url plugin: add list of hosts for which no link info should be retrieved

[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 cdbe5cc8c73b164280ebd4290cf735de2ee4cefb..0160f9039da6d8b9d2879790d6aa51682ba50e58 100644 (file)
--- a/data/rbot/plugins/url.rb
+++ b/data/rbot/plugins/url.rb
@@ -1,3 +1,8 @@
+#-- vim:sw=2:et
+#++
+#
+# :title: Url plugin
+
 define_structure :Url, :channel, :nick, :time, :url, :info
 
 class ::UrlLinkError < RuntimeError
@@ -6,13 +11,14 @@ end
 class UrlPlugin < Plugin
   TITLE_RE = /<\s*?title\s*?>(.+?)<\s*?\/title\s*?>/im
   LINK_INFO = "[Link Info]"
+  OUR_UNSAFE = Regexp.new("[^#{URI::PATTERN::UNRESERVED}#{URI::PATTERN::RESERVED}%# ]", false, 'N')
 
   BotConfig.register BotConfigIntegerValue.new('url.max_urls',
     :default => 100, :validate => Proc.new{|v| v > 0},
     :desc => "Maximum number of urls to store. New urls replace oldest ones.")
-  BotConfig.register BotConfigBooleanValue.new('url.display_link_info',
-    :default => false,
-    :desc => "Get the title of any links pasted to the channel and display it (also tells if the link is broken or the site is down)")
+  BotConfig.register BotConfigIntegerValue.new('url.display_link_info',
+    :default => 0,
+    :desc => "Get the title of links pasted to the channel and display it (also tells if the link is broken or the site is down). Do it for at most this many links per line (set to 0 to disable)")
   BotConfig.register BotConfigBooleanValue.new('url.titles_only',
     :default => false,
     :desc => "Only show info for links that have <title> tags (in other words, don't display info for jpegs, mpegs, etc.)")
@@ -22,11 +28,24 @@ class UrlPlugin < Plugin
   BotConfig.register BotConfigBooleanValue.new('url.info_on_list',
     :default => false,
     :desc => "Show link info when listing/searching for urls")
+  BotConfig.register BotConfigArrayValue.new('url.no_info_hosts',
+    :default => ['localhost', '^192\.168\.', '^10\.', '^127\.0\.0\.1', '^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")
 
 
   def initialize
     super
     @registry.set_default(Array.new)
+    unless @bot.config['url.display_link_info'].kind_of?(Integer)
+      @bot.config.items[:'url.display_link_info'].set_string(@bot.config['url.display_link_info'].to_s)
+    end
+    reset_no_info_hosts
+  end
+
+  def reset_no_info_hosts
+    @no_info_hosts = Regexp.new(@bot.config['url.no_info_hosts'].join('|'), true)
+    debug "no info hosts regexp set to #{@no_info_hosts}"
   end
 
   def help(plugin, topic="")
@@ -38,14 +57,19 @@ class UrlPlugin < Plugin
     $1.ircify_html
   end
 
-  def get_title_for_url(uri_str, nick = nil, channel = nil)
+  def get_title_for_url(uri_str, nick = nil, channel = nil, ircline = nil)
 
     url = uri_str.kind_of?(URI) ? uri_str : URI.parse(uri_str)
     return if url.scheme !~ /https?/
 
+    if url.host =~ @no_info_hosts
+      return "Sorry, info retrieval for #{url.host} is disabled"
+    end
+
     logopts = Hash.new
     logopts[:nick] = nick if nick
     logopts[:channel] = channel if channel
+    logopts[:ircline] = ircline if ircline
 
     title = nil
     extra = String.new
@@ -131,18 +155,24 @@ class UrlPlugin < Plugin
   def listen(m)
     return unless m.kind_of?(PrivMessage)
     return if m.address?
-    urls = URI.extract(m.message)
+
+    escaped = URI.escape(m.message, OUR_UNSAFE)
+    urls = URI.extract(escaped)
     return if urls.empty?
     debug "found urls #{urls.inspect}"
     list = @registry[m.target]
+    urls_displayed = 0
     urls.each { |urlstr|
       debug "working on #{urlstr}"
+      next unless urlstr =~ /^https?:/
       title = nil
-      if @bot.config['url.display_link_info']
+      debug "display link info: #{@bot.config['url.display_link_info']}"
+      if @bot.config['url.display_link_info'] > urls_displayed
+        urls_displayed += 1
         Thread.start do
           debug "Getting title for #{urlstr}..."
           begin
-            title = get_title_for_url urlstr, m.source.nick, m.channel
+            title = get_title_for_url urlstr, m.source.nick, m.channel, m.message
             if title
               m.reply "#{LINK_INFO} #{title}", :overlong => :truncate
               debug "Title found!"