ircutils: Add formatWhois function

Parsing through the various WHOIS replies to build a formatted string
isn't a trivial task, especially since there is some privacy related
information.  Consolidate this handling into a single function so
there's one place to fix bugs.

Also fix an issue with people putting (unterminated) formatted text into
the "realname" field of their IRC client (c.f., ProgVal/Limnoria#1083).

Signed-off-by: James McCoy <vega.james@gmail.com>

plugins/Network/plugin.py

@@ -1,6 +1,6 @@
 ###
 # Copyright (c) 2002-2004, Jeremiah Fincher
-# Copyright (c) 2010, James McCoy
+# Copyright (c) 2010,2015 James McCoy
 # All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
@@ -158,87 +158,9 @@ class Network(callbacks.Plugin):
         if (irc, loweredNick) not in self._whois:
             return
         (replyIrc, replyMsg, d) = self._whois[(irc, loweredNick)]
-        hostmask = '@'.join(d['311'].args[2:4])
+        d['318'] = msg
-        user = d['311'].args[-1]
+        s = ircutils.formatWhois(irc, d, caller=replyMsg.nick,
-        if '319' in d:
+                                 channel=replyMsg.args[0])
-            channels = d['319'].args[-1].split()
-            ops = []
-            voices = []
-            normal = []
-            halfops = []
-            for channel in channels:
-                origchan = channel
-                channel = channel.lstrip('@%+~!')
-                # UnrealIRCd uses & for user modes and disallows it as a
-                # channel-prefix, flying in the face of the RFC.  Have to
-                # handle this specially when processing WHOIS response.
-                testchan = channel.lstrip('&')
-                if testchan != channel and irc.isChannel(testchan):
-                    channel = testchan
-                diff = len(channel) - len(origchan)
-                modes = origchan[:diff]
-                chan = irc.state.channels.get(channel)
-                # The user is in a channel the bot is in, so the ircd may have
-                # responded with otherwise private data.
-                if chan:
-                    # Skip channels the callee isn't in.  This helps prevents
-                    # us leaking information when the channel is +s or the
-                    # target is +i
-                    if replyMsg.nick not in chan.users:
-                        continue
-                    # Skip +s channels the target is in only if the reply isn't
-                    # being sent to that channel
-                    if 's' in chan.modes and \
-                       not ircutils.strEqual(replyMsg.args[0], channel):
-                        continue
-                if not modes:
-                    normal.append(channel)
-                elif utils.iter.any(lambda c: c in modes,('@', '&', '~', '!')):
-                    ops.append(channel[1:])
-                elif utils.iter.any(lambda c: c in modes, ('%',)):
-                    halfops.append(channel[1:])
-                elif utils.iter.any(lambda c: c in modes, ('+',)):
-                    voices.append(channel[1:])
-            L = []
-            if ops:
-                L.append(format('is an op on %L', ops))
-            if halfops:
-                L.append(format('is a halfop on %L', halfops))
-            if voices:
-                L.append(format('is voiced on %L', voices))
-            if normal:
-                if L:
-                    L.append(format('is also on %L', normal))
-                else:
-                    L.append(format('is on %L', normal))
-        else:
-            L = ['isn\'t on any non-secret channels']
-        channels = format('%L', L)
-        if '317' in d:
-            idle = utils.timeElapsed(d['317'].args[2])
-            signon = time.strftime(conf.supybot.reply.format.time(),
-                                   time.localtime(float(d['317'].args[3])))
-        else:
-            idle = '<unknown>'
-            signon = '<unknown>'
-        if '312' in d:
-            server = d['312'].args[2]
-        else:
-            server = '<unknown>'
-        if '301' in d:
-            away = '  %s is away: %s.' % (nick, d['301'].args[2])
-        else:
-            away = ''
-        if '320' in d:
-            if d['320'].args[2]:
-                identify = ' identified'
-            else:
-                identify = ''
-        else:
-            identify = ''
-        s = '%s (%s) has been%s on server %s since %s (idle for %s) and ' \
-            '%s.%s' % (user, hostmask, identify, server, signon, idle,
-                       channels, away)
         replyIrc.reply(s)
         del self._whois[(irc, loweredNick)]
 

plugins/Relay/plugin.py

@@ -1,6 +1,6 @@
 ###
 # Copyright (c) 2002-2004, Jeremiah Fincher
-# Copyright (c) 2010, James McCoy
+# Copyright (c) 2010,2015 James McCoy
 # All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
@@ -196,64 +196,9 @@ class Relay(callbacks.Plugin):
         if (irc, loweredNick) not in self._whois:
             return
         (replyIrc, replyMsg, d) = self._whois[(irc, loweredNick)]
-        hostmask = '@'.join(d['311'].args[2:4])
+        d['318'] = msg
-        user = d['311'].args[-1]
+        s = ircutils.formatWhois(irc, d, caller=replyMsg.nick,
-        if '319' in d:
+                                 channel=replyMsg.args[0])
-            channels = d['319'].args[-1].split()
-            ops = []
-            voices = []
-            normal = []
-            halfops = []
-            for channel in channels:
-                if channel.startswith('@'):
-                    ops.append(channel[1:])
-                elif channel.startswith('%'):
-                    halfops.append(channel[1:])
-                elif channel.startswith('+'):
-                    voices.append(channel[1:])
-                else:
-                    normal.append(channel)
-            L = []
-            if ops:
-                L.append(format('is an op on %L', ops))
-            if halfops:
-                L.append(format('is a halfop on %L', halfups))
-            if voices:
-                L.append(format('is voiced on %L', voices))
-            if normal:
-                if L:
-                    L.append(format('is also on %L', normal))
-                else:
-                    L.append(format('is on %L', normal))
-        else:
-            L = ['isn\'t on any non-secret channels']
-        channels = format('%L', L)
-        if '317' in d:
-            idle = utils.timeElapsed(d['317'].args[2])
-            signon = time.strftime(conf.supybot.reply.format.time(),
-                                   time.localtime(float(d['317'].args[3])))
-        else:
-            idle = '<unknown>'
-            signon = '<unknown>'
-        if '312' in d:
-            server = d['312'].args[2]
-        else:
-            server = '<unknown>'
-        if '301' in d:
-            away = format('  %s is away: %s.', nick, d['301'].args[2])
-        else:
-            away = ''
-        if '320' in d:
-            if d['320'].args[2]:
-                identify = ' identified'
-            else:
-                identify = ''
-        else:
-            identify = ''
-        s = format('%s (%s) has been%s on server %s since %s (idle for %s) '
-                   'and %s.%s',
-                   user, hostmask, identify, server, signon, idle,
-                   channels, away)
         replyIrc.reply(s)
         del self._whois[(irc, loweredNick)]
 

src/ircutils.py

@@ -1,6 +1,6 @@
 ###
 # Copyright (c) 2002-2005, Jeremiah Fincher
-# Copyright (c) 2009,2011, James McCoy
+# Copyright (c) 2009,2011,2015 James McCoy
 # All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
@@ -338,6 +338,110 @@ def stripFormatting(s):
     s = stripUnderline(s)
     return s.replace('\x0f', '').replace('\x0F', '')
 
+_containsFormattingRe = re.compile(r'[\x02\x03\x16\x1f]')
+def formatWhois(irc, replies, caller='', channel=''):
+    """Returns a string describing the target of a WHOIS command.
+
+    Arguments are:
+    * irc: the irclib.Irc object on which the replies was received
+
+    * replies: a dict mapping the reply codes ('311', '312', etc.) to their
+      corresponding ircmsg.IrcMsg
+
+    * caller: an optional nick specifying who requested the whois information
+
+    * channel: an optional channel specifying where the reply will be sent
+
+    If provided, caller and channel will be used to avoid leaking information
+    that the caller/channel shouldn't be privy to.
+    """
+    hostmask = '@'.join(replies['311'].args[2:4])
+    nick = replies['318'].args[1]
+    user = replies['311'].args[-1]
+    if _containsFormattingRe.search(user) and user[-1] != '\x0f':
+        # For good measure, disable any formatting
+        user = '%s\x0f' % user
+    if '319' in replies:
+        channels = replies['319'].args[-1].split()
+        ops = []
+        voices = []
+        normal = []
+        halfops = []
+        for chan in channels:
+            origchan = chan
+            chan = chan.lstrip('@%+~!')
+            # UnrealIRCd uses & for user modes and disallows it as a
+            # channel-prefix, flying in the face of the RFC.  Have to
+            # handle this specially when processing WHOIS response.
+            testchan = chan.lstrip('&')
+            if testchan != chan and irc.isChannel(testchan):
+                chan = testchan
+            diff = len(chan) - len(origchan)
+            modes = origchan[:diff]
+            chanState = irc.state.channels.get(chan)
+            # The user is in a channel the bot is in, so the ircd may have
+            # responded with otherwise private data.
+            if chanState:
+                # Skip channels the callee isn't in.  This helps prevents
+                # us leaking information when the channel is +s or the
+                # target is +i
+                if caller not in chanState.users:
+                    continue
+                # Skip +s channels the target is in only if the reply isn't
+                # being sent to that channel
+                if 's' in chanState.modes and \
+                   not ircutils.strEqual(channel or '', chan):
+                    continue
+            if not modes:
+                normal.append(chan)
+            elif utils.iter.any(lambda c: c in modes,('@', '&', '~', '!')):
+                ops.append(chan[1:])
+            elif utils.iter.any(lambda c: c in modes, ('%',)):
+                halfops.append(chan[1:])
+            elif utils.iter.any(lambda c: c in modes, ('+',)):
+                voices.append(chan[1:])
+        L = []
+        if ops:
+            L.append(format('is an op on %L', ops))
+        if halfops:
+            L.append(format('is a halfop on %L', halfops))
+        if voices:
+            L.append(format('is voiced on %L', voices))
+        if normal:
+            if L:
+                L.append(format('is also on %L', normal))
+            else:
+                L.append(format('is on %L', normal))
+    else:
+        L = ['isn\'t on any non-secret channels']
+    channels = format('%L', L)
+    if '317' in replies:
+        idle = utils.timeElapsed(replies['317'].args[2])
+        signon = utils.str.timestamp(float(replies['317'].args[3]))
+    else:
+        idle = '<unknown>'
+        signon = '<unknown>'
+    if '312' in replies:
+        server = replies['312'].args[2]
+    else:
+        server = '<unknown>'
+    if '301' in replies:
+        away = '  %s is away: %s.' % (nick, replies['301'].args[2])
+    else:
+        away = ''
+    if '320' in replies:
+        if replies['320'].args[2]:
+            identify = ' identified'
+        else:
+            identify = ''
+    else:
+        identify = ''
+    s = utils.str.format('%s (%s) has been%s on server %s since %s '
+                         '(idle for %s) and %s.%s',
+                         user, hostmask, identify, server, signon, idle,
+                         channels, away)
+    return s
+
 class FormatContext(object):
     def __init__(self):
         self.reset()