From db81b15acb84d77316aac81cdf8de4dc445edb59 Mon Sep 17 00:00:00 2001
From: Shivaram Lingamneni <slingamn@cs.stanford.edu>
Date: Fri, 15 Jan 2021 06:19:13 -0500
Subject: [PATCH] initial work on #1483

Add the new utf8-only cap, disallow non-utf8 when websockets are enabled
---
 irc/client.go    |  9 ++++++---
 irc/config.go    |  6 ++++++
 irc/ircconn.go   | 14 +++++---------
 irc/listeners.go |  1 +
 4 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/irc/client.go b/irc/client.go
index 0243c085..d6408711 100644
--- a/irc/client.go
+++ b/irc/client.go
@@ -686,9 +686,12 @@ func (client *Client) run(session *Session) {
 		if err == errInvalidUtf8 {
 			invalidUtf8 = true // handle as normal, including labeling
 		} else if err != nil {
-			quitMessage := "connection closed"
-			if err == errReadQ {
-				quitMessage = "readQ exceeded"
+			var quitMessage string
+			switch err {
+			case errReadQ, errWSBinaryMessage:
+				quitMessage = err.Error()
+			default:
+				quitMessage = "connection closed"
 			}
 			client.Quit(quitMessage, session)
 			// since the client did not actually send us a QUIT,
diff --git a/irc/config.go b/irc/config.go
index 8ddb3da8..de1a7ca8 100644
--- a/irc/config.go
+++ b/irc/config.go
@@ -837,6 +837,9 @@ func (conf *Config) prepareListeners() (err error) {
 		}
 		lconf.RequireProxy = block.TLS.Proxy || block.Proxy
 		lconf.WebSocket = block.WebSocket
+		if lconf.WebSocket && !conf.Server.EnforceUtf8 {
+			return fmt.Errorf("enabling a websocket listener requires the use of server.enforce-utf8")
+		}
 		lconf.HideSTS = block.HideSTS
 		conf.Server.trueListeners[addr] = lconf
 	}
@@ -1446,6 +1449,9 @@ func (config *Config) generateISupport() (err error) {
 	if config.Server.Casemapping == CasemappingPRECIS {
 		isupport.Add("UTF8MAPPING", precisUTF8MappingToken)
 	}
+	if config.Server.EnforceUtf8 {
+		isupport.Add("UTF8ONLY", "")
+	}
 	isupport.Add("WHOX", "")
 
 	err = isupport.RegenerateCachedReply()
diff --git a/irc/ircconn.go b/irc/ircconn.go
index 2f8f6984..980b34f5 100644
--- a/irc/ircconn.go
+++ b/irc/ircconn.go
@@ -19,8 +19,9 @@ const (
 )
 
 var (
-	crlf     = []byte{'\r', '\n'}
-	errReadQ = errors.New("ReadQ Exceeded")
+	crlf               = []byte{'\r', '\n'}
+	errReadQ           = errors.New("ReadQ Exceeded")
+	errWSBinaryMessage = errors.New("WebSocket binary messages are unsupported")
 )
 
 // IRCConn abstracts away the distinction between a regular
@@ -148,11 +149,7 @@ func (wc IRCWSConn) UnderlyingConn() *utils.WrappedConn {
 
 func (wc IRCWSConn) WriteLine(buf []byte) (err error) {
 	buf = bytes.TrimSuffix(buf, crlf)
-	if !globalUtf8EnforcementSetting && !utf8.Valid(buf) {
-		// there's not much we can do about this;
-		// silently drop the message
-		return nil
-	}
+	// #1483: if we have websockets at all, then we're enforcing utf8
 	return wc.conn.WriteMessage(websocket.TextMessage, buf)
 }
 
@@ -172,8 +169,7 @@ func (wc IRCWSConn) ReadLine() (line []byte, err error) {
 		if messageType == websocket.TextMessage {
 			return line, nil
 		} else {
-			// for purposes of fakelag, treat non-text message as an empty line
-			return nil, nil
+			return nil, errWSBinaryMessage
 		}
 	} else if err == websocket.ErrReadLimit {
 		return line, errReadQ
diff --git a/irc/listeners.go b/irc/listeners.go
index 8b709fcf..c50138f8 100644
--- a/irc/listeners.go
+++ b/irc/listeners.go
@@ -166,6 +166,7 @@ func (wl *WSListener) handle(w http.ResponseWriter, r *http.Request) {
 			}
 			return false
 		},
+		Subprotocols: []string{"text.ircv3.net"},
 	}
 
 	conn, err := wsUpgrader.Upgrade(w, r, nil)