commands: Disallow IRIs from 'url' and 'httpUrl' converters.

urllib doesn't support IRIs, and gives out a cryptic
'UnicodeEncodeError: 'ascii' codec can't encode character ...'
if we don't validate it.

Author: progval

plugins/Web/test.py

@@ -176,5 +176,9 @@ def testNonSnarfingRegexpConfigurable(self):
         finally:
             conf.supybot.plugins.Web.nonSnarfingRegexp.set('')
 
+    def testFetchIri(self):
+        self.assertRegexp('fetch http://café.example.org/',
+            'Error: .*is not a valid')
+
 
 # vim:set shiftwidth=4 softtabstop=4 expandtab textwidth=79:

src/commands.py

@@ -667,8 +667,15 @@ def getGlob(irc, msg, args, state):
         glob = '*%s*' % glob
     state.args.append(glob)
 
+def _checkUrl(url):
+    try:
+        args[0].encode('ascii')
+    except UnicodeEncodeError:
+        state.errorInvalid(_('url'), args[0])
+
 def getUrl(irc, msg, args, state):
     if utils.web.urlRe.match(args[0]):
+        _checkUrl(args[0])
         state.args.append(args.pop(0))
     else:
         state.errorInvalid(_('url'), args[0])
@@ -681,8 +688,10 @@ def getEmail(irc, msg, args, state):
 
 def getHttpUrl(irc, msg, args, state):
     if utils.web.httpUrlRe.match(args[0]):
+        _checkUrl(args[0])
         state.args.append(args.pop(0))
     elif utils.web.httpUrlRe.match('http://' + args[0]):
+        _checkUrl('http://' + args[0])
         state.args.append('http://' + args.pop(0))
     else:
         state.errorInvalid(_('http url'), args[0])