change"\r
            failures -= 1 if failures > 0\r
          else\r
            if not feed.items\r
              debug "no previous items in feed #{feed}"\r
              parseRss(feed)\r
              failures -= 1 if failures > 0\r
            else\r
              # This one is used for debugging\r
              otxt = []\r
\r
              # These are used for checking new items vs old ones\r
              uid_opts = { :show_updated => @bot.config['rss.show_updated'] }\r
              oids = Set.new feed.items.map { |item|\r
                uid = RSS.item_uid_for_bot(item, uid_opts)\r
                otxt << item.to_s\r
                debug [uid, item].inspect\r
                debug [uid, otxt.last].inspect\r
                uid\r
              }\r
\r
              unless parseRss(feed)\r
                debug "no items in feed #{feed}"\r
                failures += 1\r
              else\r
                debug "Checking if new items are available for #{feed}"\r
                failures -= 1 if failures > 0\r
                # debug "Old:"\r
                # debug oldxml\r
                # debug "New:"\r
                # debug feed.xml\r
\r
                dispItems = feed.items.reject { |item|\r
                  uid = RSS.item_uid_for_bot(item, uid_opts)\r
                  txt = item.to_s\r
                  if oids.include?(uid)\r
                    debug "rejecting old #{uid} #{item.inspect}"\r
                    debug [uid, txt].inspect\r
                    true\r
                  else\r
                    debug "accepting new #{uid} #{item.inspect}"\r
                    debug [uid, txt].inspect\r
                    warning "same text! #{txt}" if otxt.include?(txt)\r
                    false\r
                  end\r
                }\r
\r
                if dispItems.length > 0\r
                  debug "Found #{dispItems.length} new items in #{feed}"\r
                  # When displaying watched feeds, publish them from older to newer\r
                  dispItems.reverse.each { |item|\r
                    printFormattedRss(feed, item)\r
                  }\r
                else\r
                  debug "No new items found in #{feed}"\r
                end\r
              end\r
            end\r
          end\r
        end\r
      rescue Exception => e\r
        error "Error watching #{feed}: #{e.inspect}"\r
        debug e.backtrace.join("\n")\r
        failures += 1\r
      end\r
\r
      status[:failures] = failures\r
\r
      seconds = calculate_timeout(feed, failures)\r
      debug "watcher for #{feed} going to sleep #{seconds} seconds.."\r
      begin\r
        @bot.timer.reschedule(@watch[feed.handle], seconds)\r
      rescue\r
        warning "watcher for #{feed} failed to reschedule: #{$!.inspect}"\r
      end\r
    }\r
    debug "watcher for #{feed} added"\r
  end\r
\r
  def calculate_timeout(feed, failures = 0)\r
      seconds = @bot.config['rss.thread_sleep']\r
      feed.mutex.synchronize do\r
        seconds = feed.refresh_rate if feed.refresh_rate\r
      end\r
      seconds *= failures + 1\r
      seconds += seconds * (rand(100)-50)/100\r
      return seconds\r
  end\r
\r
  def select_nonempty(*ar)\r
    debug ar\r
    ret = ar.map { |i| (i && i.empty?) ? nil : i }.compact.first\r
    (ret && ret.empty?) ? nil : ret\r
  end\r
\r
  def printFormattedRss(feed, item, opts=nil)\r
    debug item\r
    places = feed.watchers\r
    handle = "::#{feed.handle}:: "\r
    date = String.new\r
    if opts\r
      places = opts[:places] if opts.key?(:places)\r
      handle = opts[:handle].to_s if opts.key?(:handle)\r
      if opts.key?(:date) && opts[:date]\r
        if item.respond_to?(:updated)\r
          if item.updated.content.class <= Time\r
            date = item.updated.content.strftime("%Y/%m/%d %H:%M")\r
          else\r
            date = item.updated.content.to_s\r
          end\r
        elsif item.respond_to?(:source) and item.source.respond_to?(:updated)\r
          if item.source.updated.content.class <= Time\r
            date = item.source.updated.content.strftime("%Y/%m/%d %H:%M")\r
          else\r
            date = item.source.updated.content.to_s\r
          end\r
        elsif item.respond_to?(:pubDate) \r
          if item.pubDate.class <= Time\r
            date = item.pubDate.strftime("%Y/%m/%d %H:%M")\r
          else\r
            date = item.pubDate.to_s\r
          end\r
        elsif item.respond_to?(:date)\r
          if item.date.class <= Time\r
            date = item.date.strftime("%Y/%m/%d %H:%M")\r
          else\r
            date = item.date.to_s\r
          end\r
        else\r
          date = "(no date)"\r
        end\r
        date += " :: "\r
      end\r
    end\r
\r
    tit_opt = {}\r
    # Twitters don't need a cap on the title length since they have a hard\r
    # limit to 160 characters, and most of them are under 140 characters\r
    tit_opt[:limit] = @bot.config['rss.head_max'] unless feed.type == 'twitter'\r
\r
    if item.title\r
      base_title = item.title.to_s.dup\r
      # git changesets are SHA1 hashes (40 hex digits), way too long, get rid of them, as they are\r
      # visible in the URL anyway\r
      # TODO make this optional?\r
      base_title.sub!(/^Changeset \[([\da-f]{40})\]:/) { |c| "(git commit)"} if feed.type == 'trac'\r
      title = "#{Bold}#{base_title.ircify_html(tit_opt)}#{Bold}"\r
    end\r
\r
    desc_opt = {}\r
    desc_opt[:limit] = @bot.config['rss.text_max']\r
    desc_opt[:a_href] = :link_out if @bot.config['rss.show_links']\r
\r
    # We prefer content_encoded here as it tends to provide more html formatting \r
    # for use with ircify_html.\r
    if item.respond_to?(:content_encoded) && item.content_encoded\r
      desc = item.content_encoded.ircify_html(desc_opt)\r
    elsif item.respond_to?(:description) && item.description\r
      desc = item.description.ircify_html(desc_opt)\r
    else\r
      if item.content.type == "html"\r
        desc = item.content.content.ircify_html(desc_opt)\r
      else\r
        desc = item.content.content\r
        if desc.size > desc_opt[:limit]\r
          desc = desc.slice(0, desc_opt[:limit]) + "#{Reverse}...#{Reverse}"\r
        end\r
      end\r
    end\r
\r
    link = item.link.href rescue item.link.chomp rescue nil\r
\r
    category = select_nonempty((item.category.content rescue nil), (item.dc_subject rescue nil))\r
    author = select_nonempty((item.author.name.content rescue nil), (item.dc_creator rescue nil), (item.author rescue nil))\r
\r
    line1 = nil\r
    line2 = nil\r
\r
    at = ((item.title && item.link) ? ' @ ' : '')\r
\r
    case feed.type\r
    when 'blog'\r
      author += " " if author\r
      abt = category ? "about #{category} " : ""\r
      line1 = "#{handle}#{date}#{author}blogged #{abt}at #{link}"\r
      line2 = "#{handle}#{title} - #{desc}"\r
    when 'git'\r
      author += " " if author\r
      line1 = "#{handle}#{date}#{author}commited #{title} @ #{link}"\r
    when 'forum'\r
      line1 = "#{handle}#{date}#{title}#{at}#{link}"\r
    when 'wiki'\r
      line1 = "#{handle}#{date}#{title}#{at}#{link} has been edited by #{author}. #{desc}"\r
    when 'gmane'\r
      line1 = "#{handle}#{date}Message #{title} sent by #{author}. #{desc}"\r
    when 'trac'\r
      line1 = "#{handle}#{date}#{title} @ #{link}"\r
      unless item.title =~ /^(?:Changeset \[(?:[\da-f]+)\]|\(git commit\))/\r
        line2 = "#{handle}#{date}#{desc}"\r
      end\r
    when '/.'\r
      dept = "(from the #{item.slash_department} dept) " rescue nil\r
      sec = " in section #{item.slash_section}" rescue nil\r
\r
      line1 = "#{handle}#{date}#{dept}#{title}#{at}#{link} (posted by #{author}#{sec})"\r
    else\r
      line1 = "#{handle}#{date}#{title}#{at}#{link}"\r
      line1 << " (by #{author})" if author\r
    end\r
    places.each { |loc|\r
      @bot.say loc, line1, :overlong => :truncate\r
      next unless line2\r
      @bot.say loc, line2, :overlong => :truncate\r
    }\r
  end\r
\r
  def fetchRss(feed, m=nil, cache=true)\r
    feed.last_fetched = Time.now\r
    begin\r
      # Use 60 sec timeout, cause the default is too low\r
      xml = @bot.httputil.get(feed.url,\r
                              :read_timeout => 60,\r
                              :open_timeout => 60,\r
                              :cache => cache)\r
    rescue URI::InvalidURIError, URI::BadURIError => e\r
      report_problem("invalid rss feed #{feed.url}", e, m)\r
      return nil\r
    rescue => e\r
      report_problem("error getting #{feed.url}", e, m)\r
      return nil\r
    end\r
    debug "fetched #{feed}"\r
    unless xml\r
      report_problem("reading feed #{feed} failed", nil, m)\r
      return nil\r
    end\r
    # Ok, 0.9 feeds are not supported, maybe because\r
    # Netscape happily removed the DTD. So what we do is just to\r
    # reassign the 0.9 RDFs to 1.0, and hope it goes right.\r
    xml.gsub!("xmlns=\"http://my.netscape.com/rdf/simple/0.9/\"",\r
              "xmlns=\"http://purl.org/rss/1.0/\"")\r
    feed.mutex.synchronize do\r
      feed.xml = xml\r
    end\r
    return true\r
  end\r
\r
  def parseRss(feed, m=nil)\r
    return nil unless feed.xml\r
    feed.mutex.synchronize do\r
      xml = feed.xml\r
      begin\r
        ## do validate parse\r
        rss = RSS::Parser.parse(xml)\r
        debug "parsed and validated #{feed}"\r
      rescue RSS::InvalidRSSError\r
        ## do non validate parse for invalid RSS 1.0\r
        begin\r
          rss = RSS::Parser.parse(xml, false)\r
          debug "parsed but not validated #{feed}"\r
        rescue RSS::Error => e\r
          report_problem("parsing rss stream failed, whoops =(", e, m)\r
          return nil\r
        end\r
      rescue RSS::Error => e\r
        report_problem("parsing rss stream failed, oioi", e, m)\r
        return nil\r
      rescue => e\r
        report_problem("processing error occured, sorry =(", e, m)\r
        return nil\r
      end\r
      items = []\r
      if rss.nil?\r
        report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)\r
      else\r
        begin\r
          rss.output_encoding = 'UTF-8'\r
        rescue RSS::UnknownConvertMethod => e\r
          report_problem("bah! something went wrong =(", e, m)\r
          return nil\r
        end\r
        if rss.respond_to? :channel\r
          rss.channel.title ||= "Unknown"\r
          title = rss.channel.title\r
        else\r
          title = rss.title.content\r
        end\r
        rss.items.each do |item|\r
          item.title ||= "Unknown"\r
          items << item\r
        end\r
      end\r
\r
      if items.empty?\r
        report_problem("no items found in the feed, maybe try weed?", e, m)\r
        return nil\r
      end\r
      feed.title = title\r
      feed.items = items\r
      return true\r
    end\r
  end\r
end\r
\r
plugin = RSSFeedsPlugin.new\r
\r
plugin.default_auth( 'edit', false )\r
plugin.default_auth( 'edit:add', true)\r
\r
plugin.map 'rss show :handle :limit',\r
  :action => 'show_rss',\r
  :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},\r
  :defaults => {:limit => 5}\r
plugin.map 'rss list :handle',\r
  :action => 'list_rss',\r
  :defaults => {:handle => nil}\r
plugin.map 'rss watched :handle [in :chan]',\r
  :action => 'watched_rss',\r
  :defaults => {:handle => nil}\r
plugin.map 'rss who watches :handle',\r
  :action => 'who_watches',\r
  :defaults => {:handle => nil}\r
plugin.map 'rss add :handle :url :type',\r
  :action => 'add_rss',\r
  :auth_path => 'edit',\r
  :defaults => {:type => nil}\r
plugin.map 'rss change :what of :handle to :new',\r
  :action => 'change_rss',\r
  :auth_path => 'edit',\r
  :requirements => { :what => /handle|url|format|type|refresh/ }\r
plugin.map 'rss change :what for :handle to :new',\r
  :action => 'change_rss',\r
  :auth_path => 'edit',\r
  :requirements => { :what => /handle|url|format|type|refesh/ }\r
plugin.map 'rss del :handle',\r
  :auth_path => 'edit:rm!',\r
  :action => 'del_rss'\r
plugin.map 'rss delete :handle',\r
  :auth_path => 'edit:rm!',\r
  :action => 'del_rss'\r
plugin.map 'rss rm :handle',\r
  :auth_path => 'edit:rm!',\r
  :action => 'del_rss'\r
plugin.map 'rss replace :handle :url :type',\r
  :auth_path => 'edit',\r
  :action => 'replace_rss',\r
  :defaults => {:type => nil}\r
plugin.map 'rss forcereplace :handle :url :type',\r
  :auth_path => 'edit',\r
  :action => 'forcereplace_rss',\r
  :defaults => {:type => nil}\r
plugin.map 'rss watch :handle [in :chan]',\r
  :action => 'watch_rss',\r
  :defaults => {:url => nil, :type => nil}\r
plugin.map 'rss watch :handle :url :type [in :chan]',\r
  :action => 'watch_rss',\r
  :defaults => {:url => nil, :type => nil}\r
plugin.map 'rss unwatch :handle [in :chan]',\r
  :action => 'unwatch_rss'\r
plugin.map 'rss rmwatch :handle [in :chan]',\r
  :action => 'unwatch_rss'\r
plugin.map 'rss rewatch [:handle]',\r
  :action => 'rewatch_rss'\r