From cfe5c4a43bd47c21c377dc15acee43771b66bc54 Mon Sep 17 00:00:00 2001
From: levlam <levlam@telegram.org>
Date: Sat, 4 Jun 2022 14:57:29 +0300
Subject: [PATCH] Ensure that strings passed to TDLib's log message callback
 are UTF-8 encoded.

---
 td/telegram/Client.cpp       | 15 ++++++++++++++-
 td/telegram/Client.h         |  2 +-
 td/telegram/td_json_client.h |  2 +-
 3 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/td/telegram/Client.cpp b/td/telegram/Client.cpp
index e9f9bf548..a102efbfd 100644
--- a/td/telegram/Client.cpp
+++ b/td/telegram/Client.cpp
@@ -24,6 +24,8 @@
 #include "td/utils/port/RwMutex.h"
 #include "td/utils/port/thread.h"
 #include "td/utils/Slice.h"
+#include "td/utils/StringBuilder.h"
+#include "td/utils/utf8.h"
 
 #include <algorithm>
 #include <atomic>
@@ -679,7 +681,18 @@ static std::atomic<ClientManager::LogMessageCallbackPtr> log_message_callback;
 static void log_message_callback_wrapper(int verbosity_level, CSlice message) {
   auto callback = log_message_callback.load(std::memory_order_relaxed);
   if (callback != nullptr) {
-    callback(verbosity_level, message.c_str());
+    if (check_utf8(message)) {
+      callback(verbosity_level, message.c_str());
+    } else {
+      size_t pos = 0;
+      while (1 <= message[pos] && message[pos] <= 126) {
+        pos++;
+      }
+      CHECK(pos + 1 < message.size());
+      auto utf8_message = PSTRING() << message.substr(0, pos)
+                                    << url_encode(message.substr(pos, message.size() - pos - 1)) << '\n';
+      callback(verbosity_level, utf8_message.c_str());
+    }
   }
 }
 
diff --git a/td/telegram/Client.h b/td/telegram/Client.h
index f765d0264..b8c1e814f 100644
--- a/td/telegram/Client.h
+++ b/td/telegram/Client.h
@@ -129,7 +129,7 @@ class ClientManager final {
    * \param verbosity_level Log verbosity level with which the message was added (-1 - 1024).
    *                        If 0, then TDLib will crash as soon as the callback returns.
    *                        None of the TDLib methods can be called from the callback.
-   * \param message Null-terminated string with the message added to the log.
+   * \param message Null-terminated UTF-8-encoded string with the message added to the log.
    */
   using LogMessageCallbackPtr = void (*)(int verbosity_level, const char *message);
 
diff --git a/td/telegram/td_json_client.h b/td/telegram/td_json_client.h
index e27a3bae7..e6e0c2ded 100644
--- a/td/telegram/td_json_client.h
+++ b/td/telegram/td_json_client.h
@@ -91,7 +91,7 @@ TDJSON_EXPORT const char *td_execute(const char *request);
  * \param verbosity_level Log verbosity level with which the message was added (-1 - 1024).
  *                        If 0, then TDLib will crash as soon as the callback returns.
  *                        None of the TDLib methods can be called from the callback.
- * \param message Null-terminated string with the logged message. The string isn't guaranteed to be encoded in UTF-8.
+ * \param message Null-terminated UTF-8-encoded string with the message added to the log.
  */
 typedef void (*td_log_message_callback_ptr)(int verbosity_level, const char *message);