X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fmarkov.rb;h=1aa1a0b236d0e89610a7ffbfe090858b8cf0bb7f;hb=c3aec9254dd9e48c0fcc7bfd5432d6555b881071;hp=6c51af5179f696a35d8cf749e31432042c236db2;hpb=bd68a53c74b1e8e9877af68743ca382fadf26d9b;p=user%2Fhenk%2Fcode%2Fruby%2Frbot.git

diff --git a/data/rbot/plugins/markov.rb b/data/rbot/plugins/markov.rb
old mode 100644
new mode 100755
index 6c51af51..1aa1a0b2
--- a/data/rbot/plugins/markov.rb
+++ b/data/rbot/plugins/markov.rb
@@ -1,28 +1,353 @@
+#-- vim:sw=2:et
+#++
+#
+# :title: Markov plugin
+#
+# Author:: Tom Gilbert <tom@linuxbrit.co.uk>
+# Copyright:: (C) 2005 Tom Gilbert
+#
+# Contribute to chat with random phrases built from word sequences learned
+# by listening to chat
+
 class MarkovPlugin < Plugin
+  Config.register Config::BooleanValue.new('markov.enabled',
+    :default => false,
+    :desc => "Enable and disable the plugin")
+  Config.register Config::IntegerValue.new('markov.probability',
+    :default => 25,
+    :validate => Proc.new { |v| (0..100).include? v },
+    :desc => "Percentage chance of markov plugin chipping in")
+  Config.register Config::ArrayValue.new('markov.ignore',
+    :default => [],
+    :desc => "Hostmasks and channel names markov should NOT learn from (e.g. idiot*!*@*, #privchan).")
+  Config.register Config::IntegerValue.new('markov.max_words',
+    :default => 50,
+    :validate => Proc.new { |v| (0..100).include? v },
+    :desc => "Maximum number of words the bot should put in a sentence")
+  Config.register Config::IntegerValue.new('markov.learn_delay',
+    :default => 0.5,
+    :validate => Proc.new { |v| v >= 0 },
+    :desc => "Time the learning thread spends sleeping after learning a line. If set to zero, learning from files can be very CPU intensive, but also faster.")
+
+  MARKER = :"\r\n"
+
+  # upgrade a registry entry from 0.9.14 and earlier, converting the Arrays
+  # into Hashes of weights
+  def upgrade_entry(k, logfile)
+    logfile.puts "\t#{k.inspect}"
+    logfile.flush
+    logfile.fsync
+
+    ar = @registry[k]
+
+    # wipe the current key
+    @registry.delete(k)
+
+    # discard empty keys
+    if ar.empty?
+      logfile.puts "\tEMPTY"
+      return
+    end
+
+    # otherwise, proceed
+    logfile.puts "\t#{ar.inspect}"
+
+    # re-encode key to UTF-8 and cleanup as needed
+    words = k.split.map do |w|
+      BasicUserMessage.strip_formatting(
+        @bot.socket.filter.in(w)
+      ).sub(/\001$/,'')
+    end
+
+    # old import that failed to split properly?
+    if words.length == 1 and words.first.include? '/'
+      # split at the last /
+      unsplit = words.first
+      at = unsplit.rindex('/')
+      words = [unsplit[0,at], unsplit[at+1..-1]]
+    end
+
+    # if any of the re-split/re-encoded words have spaces,
+    # or are empty, we would get a chain we can't convert,
+    # so drop it
+    if words.first.empty? or words.first.include?(' ') or
+      words.last.empty? or words.last.include?(' ')
+      logfile.puts "\tSKIPPED"
+      return
+    end
+
+    # former unclean CTCP, we can't convert this
+    if words.first[0] == 1
+      logfile.puts "\tSKIPPED"
+      return
+    end
+
+    # nonword CTCP => SKIP
+    # someword CTCP => nonword someword
+    if words.last[0] == 1
+      if words.first == "nonword"
+        logfile.puts "\tSKIPPED"
+        return
+      end
+      words.unshift MARKER
+      words.pop
+    end
+
+    # intern the old keys
+    words.map! do |w|
+      ['nonword', MARKER].include?(w) ? MARKER : w.chomp("\001")
+    end
+
+    newkey = words.join(' ')
+    logfile.puts "\t#{newkey.inspect}"
+
+    # the new key exists already, so we want to merge
+    if k != newkey and @registry.key? newkey
+      ar2 = @registry[newkey]
+      logfile.puts "\tMERGE"
+      logfile.puts "\t\t#{ar2.inspect}"
+      ar.push(*ar2)
+      # and get rid of the key
+      @registry.delete(newkey)
+    end
+
+    total = 0
+    hash = Hash.new(0)
+
+    @chains_mutex.synchronize do
+      if @chains.key? newkey
+        ar2 = @chains[newkey]
+        total += ar2.first
+        hash.update ar2.last
+      end
+
+      ar.each do |word|
+        case word
+        when :nonword
+          # former marker
+          sym = MARKER
+        else
+          # we convert old words into UTF-8, cleanup, resplit if needed,
+          # and only get the first word. we may lose some data for old
+          # missplits, but this is the best we can do
+          w = BasicUserMessage.strip_formatting(
+            @bot.socket.filter.in(word).split.first
+          )
+          case w
+          when /^\001\S+$/, "\001", ""
+            # former unclean CTCP or end of CTCP
+            next
+          else
+            # intern after clearing leftover end-of-actions if present
+            sym = w.chomp("\001").intern
+          end
+        end
+        hash[sym] += 1
+        total += 1
+      end
+      if hash.empty?
+        logfile.puts "\tSKIPPED"
+        return
+      end
+      logfile.puts "\t#{[total, hash].inspect}"
+      @chains[newkey] = [total, hash]
+    end
+  end
+
+  def upgrade_registry
+    # we load all the keys and then iterate over this array because
+    # running each() on the registry and updating it at the same time
+    # doesn't work
+    keys = @registry.keys
+    # no registry, nothing to do
+    return if keys.empty?
+
+    ki = 0
+    log "starting markov database conversion thread (v1 to v2, #{keys.length} keys)"
+
+    keys.each { |k| @upgrade_queue.push k }
+    @upgrade_queue.push nil
+
+    @upgrade_thread = Thread.new do
+      logfile = File.open(@bot.path('markov-conversion.log'), 'a')
+      logfile.puts "=== conversion thread started #{Time.now} ==="
+      while k = @upgrade_queue.pop
+        ki += 1
+        logfile.puts "Key #{ki} (#{@upgrade_queue.length} in queue):"
+        begin
+          upgrade_entry(k, logfile)
+        rescue Exception => e
+          logfile.puts "=== ERROR ==="
+          logfile.puts e.pretty_inspect
+          logfile.puts "=== EREND ==="
+        end
+        sleep @bot.config['markov.learn_delay'] unless @bot.config['markov.learn_delay'].zero?
+      end
+      logfile.puts "=== conversion thread stopped #{Time.now} ==="
+      logfile.close
+    end
+    @upgrade_thread.priority = -1
+  end
+
+  attr_accessor :chains
+
   def initialize
     super
     @registry.set_default([])
-    @lastline = false
-  end
-
-  def generate_string(seedline)
-    # limit to max of 50 words
-    return unless seedline
-    word1, word2 = seedline.split(/\s+/)
-    output = word1 + " " + word2
-    50.times do
-      wordlist = @registry["#{word1}/#{word2}"]
-      break if wordlist.empty?
-      word3 = wordlist[rand(wordlist.length)]
-      break if word3 == :nonword
-      output = output + " " + word3
+    if @registry.has_key?('enabled')
+      @bot.config['markov.enabled'] = @registry['enabled']
+      @registry.delete('enabled')
+    end
+    if @registry.has_key?('probability')
+      @bot.config['markov.probability'] = @registry['probability']
+      @registry.delete('probability')
+    end
+    if @bot.config['markov.ignore_users']
+      debug "moving markov.ignore_users to markov.ignore"
+      @bot.config['markov.ignore'] = @bot.config['markov.ignore_users'].dup
+      @bot.config.delete('markov.ignore_users'.to_sym)
+    end
+
+    @chains = @registry.sub_registry('v2')
+    @chains.set_default([])
+    @chains_mutex = Mutex.new
+
+    @upgrade_queue = Queue.new
+    @upgrade_thread = nil
+    upgrade_registry
+
+    @learning_queue = Queue.new
+    @learning_thread = Thread.new do
+      while s = @learning_queue.pop
+        learn_line s
+        sleep @bot.config['markov.learn_delay'] unless @bot.config['markov.learn_delay'].zero?
+      end
+    end
+    @learning_thread.priority = -1
+  end
+
+  def cleanup
+    if @upgrade_thread and @upgrade_thread.alive?
+      debug 'closing conversion thread'
+      @upgrade_queue.clear
+      @upgrade_queue.push nil
+      @upgrade_thread.join
+      debug 'conversion thread closed'
+    end
+
+    debug 'closing learning thread'
+    @learning_queue.push nil
+    @learning_thread.join
+    debug 'learning thread closed'
+  end
+
+  # if passed a pair, pick a word from the registry using the pair as key.
+  # otherwise, pick a word from an given list
+  def pick_word(word1, word2=MARKER)
+    if word1.kind_of? Array
+      wordlist = word1
+    else
+      k = "#{word1} #{word2}"
+      return MARKER unless @chains.key? k
+      wordlist = @chains[k]
+    end
+    total = wordlist.first
+    hash = wordlist.last
+    return MARKER if total == 0
+    return hash.keys.first if hash.length == 1
+    hit = rand(total)
+    ret = MARKER
+    hash.each do |k, w|
+      hit -= w
+      if hit < 0
+        ret = k
+        break
+      end
+    end
+    return ret
+  end
+
+  def generate_string(word1, word2)
+    # limit to max of markov.max_words words
+    if word2
+      output = "#{word1} #{word2}"
+    else
+      output = word1.to_s
+    end
+
+    if @chains.key? output
+      wordlist = @chains[output]
+      wordlist.last.delete(MARKER)
+    else
+      output.downcase!
+      keys = []
+      @chains.each_key(output) do |key|
+        if key.downcase.include? output
+          keys << key
+        else
+          break
+        end
+      end
+      if keys.empty?
+        keys = @chains.keys.select { |k| k.downcase.include? output }
+      end
+      return nil if keys.empty?
+      while key = keys.delete_one
+        wordlist = @chains[key]
+        wordlist.last.delete(MARKER)
+        unless wordlist.empty?
+          output = key
+          # split using / / so that we can properly catch the marker
+          word1, word2 = output.split(/ /).map {|w| w.intern}
+          break
+        end
+      end
+    end
+
+    word3 = pick_word(wordlist)
+    return nil if word3 == MARKER
+
+    output << " #{word3}"
+    word1, word2 = word2, word3
+
+    (@bot.config['markov.max_words'] - 1).times do
+      word3 = pick_word(word1, word2)
+      break if word3 == MARKER
+      output << " #{word3}"
       word1, word2 = word2, word3
     end
     return output
   end
 
   def help(plugin, topic="")
-    "markov plugin: listens to chat to build a markov chain, with which it can (perhaps) attempt to (inanely) contribute to 'discussion'. Sort of.. Will get a *lot* better after listening to a lot of chat. usage: 'markov' to attempt to say something relevant to the last line of chat, if it can.  other options to markov: 'ignore' => ignore a hostmask (accept no input), 'status' => show current status, 'probability' => set the % chance of rbot responding to input, 'chat' => try and say something intelligent, 'chat about <foo> <bar>' => riff on a word pair (if possible)"
+    topic, subtopic = topic.split
+
+    case topic
+    when "ignore"
+      case subtopic
+      when "add"
+        "markov ignore add <hostmask|channel> => ignore a hostmask or a channel"
+      when "list"
+        "markov ignore list => show ignored hostmasks and channels"
+      when "remove"
+        "markov ignore remove <hostmask|channel> => unignore a hostmask or channel"
+      else
+        "ignore hostmasks or channels -- topics: add, remove, list"
+      end
+    when "status"
+      "markov status => show if markov is enabled, probability and amount of messages in queue for learning"
+    when "probability"
+      "markov probability [<percent>] => set the % chance of rbot responding to input, or display the current probability"
+    when "chat"
+      case subtopic
+      when "about"
+        "markov chat about <word> [<another word>] => talk about <word> or riff on a word pair (if possible)"
+      else
+        "markov chat => try to say something intelligent"
+      end
+    else
+      "markov plugin: listens to chat to build a markov chain, with which it can (perhaps) attempt to (inanely) contribute to 'discussion'. Sort of.. Will get a *lot* better after listening to a lot of chat. Usage: 'chat' to attempt to say something relevant to the last line of chat, if it can -- help topics: ignore, status, probability, chat, chat about"
+    end
   end
 
   def clean_str(s)
@@ -33,139 +358,246 @@ class MarkovPlugin < Plugin
   end
 
   def probability?
-    prob = @registry['probability']
-    prob = 25 if prob.kind_of? Array;
-    prob = 0 if prob < 0
-    prob = 100 if prob > 100
-    return prob
+    return @bot.config['markov.probability']
   end
 
   def status(m,params)
-    enabled = @registry['enabled']
-    if (enabled)
-      m.reply "markov is currently enabled, #{probability?}% chance of chipping in"
+    if @bot.config['markov.enabled']
+      reply = _("markov is currently enabled, %{p}% chance of chipping in") % { :p => probability? }
+      l = @learning_queue.length
+      reply << (_(", %{l} messages in queue") % {:l => l}) if l > 0
+      l = @upgrade_queue.length
+      reply << (_(", %{l} chains to upgrade") % {:l => l}) if l > 0
     else
-      m.reply "markov is currently disabled"
+      reply = _("markov is currently disabled")
     end
+    m.reply reply
   end
 
-  def ignore?(user=nil)
-    return @registry['ignore_users'].include?(user)
+  def ignore?(m=nil)
+    return false unless m
+    return true if m.address? or m.private?
+    @bot.config['markov.ignore'].each do |mask|
+      return true if m.channel.downcase == mask.downcase
+      return true if m.source.matches?(mask)
+    end
+    return false
   end
 
   def ignore(m, params)
-    if @registry['ignore_users'].nil?
-      @registry['ignore_users'] = []
-    end
     action = params[:action]
     user = params[:option]
     case action
     when 'remove':
-      if @registry['ignore_users'].include? user
-        s = @registry['ignore_users']
+      if @bot.config['markov.ignore'].include? user
+        s = @bot.config['markov.ignore']
         s.delete user
-        @registry['ignore_users'] = s
-        m.reply "#{user} removed"
+        @bot.config['ignore'] = s
+        m.reply _("%{u} removed") % { :u => user }
       else
-        m.reply "not found in list"
+        m.reply _("not found in list")
       end
     when 'add':
       if user
-        if @registry['ignore_users'].include?(user)
-          m.reply "#{user} already in list"
+        if @bot.config['markov.ignore'].include?(user)
+          m.reply _("%{u} already in list") % { :u => user }
         else
-          @registry['ignore_users'] = @registry['ignore_users'].push user 
-          m.reply "#{user} added to markov ignore list"
+          @bot.config['markov.ignore'] = @bot.config['markov.ignore'].push user
+          m.reply _("%{u} added to markov ignore list") % { :u => user }
         end
       else
-        m.reply "give the name of a person to ignore"
+        m.reply _("give the name of a person or channel to ignore")
       end
     when 'list':
-      m.reply "I'm ignoring #{@registry['ignore_users'].join(", ")}"
+      m.reply _("I'm ignoring %{ignored}") % { :ignored => @bot.config['markov.ignore'].join(", ") }
     else
-      m.reply "have markov ignore the input from a hostmask.  usage: markov ignore add <mask>; markov ignore remove <mask>; markov ignore list"
+      m.reply _("have markov ignore the input from a hostmask or a channel. usage: markov ignore add <mask or channel>; markov ignore remove <mask or channel>; markov ignore list")
     end
   end
 
   def enable(m, params)
-    @registry['enabled'] = true
+    @bot.config['markov.enabled'] = true
     m.okay
   end
 
   def probability(m, params)
-    @registry['probability'] = params[:probability].to_i
-    m.okay
+    if params[:probability]
+      @bot.config['markov.probability'] = params[:probability].to_i
+      m.okay
+    else
+      m.reply _("markov has a %{prob}% chance of chipping in") % { :prob => probability? }
+    end
   end
 
   def disable(m, params)
-    @registry['enabled'] = false
+    @bot.config['markov.enabled'] = false
     m.okay
   end
 
   def should_talk
-    return false unless @registry['enabled']
+    return false unless @bot.config['markov.enabled']
     prob = probability?
     return true if prob > rand(100)
     return false
   end
 
+  def delay
+    1 + rand(5)
+  end
+
   def random_markov(m, message)
     return unless should_talk
-    line = generate_string(message)
+
+    word1, word2 = clean_str(message).split(/\s+/)
+    return unless word1 and word2
+    line = generate_string(word1.intern, word2.intern)
     return unless line
-    m.reply line unless line == message
+    # we do nothing if the line we return is just an initial substring
+    # of the line we received
+    return if message.index(line) == 0
+    @bot.timer.add_once(delay) {
+      m.reply line, :nick => false, :to => :public
+    }
   end
 
   def chat(m, params)
-    seed = "#{params[:seed1]} #{params[:seed2]}"
-    line = generate_string seed
-    if line != seed
-      m.reply line 
+    line = generate_string(params[:seed1], params[:seed2])
+    if line and line != [params[:seed1], params[:seed2]].compact.join(" ")
+      m.reply line
     else
-      m.reply "I can't :("
+      m.reply _("I can't :(")
     end
   end
 
   def rand_chat(m, params)
     # pick a random pair from the db and go from there
-    word1, word2 = :nonword, :nonword
+    word1, word2 = MARKER, MARKER
     output = Array.new
-    50.times do
-      wordlist = @registry["#{word1}/#{word2}"]
-      break if wordlist.empty?
-      word3 = wordlist[rand(wordlist.length)]
-      break if word3 == :nonword
+    @bot.config['markov.max_words'].times do
+      word3 = pick_word(word1, word2)
+      break if word3 == MARKER
       output << word3
       word1, word2 = word2, word3
     end
     if output.length > 1
       m.reply output.join(" ")
     else
-      m.reply "I can't :("
+      m.reply _("I can't :(")
     end
   end
-  
-  def listen(m)
-    return unless m.kind_of?(PrivMessage) && m.public?
-    return if m.address?
-    return if ignore? m.source
+
+  def learn(*lines)
+    lines.each { |l| @learning_queue.push l }
+  end
+
+  def unreplied(m)
+    return if ignore? m
 
     # in channel message, the kind we are interested in
-    message = clean_str m.message
-    
-    wordlist = message.split(/\s+/)
-    return unless wordlist.length > 2
-    @lastline = message
-    word1, word2 = :nonword, :nonword
+    message = m.plainmessage
+
+    if m.action?
+      message = "#{m.sourcenick} #{message}"
+    end
+
+    learn message
+    random_markov(m, message) unless m.replied?
+  end
+
+  def learn_triplet(word1, word2, word3)
+      k = "#{word1} #{word2}"
+      @chains_mutex.synchronize do
+        total = 0
+        hash = Hash.new(0)
+        if @chains.key? k
+          t2, h2 = @chains[k]
+          total += t2
+          hash.update h2
+        end
+        hash[word3] += 1
+        total += 1
+        @chains[k] = [total, hash]
+      end
+  end
+
+  def learn_line(message)
+    # debug "learning #{message.inspect}"
+    wordlist = clean_str(message).split(/\s+/).map { |w| w.intern }
+    return unless wordlist.length >= 2
+    word1, word2 = MARKER, MARKER
+    wordlist << MARKER
     wordlist.each do |word3|
-      @registry["#{word1}/#{word2}"] = @registry["#{word1}/#{word2}"].push(word3)
+      learn_triplet(word1, word2, word3)
       word1, word2 = word2, word3
     end
-    @registry["#{word1}/#{word2}"] = [:nonword]
+  end
+
+  # TODO allow learning from URLs
+  def learn_from(m, params)
+    begin
+      path = params[:file]
+      file = File.open(path, "r")
+      pattern = params[:pattern].empty? ? nil : Regexp.new(params[:pattern].to_s)
+    rescue Errno::ENOENT
+      m.reply _("no such file")
+      return
+    end
+
+    if file.eof?
+      m.reply _("the file is empty!")
+      return
+    end
+
+    if params[:testing]
+      lines = []
+      range = case params[:lines]
+      when /^\d+\.\.\d+$/
+        Range.new(*params[:lines].split("..").map { |e| e.to_i })
+      when /^\d+$/
+        Range.new(1, params[:lines].to_i)
+      else
+        Range.new(1, [@bot.config['send.max_lines'], 3].max)
+      end
+
+      file.each do |line|
+        next unless file.lineno >= range.begin
+        lines << line.chomp
+        break if file.lineno == range.end
+      end
+
+      lines = lines.map do |l|
+        pattern ? l.scan(pattern).to_s : l
+      end.reject { |e| e.empty? }
+
+      if pattern
+        unless lines.empty?
+          m.reply _("example matches for that pattern at lines %{range} include: %{lines}") % {
+            :lines => lines.map { |e| Underline+e+Underline }.join(", "),
+            :range => range.to_s
+          }
+        else
+          m.reply _("the pattern doesn't match anything at lines %{range}") % {
+            :range => range.to_s
+          }
+        end
+      else
+        m.reply _("learning from the file without a pattern would learn, for example: ")
+        lines.each { |l| m.reply l }
+      end
+
+      return
+    end
+
+    if pattern
+      file.each { |l| learn(l.scan(pattern).to_s) }
+    else
+      file.each { |l| learn(l.chomp) }
+    end
 
-    random_markov(m, message)
+    m.okay
   end
 end
+
 plugin = MarkovPlugin.new
 plugin.map 'markov ignore :action :option', :action => "ignore"
 plugin.map 'markov ignore :action', :action => "ignore"
@@ -173,7 +605,16 @@ plugin.map 'markov ignore', :action => "ignore"
 plugin.map 'markov enable', :action => "enable"
 plugin.map 'markov disable', :action => "disable"
 plugin.map 'markov status', :action => "status"
-plugin.map 'chat about :seed1 :seed2', :action => "chat"
+plugin.map 'chat about :seed1 [:seed2]', :action => "chat"
 plugin.map 'chat', :action => "rand_chat"
-plugin.map 'markov probability :probability', :action => "probability",
-           :requirements => {:probability => /^\d+$/}
+plugin.map 'markov probability [:probability]', :action => "probability",
+           :requirements => {:probability => /^\d+%?$/}
+plugin.map 'markov learn from :file [:testing [:lines lines]] [using pattern *pattern]', :action => "learn_from", :thread => true,
+           :requirements => {
+             :testing => /^testing$/,
+             :lines   => /^(?:\d+\.\.\d+|\d+)$/ }
+
+plugin.default_auth('ignore', false)
+plugin.default_auth('probability', false)
+plugin.default_auth('learn', false)
+