X-Git-Url: https://git.netwichtig.de/gitweb/?a=blobdiff_plain;f=data%2Frbot%2Fplugins%2Fmarkov.rb;h=deeb9cae4cb3bdbdf0b67d306b7a9729bbefa3c6;hb=57f03b3cd65ff75684e929553d8a3d64713afd28;hp=089d939decc4ccd32e42236a5d02ee1d5c41b969;hpb=041ceb742df2076c93e3fb88fe7a06b5e9c79810;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 089d939d..deeb9cae
--- a/data/rbot/plugins/markov.rb
+++ b/data/rbot/plugins/markov.rb
@@ -44,7 +44,7 @@ class MarkovPlugin < Plugin
     @learning_queue = Queue.new
     @learning_thread = Thread.new do
       while s = @learning_queue.pop
-        learn s
+        learn_line s
         sleep 0.5
       end
     end
@@ -60,30 +60,85 @@ class MarkovPlugin < Plugin
 
   def generate_string(word1, word2)
     # limit to max of markov.max_words words
-    output = word1 + " " + word2
+    if word2
+      output = "#{word1} #{word2}"
+    else
+      output = word1.to_s
+    end
 
-    # try to avoid :nonword in the first iteration
-    wordlist = @registry["#{word1} #{word2}"]
-    wordlist.delete(:nonword)
-    if not wordlist.empty?
-      word3 = wordlist[rand(wordlist.length)]
-      output = output + " " + word3
-      word1, word2 = word2, word3
+    if @registry.key? output
+      wordlist = @registry[output]
+      wordlist.delete(:nonword)
+    else
+      output.downcase!
+      keys = []
+      @registry.each_key(output) do |key|
+        if key.downcase.include? output
+          keys << key
+        else
+          break
+        end
+      end
+      if keys.empty?
+        keys = @registry.keys.select { |k| k.downcase.include? output }
+      end
+      return nil if keys.empty?
+      while key = keys.delete_one
+        wordlist = @registry[key]
+        wordlist.delete(:nonword)
+        unless wordlist.empty?
+          output = key
+          word1, word2 = output.split
+          break
+        end
+      end
     end
+    return nil if wordlist.empty?
+
+    word3 = wordlist.pick_one
+    output << " #{word3}"
+    word1, word2 = word2, word3
 
     (@bot.config['markov.max_words'] - 1).times do
       wordlist = @registry["#{word1} #{word2}"]
       break if wordlist.empty?
-      word3 = wordlist[rand(wordlist.length)]
+      word3 = wordlist.pick_one
       break if word3 == :nonword
-      output = output + " " + word3
+      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 [<chance>]' => set the % chance of rbot responding to input, or display the current probability, '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)
@@ -127,25 +182,25 @@ class MarkovPlugin < Plugin
         s = @bot.config['markov.ignore']
         s.delete user
         @bot.config['ignore'] = s
-        m.reply "#{user} removed"
+        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 @bot.config['markov.ignore'].include?(user)
-          m.reply "#{user} already in list"
+          m.reply _("%{u} already in list") % { :u => user }
         else
           @bot.config['markov.ignore'] = @bot.config['markov.ignore'].push user
-          m.reply "#{user} added to markov ignore list"
+          m.reply _("%{u} added to markov ignore list") % { :u => user }
         end
       else
-        m.reply "give the name of a person or channel to ignore"
+        m.reply _("give the name of a person or channel to ignore")
       end
     when 'list':
-      m.reply "I'm ignoring #{@bot.config['markov.ignore'].join(", ")}"
+      m.reply _("I'm ignoring %{ignored}") % { :ignored => @bot.config['markov.ignore'].join(", ") }
     else
-      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"
+      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
 
@@ -196,10 +251,10 @@ class MarkovPlugin < Plugin
 
   def chat(m, params)
     line = generate_string(params[:seed1], params[:seed2])
-    if line != "#{params[:seed1]} #{params[:seed2]}"
-      m.reply line 
+    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
 
@@ -218,11 +273,15 @@ class MarkovPlugin < Plugin
     if output.length > 1
       m.reply output.join(" ")
     else
-      m.reply "I can't :("
+      m.reply _("I can't :(")
     end
   end
-  
-  def message(m)
+
+  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
@@ -231,12 +290,12 @@ class MarkovPlugin < Plugin
     if m.action?
       message = "#{m.sourcenick} #{message}"
     end
-    
-    @learning_queue.push message
+
+    learn message
     random_markov(m, message) unless m.replied?
   end
 
-  def learn(message)
+  def learn_line(message)
     # debug "learning #{message}"
     wordlist = message.split(/\s+/)
     return unless wordlist.length >= 2
@@ -249,6 +308,71 @@ class MarkovPlugin < Plugin
     k = "#{word1} #{word2}"
     @registry[k] = @registry[k].push(: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
+
+    m.okay
+  end
 end
 
 plugin = MarkovPlugin.new
@@ -258,11 +382,16 @@ 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+%?$/}
+plugin.map 'markov learn from :file [:testing [:lines lines]] [using pattern *pattern]', :action => "learn_from", :thread => true,
+           :requirements => {
+             :testing => /^testing$/,
+             :lines   => /^(?:\d+\.\.\d+|\d+)$/ }
 
 plugin.default_auth('ignore', false)
 plugin.default_auth('probability', false)
+plugin.default_auth('learn', false)