4 # :title: RSS feed plugin for rbot
6 # Author:: Stanislav Karchebny <berkus@madfire.net>
7 # Author:: Ian Monroe <ian@monroe.nu>
8 # Author:: Mark Kretschmann <markey@web.de>
9 # Author:: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
11 # Copyright:: (C) 2004 Stanislav Karchebny
12 # Copyright:: (C) 2005 Ian Monroe, Mark Kretschmann
13 # Copyright:: (C) 2006-2007 Giuseppe Bilotta
15 # License:: MIT license
19 # Try to load rss/content/2.0 so we can access the data in <content:encoded>
22 require 'rss/content/2.0'
28 # Add support for Slashdot namespace in RDF. The code is just an adaptation
29 # of the DublinCore code.
30 unless defined?(SLASH_PREFIX)
31 SLASH_PREFIX = 'slash'
32 SLASH_URI = "http://purl.org/rss/1.0/modules/slash/"
34 RDF.install_ns(SLASH_PREFIX, SLASH_URI)
37 def append_features(klass)
40 return if klass.instance_of?(Module)
41 SlashModel::ELEMENT_NAME_INFOS.each do |name, plural_name|
42 plural = plural_name || "#{name}s"
43 full_name = "#{SLASH_PREFIX}_#{name}"
44 full_plural_name = "#{SLASH_PREFIX}_#{plural}"
45 klass_name = "Slash#{Utils.to_class_name(name)}"
47 # This will fail with older version of the Ruby RSS module
49 klass.install_have_children_element(name, SLASH_URI, "*",
50 full_name, full_plural_name)
51 klass.install_must_call_validator(SLASH_PREFIX, SLASH_URI)
53 klass.module_eval("install_have_children_element(#{full_name.dump}, #{full_plural_name.dump})")
56 klass.module_eval(<<-EOC, *get_file_and_line_from_caller(0))
57 remove_method :#{full_name} if method_defined? :#{full_name}
58 remove_method :#{full_name}= if method_defined? :#{full_name}=
59 remove_method :set_#{full_name} if method_defined? :set_#{full_name}
62 @#{full_name}.first and @#{full_name}.first.value
65 def #{full_name}=(new_value)
66 @#{full_name}[0] = Utils.new_with_value_if_need(#{klass_name}, new_value)
68 alias set_#{full_name} #{full_name}=
85 ELEMENT_NAME_INFOS = SlashModel::TEXT_ELEMENTS.to_a
87 ELEMENTS = TEXT_ELEMENTS.keys
89 ELEMENTS.each do |name, plural_name|
90 module_eval(<<-EOC, *get_file_and_line_from_caller(0))
91 class Slash#{Utils.to_class_name(name)} < Element
106 @tag_name = #{name.dump}
108 alias_method(:value, :content)
109 alias_method(:value=, :content=)
111 def initialize(*args)
113 if Utils.element_initialize_arguments?(args)
117 self.content = args[0]
119 # Older Ruby RSS module
122 self.content = args[0]
127 tag_name_with_prefix(SLASH_PREFIX)
130 def maker_target(target)
134 def setup_maker_attributes(#{name})
135 #{name}.content = content
143 class Item; include SlashModel; end
146 SlashModel::ELEMENTS.each do |name|
147 class_name = Utils.to_class_name(name)
148 BaseListener.install_class_name(SLASH_URI, name, "Slash#{class_name}")
151 SlashModel::ELEMENTS.collect! {|name| "#{SLASH_PREFIX}_#{name}"}
157 attr_accessor :url, :handle, :type, :refresh_rate, :xml, :title, :items,
158 :mutex, :watchers, :last_fetched
160 def initialize(url,handle=nil,type=nil,watchers=[], xml=nil, lf = nil)
175 sanitize_watchers(watchers)
179 @mutex.synchronize do
182 @type ? @type.dup : nil,
184 @xml ? @xml.dup : nil,
189 # Downcase all watchers, possibly turning them into Strings if they weren't
190 def sanitize_watchers(list=@watchers)
203 @watchers.include?(who.downcase)
210 @mutex.synchronize do
211 @watchers << who.downcase
217 @mutex.synchronize do
218 @watchers.delete(who.downcase)
223 [@handle,@url,@type,@refresh_rate,@watchers]
226 def to_s(watchers=false)
228 a = self.to_a.flatten
232 a.compact.join(" | ")
236 class RSSFeedsPlugin < Plugin
237 Config.register Config::IntegerValue.new('rss.head_max',
238 :default => 100, :validate => Proc.new{|v| v > 0 && v < 200},
239 :desc => "How many characters to use of a RSS item header")
241 Config.register Config::IntegerValue.new('rss.text_max',
242 :default => 200, :validate => Proc.new{|v| v > 0 && v < 400},
243 :desc => "How many characters to use of a RSS item text")
245 Config.register Config::IntegerValue.new('rss.thread_sleep',
246 :default => 300, :validate => Proc.new{|v| v > 30},
247 :desc => "How many seconds to sleep before checking RSS feeds again")
249 Config.register Config::BooleanValue.new('rss.show_updated',
251 :desc => "Whether feed items for which the description was changed should be shown as new")
253 Config.register Config::BooleanValue.new('rss.show_links',
255 :desc => "Whether to display links from the text of a feed item.")
257 # Make an 'unique' ID for a given item, based on appropriate bot options
258 # Currently only suppored is bot.config['rss.show_updated']: when false,
259 # only the guid/link is accounted for.
261 def block_rescue(df = nil, &block)
262 v = block.call rescue nil
263 (String === v && '' != v) ? v : nil
268 (block_rescue do item.guid.content end ||
269 block_rescue do item.guid end ||
270 block_rescue do item.link.href end ||
271 block_rescue do item.link end
274 if @bot.config['rss.show_updated']
276 block_rescue do item.content.content end ||
277 block_rescue do item.description end
280 block_rescue do item.title.content end ||
281 block_rescue do item.title end
284 # debug "taking hash of #{uid.inspect}"
289 # We used to save the Mutex with the RssBlob, which was idiotic. And
290 # since Mutexes dumped in one version might not be resotrable in another,
291 # we need a few tricks to be able to restore data from other versions of Ruby
293 # When migrating 1.8.6 => 1.8.5, all we need to do is define an empty
294 # #marshal_load() method for Mutex. For 1.8.5 => 1.8.6 we need something
295 # dirtier, as seen later on in the initialization code.
296 unless Mutex.new.respond_to?(:marshal_load)
298 def marshal_load(str)
304 # Auxiliary method used to collect two lines for rss output filters,
305 # running substitutions against DataStream _s_ optionally joined
307 def make_stream(line1, line2, s, h={})
309 DataStream.new([line1, line2].compact.join("\n") % ss, ss)
312 # Define default RSS filters
314 # TODO: load personal ones
317 @bot.register_filter(:headlines, @outkey) { |s|
318 line1 = (s[:handle].empty? ? "%{date}" : "%{handle}") << "%{title}"
319 make_stream(line1, nil, s)
321 @bot.register_filter(:blog, @outkey) { |s|
322 author = s[:author] ? (s[:author] + " ") : ""
323 abt = s[:category] ? "about #{s[:category]} " : ""
324 line1 = "%{handle}%{date}%{author}blogged %{abt}at %{link}"
325 line2 = "%{handle}%{title} - %{desc}"
326 make_stream(line1, line2, s, :author => author, :abt => abt)
328 @bot.register_filter(:photoblog, @outkey) { |s|
329 author = s[:author] ? (s[:author] + " ") : ""
330 abt = s[:category] ? "under #{s[:category]} " : ""
331 line1 = "%{handle}%{date}%{author}added an image %{abt}at %{link}"
332 line2 = "%{handle}%{title} - %{desc}"
333 make_stream(line1, line2, s, :author => author, :abt => abt)
335 @bot.register_filter(:news, @outkey) { |s|
336 line1 = "%{handle}%{date}%{title}%{at}%{link}" % s
337 line2 = "%{handle}%{date}%{desc}" % s
338 make_stream(line1, line2, s)
340 @bot.register_filter(:git, @outkey) { |s|
341 author = s[:author] ? (s[:author] + " ") : ""
342 line1 = "%{handle}%{date}%{author}committed %{title}%{at}%{link}"
343 make_stream(line1, nil, s, :author => author)
345 @bot.register_filter(:forum, @outkey) { |s|
346 line1 = "%{handle}%{date}%{title}%{at}%{link}"
347 make_stream(line1, nil, s)
349 @bot.register_filter(:wiki, @outkey) { |s|
350 line1 = "%{handle}%{date}%{title}%{at}%{link}"
351 line1 << "has been edited by %{author}. %{desc}"
352 make_stream(line1, nil, s)
354 @bot.register_filter(:gmane, @outkey) { |s|
355 line1 = "%{handle}%{date}Message %{title} sent by %{author}. %{desc}"
356 make_stream(line1, nil, s)
358 @bot.register_filter(:trac, @outkey) { |s|
359 author = s[:author].sub(/@\S+?\s*>/, "@...>") + ": " if s[:author]
360 line1 = "%{handle}%{date}%{author}%{title}%{at}%{link}"
362 unless s[:item].title =~ /^(?:Changeset \[(?:[\da-f]+)\]|\(git commit\))/
363 line2 = "%{handle}%{date}%{desc}"
365 make_stream(line1, line2, s, :author => author)
367 @bot.register_filter(:"/.", @outkey) { |s|
368 dept = "(from the #{s[:item].slash_department} dept) " rescue nil
369 sec = " in section #{s[:item].slash_section}" rescue nil
370 line1 = "%{handle}%{date}%{dept}%{title}%{at}%{link} "
371 line1 << "(posted by %{author}%{sec})"
372 make_stream(line1, nil, s, :dept => dept, :sec => sec)
374 @bot.register_filter(:default, @outkey) { |s|
375 line1 = "%{handle}%{date}%{title}%{at}%{link}"
376 line1 << " (by %{author})" if s[:author]
377 make_stream(line1, nil, s)
380 # Define an HTML info filter too
381 @bot.register_filter(:rss, :htmlinfo) { |s| htmlinfo_filter(s) }
383 # This is the output format used by the input filter
384 @bot.register_filter(:htmlinfo, @outkey) { |s|
385 line1 = "%{title}%{at}%{link}"
386 make_stream(line1, nil, s)
390 FEED_NS = %r{xmlns.*http://(purl\.org/rss|www.w3c.org/1999/02/22-rdf)}
391 def htmlinfo_filter(s)
392 return nil unless s[:headers] and s[:headers]['x-rbot-location']
393 return nil unless s[:headers]['content-type'].first.match(/xml|rss|atom|rdf/i) or
394 (s[:text].include?("<rdf:RDF") and s[:text].include?("<channel")) or
395 s[:text].include?("<rss") or s[:text].include?("<feed") or
396 s[:text].match(FEED_NS)
397 blob = RssBlob.new(s[:headers]['x-rbot-location'],"", :htmlinfo)
398 unless (fetchRss(blob, nil) and parseRss(blob, nil) rescue nil)
399 debug "#{s.pretty_inspect} is not an RSS feed, despite the appearances"
403 blob.items.each { |it|
404 output << printFormattedRss(blob, it)[:text]
406 return {:title => blob.title, :content => output.join(" | ")}
409 # Display the known rss types
410 def rss_types(m, params)
411 ar = @bot.filter_names(@outkey)
413 m.reply ar.map { |k| k.to_s }.sort!.join(", ")
423 if @registry.has_key?(:feeds)
424 # When migrating from Ruby 1.8.5 to 1.8.6, dumped Mutexes may render the
425 # data unrestorable. If this happens, we patch the data, thus allowing
426 # the restore to work.
428 # This is actually pretty safe for a number of reasons:
429 # * the code is only called if standard marshalling fails
430 # * the string we look for is quite unlikely to appear randomly
431 # * if the string appears somewhere and the patched string isn't recoverable
432 # either, we'll get another (unrecoverable) error, which makes the rss
433 # plugin unsable, just like it was if no recovery was attempted
434 # * if the string appears somewhere and the patched string is recoverable,
435 # we may get a b0rked feed, which is eventually overwritten by a clean
436 # one, so the worst thing that can happen is that a feed update spams
438 @registry.recovery = Proc.new { |val|
439 patched = val.sub(":\v@mutexo:\nMutex", ":\v@mutexo:\vObject")
440 ret = Marshal.restore(patched)
441 ret.each_value { |blob|
447 @feeds = @registry[:feeds]
450 @registry.recovery = nil
452 @feeds.keys.grep(/[A-Z]/) { |k|
453 @feeds[k.downcase] = @feeds[k]
473 @feeds.select { |h, f| f.watched? }
482 unparsed = Hash.new()
485 # we don't want to save the mutex
486 unparsed[k].mutex = nil
488 @registry[:feeds] = unparsed
491 def stop_watch(handle)
492 if @watch.has_key?(handle)
494 debug "Stopping watch #{handle}"
495 @bot.timer.remove(@watch[handle])
496 @watch.delete(handle)
497 rescue Exception => e
498 report_problem("Failed to stop watch for #{handle}", e, nil)
504 @watch.each_key { |k|
509 def help(plugin,topic="")
512 "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"
514 "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
516 "rss watched [#{Bold}handle#{Bold}] [in #{Bold}chan#{Bold}]: list all watched rss feeds (matching #{Bold}handle#{Bold}) (in channel #{Bold}chan#{Bold})"
517 when "who", "watches", "who watches"
518 "rss who watches [#{Bold}handle#{Bold}]]: list all watchers for rss feeds (matching #{Bold}handle#{Bold})"
520 "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})"
522 "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}"
523 when /^(del(ete)?|rm)$/
524 "rss del(ete)|rm #{Bold}handle#{Bold} : delete rss feed #{Bold}handle#{Bold}"
526 "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"
528 "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})"
530 "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"
532 "rss unwatch|rmwatch #{Bold}handle#{Bold} [in #{Bold}chan#{Bold}]: stop watching rss #{Bold}handle#{Bold} (in channel #{Bold}chan#{Bold}) for changes"
533 when /who(?: watche?s?)?/
534 "rss who watches #{Bold}handle#{Bold}: lists watches for rss #{Bold}handle#{Bold}"
536 "rss rewatch : restart threads that watch for changes in watched rss"
538 "rss types : show the rss types for which an output format existi (all other types will use the default one)"
540 "manage RSS feeds: rss types|show|list|watched|add|change|del(ete)|rm|(force)replace|watch|unwatch|rmwatch|rewatch|who watches"
544 def report_problem(report, e=nil, m=nil)
545 if m && m.respond_to?(:reply)
552 debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
556 def show_rss(m, params)
557 handle = params[:handle]
558 lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
559 debug lims.to_a.inspect
561 ll = [[lims[1].to_i-1,lims[2].to_i-1].min, 0].max
562 ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
563 rev = lims[1].to_i > lims[2].to_i
566 ul = [[lims[1].to_i-1, 0].max, 14].min
570 feed = @feeds.fetch(handle.downcase, nil)
572 m.reply "I don't know any feeds named #{handle}"
576 m.reply "lemme fetch it..."
578 we_were_watching = false
580 if @watch.key?(feed.handle)
581 # If a feed is being watched, we run the watcher thread
582 # so that all watchers can be informed of changes to
583 # the feed. Before we do that, though, we remove the
584 # show requester from the watchlist, if present, lest
585 # he gets the update twice.
586 if feed.watched_by?(m.replyto)
587 we_were_watching = true
588 feed.rm_watch(m.replyto)
590 @bot.timer.reschedule(@watch[feed.handle], 0)
592 feed.add_watch(m.replyto)
595 fetched = fetchRss(feed, m, false)
597 return unless fetched or feed.xml
598 if fetched or not feed.items
599 parsed = parseRss(feed, m)
601 return unless feed.items
602 m.reply "using old data" unless fetched and parsed
607 # We sort the feeds in freshness order (newer ones first)
608 items = freshness_sort(items)
612 m.reply "Channel : #{title}"
614 printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
618 def itemDate(item,ex=nil)
619 return item.pubDate if item.respond_to?(:pubDate) and item.pubDate
620 return item.date if item.respond_to?(:date) and item.date
624 def freshness_sort(items)
627 itemDate(b, notime) <=> itemDate(a, notime)
631 def list_rss(m, params)
632 wanted = params[:handle]
634 @feeds.each { |handle, feed|
635 next if wanted and !handle.match(/#{wanted}/i)
636 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
637 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
638 (reply << " (watched)") if feed.watched_by?(m.replyto)
642 reply = "no feeds found"
643 reply << " matching #{wanted}" if wanted
645 m.reply reply, :max_lines => reply.length
648 def watched_rss(m, params)
649 wanted = params[:handle]
650 chan = params[:chan] || m.replyto
652 watchlist.each { |handle, feed|
653 next if wanted and !handle.match(/#{wanted}/i)
654 next unless feed.watched_by?(chan)
655 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
656 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
660 reply = "no watched feeds"
661 reply << " matching #{wanted}" if wanted
666 def who_watches(m, params)
667 wanted = params[:handle]
669 watchlist.each { |handle, feed|
670 next if wanted and !handle.match(/#{wanted}/i)
671 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
672 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
673 reply << ": watched by #{feed.watchers.join(', ')}"
677 reply = "no watched feeds"
678 reply << " matching #{wanted}" if wanted
683 def add_rss(m, params, force=false)
684 handle = params[:handle]
686 unless url.match(/https?/)
687 m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
691 if @feeds.fetch(handle.downcase, nil) && !force
692 m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"
696 m.reply "You must specify both a handle and an url to add an RSS feed"
699 @feeds[handle.downcase] = RssBlob.new(url,handle,type)
700 reply = "Added RSS #{url} named #{handle}"
702 reply << " (format: #{type})"
708 def change_rss(m, params)
709 handle = params[:handle].downcase
710 feed = @feeds.fetch(handle, nil)
712 m.reply "No such feed with handle #{handle}"
715 case params[:what].intern
717 new = params[:new].downcase
718 if @feeds.key?(new) and @feeds[new]
719 m.reply "There already is a feed with handle #{new}"
722 feed.mutex.synchronize do
724 @feeds.delete(handle)
731 feed.mutex.synchronize do
736 new = nil if new == 'default'
737 feed.mutex.synchronize do
741 new = params[:new].to_i
742 new = nil if new == 0
743 feed.mutex.synchronize do
744 feed.refresh_rate = new
747 m.reply "Don't know how to change #{params[:what]} for feeds"
750 m.reply "Feed changed:"
751 list_rss(m, {:handle => handle})
754 def del_rss(m, params, pass=false)
755 feed = unwatch_rss(m, params, true)
758 m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
761 @feeds.delete(feed.handle.downcase)
766 def replace_rss(m, params)
767 handle = params[:handle]
768 if @feeds.key?(handle.downcase)
769 del_rss(m, {:handle => handle}, true)
771 if @feeds.key?(handle.downcase)
772 m.reply "can't replace #{feed.handle}"
774 add_rss(m, params, true)
778 def forcereplace_rss(m, params)
779 add_rss(m, params, true)
782 def watch_rss(m, params)
783 handle = params[:handle]
784 chan = params[:chan] || m.replyto
790 feed = @feeds.fetch(handle.downcase, nil)
792 if feed.add_watch(chan)
796 m.reply "Already watching #{feed.handle} in #{chan}"
799 m.reply "Couldn't watch feed #{handle} (no such feed found)"
803 def unwatch_rss(m, params, pass=false)
804 handle = params[:handle].downcase
805 chan = params[:chan] || m.replyto
806 unless @feeds.has_key?(handle)
807 m.reply("dunno that feed")
810 feed = @feeds[handle]
811 if feed.rm_watch(chan)
812 m.reply "#{chan} has been removed from the watchlist for #{feed.handle}"
814 m.reply("#{chan} wasn't watching #{feed.handle}") unless pass
822 def rewatch_rss(m=nil, params=nil)
823 if params and handle = params[:handle]
824 feed = @feeds.fetch(handle.downcase, nil)
826 @bot.timer.reschedule(@watch[feed.handle], (params[:delay] || 0).to_f)
829 m.reply _("no such feed %{handle}") % { :handle => handle } if m
834 # Read watches from list.
835 watchlist.each{ |handle, feed|
843 def watchRss(feed, m=nil)
844 if @watch.has_key?(feed.handle)
845 report_problem("watcher thread for #{feed.handle} is already running", nil, m)
849 status[:failures] = 0
852 tmout = feed.last_fetched + calculate_timeout(feed) - Time.now
853 tmout = 0 if tmout < 0
855 debug "scheduling a watcher for #{feed} in #{tmout} seconds"
856 @watch[feed.handle] = @bot.timer.add(tmout) {
857 debug "watcher for #{feed} wakes up"
858 failures = status[:failures]
860 debug "fetching #{feed}"
861 first_run = !feed.last_fetched
862 oldxml = feed.xml ? feed.xml.dup : nil
863 unless fetchRss(feed)
867 debug "first run for #{feed}, getting items"
869 elsif oldxml and oldxml == feed.xml
870 debug "xml for #{feed} didn't change"
871 failures -= 1 if failures > 0
874 debug "no previous items in feed #{feed}"
876 failures -= 1 if failures > 0
878 # This one is used for debugging
881 # These are used for checking new items vs old ones
882 oids = Set.new feed.items.map { |item|
885 debug [uid, item].inspect
886 debug [uid, otxt.last].inspect
890 unless parseRss(feed)
891 debug "no items in feed #{feed}"
894 debug "Checking if new items are available for #{feed}"
895 failures -= 1 if failures > 0
901 dispItems = feed.items.reject { |item|
904 if oids.include?(uid)
905 debug "rejecting old #{uid} #{item.inspect}"
906 debug [uid, txt].inspect
909 debug "accepting new #{uid} #{item.inspect}"
910 debug [uid, txt].inspect
911 warning "same text! #{txt}" if otxt.include?(txt)
916 if dispItems.length > 0
917 debug "Found #{dispItems.length} new items in #{feed}"
918 # When displaying watched feeds, publish them from older to newer
919 dispItems.reverse.each { |item|
920 printFormattedRss(feed, item)
923 debug "No new items found in #{feed}"
929 rescue Exception => e
930 error "Error watching #{feed}: #{e.inspect}"
931 debug e.backtrace.join("\n")
935 status[:failures] = failures
937 seconds = calculate_timeout(feed, failures)
938 debug "watcher for #{feed} going to sleep #{seconds} seconds.."
940 @bot.timer.reschedule(@watch[feed.handle], seconds)
942 warning "watcher for #{feed} failed to reschedule: #{$!.inspect}"
945 debug "watcher for #{feed} added"
948 def calculate_timeout(feed, failures = 0)
949 seconds = @bot.config['rss.thread_sleep']
950 feed.mutex.synchronize do
951 seconds = feed.refresh_rate if feed.refresh_rate
953 seconds *= failures + 1
954 seconds += seconds * (rand(100)-50)/100
958 def select_nonempty(*ar)
960 ar.each { |i| return i unless i.nil_or_empty? }
964 def printFormattedRss(feed, item, opts=nil)
966 places = feed.watchers
967 handle = feed.handle.empty? ? "" : "::#{feed.handle}:: "
970 places = opts[:places] if opts.key?(:places)
971 handle = opts[:handle].to_s if opts.key?(:handle)
972 if opts.key?(:date) && opts[:date]
973 if item.respond_to?(:updated)
974 if item.updated.content.class <= Time
975 date = item.updated.content.strftime("%Y/%m/%d %H:%M")
977 date = item.updated.content.to_s
979 elsif item.respond_to?(:source) and item.source.respond_to?(:updated)
980 if item.source.updated.content.class <= Time
981 date = item.source.updated.content.strftime("%Y/%m/%d %H:%M")
983 date = item.source.updated.content.to_s
985 elsif item.respond_to?(:pubDate)
986 if item.pubDate.class <= Time
987 date = item.pubDate.strftime("%Y/%m/%d %H:%M")
989 date = item.pubDate.to_s
991 elsif item.respond_to?(:date)
992 if item.date.class <= Time
993 date = item.date.strftime("%Y/%m/%d %H:%M")
995 date = item.date.to_s
1005 # Twitters don't need a cap on the title length since they have a hard
1006 # limit to 160 characters, and most of them are under 140 characters
1007 tit_opt[:limit] = @bot.config['rss.head_max'] unless feed.type == 'twitter'
1010 base_title = item.title.to_s.dup
1011 # git changesets are SHA1 hashes (40 hex digits), way too long, get rid of them, as they are
1012 # visible in the URL anyway
1013 # TODO make this optional?
1014 base_title.sub!(/^Changeset \[([\da-f]{40})\]:/) { |c| "(git commit)"} if feed.type == 'trac'
1015 title = "#{Bold}#{base_title.ircify_html(tit_opt)}#{Bold}"
1019 desc_opt[:limit] = @bot.config['rss.text_max']
1020 desc_opt[:a_href] = :link_out if @bot.config['rss.show_links']
1022 # We prefer content_encoded here as it tends to provide more html formatting
1023 # for use with ircify_html.
1024 if item.respond_to?(:content_encoded) && item.content_encoded
1025 desc = item.content_encoded.ircify_html(desc_opt)
1026 elsif item.respond_to?(:description) && item.description
1027 desc = item.description.ircify_html(desc_opt)
1028 elsif item.respond_to?(:content) && item.content
1029 if item.content.type == "html"
1030 desc = item.content.content.ircify_html(desc_opt)
1032 desc = item.content.content
1033 if desc.size > desc_opt[:limit]
1034 desc = desc.slice(0, desc_opt[:limit]) + "#{Reverse}...#{Reverse}"
1041 link = item.link.href rescue item.link rescue nil
1044 category = select_nonempty((item.category.content rescue nil), (item.dc_subject rescue nil))
1045 category.strip! if category
1046 author = select_nonempty((item.author.name.content rescue nil), (item.dc_creator rescue nil), (item.author rescue nil))
1047 author.strip! if author
1052 at = ((item.title && item.link) ? ' @ ' : '')
1054 key = @bot.global_filter_name(feed.type, @outkey)
1055 key = @bot.global_filter_name(:default, @outkey) unless @bot.has_filter?(key)
1057 output = @bot.filter(key, :item => item, :handle => handle, :date => date,
1058 :title => title, :desc => desc, :link => link,
1059 :category => category, :author => author, :at => at)
1061 return output if places.empty?
1064 output.to_s.each_line { |line|
1065 @bot.say loc, line, :overlong => :truncate
1070 def fetchRss(feed, m=nil, cache=true)
1071 feed.last_fetched = Time.now
1073 # Use 60 sec timeout, cause the default is too low
1074 xml = @bot.httputil.get(feed.url,
1075 :read_timeout => 60,
1076 :open_timeout => 60,
1078 rescue URI::InvalidURIError, URI::BadURIError => e
1079 report_problem("invalid rss feed #{feed.url}", e, m)
1082 report_problem("error getting #{feed.url}", e, m)
1085 debug "fetched #{feed}"
1087 report_problem("reading feed #{feed} failed", nil, m)
1090 # Ok, 0.9 feeds are not supported, maybe because
1091 # Netscape happily removed the DTD. So what we do is just to
1092 # reassign the 0.9 RDFs to 1.0, and hope it goes right.
1093 xml.gsub!("xmlns=\"http://my.netscape.com/rdf/simple/0.9/\"",
1094 "xmlns=\"http://purl.org/rss/1.0/\"")
1095 feed.mutex.synchronize do
1101 def parseRss(feed, m=nil)
1102 return nil unless feed.xml
1103 feed.mutex.synchronize do
1106 ## do validate parse
1107 rss = RSS::Parser.parse(xml)
1108 debug "parsed and validated #{feed}"
1109 rescue RSS::InvalidRSSError
1110 ## do non validate parse for invalid RSS 1.0
1112 rss = RSS::Parser.parse(xml, false)
1113 debug "parsed but not validated #{feed}"
1114 rescue RSS::Error => e
1115 report_problem("parsing rss stream failed, whoops =(", e, m)
1118 rescue RSS::Error => e
1119 report_problem("parsing rss stream failed, oioi", e, m)
1122 report_problem("processing error occured, sorry =(", e, m)
1127 if xml.match(/xmlns\s*=\s*(['"])http:\/\/www.w3.org\/2005\/Atom\1/) and not defined?(RSS::Atom)
1128 report_problem("#{feed.handle} @ #{feed.url} looks like an Atom feed, but your Ruby/RSS library doesn't seem to support it. Consider getting the latest version from http://raa.ruby-lang.org/project/rss/", nil, m)
1130 report_problem("#{feed.handle} @ #{feed.url} doesn't seem to contain an RSS or Atom feed I can read", nil, m)
1135 rss.output_encoding = 'UTF-8'
1136 rescue RSS::UnknownConvertMethod => e
1137 report_problem("bah! something went wrong =(", e, m)
1140 if rss.respond_to? :channel
1141 rss.channel.title ||= "(?)"
1142 title = rss.channel.title
1144 title = rss.title.content
1146 rss.items.each do |item|
1147 item.title ||= "(?)"
1153 report_problem("no items found in the feed, maybe try weed?", e, m)
1156 feed.title = title.strip
1163 plugin = RSSFeedsPlugin.new
1165 plugin.default_auth( 'edit', false )
1166 plugin.default_auth( 'edit:add', true)
1168 plugin.map 'rss show :handle :limit',
1169 :action => 'show_rss',
1170 :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
1171 :defaults => {:limit => 5}
1172 plugin.map 'rss list :handle',
1173 :action => 'list_rss',
1174 :defaults => {:handle => nil}
1175 plugin.map 'rss watched :handle [in :chan]',
1176 :action => 'watched_rss',
1177 :defaults => {:handle => nil}
1178 plugin.map 'rss who watches :handle',
1179 :action => 'who_watches',
1180 :defaults => {:handle => nil}
1181 plugin.map 'rss add :handle :url :type',
1182 :action => 'add_rss',
1183 :auth_path => 'edit',
1184 :defaults => {:type => nil}
1185 plugin.map 'rss change :what of :handle to :new',
1186 :action => 'change_rss',
1187 :auth_path => 'edit',
1188 :requirements => { :what => /handle|url|format|type|refresh/ }
1189 plugin.map 'rss change :what for :handle to :new',
1190 :action => 'change_rss',
1191 :auth_path => 'edit',
1192 :requirements => { :what => /handle|url|format|type|refesh/ }
1193 plugin.map 'rss del :handle',
1194 :auth_path => 'edit:rm!',
1195 :action => 'del_rss'
1196 plugin.map 'rss delete :handle',
1197 :auth_path => 'edit:rm!',
1198 :action => 'del_rss'
1199 plugin.map 'rss rm :handle',
1200 :auth_path => 'edit:rm!',
1201 :action => 'del_rss'
1202 plugin.map 'rss replace :handle :url :type',
1203 :auth_path => 'edit',
1204 :action => 'replace_rss',
1205 :defaults => {:type => nil}
1206 plugin.map 'rss forcereplace :handle :url :type',
1207 :auth_path => 'edit',
1208 :action => 'forcereplace_rss',
1209 :defaults => {:type => nil}
1210 plugin.map 'rss watch :handle [in :chan]',
1211 :action => 'watch_rss',
1212 :defaults => {:url => nil, :type => nil}
1213 plugin.map 'rss watch :handle :url :type [in :chan]',
1214 :action => 'watch_rss',
1215 :defaults => {:url => nil, :type => nil}
1216 plugin.map 'rss unwatch :handle [in :chan]',
1217 :action => 'unwatch_rss'
1218 plugin.map 'rss rmwatch :handle [in :chan]',
1219 :action => 'unwatch_rss'
1220 plugin.map 'rss rewatch [:handle] [:delay]',
1221 :action => 'rewatch_rss'
1222 plugin.map 'rss types',
1223 :action => 'rss_types'