From 2ae1bb7ea11f67217c87fb8392bb3c206aa7bd99 Mon Sep 17 00:00:00 2001
From: Wiebe Cazemier <wiebe@halfgaar.net>
Date: Sat, 27 Mar 2021 15:29:38 +0100
Subject: [PATCH] Integrate valid UTF-8 and publish char checking

---
 mqttpacket.cpp | 10 ++--------
 utils.cpp      |  5 ++++-
 utils.h        |  2 +-
 3 files changed, 7 insertions(+), 10 deletions(-)

diff --git a/mqttpacket.cpp b/mqttpacket.cpp
index bb7dd5d..6fde0dd 100644
--- a/mqttpacket.cpp
+++ b/mqttpacket.cpp
@@ -439,15 +439,9 @@ void MqttPacket::handlePublish()
 
     topic = std::string(readBytes(variable_header_length), variable_header_length);
 
-    if (!isValidUtf8(topic))
+    if (!isValidUtf8(topic, true))
     {
-        logger->logf(LOG_WARNING, "Client '%s' published a message with invalid UTF8. Dropping.", sender->repr().c_str());
-        return;
-    }
-
-    if (!isValidPublishPath(topic))
-    {
-        logger->logf(LOG_WARNING, "Client '%s' published a message with invalid publish path '%s'. Dropping.", sender->repr().c_str(), topic.c_str());
+        logger->logf(LOG_WARNING, "Client '%s' published a message with invalid UTF8 or +/# in it. Dropping.", sender->repr().c_str());
         return;
     }
 
diff --git a/utils.cpp b/utils.cpp
index 2e3b23a..8a1ed76 100644
--- a/utils.cpp
+++ b/utils.cpp
@@ -77,7 +77,7 @@ bool topicsMatch(const std::string &subscribeTopic, const std::string &publishTo
     return result;
 }
 
-bool isValidUtf8(const std::string &s)
+bool isValidUtf8(const std::string &s, bool alsoCheckInvalidPublishChars)
 {
     int multibyte_remain = 0;
     int cur_code_point = 0;
@@ -86,6 +86,9 @@ bool isValidUtf8(const std::string &s)
         if (x == 0)
             return false;
 
+        if (alsoCheckInvalidPublishChars && (x == '#' || x == '+'))
+            return false;
+
         if(!multibyte_remain)
         {
             cur_code_point = 0;
diff --git a/utils.h b/utils.h
index 568ff7e..8ab3af7 100644
--- a/utils.h
+++ b/utils.h
@@ -49,7 +49,7 @@ std::vector<std::string> splitToVector(const std::string &input, const char sep,
 
 bool topicsMatch(const std::string &subscribeTopic, const std::string &publishTopic);
 
-bool isValidUtf8(const std::string &s);
+bool isValidUtf8(const std::string &s, bool alsoCheckInvalidPublishChars = false);
 
 bool strContains(const std::string &s, const std::string &needle);
 
--
libgit2 0.21.4