From ddbe1833c44cc67a2a70d6548960a25b80166633 Mon Sep 17 00:00:00 2001
From: Tom Gilbert <tom@linuxbrit.co.uk>
Date: Fri, 20 Jan 2006 16:03:48 +0000
Subject: new submission from Chris

---
 data/rbot/plugins/url.rb | 42 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 39 insertions(+), 3 deletions(-)

diff --git a/data/rbot/plugins/url.rb b/data/rbot/plugins/url.rb
index 7972037a..286e8c4a 100644
--- a/data/rbot/plugins/url.rb
+++ b/data/rbot/plugins/url.rb
@@ -1,17 +1,42 @@
+require 'open-uri'
+
 Url = Struct.new("Url", :channel, :nick, :time, :url)
+TITLE_RE = /<\s*title\s*>(.+)<\s*\/title\s*>/im
 
 class UrlPlugin < Plugin
   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.say_titles',
+    :default => true, 
+    :desc => "Get the title of any links pasted to the channel and display it (Also, tells if the link is broken)")
   
   def initialize
     super
     @registry.set_default(Array.new)
   end
+
   def help(plugin, topic="")
     "urls [<max>=4] => list <max> last urls mentioned in current channel, urls search [<max>=4] <regexp> => search for matching urls. In a private message, you must specify the channel to query, eg. urls <channel> [max], urls search <channel> [max] <regexp>"
   end
+
+  def get_title_for_url(url)
+    begin
+      pagedata = open(url).read
+      return unless TITLE_RE.match(pagedata)
+      title = $1.strip.gsub(/\s*\n\s*/, " ")
+      title = title[0..255] if title.length > 255 
+      "Title: #{title}"
+
+    rescue SocketError => e
+      "Title: ^- Error connecting to site (#{e.message})"
+    rescue OpenURI::HTTPError => e
+      "Title: ^- Error getting page (#{e.message})"
+    rescue Exception => e
+      "Title: ^- Error: #{e.inspect}"
+    end
+  end
+
   def listen(m)
     return unless m.kind_of?(PrivMessage)
     return if m.address?
@@ -20,10 +45,21 @@ class UrlPlugin < Plugin
       if m.message =~ /((f|ht)tps?:\/\/.*?)(?:\s+|$)/
         urlstr = $1
         list = @registry[m.target]
+
+        if @bot.config['url.say_titles']
+          debug "Getting title for #{urlstr}..."
+          title = get_title_for_url urlstr
+          if title
+            m.reply title
+            debug "Title found!"
+          else
+            debug "Title not found!"
+          end        
+        end
+    
         # check to see if this url is already listed
-        return if list.find {|u|
-          u.url == urlstr
-        }
+        return if list.find {|u| u.url == urlstr }
+        
         url = Url.new(m.target, m.sourcenick, Time.new, urlstr)
         debug "#{list.length} urls so far"
         if list.length > @bot.config['url.max_urls']
-- 
cgit v1.2.3