X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Frss.rb;h=13689bf14efdfeb63136de3a437ecbd64a71a6a3;hb=9d29f400bb3a354779185d61049ce7cdfa7744ee;hp=0e6e5530730a06d104eaf8ed6f06943797d251f5;hpb=a325446ce0a058d4e561aabcff11ceb0ad3e3cf1;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/rss.rb b/data/rbot/plugins/rss.rb
index 0e6e5530..13689bf1 100644
--- a/data/rbot/plugins/rss.rb
+++ b/data/rbot/plugins/rss.rb
@@ -1,18 +1,23 @@
+#-- vim:sw=2:et
+#++
+#
 # RSS feed plugin for RubyBot
 # (c) 2004 Stanislav Karchebny <berkus@madfire.net>
 # (c) 2005 Ian Monroe <ian@monroe.nu>
 # (c) 2005 Mark Kretschmann <markey@web.de>
+# (c) 2006 Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
+#
 # Licensed under MIT License.
 
 require 'rss/parser'
 require 'rss/1.0'
 require 'rss/2.0'
 require 'rss/dublincore'
-begin
-  # require 'rss/dublincore/2.0'
-rescue
-  warning "Unable to load RSS libraries, RSS plugin functionality crippled"
-end
+# begin
+#   require 'rss/dublincore/2.0'
+# rescue
+#   warning "Unable to load RSS libraries, RSS plugin functionality crippled"
+# end
 
 class ::String
   def shorten(limit)
@@ -26,10 +31,6 @@ class ::String
   def riphtml
     self.gsub(/<[^>]+>/, '').gsub(/&amp;/,'&').gsub(/&quot;/,'"').gsub(/&lt;/,'<').gsub(/&gt;/,'>').gsub(/&ellip;/,'...').gsub(/&apos;/, "'").gsub("\n",'')
   end
-
-  def mysqlize
-    self.gsub(/'/, "''")
-  end
 end
 
 class ::RssBlob
@@ -46,7 +47,17 @@ class ::RssBlob
       @handle = url
     end
     @type = type
-    @watchers = watchers
+    @watchers=[]
+    sanitize_watchers(watchers)
+  end
+
+  # Downcase all watchers, possibly turning them into Strings if they weren't
+  def sanitize_watchers(list=@watchers)
+    ls = list.dup
+    @watchers.clear
+    ls.each { |w|
+      add_watch(w)
+    }
   end
 
   def watched?
@@ -54,68 +65,104 @@ class ::RssBlob
   end
 
   def watched_by?(who)
-    @watchers.include?(who)
+    @watchers.include?(who.downcase)
   end
 
   def add_watch(who)
     if watched_by?(who)
       return nil
     end
-    @watchers << who unless watched_by?(who)
+    @watchers << who.downcase
     return who
   end
 
   def rm_watch(who)
-    @watchers.delete(who)
+    @watchers.delete(who.downcase)
+  end
+
+  def to_a
+    [@handle,@url,@type,@watchers]
   end
 
-  #  def to_ary
-  #    [@handle,@url,@type,@watchers]
-  #  end
+  def to_s(watchers=false)
+    if watchers
+      a = self.to_a.flatten
+    else
+      a = self.to_a[0,3]
+    end
+    a.compact.join(" | ")
+  end
 end
 
 class RSSFeedsPlugin < Plugin
-  @@watchThreads = Hash.new
-  @@mutex = Mutex.new
+  BotConfig.register BotConfigIntegerValue.new('rss.head_max',
+    :default => 30, :validate => Proc.new{|v| v > 0 && v < 200},
+    :desc => "How many characters to use of a RSS item header")
+
+  BotConfig.register BotConfigIntegerValue.new('rss.text_max',
+    :default => 90, :validate => Proc.new{|v| v > 0 && v < 400},
+    :desc => "How many characters to use of a RSS item text")
+
+  BotConfig.register BotConfigIntegerValue.new('rss.thread_sleep',
+    :default => 300, :validate => Proc.new{|v| v > 30},
+    :desc => "How many seconds to sleep before checking RSS feeds again")
 
   def initialize
     super
-    kill_threads
     if @registry.has_key?(:feeds)
       @feeds = @registry[:feeds]
+      @feeds.keys.grep(/[A-Z]/) { |k|
+        @feeds[k.downcase] = @feeds[k]
+        @feeds.delete(k)
+      }
+      @feeds.each { |k, f|
+        f.sanitize_watchers
+      }
     else
       @feeds = Hash.new
     end
+    @watch = Hash.new
     rewatch_rss
   end
 
+  def name
+    "rss"
+  end
+
   def watchlist
     @feeds.select { |h, f| f.watched? }
   end
 
   def cleanup
-    kill_threads
+    stop_watches
   end
 
   def save
     @registry[:feeds] = @feeds
   end
 
-  def kill_threads
-    @@mutex.synchronize {
-      # Abort all running threads.
-      @@watchThreads.each { |url, thread|
-        debug "Killing thread for #{url}"
-        thread.kill
-      }
-      @@watchThreads = Hash.new
+  def stop_watch(handle)
+    if @watch.has_key?(handle)
+      begin
+        debug "Stopping watch #{handle}"
+        @bot.timer.remove(@watch[handle])
+        @watch.delete(handle)
+      rescue => e
+        report_problem("Failed to stop watch for #{handle}", e, nil)
+      end
+    end
+  end
+
+  def stop_watches
+    @watch.each_key { |k|
+      stop_watch(k)
     }
   end
 
   def help(plugin,topic="")
     case topic
     when "show"
-      "rss show #{Bold}handle#{Bold} [#{Bold}limit#{Bold}] : show #{Bold}limit#{Bold} (default: 5, max: 15) entries from rss #{Bold}handle#{Bold}"
+      "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"
     when "list"
       "rss list [#{Bold}handle#{Bold}] : list all rss feeds (matching #{Bold}handle#{Bold})"
     when "watched"
@@ -139,47 +186,75 @@ class RSSFeedsPlugin < Plugin
     end
   end
 
-  def report_problem(report, m=nil)
-    if m
+  def report_problem(report, e=nil, m=nil)
+    if m && m.respond_to?(:reply)
       m.reply report
     else
       warning report
     end
+    if e
+      debug e.inspect
+      debug e.backtrace.join("\n") if e.respond_to?(:backtrace)
+    end
   end
 
   def show_rss(m, params)
     handle = params[:handle]
-    limit = params[:limit].to_i
-    limit = 15 if limit > 15
-    limit = 1 if limit <= 0
-    feed = @feeds.fetch(handle, nil)
+    lims = params[:limit].to_s.match(/(\d+)(?:..(\d+))?/)
+    debug lims.to_a.inspect
+    if lims[2]
+      ll = [[lims[1].to_i-1,lims[2].to_i-1].min,  0].max
+      ul = [[lims[1].to_i-1,lims[2].to_i-1].max, 14].min
+      rev = lims[1].to_i > lims[2].to_i
+    else
+      ll = 0
+      ul = [[lims[1].to_i-1, 0].max, 14].min
+      rev = false
+    end
+
+    feed = @feeds.fetch(handle.downcase, nil)
     unless feed
       m.reply "I don't know any feeds named #{handle}"
       return
     end
-    m.reply("Please wait, querying...")
+
+    m.reply "lemme fetch it..."
     title = items = nil
-    @@mutex.synchronize {
-      title, items = fetchRss(feed, m)
-    }
+    title, items = fetchRss(feed, m)
     return unless items
-    m.reply("Channel : #{title}")
-    # TODO: optional by-date sorting if dates present
-    items[0...limit].reverse.each do |item|
-      printRssItem(m.replyto,item)
+
+    # We sort the feeds in freshness order (newer ones first)
+    items = freshness_sort(items)
+    disp = items[ll..ul]
+    disp.reverse! if rev
+
+    m.reply "Channel : #{title}"
+    disp.each do |item|
+      printFormattedRss(feed, item, {:places=>[m.replyto],:handle=>nil,:date=>true})
     end
   end
 
+  def itemDate(item,ex=nil)
+    return item.pubDate if item.respond_to?(:pubDate) and item.pubDate
+    return item.date if item.respond_to?(:date) and item.date
+    return ex
+  end
+
+  def freshness_sort(items)
+    notime = Time.at(0)
+    items.sort { |a, b|
+      itemDate(b, notime) <=> itemDate(a, notime)
+    }
+  end
+
   def list_rss(m, params)
     wanted = params[:handle]
     reply = String.new
-    @@mutex.synchronize {
-      @feeds.each { |handle, feed|
-        next if wanted and !handle.match(wanted)
-        reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
-        (reply << " (watched)") if feed.watched_by?(m.replyto)
-        reply << "\n"
-      }
+    @feeds.each { |handle, feed|
+      next if wanted and !handle.match(/#{wanted}/i)
+      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})"
+      (reply << " (watched)") if feed.watched_by?(m.replyto)
+      reply << "\n"
     }
     if reply.empty?
       reply = "no feeds found"
@@ -191,12 +266,10 @@ class RSSFeedsPlugin < Plugin
   def watched_rss(m, params)
     wanted = params[:handle]
     reply = String.new
-    @@mutex.synchronize {
-      watchlist.each { |handle, feed|
-        next if wanted and !handle.match(wanted)
-        next unless feed.watched_by?(m.replyto)
-        reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})\n"
-      }
+    watchlist.each { |handle, feed|
+      next if wanted and !handle.match(/#{wanted}/i)
+      next unless feed.watched_by?(m.replyto)
+      reply << "#{feed.handle}: #{feed.url} (in format: #{feed.type ? feed.type : 'default'})\n"
     }
     if reply.empty?
       reply = "no watched feeds"
@@ -208,18 +281,20 @@ class RSSFeedsPlugin < Plugin
   def add_rss(m, params, force=false)
     handle = params[:handle]
     url = params[:url]
+    unless url.match(/https?/)
+      m.reply "I only deal with feeds from HTTP sources, so I can't use #{url} (maybe you forgot the handle?)"
+      return
+    end
     type = params[:type]
-    if @feeds.fetch(handle, nil) && !force
-      m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle].url})"
+    if @feeds.fetch(handle.downcase, nil) && !force
+      m.reply "There is already a feed named #{handle} (URL: #{@feeds[handle.downcase].url})"
       return
     end
     unless url
       m.reply "You must specify both a handle and an url to add an RSS feed"
       return
     end
-    @@mutex.synchronize {
-      @feeds[handle] = RssBlob.new(url,handle,type)
-    }
+    @feeds[handle.downcase] = RssBlob.new(url,handle,type)
     reply = "Added RSS #{url} named #{handle}"
     if type
       reply << " (format: #{type})"
@@ -234,19 +309,17 @@ class RSSFeedsPlugin < Plugin
       m.reply "someone else is watching #{feed.handle}, I won't remove it from my list"
       return
     end
-    @@mutex.synchronize {
-      @feeds.delete(feed.handle)
-    }
+    @feeds.delete(feed.handle.downcase)
     m.okay unless pass
     return
   end
 
   def replace_rss(m, params)
     handle = params[:handle]
-    if @feeds.key?(handle)
+    if @feeds.key?(handle.downcase)
       del_rss(m, {:handle => handle}, true)
     end
-    if @feeds.key?(handle)
+    if @feeds.key?(handle.downcase)
       m.reply "can't replace #{feed.handle}"
     else
       add_rss(m, params, true)
@@ -264,26 +337,21 @@ class RSSFeedsPlugin < Plugin
     if url
       add_rss(m, params)
     end
-    feed = nil
-    @@mutex.synchronize {
-      feed = @feeds.fetch(handle, nil)
-    }
+    feed = @feeds.fetch(handle.downcase, nil)
     if feed
-      @@mutex.synchronize {
-        if feed.add_watch(m.replyto)
-          watchRss(feed, m)
-          m.okay
-        else
-          m.reply "Already watching #{feed.handle}"
-        end
-      }
+      if feed.add_watch(m.replyto)
+        watchRss(feed, m)
+        m.okay
+      else
+        m.reply "Already watching #{feed.handle}"
+      end
     else
       m.reply "Couldn't watch feed #{handle} (no such feed found)"
     end
   end
 
   def unwatch_rss(m, params, pass=false)
-    handle = params[:handle]
+    handle = params[:handle].downcase
     unless @feeds.has_key?(handle)
       m.reply("dunno that feed")
       return
@@ -295,19 +363,13 @@ class RSSFeedsPlugin < Plugin
       m.reply("#{m.replyto} wasn't watching #{feed.handle}") unless pass
     end
     if !feed.watched?
-      @@mutex.synchronize {
-        if @@watchThreads[handle].kind_of? Thread
-          @@watchThreads[handle].kill
-          debug "rmwatch: Killed thread for #{handle}"
-          @@watchThreads.delete(handle)
-        end
-      }
+      stop_watch(handle)
     end
     return feed
   end
 
-  def rewatch_rss(m=nil)
-    kill_threads
+  def rewatch_rss(m=nil, params=nil)
+    stop_watches
 
     # Read watches from list.
     watchlist.each{ |handle, feed|
@@ -318,26 +380,28 @@ class RSSFeedsPlugin < Plugin
 
   private
   def watchRss(feed, m=nil)
-    if @@watchThreads.has_key?(feed.handle)
-      report_problem("watcher thread for #{feed.handle} is already running", m)
+    if @watch.has_key?(feed.handle)
+      report_problem("watcher thread for #{feed.handle} is already running", nil, m)
       return
     end
-    @@watchThreads[feed.handle] = Thread.new do
-      debug 'watchRss thread started.'
-      oldItems = []
-      firstRun = true
-      loop do
-        begin
-          debug 'Fetching rss feed...'
-          title = newItems = nil
-          @@mutex.synchronize {
-            title, newItems = fetchRss(feed)
-          }
-          unless newItems
-            m.reply "no items in feed"
-            break
-          end
-          debug "Checking if new items are available"
+    status = Hash.new
+    status[:oldItems] = []
+    status[:firstRun] = true
+    status[:failures] = 0
+    @watch[feed.handle] = @bot.timer.add(0, status) {
+      debug "watcher for #{feed} started"
+      oldItems = status[:oldItems]
+      firstRun = status[:firstRun]
+      failures = status[:failures]
+      begin
+        debug "fetching #{feed}"
+        title = newItems = nil
+        title, newItems = fetchRss(feed)
+        unless newItems
+          debug "no items in feed #{feed}"
+          failures +=1
+        else
+          debug "Checking if new items are available for #{feed}"
           if firstRun
             debug "First run, we'll see next time"
             firstRun = false
@@ -347,53 +411,82 @@ class RSSFeedsPlugin < Plugin
               otxt.include?(item.to_s)
             }
             if dispItems.length > 0
-              debug "Found #{dispItems.length} new items"
-              dispItems.each { |item|
-                debug "showing #{item.title}"
-                @@mutex.synchronize {
-                  printFormattedRss(feed.watchers, item, feed.type)
-                }
+              debug "Found #{dispItems.length} new items in #{feed}"
+              # When displaying watched feeds, publish them from older to newer
+              dispItems.reverse.each { |item|
+                printFormattedRss(feed, item)
               }
             else
-              debug "No new items found"
+              debug "No new items found in #{feed}"
             end
           end
           oldItems = newItems.dup
-        rescue Exception => e
-          error "IO failed: #{e.inspect}"
-          debug e.backtrace.join("\n")
         end
-
-        seconds = 150 + rand(100)
-        debug "Thread going to sleep #{seconds} seconds.."
-        sleep seconds
+      rescue Exception => e
+        error "Error watching #{feed}: #{e.inspect}"
+        debug e.backtrace.join("\n")
+        failures += 1
       end
-    end
+
+      status[:oldItems] = oldItems
+      status[:firstRun] = firstRun
+      status[:failures] = failures
+
+      seconds = @bot.config['rss.thread_sleep'] * (failures + 1)
+      seconds += seconds * (rand(100)-50)/100
+      debug "watcher for #{feed} going to sleep #{seconds} seconds.."
+      @bot.timer.reschedule(@watch[feed.handle], seconds)
+    }
+    debug "watcher for #{feed} added"
   end
 
-  def printRssItem(loc,item)
-    if item.kind_of?(RSS::RDF::Item)
-      @bot.say loc, item.title.chomp.riphtml.shorten(20) + " @ " + item.link
-    else
-      @bot.say loc, "#{item.pubDate.to_s.chomp+": " if item.pubDate}#{item.title.chomp.riphtml.shorten(20)+" :: " if item.title}#{" @ "+item.link.chomp if item.link}"
-    end
-  end
-
-  def printFormattedRss(locs, item, type)
-    locs.each { |loc|
-      case type
-      when 'amarokblog'
-        @bot.say loc, "::#{item.category.content} just blogged at #{item.link}::"
-        @bot.say loc, "::#{item.title.chomp.riphtml} - #{item.description.chomp.riphtml.shorten(60)}::"
-      when 'amarokforum'
-        @bot.say loc, "::Forum:: #{item.pubDate.to_s.chomp+": " if item.pubDate}#{item.title.chomp.riphtml+" :: " if item.title}#{" @ "+item.link.chomp if item.link}"
-      when 'mediawiki'
-        @bot.say loc, "::Wiki:: #{item.title} has been edited by #{item.dc_creator}. #{item.description.split("\n")[0].chomp.riphtml.shorten(60)} #{item.link} ::"
-        debug "mediawiki #{item.title}"
-      when "gmame"
-        @bot.say loc, "::amarok-devel:: Message #{item.title} sent by #{item.dc_creator}. #{item.description.split("\n")[0].chomp.riphtml.shorten(60)}::"
+  def printFormattedRss(feed, item, opts=nil)
+    places = feed.watchers
+    handle = "::#{feed.handle}:: "
+    date = String.new
+    if opts
+      places = opts[:places] if opts.key?(:places)
+      handle = opts[:handle].to_s if opts.key?(:handle)
+      if opts.key?(:date) && opts[:date]
+        if item.respond_to?(:pubDate) 
+          if item.pubDate.class <= Time
+            date = item.pubDate.strftime("%Y/%m/%d %H.%M.%S")
+          else
+            date = item.pubDate.to_s
+          end
+        elsif  item.respond_to?(:date)
+          if item.date.class <= Time
+            date = item.date.strftime("%Y/%m/%d %H.%M.%S")
+          else
+            date = item.date.to_s
+          end
+        else
+          date = "(no date)"
+        end
+        date += " :: "
+      end
+    end
+    title = "#{Bold}#{item.title.chomp.riphtml}#{Bold}" if item.title
+    desc = item.description.gsub(/\s+/,' ').strip.riphtml.shorten(@bot.config['rss.text_max']) if item.description
+    link = item.link.chomp if item.link
+    places.each { |loc|
+      case feed.type
+      when 'blog'
+        @bot.say loc, "#{handle}#{date}#{item.category.content} blogged at #{link}"
+        @bot.say loc, "#{handle}#{title} - #{desc}"
+      when 'forum'
+        @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
+      when 'wiki'
+        @bot.say loc, "#{handle}#{date}#{item.title} has been edited by #{item.dc_creator}. #{desc} #{link}"
+      when 'gmame'
+        @bot.say loc, "#{handle}#{date}Message #{title} sent by #{item.dc_creator}. #{desc}"
+      when 'trac'
+        @bot.say loc, "#{handle}#{date}#{title} @ #{link}"
+        unless item.title =~ /^Changeset \[(\d+)\]/
+          @bot.say loc, "#{handle}#{date}#{desc}"
+        end
       else
-        printRssItem(loc,item)
+        @bot.say loc, "#{handle}#{date}#{title}#{' @ ' if item.title && item.link}#{link}"
       end
     }
   end
@@ -401,46 +494,47 @@ class RSSFeedsPlugin < Plugin
   def fetchRss(feed, m=nil)
     begin
       # Use 60 sec timeout, cause the default is too low
-      xml = @bot.httputil.get_cached(feed.url,60,60)
+      xml = @bot.httputil.get_cached(feed.url, 60, 60)
     rescue URI::InvalidURIError, URI::BadURIError => e
-      report_problem("invalid rss feed #{feed.url}", m)
+      report_problem("invalid rss feed #{feed.url}", e, m)
+      return
+    rescue => e
+      report_problem("error getting #{feed.url}", e, m)
       return
     end
-    debug 'fetched'
+    debug "fetched #{feed}"
     unless xml
-      report_problem("reading feed #{url} failed", m)
+      report_problem("reading feed #{feed} failed", nil, m)
       return
     end
 
     begin
       ## do validate parse
       rss = RSS::Parser.parse(xml)
-      debug 'parsed'
+      debug "parsed #{feed}"
     rescue RSS::InvalidRSSError
       ## do non validate parse for invalid RSS 1.0
       begin
         rss = RSS::Parser.parse(xml, false)
-      rescue RSS::Error
-        report_problem("parsing rss stream failed, whoops =(", m)
+      rescue RSS::Error => e
+        report_problem("parsing rss stream failed, whoops =(", e, m)
         return
       end
-    rescue RSS::Error
-      report_problem("parsing rss stream failed, oioi", m)
+    rescue RSS::Error => e
+      report_problem("parsing rss stream failed, oioi", e, m)
       return
     rescue => e
-      report_problem("processing error occured, sorry =(", m)
-      debug e.inspect
-      debug e.backtrace.join("\n")
+      report_problem("processing error occured, sorry =(", e, m)
       return
     end
     items = []
     if rss.nil?
-      report_problem("#{feed.url} does not include RSS 1.0 or 0.9x/2.0",m)
+      report_problem("#{feed} does not include RSS 1.0 or 0.9x/2.0", nil, m)
     else
       begin
-        rss.output_encoding = "euc-jp"
-      rescue RSS::UnknownConvertMethod
-        report_problem("bah! something went wrong =(",m)
+        rss.output_encoding = 'UTF-8'
+      rescue RSS::UnknownConvertMethod => e
+        report_problem("bah! something went wrong =(", e, m)
         return
       end
       rss.channel.title ||= "Unknown"
@@ -452,7 +546,7 @@ class RSSFeedsPlugin < Plugin
     end
 
     if items.empty?
-      report_problem("no items found in the feed, maybe try weed?",m)
+      report_problem("no items found in the feed, maybe try weed?", e, m)
       return
     end
     return [title, items]
@@ -463,7 +557,7 @@ plugin = RSSFeedsPlugin.new
 
 plugin.map 'rss show :handle :limit',
   :action => 'show_rss',
-  :requirements => {:limit => /^\d+$/},
+  :requirements => {:limit => /^\d+(?:\.\.\d+)?$/},
   :defaults => {:limit => 5}
 plugin.map 'rss list :handle',
   :action => 'list_rss',
@@ -493,5 +587,6 @@ plugin.map 'rss unwatch :handle',
   :action => 'unwatch_rss'
 plugin.map 'rss rmwatch :handle',
   :action => 'unwatch_rss'
-plugin.map 'rss rewatch :handle',
+plugin.map 'rss rewatch',
   :action => 'rewatch_rss'
+