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
19 # Try to load rss/content/2.0 so we can access the data in <content:encoded>
\r
22 require 'rss/content/2.0'
\r
28 # Make an 'unique' ID for a given item, based on appropriate bot options
\r
29 # Currently only suppored is bot.config['rss.show_updated']: when true, the
\r
30 # description is included in the uid hashing, otherwise it's not
\r
32 def RSS.item_uid_for_bot(item, opts={})
\r
33 options = { :show_updated => true}.merge(opts)
\r
35 if options[:show_updated]
\r
36 desc = item.content.content rescue item.description rescue nil
\r
38 [(item.title.content rescue item.title rescue nil),
\r
39 (item.link.href rescue item.link),
\r
43 # Add support for Slashdot namespace in RDF. The code is just an adaptation
\r
44 # of the DublinCore code.
\r
45 unless defined?(SLASH_PREFIX)
\r
46 SLASH_PREFIX = 'slash'
\r
47 SLASH_URI = "http://purl.org/rss/1.0/modules/slash/"
\r
49 RDF.install_ns(SLASH_PREFIX, SLASH_URI)
\r
51 module BaseSlashModel
\r
52 def append_features(klass)
\r
55 return if klass.instance_of?(Module)
\r
56 SlashModel::ELEMENT_NAME_INFOS.each do |name, plural_name|
\r
57 plural = plural_name || "#{name}s"
\r
58 full_name = "#{SLASH_PREFIX}_#{name}"
\r
59 full_plural_name = "#{SLASH_PREFIX}_#{plural}"
\r
60 klass_name = "Slash#{Utils.to_class_name(name)}"
\r
62 # This will fail with older version of the Ruby RSS module
\r
64 klass.install_have_children_element(name, SLASH_URI, "*",
\r
65 full_name, full_plural_name)
\r
66 klass.install_must_call_validator(SLASH_PREFIX, SLASH_URI)
\r
67 rescue ArgumentError
\r
68 klass.module_eval("install_have_children_element(#{full_name.dump}, #{full_plural_name.dump})")
\r
71 klass.module_eval(<<-EOC, *get_file_and_line_from_caller(0))
\r
72 remove_method :#{full_name} if method_defined? :#{full_name}
\r
73 remove_method :#{full_name}= if method_defined? :#{full_name}=
\r
74 remove_method :set_#{full_name} if method_defined? :set_#{full_name}
\r
77 @#{full_name}.first and @#{full_name}.first.value
\r
80 def #{full_name}=(new_value)
\r
81 @#{full_name}[0] = Utils.new_with_value_if_need(#{klass_name}, new_value)
\r
83 alias set_#{full_name} #{full_name}=
\r
91 extend BaseSlashModel
\r
94 "department" => nil,
\r
100 ELEMENT_NAME_INFOS = SlashModel::TEXT_ELEMENTS.to_a
\r
102 ELEMENTS = TEXT_ELEMENTS.keys
\r
104 ELEMENTS.each do |name, plural_name|
\r
105 module_eval(<<-EOC, *get_file_and_line_from_caller(0))
\r
106 class Slash#{Utils.to_class_name(name)} < Element
\r
112 def required_prefix
\r
121 @tag_name = #{name.dump}
\r
123 alias_method(:value, :content)
\r
124 alias_method(:value=, :content=)
\r
126 def initialize(*args)
\r
128 if Utils.element_initialize_arguments?(args)
\r
132 self.content = args[0]
\r
134 # Older Ruby RSS module
\r
135 rescue NoMethodError
\r
137 self.content = args[0]
\r
142 tag_name_with_prefix(SLASH_PREFIX)
\r
145 def maker_target(target)
\r
149 def setup_maker_attributes(#{name})
\r
150 #{name}.content = content
\r
158 class Item; include SlashModel; end
\r
161 SlashModel::ELEMENTS.each do |name|
\r
162 class_name = Utils.to_class_name(name)
\r
163 BaseListener.install_class_name(SLASH_URI, name, "Slash#{class_name}")
\r
166 SlashModel::ELEMENTS.collect! {|name| "#{SLASH_PREFIX}_#{name}"}
\r
172 attr_accessor :url, :handle, :type, :refresh_rate, :xml, :title, :items,
\r
173 :mutex, :watchers, :last_fetched
\r
175 def initialize(url,handle=nil,type=nil,watchers=[], xml=nil, lf = nil)
\r
184 @refresh_rate = nil
\r
190 sanitize_watchers(watchers)
\r
194 @mutex.synchronize do
\r
195 self.class.new(@url,
\r
197 @type ? @type.dup : nil,
\r
199 @xml ? @xml.dup : nil,
\r
204 # Downcase all watchers, possibly turning them into Strings if they weren't
\r
205 def sanitize_watchers(list=@watchers)
\r
217 def watched_by?(who)
\r
218 @watchers.include?(who.downcase)
\r
222 if watched_by?(who)
\r
225 @mutex.synchronize do
\r
226 @watchers << who.downcase
\r
232 @mutex.synchronize do
\r
233 @watchers.delete(who.downcase)
\r
238 [@handle,@url,@type,@refresh_rate,@watchers]
\r
241 def to_s(watchers=false)
\r
243 a = self.to_a.flatten
\r
247 a.compact.join(" | ")
\r
251 class RSSFeedsPlugin < Plugin
\r
252 Config.register Config::IntegerValue.new('rss.head_max',
\r
253 :default => 100, :validate => Proc.new{|v| v > 0 && v < 200},
\r
254 :desc => "How many characters to use of a RSS item header")
\r
256 Config.register Config::IntegerValue.new('rss.text_max',
\r
257 :default => 200, :validate => Proc.new{|v| v > 0 && v < 400},
\r
258 :desc => "How many characters to use of a RSS item text")
\r
260 Config.register Config::IntegerValue.new('rss.thread_sleep',
\r
261 :default => 300, :validate => Proc.new{|v| v > 30},
\r
262 :desc => "How many seconds to sleep before checking RSS feeds again")
\r
264 Config.register Config::BooleanValue.new('rss.show_updated',
\r
266 :desc => "Whether feed items for which the description was changed should be shown as new")
\r
268 # We used to save the Mutex with the RssBlob, which was idiotic. And
\r
269 # since Mutexes dumped in one version might not be resotrable in another,
\r
270 # we need a few tricks to be able to restore data from other versions of Ruby
\r
272 # When migrating 1.8.6 => 1.8.5, all we need to do is define an empty
\r
273 # #marshal_load() method for Mutex. For 1.8.5 => 1.8.6 we need something
\r
274 # dirtier, as seen later on in the initialization code.
\r
275 unless Mutex.new.respond_to?(:marshal_load)
\r
277 def marshal_load(str)
\r
287 if @registry.has_key?(:feeds)
\r
288 # When migrating from Ruby 1.8.5 to 1.8.6, dumped Mutexes may render the
\r
289 # data unrestorable. If this happens, we patch the data, thus allowing
\r
290 # the restore to work.
\r
292 # This is actually pretty safe for a number of reasons:
\r
293 # * the code is only called if standard marshalling fails
\r
294 # * the string we look for is quite unlikely to appear randomly
\r
295 # * if the string appears somewhere and the patched string isn't recoverable
\r
296 # either, we'll get another (unrecoverable) error, which makes the rss
\r
297 # plugin unsable, just like it was if no recovery was attempted
\r
298 # * if the string appears somewhere and the patched string is recoverable,
\r
299 # we may get a b0rked feed, which is eventually overwritten by a clean
\r
300 # one, so the worst thing that can happen is that a feed update spams
\r
301 # the watchers once
\r
302 @registry.recovery = Proc.new { |val|
\r
303 patched = val.sub(":\v@mutexo:\nMutex", ":\v@mutexo:\vObject")
\r
304 ret = Marshal.restore(patched)
\r
305 ret.each_value { |blob|
\r
311 @feeds = @registry[:feeds]
\r
312 raise unless @feeds
\r
314 @registry.recovery = nil
\r
316 @feeds.keys.grep(/[A-Z]/) { |k|
\r
317 @feeds[k.downcase] = @feeds[k]
\r
320 @feeds.each { |k, f|
\r
321 f.mutex = Mutex.new
\r
322 f.sanitize_watchers
\r
323 parseRss(f) if f.xml
\r
337 @feeds.select { |h, f| f.watched? }
\r
346 unparsed = Hash.new()
\r
347 @feeds.each { |k, f|
\r
348 unparsed[k] = f.dup
\r
349 # we don't want to save the mutex
\r
350 unparsed[k].mutex = nil
\r
352 @registry[:feeds] = unparsed
\r
355 def stop_watch(handle)
\r
356 if @watch.has_key?(handle)
\r
358 debug "Stopping watch #{handle}"
\r
359 @bot.timer.remove(@watch[handle])
\r
360 @watch.delete(handle)
\r
361 rescue Exception => e
\r
362 report_problem("Failed to stop watch for #{handle}", e, nil)
\r
368 @watch.each_key { |k|
\r
373 def help(plugin,topic="")
\r
376 "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
378 "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
\r
380 "rss watched [#{Bold}handle#{Bold}] [in #{Bold}chan#{Bold}]: list all watched rss feeds (matching #{Bold}handle#{Bold}) (in channel #{Bold}chan#{Bold})"
\r
381 when "who", "watches", "who watches"
\r
382 "rss who watches [#{Bold}handle#{Bold}]]: list all watchers for rss feeds (matching #{Bold}handle#{Bold})"
\r
384 "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
386 "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
387 when /^(del(ete)?|rm)$/
\r
388 "rss del(ete)|rm #{Bold}handle#{Bold} : delete rss feed #{Bold}handle#{Bold}"
\r
390 "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
391 when "forcereplace"
\r
392 "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
394 "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
395 when /(un|rm)watch/
\r
396 "rss unwatch|rmwatch #{Bold}handle#{Bold} [in #{Bold}chan#{Bold}]: stop watching rss #{Bold}handle#{Bold} (in channel #{Bold}chan#{Bold}) for changes"
\r
397 when /who(?: watche?s?)?/
\r
398 "rss who watches #{Bold}handle#{Bold}: lists watches for rss #{Bold}handle#{Bold}"
\r
400 "rss rewatch : restart threads that watch for changes in watched rss"
\r
402 "manage RSS feeds: rss show|list|watched|add|change|del(ete)|rm|(force)replace|watch|unwatch|rmwatch|rewatch|who watches"
\r
406 def report_problem(report, e=nil, m=nil)
\r
407 if m && m.respond_to?(:reply)
\r
414 debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
\r
418 def show_rss(m, params)
\r
419 handle = params[:handle]
\r
420 lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
\r
421 debug lims.to_a.inspect
\r
423 ll = [[lims[1].to_i-1,lims[2].to_i-1].min, 0].max
\r
424 ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
\r
425 rev = lims[1].to_i > lims[2].to_i
\r
428 ul = [[lims[1].to_i-1, 0].max, 14].min
\r
432 feed = @feeds.fetch(handle.downcase, nil)
\r
434 m.reply "I don't know any feeds named #{handle}"
\r
438 m.reply "lemme fetch it..."
\r
439 title = items = nil
\r
440 we_were_watching = false
\r
442 if @watch.key?(feed.handle)
\r
443 # If a feed is being watched, we run the watcher thread
\r
444 # so that all watchers can be informed of changes to
\r
445 # the feed. Before we do that, though, we remove the
\r
446 # show requester from the watchlist, if present, lest
\r
447 # he gets the update twice.
\r
448 if feed.watched_by?(m.replyto)
\r
449 we_were_watching = true
\r
450 feed.rm_watch(m.replyto)
\r
452 @bot.timer.reschedule(@watch[feed.handle], 0)
\r
453 if we_were_watching
\r
454 feed.add_watch(m.replyto)
\r
457 fetched = fetchRss(feed, m, false)
\r
459 return unless fetched or feed.xml
\r
460 if not fetched and feed.items
\r
461 m.reply "using old data"
\r
463 parsed = parseRss(feed, m)
\r
464 m.reply "using old data" unless parsed
\r
466 return unless feed.items
\r
470 # We sort the feeds in freshness order (newer ones first)
\r
471 items = freshness_sort(items)
\r
472 disp = items[ll..ul]
\r
473 disp.reverse! if rev
\r
475 m.reply "Channel : #{title}"
\r
476 disp.each do |item|
\r
477 printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
\r
481 def itemDate(item,ex=nil)
\r
482 return item.pubDate if item.respond_to?(:pubDate) and item.pubDate
\r
483 return item.date if item.respond_to?(:date) and item.date
\r
487 def freshness_sort(items)
\r
488 notime = Time.at(0)
\r
489 items.sort { |a, b|
\r
490 itemDate(b, notime) <=> itemDate(a, notime)
\r
494 def list_rss(m, params)
\r
495 wanted = params[:handle]
\r
497 @feeds.each { |handle, feed|
\r
498 next if wanted and !handle.match(/#{wanted}/i)
\r
499 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
500 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
501 (reply << " (watched)") if feed.watched_by?(m.replyto)
\r
505 reply = "no feeds found"
\r
506 reply << " matching #{wanted}" if wanted
\r
508 m.reply reply, :max_lines => reply.length
\r
511 def watched_rss(m, params)
\r
512 wanted = params[:handle]
\r
513 chan = params[:chan] || m.replyto
\r
515 watchlist.each { |handle, feed|
\r
516 next if wanted and !handle.match(/#{wanted}/i)
\r
517 next unless feed.watched_by?(chan)
\r
518 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
519 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
523 reply = "no watched feeds"
\r
524 reply << " matching #{wanted}" if wanted
\r
529 def who_watches(m, params)
\r
530 wanted = params[:handle]
\r
532 watchlist.each { |handle, feed|
\r
533 next if wanted and !handle.match(/#{wanted}/i)
\r
534 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
535 (reply << " refreshing every #{Utils.secs_to_string(feed.refresh_rate)}") if feed.refresh_rate
\r
536 reply << ": watched by #{feed.watchers.join(', ')}"
\r
540 reply = "no watched feeds"
\r
541 reply << " matching #{wanted}" if wanted
\r
546 def add_rss(m, params, force=false)
\r
547 handle = params[:handle]
\r
549 unless url.match(/https?/)
\r
550 m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
\r
553 type = params[:type]
\r
554 if @feeds.fetch(handle.downcase, nil) && !force
\r
555 m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"
\r
559 m.reply "You must specify both a handle and an url to add an RSS feed"
\r
562 @feeds[handle.downcase] = RssBlob.new(url,handle,type)
\r
563 reply = "Added RSS #{url} named #{handle}"
\r
565 reply << " (format: #{type})"
\r
571 def change_rss(m, params)
\r
572 handle = params[:handle].downcase
\r
573 feed = @feeds.fetch(handle, nil)
\r
575 m.reply "No such feed with handle #{handle}"
\r
578 case params[:what].intern
\r
580 new = params[:new].downcase
\r
581 if @feeds.key?(new) and @feeds[new]
\r
582 m.reply "There already is a feed with handle #{new}"
\r
585 feed.mutex.synchronize do
\r
587 @feeds.delete(handle)
\r
594 feed.mutex.synchronize do
\r
597 when :format, :type
\r
599 new = nil if new == 'default'
\r
600 feed.mutex.synchronize do
\r
604 new = params[:new].to_i
\r
605 new = nil if new == 0
\r
606 feed.mutex.synchronize do
\r
607 feed.refresh_rate = new
\r
610 m.reply "Don't know how to change #{params[:what]} for feeds"
\r
613 m.reply "Feed changed:"
\r
614 list_rss(m, {:handle => handle})
\r
617 def del_rss(m, params, pass=false)
\r
618 feed = unwatch_rss(m, params, true)
\r
621 m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
\r
624 @feeds.delete(feed.handle.downcase)
\r
629 def replace_rss(m, params)
\r
630 handle = params[:handle]
\r
631 if @feeds.key?(handle.downcase)
\r
632 del_rss(m, {:handle => handle}, true)
\r
634 if @feeds.key?(handle.downcase)
\r
635 m.reply "can't replace #{feed.handle}"
\r
637 add_rss(m, params, true)
\r
641 def forcereplace_rss(m, params)
\r
642 add_rss(m, params, true)
\r
645 def watch_rss(m, params)
\r
646 handle = params[:handle]
\r
647 chan = params[:chan] || m.replyto
\r
649 type = params[:type]
\r
653 feed = @feeds.fetch(handle.downcase, nil)
\r
655 if feed.add_watch(chan)
\r
659 m.reply "Already watching #{feed.handle} in #{chan}"
\r
662 m.reply "Couldn't watch feed #{handle} (no such feed found)"
\r
666 def unwatch_rss(m, params, pass=false)
\r
667 handle = params[:handle].downcase
\r
668 chan = params[:chan] || m.replyto
\r
669 unless @feeds.has_key?(handle)
\r
670 m.reply("dunno that feed")
\r
673 feed = @feeds[handle]
\r
674 if feed.rm_watch(chan)
\r
675 m.reply "#{chan} has been removed from the watchlist for #{feed.handle}"
\r
677 m.reply("#{chan} wasn't watching #{feed.handle}") unless pass
\r
685 def rewatch_rss(m=nil, params=nil)
\r
686 if params and handle = params[:handle]
\r
687 feed = @feeds.fetch(handle.downcase, nil)
\r
689 @bot.timer.reschedule(@watch[feed.handle], 0)
\r
692 m.reply _("no such feed %{handle}") % { :handle => handle } if m
\r
697 # Read watches from list.
\r
698 watchlist.each{ |handle, feed|
\r
706 def watchRss(feed, m=nil)
\r
707 if @watch.has_key?(feed.handle)
\r
708 report_problem("watcher thread for #{feed.handle} is already running", nil, m)
\r
712 status[:failures] = 0
\r
714 if feed.last_fetched
\r
715 tmout = feed.last_fetched + calculate_timeout(feed) - Time.now
\r
716 tmout = 0 if tmout < 0
\r
718 debug "scheduling a watcher for #{feed} in #{tmout} seconds"
\r
719 @watch[feed.handle] = @bot.timer.add(tmout) {
\r
720 debug "watcher for #{feed} wakes up"
\r
721 failures = status[:failures]
\r
723 debug "fetching #{feed}"
\r
724 first_run = !feed.last_fetched
\r
725 oldxml = feed.xml ? feed.xml.dup : nil
\r
726 unless fetchRss(feed)
\r
730 debug "first run for #{feed}, getting items"
\r
732 elsif oldxml and oldxml == feed.xml
\r
733 debug "xml for #{feed} didn't change"
\r
734 failures -= 1 if failures > 0
\r
737 debug "no previous items in feed #{feed}"
\r
739 failures -= 1 if failures > 0
\r
741 # This one is used for debugging
\r
744 # These are used for checking new items vs old ones
\r
745 uid_opts = { :show_updated => @bot.config['rss.show_updated'] }
\r
746 oids = Set.new feed.items.map { |item|
\r
747 uid = RSS.item_uid_for_bot(item, uid_opts)
\r
749 debug [uid, item].inspect
\r
750 debug [uid, otxt.last].inspect
\r
754 unless parseRss(feed)
\r
755 debug "no items in feed #{feed}"
\r
758 debug "Checking if new items are available for #{feed}"
\r
759 failures -= 1 if failures > 0
\r
765 dispItems = feed.items.reject { |item|
\r
766 uid = RSS.item_uid_for_bot(item, uid_opts)
\r
768 if oids.include?(uid)
\r
769 debug "rejecting old #{uid} #{item.inspect}"
\r
770 debug [uid, txt].inspect
\r
773 debug "accepting new #{uid} #{item.inspect}"
\r
774 debug [uid, txt].inspect
\r
775 warning "same text! #{txt}" if otxt.include?(txt)
\r
780 if dispItems.length > 0
\r
781 debug "Found #{dispItems.length} new items in #{feed}"
\r
782 # When displaying watched feeds, publish them from older to newer
\r
783 dispItems.reverse.each { |item|
\r
784 printFormattedRss(feed, item)
\r
787 debug "No new items found in #{feed}"
\r
793 rescue Exception => e
\r
794 error "Error watching #{feed}: #{e.inspect}"
\r
795 debug e.backtrace.join("\n")
\r
799 status[:failures] = failures
\r
801 seconds = calculate_timeout(feed, failures)
\r
802 debug "watcher for #{feed} going to sleep #{seconds} seconds.."
\r
804 @bot.timer.reschedule(@watch[feed.handle], seconds)
\r
806 warning "watcher for #{feed} failed to reschedule: #{$!.inspect}"
\r
809 debug "watcher for #{feed} added"
\r
812 def calculate_timeout(feed, failures = 0)
\r
813 seconds = @bot.config['rss.thread_sleep']
\r
814 feed.mutex.synchronize do
\r
815 seconds = feed.refresh_rate if feed.refresh_rate
\r
817 seconds *= failures + 1
\r
818 seconds += seconds * (rand(100)-50)/100
\r
822 def select_nonempty(*ar)
\r
824 ret = ar.map { |i| (i && i.empty?) ? nil : i }.compact.first
\r
825 (ret && ret.empty?) ? nil : ret
\r
828 def printFormattedRss(feed, item, opts=nil)
\r
830 places = feed.watchers
\r
831 handle = "::#{feed.handle}:: "
\r
834 places = opts[:places] if opts.key?(:places)
\r
835 handle = opts[:handle].to_s if opts.key?(:handle)
\r
836 if opts.key?(:date) && opts[:date]
\r
837 if item.respond_to?(:updated)
\r
838 if item.updated.content.class <= Time
\r
839 date = item.updated.content.strftime("%Y/%m/%d %H:%M")
\r
841 date = item.updated.content.to_s
\r
843 elsif item.respond_to?(:source) and item.source.respond_to?(:updated)
\r
844 if item.source.updated.content.class <= Time
\r
845 date = item.source.updated.content.strftime("%Y/%m/%d %H:%M")
\r
847 date = item.source.updated.content.to_s
\r
849 elsif item.respond_to?(:pubDate)
\r
850 if item.pubDate.class <= Time
\r
851 date = item.pubDate.strftime("%Y/%m/%d %H:%M")
\r
853 date = item.pubDate.to_s
\r
855 elsif item.respond_to?(:date)
\r
856 if item.date.class <= Time
\r
857 date = item.date.strftime("%Y/%m/%d %H:%M")
\r
859 date = item.date.to_s
\r
869 # Twitters don't need a cap on the title length since they have a hard
\r
870 # limit to 160 characters, and most of them are under 140 characters
\r
871 tit_opt[:limit] = @bot.config['rss.head_max'] unless feed.type == 'twitter'
\r
874 base_title = item.title.to_s.dup
\r
875 # git changesets are SHA1 hashes (40 hex digits), way too long, get rid of them, as they are
\r
876 # visible in the URL anyway
\r
877 # TODO make this optional?
\r
878 base_title.sub!(/^Changeset \[([\da-f]{40})\]:/) { |c| "(git commit)"} if feed.type == 'trac'
\r
879 title = "#{Bold}#{base_title.ircify_html(tit_opt)}#{Bold}"
\r
883 :limit => @bot.config['rss.text_max'],
\r
884 :a_href => :link_out
\r
887 # We prefer content_encoded here as it tends to provide more html formatting
\r
888 # for use with ircify_html.
\r
889 if item.respond_to?(:content_encoded) && item.content_encoded
\r
890 desc = item.content_encoded.ircify_html(desc_opt)
\r
891 elsif item.respond_to?(:description) && item.description
\r
892 desc = item.description.ircify_html(desc_opt)
\r
894 if item.content.type == "html"
\r
895 desc = item.content.content.ircify_html(desc_opt)
\r
897 desc = item.content.content
\r
898 if desc.size > desc_opt[:limit]
\r
899 desc = desc.slice(0, desc_opt[:limit]) + "#{Reverse}...#{Reverse}"
\r
904 link = item.link.href rescue item.link.chomp rescue nil
\r
906 category = select_nonempty((item.category.content rescue nil), (item.dc_subject rescue nil))
\r
907 author = select_nonempty((item.author.name.content rescue nil), (item.dc_creator rescue nil), (item.author rescue nil))
\r
912 at = ((item.title && item.link) ? ' @ ' : '')
\r
916 author += " " if author
\r
917 abt = category ? "about #{category} " : ""
\r
918 line1 = "#{handle}#{date}#{author}blogged #{abt}at #{link}"
\r
919 line2 = "#{handle}#{title} - #{desc}"
\r
921 author += " " if author
\r
922 line1 = "#{handle}#{date}#{author}commited #{title} @ #{link}"
\r
924 line1 = "#{handle}#{date}#{title}#{at}#{link}"
\r
926 line1 = "#{handle}#{date}#{title}#{at}#{link} has been edited by #{author}. #{desc}"
\r
928 line1 = "#{handle}#{date}Message #{title} sent by #{author}. #{desc}"
\r
930 line1 = "#{handle}#{date}#{title} @ #{link}"
\r
931 unless item.title =~ /^(?:Changeset \[(?:[\da-f]+)\]|\(git commit\))/
\r
932 line2 = "#{handle}#{date}#{desc}"
\r
935 dept = "(from the #{item.slash_department} dept) " rescue nil
\r
936 sec = " in section #{item.slash_section}" rescue nil
\r
938 line1 = "#{handle}#{date}#{dept}#{title}#{at}#{link} (posted by #{author}#{sec})"
\r
940 line1 = "#{handle}#{date}#{title}#{at}#{link}"
\r
941 line1 << " (by #{author})" if author
\r
943 places.each { |loc|
\r
944 @bot.say loc, line1, :overlong => :truncate
\r
946 @bot.say loc, line2, :overlong => :truncate
\r
950 def fetchRss(feed, m=nil, cache=true)
\r
951 feed.last_fetched = Time.now
\r
953 # Use 60 sec timeout, cause the default is too low
\r
954 xml = @bot.httputil.get(feed.url,
\r
955 :read_timeout => 60,
\r
956 :open_timeout => 60,
\r
958 rescue URI::InvalidURIError, URI::BadURIError => e
\r
959 report_problem("invalid rss feed #{feed.url}", e, m)
\r
962 report_problem("error getting #{feed.url}", e, m)
\r
965 debug "fetched #{feed}"
\r
967 report_problem("reading feed #{feed} failed", nil, m)
\r
970 # Ok, 0.9 feeds are not supported, maybe because
\r
971 # Netscape happily removed the DTD. So what we do is just to
\r
972 # reassign the 0.9 RDFs to 1.0, and hope it goes right.
\r
973 xml.gsub!("xmlns=\"http://my.netscape.com/rdf/simple/0.9/\"",
\r
974 "xmlns=\"http://purl.org/rss/1.0/\"")
\r
975 feed.mutex.synchronize do
\r
981 def parseRss(feed, m=nil)
\r
982 return nil unless feed.xml
\r
983 feed.mutex.synchronize do
\r
986 ## do validate parse
\r
987 rss = RSS::Parser.parse(xml)
\r
988 debug "parsed and validated #{feed}"
\r
989 rescue RSS::InvalidRSSError
\r
990 ## do non validate parse for invalid RSS 1.0
\r
992 rss = RSS::Parser.parse(xml, false)
\r
993 debug "parsed but not validated #{feed}"
\r
994 rescue RSS::Error => e
\r
995 report_problem("parsing rss stream failed, whoops =(", e, m)
\r
998 rescue RSS::Error => e
\r
999 report_problem("parsing rss stream failed, oioi", e, m)
\r
1002 report_problem("processing error occured, sorry =(", e, m)
\r
1007 report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
\r
1010 rss.output_encoding = 'UTF-8'
\r
1011 rescue RSS::UnknownConvertMethod => e
\r
1012 report_problem("bah! something went wrong =(", e, m)
\r
1015 if rss.respond_to? :channel
\r
1016 rss.channel.title ||= "Unknown"
\r
1017 title = rss.channel.title
\r
1019 title = rss.title.content
\r
1021 rss.items.each do |item|
\r
1022 item.title ||= "Unknown"
\r
1028 report_problem("no items found in the feed, maybe try weed?", e, m)
\r
1031 feed.title = title
\r
1032 feed.items = items
\r
1038 plugin = RSSFeedsPlugin.new
\r
1040 plugin.default_auth( 'edit', false )
\r
1041 plugin.default_auth( 'edit:add', true)
\r
1043 plugin.map 'rss show :handle :limit',
\r
1044 :action => 'show_rss',
\r
1045 :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
\r
1046 :defaults => {:limit => 5}
\r
1047 plugin.map 'rss list :handle',
\r
1048 :action => 'list_rss',
\r
1049 :defaults => {:handle => nil}
\r
1050 plugin.map 'rss watched :handle [in :chan]',
\r
1051 :action => 'watched_rss',
\r
1052 :defaults => {:handle => nil}
\r
1053 plugin.map 'rss who watches :handle',
\r
1054 :action => 'who_watches',
\r
1055 :defaults => {:handle => nil}
\r
1056 plugin.map 'rss add :handle :url :type',
\r
1057 :action => 'add_rss',
\r
1058 :auth_path => 'edit',
\r
1059 :defaults => {:type => nil}
\r
1060 plugin.map 'rss change :what of :handle to :new',
\r
1061 :action => 'change_rss',
\r
1062 :auth_path => 'edit',
\r
1063 :requirements => { :what => /handle|url|format|type|refresh/ }
\r
1064 plugin.map 'rss change :what for :handle to :new',
\r
1065 :action => 'change_rss',
\r
1066 :auth_path => 'edit',
\r
1067 :requirements => { :what => /handle|url|format|type|refesh/ }
\r
1068 plugin.map 'rss del :handle',
\r
1069 :auth_path => 'edit:rm!',
\r
1070 :action => 'del_rss'
\r
1071 plugin.map 'rss delete :handle',
\r
1072 :auth_path => 'edit:rm!',
\r
1073 :action => 'del_rss'
\r
1074 plugin.map 'rss rm :handle',
\r
1075 :auth_path => 'edit:rm!',
\r
1076 :action => 'del_rss'
\r
1077 plugin.map 'rss replace :handle :url :type',
\r
1078 :auth_path => 'edit',
\r
1079 :action => 'replace_rss',
\r
1080 :defaults => {:type => nil}
\r
1081 plugin.map 'rss forcereplace :handle :url :type',
\r
1082 :auth_path => 'edit',
\r
1083 :action => 'forcereplace_rss',
\r
1084 :defaults => {:type => nil}
\r
1085 plugin.map 'rss watch :handle [in :chan]',
\r
1086 :action => 'watch_rss',
\r
1087 :defaults => {:url => nil, :type => nil}
\r
1088 plugin.map 'rss watch :handle :url :type [in :chan]',
\r
1089 :action => 'watch_rss',
\r
1090 :defaults => {:url => nil, :type => nil}
\r
1091 plugin.map 'rss unwatch :handle [in :chan]',
\r
1092 :action => 'unwatch_rss'
\r
1093 plugin.map 'rss rmwatch :handle [in :chan]',
\r
1094 :action => 'unwatch_rss'
\r
1095 plugin.map 'rss rewatch [:handle]',
\r
1096 :action => 'rewatch_rss'
\r