From cb7a39544440eb4ccbbf9fe9ee26c79ab6c19f88 Mon Sep 17 00:00:00 2001
From: Valentin Lorentz <progval+git@progval.net>
Date: Thu, 11 Mar 2021 19:02:55 +0100
Subject: [PATCH] irclib: truncate outgoing messages to 512 bytes, not 512
 chars

---
 src/irclib.py       | 17 +++++++++++++++--
 test/test_irclib.py | 23 +++++++++++++++++++++++
 2 files changed, 38 insertions(+), 2 deletions(-)

diff --git a/src/irclib.py b/src/irclib.py
index a5bcd2702..701dd1fb9 100644
--- a/src/irclib.py
+++ b/src/irclib.py
@@ -1306,7 +1306,8 @@ class Irc(IrcCommandDispatcher, log.Firewalled):
         else:
             msg_tags_str = ''
             msg_rest_str = msg_str
-        if len(msg_rest_str) > MAX_LINE_SIZE:
+        msg_rest_bytes = msg_rest_str.encode()
+        if len(msg_rest_bytes) > MAX_LINE_SIZE:
             # Yes, this violates the contract, but at this point it doesn't
             # matter.  That's why we gotta go munging in private attributes
             #
@@ -1315,7 +1316,19 @@ class Irc(IrcCommandDispatcher, log.Firewalled):
             # this issue, there's no fundamental reason to make it a
             # warning.
             log.debug('Truncating %r, message is too long.', msg)
-            msg._str = msg_tags_str + msg_rest_str[:MAX_LINE_SIZE-2] + '\r\n'
+
+            # Truncate to 512 bytes (minus 2 for '\r\n')
+            msg_rest_bytes = msg_rest_bytes[:MAX_LINE_SIZE-2]
+
+            # The above truncation may have truncated in the middle of a
+            # multi-byte character.
+            # I was about to write a UTF-8 decoder here just to trim them
+            # properly, but fortunately there is a neat trick to trim it
+            # while decoding: just ignore invalid bytes!
+            # https://stackoverflow.com/a/1820949/539465
+            msg_rest_str = msg_rest_bytes.decode(errors="ignore")
+
+            msg._str = msg_tags_str + msg_rest_str + '\r\n'
             msg._len = len(str(msg))
         # TODO: truncate tags
 
diff --git a/test/test_irclib.py b/test/test_irclib.py
index 0e7851d6c..0d870e93b 100644
--- a/test/test_irclib.py
+++ b/test/test_irclib.py
@@ -1084,6 +1084,29 @@ class IrcTestCase(SupyTestCase):
         self.assertEqual(msg5.tagged('batch'), outer)
         self.assertEqual(self.irc.state.getParentBatches(msg5), [outer])
 
+    def testTruncate(self):
+        self.irc = irclib.Irc('test')
+
+        while self.irc.takeMsg():
+            pass
+
+        # over 512 bytes
+        msg = ircmsgs.IrcMsg(command='PRIVMSG', args=('#test2', 'é'*256))
+        self.irc.sendMsg(msg)
+        m = self.irc.takeMsg()
+        remaining_payload = 'é' * ((512 - len('PRIVMSG #test2 :\r\n'))//2)
+        self.assertEqual(
+                str(m), 'PRIVMSG #test2 :%s\r\n' % remaining_payload)
+
+        # over 512 bytes, make sure it doesn't truncate in the middle of
+        # a character
+        msg = ircmsgs.IrcMsg(command='PRIVMSG', args=('#test', 'é'*256))
+        self.irc.sendMsg(msg)
+        m = self.irc.takeMsg()
+        remaining_payload = 'é' * ((512 - len('PRIVMSG #test :\r\n'))//2)
+        self.assertEqual(
+                str(m), 'PRIVMSG #test :%s\r\n' % remaining_payload)
+
 
 class SaslTestCase(SupyTestCase, CapNegMixin):
     def setUp(self):