send tweets as UTF-8

[irc.git] / commands.py

diff --git a/commands.py b/commands.py
index 0f958e277ac8c47b7976446fdfe4234e1deb3093..6049640b62fddce7749fa01f1d5b5b3c83b3099b 100644 (file)
--- a/commands.py
+++ b/commands.py
@@ -324,14 +324,15 @@ class UrlLog:
     def urltype(self):
         z=min(len(urlcomplaints)-1, self.count-1)
         return urlcomplaints[z]
-        
-urlre = re.compile("(https?://[^ ]+)( |$)")
+
+#(?:) is a regexp that doesn't group        
+urlre = re.compile("((?:(?:http)|(?:nsfw))s?://[^ ]+)( |$)")
 urlcomplaints = ["a contemporary","an interesting","a fascinating","an overused","a vastly overused"]
 
 ### Deal with /msg bot url or ~url in channel
 def urlq(bot, cmd, nick, conn, public,urldb):
   if (not urlre.search(cmd)):
-    bot.automsg(False,nick,"Please use 'url' only with http URLs")
+    bot.automsg(False,nick,"Please use 'url' only with http, https, nsfw, or nsfws URLs")
     return
 
   urlstring=urlre.search(cmd).group(1)
@@ -380,6 +381,8 @@ def urlexpire(urldb,expire):
 
 # canonicalise BBC URLs (internal use only)
 def canonical_url(urlstring):
+  if "nsfw://" in urlstring or "nsfws://" in urlstring:
+      urlstring=urlstring.replace("nsfw","http",1)
   if (urlstring.find("news.bbc.co.uk") != -1):
     for middle in ("/low/","/mobile/"):
       x = urlstring.find(middle)
@@ -398,17 +401,18 @@ def twitterq(bot,cmd,nick,conn,public,twitapi):
   urlstring = urlre.search(cmd).group(1)
   if (urlstring.find("twitter.com") !=-1):
     stringout = getTweet(urlstring,twitapi)
-    try:
-        bot.automsg(public, nick, stringout)
-    except UnicodeEncodeError:
-        bot.automsg(public, nick, "Sorry, that tweet contained non-ASCII characters")
+    bot.automsg(public, nick, stringout)
   
 def getTweet(urlstring,twitapi):
   parts = string.split(urlstring,'/')
   tweetID = parts[-1]
-  status = twitapi.GetStatus(tweetID)
-  tweeter_screen = status.user.screen_name
-  tweeter_name = status.user.name
-  tweetText = status.text
-  stringout = "tweet by %s (%s): %s" %(tweeter_screen,tweeter_name,tweetText)
+  try:
+    status = twitapi.GetStatus(tweetID)
+    tweeter_screen = status.user.screen_name.encode('UTF-8', 'replace')
+    tweeter_name = status.user.name.encode('UTF-8', 'replace')
+    tweetText = status.text.encode('UTF-8', 'replace')
+    stringout = "tweet by %s (%s): %s" %(tweeter_screen,tweeter_name,tweetText)
+  except twitter.TwitterError:
+    terror = sys.exc_info()
+    stringout = "Twitter error: %s" % terror[1].__str__()
   return stringout