markov: refactor triplet learning

[user/henk/code/ruby/rbot.git] / data / rbot / plugins / url.rb

diff --git a/data/rbot/plugins/url.rb b/data/rbot/plugins/url.rb
index b048d3f09e7a944cfaa191f701e7ba83607a6a6e..e380f29b6d428a771a9361b4ee2d8af66d8b876c 100644 (file)
--- a/data/rbot/plugins/url.rb
+++ b/data/rbot/plugins/url.rb
@@ -28,7 +28,9 @@ class UrlPlugin < Plugin
     :default => ['localhost', '^192\.168\.', '^10\.', '^127\.', '^172\.(1[6-9]|2\d|31)\.'],
     :on_change => Proc.new { |bot, v| bot.plugins['url'].reset_no_info_hosts },
     :desc => "A list of regular expressions matching hosts for which no info should be provided")
-
+  Config.register Config::ArrayValue.new('url.only_on_channels',
+    :desc => "Show link info only on these channels",
+    :default => [])
 
   def initialize
     super
@@ -69,7 +71,7 @@ class UrlPlugin < Plugin
     checks.flatten!
 
     unless checks.grep(@no_info_hosts).empty?
-      return "Sorry, info retrieval for #{url.host} (#{checks.first}) is disabled"
+      return ( opts[:always_reply] ? "Sorry, info retrieval for #{url.host} (#{checks.first}) is disabled" : false )
     end
 
     logopts = opts.dup
@@ -79,7 +81,7 @@ class UrlPlugin < Plugin
 
     begin
       debug "+ getting info for #{url.request_uri}"
-      info = Utils.get_html_info(url)
+      info = @bot.filter(:htmlinfo, url)
       debug info
       resp = info[:headers]
 
@@ -120,7 +122,18 @@ class UrlPlugin < Plugin
     return extra.join(", ") if title or not @bot.config['url.titles_only']
   end
 
-  def handle_urls(m, urls, display_info=@bot.config['url.display_link_info'])
+  def handle_urls(m, params={})
+    opts = {
+      :display_info => @bot.config['url.display_link_info'],
+      :channels => @bot.config['url.only_on_channels']
+    }.merge params
+    urls = opts[:urls]
+    display_info= opts[:display_info]
+    channels = opts[:channels]
+    unless channels.empty?
+      return unless channels.map { |c| c.downcase }.include?(m.channel.downcase)
+    end
+
     return if urls.empty?
     debug "found urls #{urls.inspect}"
     list = m.public? ? @registry[m.target] : nil
@@ -128,12 +141,13 @@ class UrlPlugin < Plugin
     urls_displayed = 0
     urls.each do |urlstr|
       debug "working on #{urlstr}"
-      next unless urlstr =~ /^https?:/
+      next unless urlstr =~ /^https?:\/\/./
       title = nil
       debug "Getting title for #{urlstr}..."
       reply = nil
       begin
         title = get_title_for_url(urlstr,
+                                  :always_reply => m.address?,
                                   :nick => m.source.nick,
                                   :channel => m.channel,
                                   :ircline => m.message)
@@ -154,7 +168,8 @@ class UrlPlugin < Plugin
 
       if display_info > urls_displayed
         if reply
-          m.plainreply(reply, :overlong => :truncate)
+          m.reply reply, :overlong => :truncate, :to => :public,
+            :nick => (m.address? ? :auto : false)
           urls_displayed += 1
         end
       end
@@ -177,17 +192,21 @@ class UrlPlugin < Plugin
   def info(m, params)
     escaped = URI.escape(params[:urls].to_s, OUR_UNSAFE)
     urls = URI.extract(escaped)
-    Thread.new { handle_urls(m, urls, params[:urls].length) }
+    Thread.new do
+      handle_urls(m,
+                  :urls => urls,
+                  :display_info => params[:urls].length,
+                  :channels => [])
+    end
   end
 
-  def listen(m)
-    return unless m.kind_of?(PrivMessage)
+  def message(m)
     return if m.address?
 
     escaped = URI.escape(m.message, OUR_UNSAFE)
     urls = URI.extract(escaped, ['http', 'https'])
     return if urls.empty?
-    Thread.new { handle_urls(m, urls) }
+    Thread.new { handle_urls(m, :urls => urls) }
   end
 
   def reply_urls(opts={})