Collect URLs in a dictionary. Fetch page title.

lolbot.py

@@ -13,11 +13,10 @@ import sys, string, random, time
 from ircbot import SingleServerIRCBot
 from irclib import nm_to_n, nm_to_h, irc_lower
 import botcommon
-import urllib
-import libxml2doc
+import time
+from mechanize import Browser
 
 # Exclamations - wrong input
-
 exclamations = [
     "Zing!",
     "Burns!",
@@ -28,7 +27,6 @@ exclamations = [
 ]
 
 # Ponderings
-
 ponderings = [
     "Hi, can I have a medium lamb roast, with just potatoes.",
     "Can I slurp on your Big Cock?",
@@ -41,7 +39,8 @@ class LolBot(SingleServerIRCBot):
 
     self.channel = channel
     self.nickname = nickname
-    self.urls = []
+    self.urls = {}
+    self.logfile = open('/tmp/lolbot.log', 'a')
 
     self.helptext = "Adds URLs to a list. Commands: list - prints a bunch of URLs; clear - clears the list; lol - say something funny; <url> - adds the URL to the list; help - this message."
 
@@ -49,6 +48,17 @@ class LolBot(SingleServerIRCBot):
     self.queue.start()
     self.start()
 
+  def save_url(self, url):
+    try:
+      br = Browser()
+      br.open(url)
+      title = br.title()
+    except Exception as ex:
+      print ex
+      title = ''
+    self.urls[url] = title
+    return title
+
   def on_nicknameinuse(self, connection, event):
     self.nickname = connection.get_nickname() + "_"
     connection.nick(self.nickname)
@@ -64,14 +74,20 @@ class LolBot(SingleServerIRCBot):
   def on_pubmsg(self, connection, event):
     "Deal with a public message in a channel."
 
-    # TODO: log it
-
-    # TODO: parse it for links and add it to the list
+    # log it
+    self.logfile.write("%s\n" % event.arguments()[0])
 
     from_nick = nm_to_n(event.source())
     args = string.split(event.arguments()[0], ":", 1)
     if len(args) > 1 and irc_lower(args[0]) == irc_lower(self.nickname):
       self.do_command(event, string.strip(args[1]), from_nick)
+    else:
+      # parse it for links, add URLs to the list
+      words = event.arguments()[0].split(" ")
+      for w in words:
+        if w.startswith('http://') or w.startswith('https://'):
+          title = self.save_url(w)
+          self.say_public(title)
 
   def say_public(self, text):
     "Print TEXT into public channel, for all to see."
@@ -119,18 +135,21 @@ class LolBot(SingleServerIRCBot):
       self.reply(self.ponder(), target)
 
     elif cmd == 'urls' or cmd == 'list':
-      urls = ' || '.join(self.urls)
-      if urls == '':
-        urls = 'No URLs yet.'
-      self.reply(urls, target)
+      for url, title in self.urls.items():
+        line = "%s %s" % (url, title)
+        self.reply(line, target)
+        time.sleep(1)
 
     elif cmd.startswith('http:') or cmd.startswith('https:'):
-      self.urls.append(cmd)
-      self.reply("URL added.", target)
+      title = self.save_url(cmd)
+      if title == '':
+        self.reply('URL added.', target)
+      if title != '':
+        self.reply('URL added: %s' % title, target)
       
     elif cmd == 'clear':
       del self.urls
-      self.urls = []
+      self.urls = {}
       self.reply('URLs cleared.', target)
 
     else: