4 # RSS feed plugin for RubyBot
\r
5 # (c) 2004 Stanislav Karchebny <berkus@madfire.net>
\r
6 # (c) 2005 Ian Monroe <ian@monroe.nu>
\r
7 # (c) 2005 Mark Kretschmann <markey@web.de>
\r
8 # (c) 2006 Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
\r
10 # Licensed under MIT License.
\r
12 require 'rss/parser'
\r
15 require 'rss/dublincore'
\r
17 # require 'rss/dublincore/2.0'
\r
19 # warning "Unable to load RSS libraries, RSS plugin functionality crippled"
\r
24 if self.length > limit
\r
25 self+". " =~ /^(.{#{limit}}[^.!;?]*[.!;?])/mi
\r
32 self.gsub(/<[^>]+>/, '').gsub(/&/,'&').gsub(/"/,'"').gsub(/</,'<').gsub(/>/,'>').gsub(/&ellip;/,'...').gsub(/'/, "'").gsub("\n",'')
\r
36 self.gsub(/'/, "''")
\r
46 def initialize(url,handle=nil,type=nil,watchers=[])
\r
55 sanitize_watchers(watchers)
\r
58 # Downcase all watchers, possibly turning them into Strings if they weren't
\r
59 def sanitize_watchers(list=@watchers)
\r
71 def watched_by?(who)
\r
72 @watchers.include?(who.downcase)
\r
79 @watchers << who.downcase
\r
84 @watchers.delete(who.downcase)
\r
88 [@handle,@url,@type,@watchers]
\r
91 def to_s(watchers=false)
\r
93 a = self.to_a.flatten
\r
97 a.compact.join(" | ")
\r
101 class RSSFeedsPlugin < Plugin
\r
102 BotConfig.register BotConfigIntegerValue.new('rss.head_max',
\r
103 :default => 30, :validate => Proc.new{|v| v > 0 && v < 200},
\r
104 :desc => "How many characters to use of a RSS item header")
\r
106 BotConfig.register BotConfigIntegerValue.new('rss.text_max',
\r
107 :default => 90, :validate => Proc.new{|v| v > 0 && v < 400},
\r
108 :desc => "How many characters to use of a RSS item text")
\r
110 BotConfig.register BotConfigIntegerValue.new('rss.thread_sleep',
\r
111 :default => 300, :validate => Proc.new{|v| v > 30},
\r
112 :desc => "How many seconds to sleep before checking RSS feeds again")
\r
114 @@watchThreads = Hash.new
\r
115 @@mutex = Mutex.new
\r
120 if @registry.has_key?(:feeds)
\r
121 @feeds = @registry[:feeds]
\r
122 @feeds.keys.grep(/[A-Z]/) { |k|
\r
123 @feeds[k.downcase] = @feeds[k]
\r
126 @feeds.each { |k, f|
\r
127 f.sanitize_watchers
\r
140 @feeds.select { |h, f| f.watched? }
\r
148 @registry[:feeds] = @feeds
\r
152 @@mutex.synchronize {
\r
153 # Abort all running threads.
\r
154 @@watchThreads.each { |url, thread|
\r
155 debug "Killing thread for #{url}"
\r
158 @@watchThreads = Hash.new
\r
162 def help(plugin,topic="")
\r
165 "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
167 "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
\r
169 "rss watched [#{Bold}handle#{Bold}] : list all watched rss feeds (matching #{Bold}handle#{Bold})"
\r
171 "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
172 when /^(del(ete)?|rm)$/
\r
173 "rss del(ete)|rm #{Bold}handle#{Bold} : delete rss feed #{Bold}handle#{Bold}"
\r
175 "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
176 when "forcereplace"
\r
177 "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
179 "rss watch #{Bold}handle#{Bold} [#{Bold}url#{Bold} [#{Bold}type#{Bold}]] : watch rss #{Bold}handle#{Bold} for changes; when the other parameters are present, it will be created if it doesn't exist yet"
\r
180 when /(un|rm)watch/
\r
181 "rss unwatch|rmwatch #{Bold}handle#{Bold} : stop watching rss #{Bold}handle#{Bold} for changes"
\r
183 "rss rewatch : restart threads that watch for changes in watched rss"
\r
185 "manage RSS feeds: rss show|list|watched|add|del(ete)|rm|(force)replace|watch|unwatch|rmwatch|rewatch"
\r
189 def report_problem(report, e=nil, m=nil)
\r
190 if m && m.respond_to?(:reply)
\r
197 debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
\r
201 def show_rss(m, params)
\r
202 handle = params[:handle]
\r
203 lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
\r
204 debug lims.to_a.inspect
\r
206 ll = [[lims[1].to_i-1,lims[2].to_i-1].min, 0].max
\r
207 ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
\r
208 rev = lims[1].to_i > lims[2].to_i
\r
211 ul = [[lims[1].to_i-1, 0].max, 14].min
\r
215 feed = @feeds.fetch(handle.downcase, nil)
\r
217 m.reply "I don't know any feeds named #{handle}"
\r
221 m.reply "lemme fetch it..."
\r
222 title = items = nil
\r
223 @@mutex.synchronize {
\r
224 title, items = fetchRss(feed, m)
\r
226 return unless items
\r
228 # We sort the feeds in freshness order (newer ones first)
\r
229 items = freshness_sort(items)
\r
230 disp = items[ll..ul]
\r
231 disp.reverse! if rev
\r
233 m.reply "Channel : #{title}"
\r
234 disp.each do |item|
\r
235 printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
\r
239 def itemDate(item,ex=nil)
\r
240 return item.pubDate if item.respond_to?(:pubDate) and item.pubDate
\r
241 return item.date if item.respond_to?(:date) and item.date
\r
245 def freshness_sort(items)
\r
246 notime = Time.at(0)
\r
247 items.sort { |a, b|
\r
248 itemDate(b, notime) <=> itemDate(a, notime)
\r
252 def list_rss(m, params)
\r
253 wanted = params[:handle]
\r
255 @@mutex.synchronize {
\r
256 @feeds.each { |handle, feed|
\r
257 next if wanted and !handle.match(/#{wanted}/i)
\r
258 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
\r
259 (reply << " (watched)") if feed.watched_by?(m.replyto)
\r
264 reply = "no feeds found"
\r
265 reply << " matching #{wanted}" if wanted
\r
270 def watched_rss(m, params)
\r
271 wanted = params[:handle]
\r
273 @@mutex.synchronize {
\r
274 watchlist.each { |handle, feed|
\r
275 next if wanted and !handle.match(/#{wanted}/i)
\r
276 next unless feed.watched_by?(m.replyto)
\r
277 reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})\n"
\r
281 reply = "no watched feeds"
\r
282 reply << " matching #{wanted}" if wanted
\r
287 def add_rss(m, params, force=false)
\r
288 handle = params[:handle]
\r
290 unless url.match(/https?/)
\r
291 m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
\r
294 type = params[:type]
\r
295 if @feeds.fetch(handle.downcase, nil) && !force
\r
296 m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"
\r
300 m.reply "You must specify both a handle and an url to add an RSS feed"
\r
303 @@mutex.synchronize {
\r
304 @feeds[handle.downcase] = RssBlob.new(url,handle,type)
\r
306 reply = "Added RSS #{url} named #{handle}"
\r
308 reply << " (format: #{type})"
\r
314 def del_rss(m, params, pass=false)
\r
315 feed = unwatch_rss(m, params, true)
\r
317 m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
\r
320 @@mutex.synchronize {
\r
321 @feeds.delete(feed.handle.downcase)
\r
327 def replace_rss(m, params)
\r
328 handle = params[:handle]
\r
329 if @feeds.key?(handle.downcase)
\r
330 del_rss(m, {:handle => handle}, true)
\r
332 if @feeds.key?(handle.downcase)
\r
333 m.reply "can't replace #{feed.handle}"
\r
335 add_rss(m, params, true)
\r
339 def forcereplace_rss(m, params)
\r
340 add_rss(m, params, true)
\r
343 def watch_rss(m, params)
\r
344 handle = params[:handle]
\r
346 type = params[:type]
\r
351 @@mutex.synchronize {
\r
352 feed = @feeds.fetch(handle.downcase, nil)
\r
355 @@mutex.synchronize {
\r
356 if feed.add_watch(m.replyto)
\r
360 m.reply "Already watching #{feed.handle}"
\r
364 m.reply "Couldn't watch feed #{handle} (no such feed found)"
\r
368 def unwatch_rss(m, params, pass=false)
\r
369 handle = params[:handle].downcase
\r
370 unless @feeds.has_key?(handle)
\r
371 m.reply("dunno that feed")
\r
374 feed = @feeds[handle]
\r
375 if feed.rm_watch(m.replyto)
\r
376 m.reply "#{m.replyto} has been removed from the watchlist for #{feed.handle}"
\r
378 m.reply("#{m.replyto} wasn't watching #{feed.handle}") unless pass
\r
381 @@mutex.synchronize {
\r
382 if @@watchThreads[handle].kind_of? Thread
\r
383 @@watchThreads[handle].kill
\r
384 debug "rmwatch: Killed thread for #{handle}"
\r
385 @@watchThreads.delete(handle)
\r
392 def rewatch_rss(m=nil, params=nil)
\r
395 # Read watches from list.
\r
396 watchlist.each{ |handle, feed|
\r
403 def watchRss(feed, m=nil)
\r
404 if @@watchThreads.has_key?(feed.handle)
\r
405 report_problem("watcher thread for #{feed.handle} is already running", nil, m)
\r
408 @@watchThreads[feed.handle] = Thread.new do
\r
409 debug "watcher for #{feed} started"
\r
415 debug "fetching #{feed}"
\r
416 title = newItems = nil
\r
417 @@mutex.synchronize {
\r
418 title, newItems = fetchRss(feed)
\r
421 debug "no items in feed #{feed}"
\r
424 debug "Checking if new items are available for #{feed}"
\r
426 debug "First run, we'll see next time"
\r
429 otxt = oldItems.map { |item| item.to_s }
\r
430 dispItems = newItems.reject { |item|
\r
431 otxt.include?(item.to_s)
\r
433 if dispItems.length > 0
\r
434 debug "Found #{dispItems.length} new items in #{feed}"
\r
435 # When displaying watched feeds, publish them from older to newer
\r
436 dispItems.reverse.each { |item|
\r
437 @@mutex.synchronize {
\r
438 printFormattedRss(feed, item)
\r
442 debug "No new items found in #{feed}"
\r
445 oldItems = newItems.dup
\r
447 rescue Exception => e
\r
448 error "Error watching #{feed}: #{e.inspect}"
\r
449 debug e.backtrace.join("\n")
\r
453 seconds = @bot.config['rss.thread_sleep'] * (failures + 1)
\r
454 seconds += seconds * (rand(100)-50)/100
\r
455 debug "watcher for #{feed} going to sleep #{seconds} seconds.."
\r
461 def printFormattedRss(feed, item, opts=nil)
\r
462 places = feed.watchers
\r
463 handle = "::#{feed.handle}:: "
\r
466 places = opts[:places] if opts.key?(:places)
\r
467 handle = opts[:handle].to_s if opts.key?(:handle)
\r
468 if opts.key?(:date) && opts[:date]
\r
469 if item.respond_to?(:pubDate)
\r
470 if item.pubDate.class <= Time
\r
471 date = item.pubDate.strftime("%Y/%m/%d %H.%M.%S")
\r
473 date = item.pubDate.to_s
\r
475 elsif item.respond_to?(:date)
\r
476 if item.date.class <= Time
\r
477 date = item.date.strftime("%Y/%m/%d %H.%M.%S")
\r
479 date = item.date.to_s
\r
487 title = "#{Bold}#{item.title.chomp.riphtml}#{Bold}" if item.title
\r
488 desc = item.description.gsub(/\s+/,' ').strip.riphtml.shorten(@bot.config['rss.text_max']) if item.description
\r
489 link = item.link.chomp if item.link
\r
490 places.each { |loc|
\r
493 @bot.say loc, "#{handle}#{date}#{item.category.content} blogged at #{link}"
\r
494 @bot.say loc, "#{handle}#{title} - #{desc}"
\r
496 @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
\r
498 @bot.say loc, "#{handle}#{date}#{item.title} has been edited by #{item.dc_creator}. #{desc} #{link}"
\r
500 @bot.say loc, "#{handle}#{date}Message #{title} sent by #{item.dc_creator}. #{desc}"
\r
502 @bot.say loc, "#{handle}#{date}#{title} @ #{link}"
\r
503 unless item.title =~ /^Changeset \[(\d+)\]/
\r
504 @bot.say loc, "#{handle}#{date}#{desc}"
\r
507 @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
\r
512 def fetchRss(feed, m=nil)
\r
514 # Use 60 sec timeout, cause the default is too low
\r
515 # Do not use get_cached for RSS until we have proper cache handling
\r
516 # xml = @bot.httputil.get_cached(feed.url,60,60)
\r
517 xml = @bot.httputil.get_cached(feed.url, 60, 60)
\r
518 rescue URI::InvalidURIError, URI::BadURIError => e
\r
519 report_problem("invalid rss feed #{feed.url}", e, m)
\r
522 report_problem("error getting #{feed.url}", e, m)
\r
525 debug "fetched #{feed}"
\r
527 report_problem("reading feed #{feed} failed", nil, m)
\r
532 ## do validate parse
\r
533 rss = RSS::Parser.parse(xml)
\r
534 debug "parsed #{feed}"
\r
535 rescue RSS::InvalidRSSError
\r
536 ## do non validate parse for invalid RSS 1.0
\r
538 rss = RSS::Parser.parse(xml, false)
\r
539 rescue RSS::Error => e
\r
540 report_problem("parsing rss stream failed, whoops =(", e, m)
\r
543 rescue RSS::Error => e
\r
544 report_problem("parsing rss stream failed, oioi", e, m)
\r
547 report_problem("processing error occured, sorry =(", e, m)
\r
552 report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
\r
555 rss.output_encoding = 'UTF-8'
\r
556 rescue RSS::UnknownConvertMethod => e
\r
557 report_problem("bah! something went wrong =(", e, m)
\r
560 rss.channel.title ||= "Unknown"
\r
561 title = rss.channel.title
\r
562 rss.items.each do |item|
\r
563 item.title ||= "Unknown"
\r
569 report_problem("no items found in the feed, maybe try weed?", e, m)
\r
572 return [title, items]
\r
576 plugin = RSSFeedsPlugin.new
\r
578 plugin.map 'rss show :handle :limit',
\r
579 :action => 'show_rss',
\r
580 :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
\r
581 :defaults => {:limit => 5}
\r
582 plugin.map 'rss list :handle',
\r
583 :action => 'list_rss',
\r
584 :defaults => {:handle => nil}
\r
585 plugin.map 'rss watched :handle',
\r
586 :action => 'watched_rss',
\r
587 :defaults => {:handle => nil}
\r
588 plugin.map 'rss add :handle :url :type',
\r
589 :action => 'add_rss',
\r
590 :defaults => {:type => nil}
\r
591 plugin.map 'rss del :handle',
\r
592 :action => 'del_rss'
\r
593 plugin.map 'rss delete :handle',
\r
594 :action => 'del_rss'
\r
595 plugin.map 'rss rm :handle',
\r
596 :action => 'del_rss'
\r
597 plugin.map 'rss replace :handle :url :type',
\r
598 :action => 'replace_rss',
\r
599 :defaults => {:type => nil}
\r
600 plugin.map 'rss forcereplace :handle :url :type',
\r
601 :action => 'forcereplace_rss',
\r
602 :defaults => {:type => nil}
\r
603 plugin.map 'rss watch :handle :url :type',
\r
604 :action => 'watch_rss',
\r
605 :defaults => {:url => nil, :type => nil}
\r
606 plugin.map 'rss unwatch :handle',
\r
607 :action => 'unwatch_rss'
\r
608 plugin.map 'rss rmwatch :handle',
\r
609 :action => 'unwatch_rss'
\r
610 plugin.map 'rss rewatch',
\r
611 :action => 'rewatch_rss'
\r