Receiving invalid UTF-8 TEXT message should fail and close the connection (fix remaining autobahn test: 6.X UTF-8 Handling)

DOCKER_VERSION

@@ -1 +1 @@
-5.1.6
+5.1.7

docs/CHANGELOG.md

@@ -1,11 +1,16 @@
 # Changelog
 All notable changes to this project will be documented in this file.
 
+## [5.1.7] - 2019-09-03
+
+- Receiving invalid UTF-8 TEXT message should fail and close the connection (fix remaining autobahn test: 6.X UTF-8 Handling)
+
 ## [5.1.6] - 2019-09-03
 
 - Sending invalid UTF-8 TEXT message should fail and close the connection (fix remaining autobahn test: 6.X UTF-8 Handling)
 - Fix failing unittest which was sending binary data in text mode with WebSocket::send to call properly call WebSocket::sendBinary instead.
 - Validate that the reason is proper utf-8. (fix autobahn test 7.5.1)
+- Validate close codes. Autobahn 7.9.*
 
 ## [5.1.5] - 2019-09-03
 

ixwebsocket/IXWebSocketCloseConstants.cpp

@@ -27,4 +27,5 @@ namespace ix
     const std::string WebSocketCloseConstants::kProtocolErrorCodeDataOpcodeOutOfSequence("Fragmentation: data message out of sequence");
     const std::string WebSocketCloseConstants::kProtocolErrorCodeContinuationOpCodeOutOfSequence("Fragmentation: continuation opcode out of sequence");
     const std::string WebSocketCloseConstants::kInvalidFramePayloadDataMessage("Invalid frame payload data");
+    const std::string WebSocketCloseConstants::kInvalidCloseCodeMessage("Invalid close code");
 }

ixwebsocket/IXWebSocketCloseConstants.h

@@ -32,5 +32,6 @@ namespace ix
         static const std::string kProtocolErrorCodeDataOpcodeOutOfSequence;
         static const std::string kProtocolErrorCodeContinuationOpCodeOutOfSequence;
         static const std::string kInvalidFramePayloadDataMessage;
+        static const std::string kInvalidCloseCodeMessage;
     };
 } // namespace ix

ixwebsocket/IXWebSocketTransport.cpp

@@ -558,13 +558,16 @@ namespace ix
                 return;
             }
 
+            unmaskReceiveBuffer(ws);
+            std::string frameData(_rxbuf.begin()+ws.header_size,
+                                  _rxbuf.begin()+ws.header_size+(size_t) ws.N);
+
             // We got a whole message, now do something with it:
             if (
                    ws.opcode == wsheader_type::TEXT_FRAME
                 || ws.opcode == wsheader_type::BINARY_FRAME
                 || ws.opcode == wsheader_type::CONTINUATION
             ) {
-                unmaskReceiveBuffer(ws);
 
                 if (ws.opcode != wsheader_type::CONTINUATION)
                 {
@@ -593,8 +596,7 @@ namespace ix
                 if (ws.fin && _chunks.empty())
                 {
                     emitMessage(_fragmentedMessageKind,
-                                std::string(_rxbuf.begin()+ws.header_size,
-                                            _rxbuf.begin()+ws.header_size+(size_t) ws.N),
+                                frameData,
                                 ws,
                                 onMessageCallback);
                 }
@@ -607,9 +609,8 @@ namespace ix
                     // the internal buffer which is slow and can let the internal OS
                     // receive buffer fill out.
                     //
-                    _chunks.emplace_back(
-                        std::vector<uint8_t>(_rxbuf.begin()+ws.header_size,
-                                             _rxbuf.begin()+ws.header_size+(size_t)ws.N));
+                    _chunks.emplace_back(frameData);
+
                     if (ws.fin)
                     {
                         emitMessage(_fragmentedMessageKind, getMergedChunks(),
@@ -624,13 +625,8 @@ namespace ix
             }
             else if (ws.opcode == wsheader_type::PING)
             {
-                unmaskReceiveBuffer(ws);
-
-                std::string pingData(_rxbuf.begin()+ws.header_size,
-                                     _rxbuf.begin()+ws.header_size + (size_t) ws.N);
-
                 // too large
-                if (pingData.size() > 125)
+                if (frameData.size() > 125)
                 {
                     // Unexpected frame type
                     close(WebSocketCloseConstants::kProtocolErrorCode,
@@ -642,29 +638,23 @@ namespace ix
                 {
                     // Reply back right away
                     bool compress = false;
-                    sendData(wsheader_type::PONG, pingData, compress);
+                    sendData(wsheader_type::PONG, frameData, compress);
                 }
 
-                emitMessage(MessageKind::PING, pingData, ws, onMessageCallback);
+                emitMessage(MessageKind::PING, frameData, ws, onMessageCallback);
             }
             else if (ws.opcode == wsheader_type::PONG)
             {
-                unmaskReceiveBuffer(ws);
-                std::string pongData(_rxbuf.begin()+ws.header_size,
-                                     _rxbuf.begin()+ws.header_size + (size_t) ws.N);
-
                 std::lock_guard<std::mutex> lck(_lastReceivePongTimePointMutex);
                 _lastReceivePongTimePoint = std::chrono::steady_clock::now();
 
-                emitMessage(MessageKind::PONG, pongData, ws, onMessageCallback);
+                emitMessage(MessageKind::PONG, frameData, ws, onMessageCallback);
             }
             else if (ws.opcode == wsheader_type::CLOSE)
             {
                 std::string reason;
                 uint16_t code = 0;
 
-                unmaskReceiveBuffer(ws);
-
                 if (ws.N >= 2)
                 {
                     // Extract the close code first, available as the first 2 bytes
@@ -674,8 +664,7 @@ namespace ix
                     // Get the reason.
                     if (ws.N > 2)
                     {
-                        reason.assign(_rxbuf.begin()+ws.header_size + 2,
-                                      _rxbuf.begin()+ws.header_size + (size_t) ws.N);
+                        reason = frameData.substr(2, frameData.size());
                     }
 
                     // Validate that the reason is proper utf-8. Autobahn 7.5.1
@@ -684,6 +673,20 @@ namespace ix
                         code = WebSocketCloseConstants::kInvalidFramePayloadData;
                         reason = WebSocketCloseConstants::kInvalidFramePayloadDataMessage;
                     }
+
+                    // Validate close codes. Autobahn 7.9.*
+                    // 1014, 1015 are debattable. The firefox MSDN has a description for them
+                    if (code < 1000 || code == 1004 || code == 1006 ||
+                        (code > 1013 && code < 3000))
+                    {
+                        // build up an error message containing the bad error code
+                        std::stringstream ss;
+                        ss << WebSocketCloseConstants::kInvalidCloseCodeMessage
+                           << ": " << code;
+                        reason = ss.str();
+
+                        code = WebSocketCloseConstants::kProtocolErrorCode;
+                    }
                 }
                 else
                 {
@@ -768,8 +771,7 @@ namespace ix
 
         for (auto&& chunk : _chunks)
         {
-            std::string str(chunk.begin(), chunk.end());
-            msg += str;
+            msg += chunk;
         }
 
         return msg;
@@ -787,11 +789,28 @@ namespace ix
         {
             std::string decompressedMessage;
             bool success = _perMessageDeflate.decompress(message, decompressedMessage);
-            onMessageCallback(decompressedMessage, wireSize, !success, messageKind);
+
+            if (messageKind == MessageKind::MSG_TEXT && !validateUtf8(decompressedMessage))
+            {
+                close(WebSocketCloseConstants::kInvalidFramePayloadData,
+                      WebSocketCloseConstants::kInvalidFramePayloadDataMessage);
+            }
+            else
+            {
+                onMessageCallback(decompressedMessage, wireSize, !success, messageKind);
+            }
         }
         else
         {
-            onMessageCallback(message, wireSize, false, messageKind);
+            if (messageKind == MessageKind::MSG_TEXT && !validateUtf8(message))
+            {
+                close(WebSocketCloseConstants::kInvalidFramePayloadData,
+                      WebSocketCloseConstants::kInvalidFramePayloadDataMessage);
+            }
+            else
+            {
+                onMessageCallback(message, wireSize, false, messageKind);
+            }
         }
     }
 

ixwebsocket/IXWebSocketTransport.h

@@ -149,7 +149,7 @@ namespace ix
         // messages (tested messages up to 700M) and we cannot put them in a single
         // buffer that is resized, as this operation can be slow when a buffer has its
         // size increased 2 fold, while appending to a list has a fixed cost.
-        std::list<std::vector<uint8_t>> _chunks;
+        std::list<std::string> _chunks;
 
         // Record the message kind (will be TEXT or BINARY) for a fragmented
         // message, present in the first chunk, since the final chunk will be a

ixwebsocket/IXWebSocketVersion.h

@@ -6,4 +6,4 @@
 
 #pragma once
 
-#define IX_WEBSOCKET_VERSION "5.1.6"
+#define IX_WEBSOCKET_VERSION "5.1.7"