4 # :title: RSS feed plugin for rbot
\r
6 # Author:: Stanislav Karchebny <berkus@madfire.net>
\r
7 # Author:: Ian Monroe <ian@monroe.nu>
\r
8 # Author:: Mark Kretschmann <markey@web.de>
\r
9 # Author:: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
\r
11 # Copyright:: (C) 2004 Stanislav Karchebny
\r
12 # Copyright:: (C) 2005 Ian Monroe, Mark Kretschmann
\r
13 # Copyright:: (C) 2006-2007 Giuseppe Bilotta
\r
15 # License:: MIT license
\r
21 # Make an 'unique' ID for a given item, based on appropriate bot options
\r
22 # Currently only suppored is bot.config['rss.show_updated']: when true, the
\r
23 # description is included in the uid hashing, otherwise it's not
\r
25 def RSS.item_uid_for_bot(item, opts={})
\r
26 options = { :show_updated => true}.merge(opts)
\r
27 desc = options[:show_updated] ? item.description : nil
\r
28 [item.title, item.link, desc].hash
\r
31 # Add support for Slashdot namespace in RDF. The code is just an adaptation
\r
32 # of the DublinCore code.
\r
33 unless defined?(SLASH_PREFIX)
\r
34 SLASH_PREFIX = 'slash'
\r
35 SLASH_URI = "http://purl.org/rss/1.0/modules/slash/"
\r
37 RDF.install_ns(SLASH_PREFIX, SLASH_URI)
\r
39 module BaseSlashModel
\r
40 def append_features(klass)
\r
43 return if klass.instance_of?(Module)
\r
44 SlashModel::ELEMENT_NAME_INFOS.each do |name, plural_name|
\r
45 plural = plural_name || "#{name}s"
\r
46 full_name = "#{SLASH_PREFIX}_#{name}"
\r
47 full_plural_name = "#{SLASH_PREFIX}_#{plural}"
\r
48 klass_name = "Slash#{Utils.to_class_name(name)}"
\r
49 klass.install_must_call_validator(SLASH_PREFIX, SLASH_URI)
\r
50 klass.install_have_children_element(name, SLASH_URI, "*",
\r
51 full_name, full_plural_name)
\r
52 klass.module_eval(<<-EOC, *get_file_and_line_from_caller(0))
\r
53 remove_method :#{full_name}
\r
54 remove_method :#{full_name}=
\r
55 remove_method :set_#{full_name}
\r
58 @#{full_name}.first and @#{full_name}.first.value
\r
61 def #{full_name}=(new_value)
\r
62 @#{full_name}[0] = Utils.new_with_value_if_need(#{klass_name}, new_value)
\r
64 alias set_#{full_name} #{full_name}=
\r
72 extend BaseSlashModel
\r
75 "department" => nil,
\r
81 ELEMENT_NAME_INFOS = SlashModel::TEXT_ELEMENTS.to_a
\r
83 ELEMENTS = TEXT_ELEMENTS.keys
\r
85 ELEMENTS.each do |name, plural_name|
\r
86 module_eval(<<-EOC, *get_file_and_line_from_caller(0))
\r
87 class Slash#{Utils.to_class_name(name)} < Element
\r
102 @tag_name = #{name.dump}
\r
104 alias_method(:value, :content)
\r
105 alias_method(:value=, :content=)
\r
107 def initialize(*args)
\r
108 if Utils.element_initialize_arguments?(args)
\r
112 self.content = args[0]
\r
117 tag_name_with_prefix(SLASH_PREFIX)
\r
120 def maker_target(target)
\r
124 def setup_maker_attributes(#{name})
\r
125 #{name}.content = content
\r
133 class Item; include SlashModel; end
\r
136 SlashModel::ELEMENTS.each do |name|
\r
137 class_name = Utils.to_class_name(name)
\r
138 BaseListener.install_class_name(SLASH_URI, name, "Slash#{class_name}")
\r
141 SlashModel::ELEMENTS.collect! {|name| "#{SLASH_PREFIX}_#{name}"}
\r
148 attr_accessor :handle
\r
149 attr_accessor :type
\r
151 attr_accessor :refresh_rate
\r
153 attr_accessor :title
\r
154 attr_accessor :items
\r
155 attr_accessor :mutex
\r
157 def initialize(url,handle=nil,type=nil,watchers=[], xml=nil)
\r
166 @refresh_rate = nil
\r
171 sanitize_watchers(watchers)
\r
175 @mutex.synchronize do
\r
176 self.class.new(@url,
\r
178 @type ? @type.dup : nil,
\r
180 @xml ? @xml.dup : nil)
\r
184 # Downcase all watchers, possibly turning them into Strings if they weren't
\r
185 def sanitize_watchers(list=@watchers)
\r
197 def watched_by?(who)
\r
198 @watchers.include?(who.downcase)
\r
202 if watched_by?(who)
\r
205 @mutex.synchronize do
\r
206 @watchers << who.downcase
\r
212 @mutex.synchronize do
\r
213 @watchers.delete(who.downcase)
\r
218 [@handle,@url,@type,@refresh_rate,@watchers]
\r
221 def to_s(watchers=false)
\r
223 a = self.to_a.flatten
\r
227 a.compact.join(" | ")
\r
231 class RSSFeedsPlugin < Plugin
\r
232 BotConfig.register BotConfigIntegerValue.new('rss.head_max',
\r
233 :default => 100, :validate => Proc.new{|v| v > 0 && v < 200},
\r
234 :desc => "How many characters to use of a RSS item header")
\r
236 BotConfig.register BotConfigIntegerValue.new('rss.text_max',
\r
237 :default => 200, :validate => Proc.new{|v| v > 0 && v < 400},
\r
238 :desc => "How many characters to use of a RSS item text")
\r
240 BotConfig.register BotConfigIntegerValue.new('rss.thread_sleep',
\r
241 :default => 300, :validate => Proc.new{|v| v > 30},
\r
242 :desc => "How many seconds to sleep before checking RSS feeds again")
\r
244 BotConfig.register BotConfigBooleanValue.new('rss.show_updated',
\r
246 :desc => "Whether feed items for which the description was changed should be shown as new")
\r
248 # We used to save the Mutex with the RssBlob, which was idiotic. And
\r
249 # since Mutexes dumped in one version might not be resotrable in another,
\r
250 # we need a few tricks to be able to restore data from other versions of Ruby
\r
252 # When migrating 1.8.6 => 1.8.5, all we need to do is define an empty
\r
253 # #marshal_load() method for Mutex. For 1.8.5 => 1.8.6 we need something
\r
254 # dirtier, as seen later on in the initialization code.
\r
255 unless Mutex.new.respond_to?(:marshal_load)
\r
257 def marshal_load(str)
\r
267 if @registry.has_key?(:feeds)
\r
268 # When migrating from Ruby 1.8.5 to 1.8.6, dumped Mutexes may render the
\r
269 # data unrestorable. If this happens, we patch the data, thus allowing
\r
270 # the restore to work.
\r
272 # This is actually pretty safe for a number of reasons:
\r
273 # * the code is only called if standard marshalling fails
\r
274 # * the string we look for is quite unlikely to appear randomly
\r
275 # * if the string appears somewhere and the patched string isn't recoverable
\r
276 # either, we'll get another (unrecoverable) error, which makes the rss
\r
277 # plugin unsable, just like it was if no recovery was attempted
\r
278 # * if the string appears somewhere and the patched string is recoverable,
\r
279 # we may get a b0rked feed, which is eventually overwritten by a clean
\r
280 # one, so the worst thing that can happen is that a feed update spams
\r
281 # the watchers once
\r
282 @registry.recovery = Proc.new { |val|
\r
283 patched = val.sub(":\v@mutexo:\nMutex", ":\v@mutexo:\vObject")
\r
284 ret = Marshal.restore(patched)
\r
285 ret.each_value { |blob|
\r
291 @feeds = @registry[:feeds]
\r
292 raise unless @feeds
\r
294 @registry.recovery = nil
\r
296 @feeds.keys.grep(/[A-Z]/) { |k|
\r
297 @feeds[k.downcase] = @feeds[k]
\r
300 @feeds.each { |k, f|
\r
301 f.mutex = Mutex.new unless f.mutex
\r
302 f.sanitize_watchers
\r
303 parseRss(f) if f.xml
\r
317 @feeds.select { |h, f| f.watched? }
\r
325 unparsed = Hash.new()
\r
326 @feeds.each { |k, f|
\r
327 unparsed[k] = f.dup
\r
328 # we don't want to save the mutex
\r
329 unparsed[k].mutex = nil
\r
331 @registry[:feeds] = unparsed
\r
334 def stop_watch(handle)
\r
335 if @watch.has_key?(handle)
\r
337 debug "Stopping watch #{handle}"
\r
338 @bot.timer.remove(@watch[handle])
\r
339 @watch.delete(handle)
\r
340 rescue Exception => e
\r
341 report_problem("Failed to stop watch for #{handle}", e, nil)
\r
347 @watch.each_key { |k|
\r
352 def help(plugin,topic="")
\r
355 "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
357 "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
\r
359 "rss watched [#{Bold}handle#{Bold}] [in #{Bold}chan#{Bold}]: list all watched rss feeds (matching #{Bold}handle#{Bold}) (in channel #{Bold}chan#{Bold})"
\r
360 when "who", "watches", "who watches"
\r
361 "rss who watches [#{Bold}handle#{Bold}]]: list all watchers for rss feeds (matching #{Bold}handle#{Bold})"
\r
363 "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
365 "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
366 when /^(del(ete)?|rm)$/
\r
367 "rss del(ete)|rm #{Bold}handle#{Bold} : delete rss feed #{Bold}handle#{Bold}"
\r
369 "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
370 when "forcereplace"
\r
371 "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
373 "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
374 when /(un|rm)watch/
\r
375 "rss unwatch|rmwatch #{Bold}handle#{Bold} [in #{Bold}chan#{Bold}]: stop watching rss #{Bold}handle#{Bold} (in channel #{Bold}chan#{Bold}) for changes"
\r
377 "rss rewatch : restart threads that watch for changes in watched rss"
\r
379 "manage RSS feeds: rss show|list|watched|add|change|del(ete)|rm|(force)replace|watch|unwatch|rmwatch|rewatch"
\r
383 def report_problem(report, e=nil, m=nil)
\r
384 if m && m.respond_to?(:reply)
\r
391 debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
\r
395 def show_rss(m, params)
\r
396 handle = params[:handle]
\r
397 lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
\r
398 debug lims.to_a.inspect
\r
400 ll = [[lims[1].to_i-1,lims[2].to_i-1].min, 0].max
\r
401 ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
\r
402 rev = lims[1].to_i > lims[2].to_i
\r
405 ul = [[lims[1].to_i-1, 0].max, 14].min
\r
409 feed = @feeds.fetch(handle.downcase, nil)
\r
411 m.reply "I don't know any feeds named #{handle}"
\r
415 m.reply "lemme fetch it..."
\r
416 title = items = nil
\r
417 fetched = fetchRss(feed, m, false)
\r
418 return unless fetched or feed.xml
\r
419 if not fetched and feed.items
\r
420 m.reply "using old data"
\r
422 parsed = parseRss(feed, m)
\r
423 m.reply "using old data" unless parsed
\r
425 return unless feed.items
\r
429 # We sort the feeds in freshness order (newer ones first)
\r
430 items = freshness_sort(items)
\r
431 disp = items[ll..ul]
\r
432 disp.reverse! if rev
\r
434 m.reply "Channel : #{title}"
\r
435 disp.each do |item|
\r
436 printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
\r
440 def itemDate(item,ex=nil)
\r
441 return item.pubDate if item.respond_to?(:pubDate) and item.pubDate
\r
442 return item.date if item.respond_to?(:date) and item.date
\r
446 def freshness_sort(items)
\r
447 notime = Time.at(0)
\r
448 items.sort { |a, b|
\r
449 itemDate(b, notime) <=> itemDate(a, notime)
\r
453 def list_rss(m, params)
\r
454 wanted = params[:handle]
\r
456 @feeds.each { |handle, feed|
\r
457 next if wanted and !handle.match(/#{wanted}/i)
\r
458 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
459 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
460 (reply << " (watched)") if feed.watched_by?(m.replyto)
\r
464 reply = "no feeds found"
\r
465 reply << " matching #{wanted}" if wanted
\r
467 m.reply reply, :max_lines => reply.length
\r
470 def watched_rss(m, params)
\r
471 wanted = params[:handle]
\r
472 chan = params[:chan] || m.replyto
\r
474 watchlist.each { |handle, feed|
\r
475 next if wanted and !handle.match(/#{wanted}/i)
\r
476 next unless feed.watched_by?(chan)
\r
477 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
478 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
482 reply = "no watched feeds"
\r
483 reply << " matching #{wanted}" if wanted
\r
488 def who_watches(m, params)
\r
489 wanted = params[:handle]
\r
491 watchlist.each { |handle, feed|
\r
492 next if wanted and !handle.match(/#{wanted}/i)
\r
493 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
494 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
495 reply << ": watched by #{feed.watchers.join(', ')}"
\r
499 reply = "no watched feeds"
\r
500 reply << " matching #{wanted}" if wanted
\r
505 def add_rss(m, params, force=false)
\r
506 handle = params[:handle]
\r
508 unless url.match(/https?/)
\r
509 m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
\r
512 type = params[:type]
\r
513 if @feeds.fetch(handle.downcase, nil) && !force
\r
514 m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"
\r
518 m.reply "You must specify both a handle and an url to add an RSS feed"
\r
521 @feeds[handle.downcase] = RssBlob.new(url,handle,type)
\r
522 reply = "Added RSS #{url} named #{handle}"
\r
524 reply << " (format: #{type})"
\r
530 def change_rss(m, params)
\r
531 handle = params[:handle].downcase
\r
532 feed = @feeds.fetch(handle, nil)
\r
534 m.reply "No such feed with handle #{handle}"
\r
537 case params[:what].intern
\r
539 new = params[:new].downcase
\r
540 if @feeds.key?(new) and @feeds[new]
\r
541 m.reply "There already is a feed with handle #{new}"
\r
544 feed.mutex.synchronize do
\r
546 @feeds.delete(handle)
\r
553 feed.mutex.synchronize do
\r
556 when :format, :type
\r
558 new = nil if new == 'default'
\r
559 feed.mutex.synchronize do
\r
563 new = params[:new].to_i
\r
564 new = nil if new == 0
\r
565 feed.mutex.synchronize do
\r
566 feed.refresh_rate = new
\r
569 m.reply "Don't know how to change #{params[:what]} for feeds"
\r
572 m.reply "Feed changed:"
\r
573 list_rss(m, {:handle => handle})
\r
576 def del_rss(m, params, pass=false)
\r
577 feed = unwatch_rss(m, params, true)
\r
579 m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
\r
582 @feeds.delete(feed.handle.downcase)
\r
587 def replace_rss(m, params)
\r
588 handle = params[:handle]
\r
589 if @feeds.key?(handle.downcase)
\r
590 del_rss(m, {:handle => handle}, true)
\r
592 if @feeds.key?(handle.downcase)
\r
593 m.reply "can't replace #{feed.handle}"
\r
595 add_rss(m, params, true)
\r
599 def forcereplace_rss(m, params)
\r
600 add_rss(m, params, true)
\r
603 def watch_rss(m, params)
\r
604 handle = params[:handle]
\r
605 chan = params[:chan] || m.replyto
\r
607 type = params[:type]
\r
611 feed = @feeds.fetch(handle.downcase, nil)
\r
613 if feed.add_watch(chan)
\r
617 m.reply "Already watching #{feed.handle} in #{chan}"
\r
620 m.reply "Couldn't watch feed #{handle} (no such feed found)"
\r
624 def unwatch_rss(m, params, pass=false)
\r
625 handle = params[:handle].downcase
\r
626 chan = params[:chan] || m.replyto
\r
627 unless @feeds.has_key?(handle)
\r
628 m.reply("dunno that feed")
\r
631 feed = @feeds[handle]
\r
632 if feed.rm_watch(chan)
\r
633 m.reply "#{chan} has been removed from the watchlist for #{feed.handle}"
\r
635 m.reply("#{chan} wasn't watching #{feed.handle}") unless pass
\r
643 def rewatch_rss(m=nil, params=nil)
\r
646 # Read watches from list.
\r
647 watchlist.each{ |handle, feed|
\r
654 def watchRss(feed, m=nil)
\r
655 if @watch.has_key?(feed.handle)
\r
656 report_problem("watcher thread for #{feed.handle} is already running", nil, m)
\r
660 status[:failures] = 0
\r
661 status[:first_run] = true
\r
662 @watch[feed.handle] = @bot.timer.add(0, status) {
\r
663 debug "watcher for #{feed} started"
\r
664 failures = status[:failures]
\r
665 first_run = status.delete(:first_run)
\r
667 debug "fetching #{feed}"
\r
668 oldxml = feed.xml ? feed.xml.dup : nil
\r
669 unless fetchRss(feed)
\r
673 debug "first run for #{feed}, getting items"
\r
675 elsif oldxml and oldxml == feed.xml
\r
676 debug "xml for #{feed} didn't change"
\r
677 failures -= 1 if failures > 0
\r
680 debug "no previous items in feed #{feed}"
\r
682 failures -= 1 if failures > 0
\r
684 # This one is used for debugging
\r
687 # These are used for checking new items vs old ones
\r
688 uid_opts = { :show_updated => @bot.config['rss.show_updated'] }
\r
689 oids = Set.new feed.items.map { |item|
\r
690 uid = RSS.item_uid_for_bot(item, uid_opts)
\r
692 debug [uid, item].inspect
\r
693 debug [uid, otxt.last].inspect
\r
697 unless parseRss(feed)
\r
698 debug "no items in feed #{feed}"
\r
701 debug "Checking if new items are available for #{feed}"
\r
702 failures -= 1 if failures > 0
\r
708 dispItems = feed.items.reject { |item|
\r
709 uid = RSS.item_uid_for_bot(item, uid_opts)
\r
711 if oids.include?(uid)
\r
712 debug "rejecting old #{uid} #{item.inspect}"
\r
713 debug [uid, txt].inspect
\r
716 debug "accepting new #{uid} #{item.inspect}"
\r
717 debug [uid, txt].inspect
\r
718 warn "same text! #{txt}" if otxt.include?(txt)
\r
723 if dispItems.length > 0
\r
724 debug "Found #{dispItems.length} new items in #{feed}"
\r
725 # When displaying watched feeds, publish them from older to newer
\r
726 dispItems.reverse.each { |item|
\r
727 printFormattedRss(feed, item)
\r
730 debug "No new items found in #{feed}"
\r
736 rescue Exception => e
\r
737 error "Error watching #{feed}: #{e.inspect}"
\r
738 debug e.backtrace.join("\n")
\r
742 status[:failures] = failures
\r
744 feed.mutex.synchronize do
\r
745 seconds = (feed.refresh_rate || @bot.config['rss.thread_sleep']) * (failures + 1)
\r
746 seconds += seconds * (rand(100)-50)/100
\r
747 debug "watcher for #{feed} going to sleep #{seconds} seconds.."
\r
748 @bot.timer.reschedule(@watch[feed.handle], seconds)
\r
751 debug "watcher for #{feed} added"
\r
754 def printFormattedRss(feed, item, opts=nil)
\r
755 places = feed.watchers
\r
756 handle = "::#{feed.handle}:: "
\r
759 places = opts[:places] if opts.key?(:places)
\r
760 handle = opts[:handle].to_s if opts.key?(:handle)
\r
761 if opts.key?(:date) && opts[:date]
\r
762 if item.respond_to?(:pubDate)
\r
763 if item.pubDate.class <= Time
\r
764 date = item.pubDate.strftime("%Y/%m/%d %H:%M")
\r
766 date = item.pubDate.to_s
\r
768 elsif item.respond_to?(:date)
\r
769 if item.date.class <= Time
\r
770 date = item.date.strftime("%Y/%m/%d %H:%M")
\r
772 date = item.date.to_s
\r
781 title = "#{Bold}#{item.title.ircify_html :limit => @bot.config['rss.head_max']}#{Bold}" if item.title
\r
783 desc = item.description.ircify_html(:limit => @bot.config['rss.text_max'], :a_href => :link_out) if item.description
\r
785 link = item.link.chomp if item.link
\r
788 category = item.dc_subject rescue item.category.content rescue nil
\r
789 category = nil if category and category.empty?
\r
790 author = item.dc_creator rescue item.author rescue nil
\r
791 author = nil if author and author.empty?
\r
796 at = ((item.title && item.link) ? ' @ ' : '')
\r
799 author << " " if author
\r
800 abt = category ? "about #{category} " : ""
\r
801 line1 = "#{handle}#{date}#{author}blogged #{abt}at #{link}"
\r
802 line2 = "#{handle}#{title} - #{desc}"
\r
804 line1 = "#{handle}#{date}#{title}#{at}#{link}"
\r
806 line1 = "#{handle}#{date}#{title}#{at}#{link} has been edited by #{author}. #{desc}"
\r
808 line1 = "#{handle}#{date}Message #{title} sent by #{author}. #{desc}"
\r
810 line1 = "#{handle}#{date}#{title} @ #{link}"
\r
811 unless item.title =~ /^Changeset \[(\d+)\]/
\r
812 line2 = "#{handle}#{date}#{desc}"
\r
815 dept = "(from the #{item.slash_department} dept) " rescue nil
\r
816 sec = " in section #{item.slash_section}" rescue nil
\r
818 line1 = "#{handle}#{date}#{dept}#{title}#{at}#{link} (posted by #{author}#{sec})"
\r
820 line1 = "#{handle}#{date}#{title}#{at}#{link}"
\r
822 places.each { |loc|
\r
823 @bot.say loc, line1, :overlong => :truncate
\r
825 @bot.say loc, line2, :overlong => :truncate
\r
829 def fetchRss(feed, m=nil, cache=true)
\r
831 # Use 60 sec timeout, cause the default is too low
\r
832 xml = @bot.httputil.get(feed.url,
\r
833 :read_timeout => 60,
\r
834 :open_timeout => 60,
\r
836 rescue URI::InvalidURIError, URI::BadURIError => e
\r
837 report_problem("invalid rss feed #{feed.url}", e, m)
\r
840 report_problem("error getting #{feed.url}", e, m)
\r
843 debug "fetched #{feed}"
\r
845 report_problem("reading feed #{feed} failed", nil, m)
\r
848 # Ok, 0.9 feeds are not supported, maybe because
\r
849 # Netscape happily removed the DTD. So what we do is just to
\r
850 # reassign the 0.9 RDFs to 1.0, and hope it goes right.
\r
851 xml.gsub!("xmlns=\"http://my.netscape.com/rdf/simple/0.9/\"",
\r
852 "xmlns=\"http://purl.org/rss/1.0/\"")
\r
853 feed.mutex.synchronize do
\r
859 def parseRss(feed, m=nil)
\r
860 return nil unless feed.xml
\r
861 feed.mutex.synchronize do
\r
864 ## do validate parse
\r
865 rss = RSS::Parser.parse(xml)
\r
866 debug "parsed and validated #{feed}"
\r
867 rescue RSS::InvalidRSSError
\r
868 ## do non validate parse for invalid RSS 1.0
\r
870 rss = RSS::Parser.parse(xml, false)
\r
871 debug "parsed but not validated #{feed}"
\r
872 rescue RSS::Error => e
\r
873 report_problem("parsing rss stream failed, whoops =(", e, m)
\r
876 rescue RSS::Error => e
\r
877 report_problem("parsing rss stream failed, oioi", e, m)
\r
880 report_problem("processing error occured, sorry =(", e, m)
\r
885 report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
\r
888 rss.output_encoding = 'UTF-8'
\r
889 rescue RSS::UnknownConvertMethod => e
\r
890 report_problem("bah! something went wrong =(", e, m)
\r
893 rss.channel.title ||= "Unknown"
\r
894 title = rss.channel.title
\r
895 rss.items.each do |item|
\r
896 item.title ||= "Unknown"
\r
902 report_problem("no items found in the feed, maybe try weed?", e, m)
\r
912 plugin = RSSFeedsPlugin.new
\r
914 plugin.map 'rss show :handle :limit',
\r
915 :action => 'show_rss',
\r
916 :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
\r
917 :defaults => {:limit => 5}
\r
918 plugin.map 'rss list :handle',
\r
919 :action => 'list_rss',
\r
920 :defaults => {:handle => nil}
\r
921 plugin.map 'rss watched :handle [in :chan]',
\r
922 :action => 'watched_rss',
\r
923 :defaults => {:handle => nil}
\r
924 plugin.map 'rss who watches :handle',
\r
925 :action => 'who_watches',
\r
926 :defaults => {:handle => nil}
\r
927 plugin.map 'rss add :handle :url :type',
\r
928 :action => 'add_rss',
\r
929 :defaults => {:type => nil}
\r
930 plugin.map 'rss change :what of :handle to :new',
\r
931 :action => 'change_rss',
\r
932 :requirements => { :what => /handle|url|format|type|refresh/ }
\r
933 plugin.map 'rss change :what for :handle to :new',
\r
934 :action => 'change_rss',
\r
935 :requirements => { :what => /handle|url|format|type|refesh/ }
\r
936 plugin.map 'rss del :handle',
\r
937 :action => 'del_rss'
\r
938 plugin.map 'rss delete :handle',
\r
939 :action => 'del_rss'
\r
940 plugin.map 'rss rm :handle',
\r
941 :action => 'del_rss'
\r
942 plugin.map 'rss replace :handle :url :type',
\r
943 :action => 'replace_rss',
\r
944 :defaults => {:type => nil}
\r
945 plugin.map 'rss forcereplace :handle :url :type',
\r
946 :action => 'forcereplace_rss',
\r
947 :defaults => {:type => nil}
\r
948 plugin.map 'rss watch :handle [in :chan]',
\r
949 :action => 'watch_rss',
\r
950 :defaults => {:url => nil, :type => nil}
\r
951 plugin.map 'rss watch :handle :url :type [in :chan]',
\r
952 :action => 'watch_rss',
\r
953 :defaults => {:url => nil, :type => nil}
\r
954 plugin.map 'rss unwatch :handle [in :chan]',
\r
955 :action => 'unwatch_rss'
\r
956 plugin.map 'rss rmwatch :handle [in :chan]',
\r
957 :action => 'unwatch_rss'
\r
958 plugin.map 'rss rewatch',
\r
959 :action => 'rewatch_rss'
\r