#-- vim:sw=2:et\r
#++\r
#\r
-# RSS feed plugin for RubyBot\r
-# (c) 2004 Stanislav Karchebny <berkus@madfire.net>\r
-# (c) 2005 Ian Monroe <ian@monroe.nu>\r
-# (c) 2005 Mark Kretschmann <markey@web.de>\r
-# (c) 2006 Giuseppe Bilotta <giuseppe.bilotta@gmail.com>\r
+# :title: RSS feed plugin for rbot\r
#\r
-# Licensed under MIT License.\r
-\r
-require 'rss/parser'\r
-require 'rss/1.0'\r
-require 'rss/2.0'\r
-require 'rss/dublincore'\r
-# begin\r
-# require 'rss/dublincore/2.0'\r
-# rescue\r
-# warning "Unable to load RSS libraries, RSS plugin functionality crippled"\r
-# end\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
+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 = options[:show_updated] ? item.description : nil\r
+ [item.title, item.link, 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\r
end\r
\r
def dup\r
- self.class.new(@url,\r
- @handle,\r
- @type ? @type.dup : nil,\r
- @watchers.dup,\r
- @xml ? @xml.dup : nil)\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
+ end\r
end\r
\r
# Downcase all watchers, possibly turning them into Strings if they weren't\r
if watched_by?(who)\r
return nil\r
end\r
- @watchers << who.downcase\r
+ @mutex.synchronize do\r
+ @watchers << who.downcase\r
+ end\r
return who\r
end\r
\r
def rm_watch(who)\r
- @watchers.delete(who.downcase)\r
+ @mutex.synchronize do\r
+ @watchers.delete(who.downcase)\r
+ end\r
end\r
\r
def to_a\r
end\r
\r
class RSSFeedsPlugin < Plugin\r
- BotConfig.register BotConfigIntegerValue.new('rss.head_max',\r
- :default => 30, :validate => Proc.new{|v| v > 0 && v < 200},\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
- BotConfig.register BotConfigIntegerValue.new('rss.text_max',\r
- :default => 90, :validate => Proc.new{|v| v > 0 && v < 400},\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
- BotConfig.register BotConfigIntegerValue.new('rss.thread_sleep',\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
+ # 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 unless f.mutex\r
+ f.mutex = Mutex.new\r
f.sanitize_watchers\r
parseRss(f) if f.xml\r
}\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
- f.mutex.synchronize do\r
- unparsed[k] = f.dup\r
- end\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
debug "Stopping watch #{handle}"\r
@bot.timer.remove(@watch[handle])\r
@watch.delete(handle)\r
- rescue => e\r
+ rescue Exception => e\r
report_problem("Failed to stop watch for #{handle}", e, nil)\r
end\r
end\r
\r
m.reply "lemme fetch it..."\r
title = items = nil\r
- fetched = fetchRss(feed, m)\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
reply = "no feeds found"\r
reply << " matching #{wanted}" if wanted\r
end\r
- m.reply reply\r
+ m.reply reply, :max_lines => reply.length\r
end\r
\r
def watched_rss(m, params)\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
status = Hash.new\r
status[:failures] = 0\r
- @watch[feed.handle] = @bot.timer.add(0, status) {\r
+ status[:first_run] = true\r
+ @watch[feed.handle] = @bot.timer.add(0) {\r
debug "watcher for #{feed} started"\r
failures = status[:failures]\r
+ first_run = status.delete(:first_run)\r
begin\r
debug "fetching #{feed}"\r
oldxml = feed.xml ? feed.xml.dup : nil\r
unless fetchRss(feed)\r
failures += 1\r
else\r
- if oldxml and oldxml == feed.xml\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
parseRss(feed)\r
failures -= 1 if failures > 0\r
else\r
- otxt = feed.items.map { |item| item.to_s }\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
- otxt.include?(item.to_s)\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
\r
status[:failures] = failures\r
\r
+ timer = nil\r
+ seconds = @bot.config['rss.thread_sleep']\r
feed.mutex.synchronize do\r
- seconds = (feed.refresh_rate || @bot.config['rss.thread_sleep']) * (failures + 1)\r
- seconds += seconds * (rand(100)-50)/100\r
- debug "watcher for #{feed} going to sleep #{seconds} seconds.."\r
- @bot.timer.reschedule(@watch[feed.handle], seconds)\r
+ timer = @watch[feed.handle]\r
+ seconds = feed.refresh_rate if feed.refresh_rate\r
end\r
+ seconds *= failures + 1\r
+ seconds += seconds * (rand(100)-50)/100\r
+ debug "watcher for #{feed} going to sleep #{seconds} seconds.."\r
+ @bot.timer.reschedule(timer, seconds) rescue warning "watcher for #{feed} failed to reschedule: #{$!.inspect}"\r
}\r
debug "watcher for #{feed} added"\r
end\r
if opts.key?(:date) && opts[:date]\r
if item.respond_to?(:pubDate) \r
if item.pubDate.class <= Time\r
- date = item.pubDate.strftime("%Y/%m/%d %H.%M.%S")\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.%S")\r
+ date = item.date.strftime("%Y/%m/%d %H:%M")\r
else\r
date = item.date.to_s\r
end\r
date += " :: "\r
end\r
end\r
- title = "#{Bold}#{item.title.chomp.riphtml}#{Bold}" if item.title\r
- desc = item.description.gsub(/\s+/,' ').strip.riphtml if item.description\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
+ title = "#{Bold}#{item.title.ircify_html(tit_opt)}#{Bold}" if item.title\r
+\r
+ desc_opt = {\r
+ :limit => @bot.config['rss.text_max'],\r
+ :a_href => :link_out\r
+ }\r
+ desc = item.description.ircify_html(desc_opt) if item.description\r
+\r
link = item.link.chomp if item.link\r
+\r
+ debug item.inspect\r
+ category = item.dc_subject rescue item.category.content rescue nil\r
+ category = nil if category and category.empty?\r
+ author = item.dc_creator rescue item.author rescue nil\r
+ author = nil if author and author.empty?\r
+\r
line1 = nil\r
line2 = nil\r
+\r
+ at = ((item.title && item.link) ? ' @ ' : '')\r
case feed.type\r
when 'blog'\r
- line1 = "#{handle}#{date}#{item.category.content} blogged at #{link}"\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 'forum'\r
- line1 = "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"\r
+ line1 = "#{handle}#{date}#{title}#{at}#{link}"\r
when 'wiki'\r
- line1 = "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link} has been edited by #{item.dc_creator}. #{desc}"\r
+ line1 = "#{handle}#{date}#{title}#{at}#{link} has been edited by #{author}. #{desc}"\r
when 'gmane'\r
- line1 = "#{handle}#{date}Message #{title} sent by #{item.dc_creator}. #{desc}"\r
+ line1 = "#{handle}#{date}Message #{title} sent by #{author}. #{desc}"\r
when 'trac'\r
line1 = "#{handle}#{date}#{title} @ #{link}"\r
unless item.title =~ /^Changeset \[(\d+)\]/\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}#{' @ ' if item.title && item.link}#{link}"\r
+ line1 = "#{handle}#{date}#{title}#{at}#{link}"\r
end\r
places.each { |loc|\r
@bot.say loc, line1, :overlong => :truncate\r
}\r
end\r
\r
- def fetchRss(feed, m=nil)\r
+ def fetchRss(feed, m=nil, cache=true)\r
begin\r
# Use 60 sec timeout, cause the default is too low\r
- xml = @bot.httputil.get_cached(feed.url, 60, 60)\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
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
begin\r
## do validate parse\r
rss = RSS::Parser.parse(xml)\r
- debug "parsed #{feed}"\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
plugin.map 'rss forcereplace :handle :url :type',\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