From d2680fff48ea340cd1f4c81665683fe62ad006ba Mon Sep 17 00:00:00 2001
From: levlam <levlam@telegram.org>
Date: Sun, 4 Feb 2024 22:12:18 +0300
Subject: [PATCH] Move include of utf8.h to tl_parsers.cpp.

---
 tdutils/td/utils/tl_parsers.cpp | 19 +++++++++++++++++++
 tdutils/td/utils/tl_parsers.h   | 20 +++++++-------------
 2 files changed, 26 insertions(+), 13 deletions(-)

diff --git a/tdutils/td/utils/tl_parsers.cpp b/tdutils/td/utils/tl_parsers.cpp
index 8812623fd..3e2bae03f 100644
--- a/tdutils/td/utils/tl_parsers.cpp
+++ b/tdutils/td/utils/tl_parsers.cpp
@@ -6,7 +6,9 @@
 //
 #include "td/utils/tl_parsers.h"
 
+#include "td/utils/logging.h"
 #include "td/utils/misc.h"
+#include "td/utils/utf8.h"
 
 namespace td {
 
@@ -55,4 +57,21 @@ BufferSlice TlBufferParser::as_buffer_slice(Slice slice) {
   return BufferSlice(slice);
 }
 
+bool TlBufferParser::is_valid_utf8(CSlice str) const {
+  if (check_utf8(str)) {
+    return true;
+  }
+  LOG(WARNING) << "Wrong UTF-8 string [[" << str << "]] in " << format::as_hex_dump<4>(parent_->as_slice());
+  return false;
+}
+
+size_t TlBufferParser::last_utf8_character_position(Slice str) {
+  CHECK(!str.empty());
+  size_t position = str.size() - 1;
+  while (position != 0 && !is_utf8_character_first_code_unit(static_cast<unsigned char>(str[position]))) {
+    position--;
+  }
+  return position;
+}
+
 }  // namespace td
diff --git a/tdutils/td/utils/tl_parsers.h b/tdutils/td/utils/tl_parsers.h
index 2dc220738..76a74b9eb 100644
--- a/tdutils/td/utils/tl_parsers.h
+++ b/tdutils/td/utils/tl_parsers.h
@@ -9,12 +9,10 @@
 #include "td/utils/buffer.h"
 #include "td/utils/common.h"
 #include "td/utils/format.h"
-#include "td/utils/logging.h"
 #include "td/utils/Slice.h"
 #include "td/utils/SliceBuilder.h"
 #include "td/utils/Status.h"
 #include "td/utils/UInt.h"
-#include "td/utils/utf8.h"
 
 #include <array>
 #include <cstring>
@@ -207,19 +205,11 @@ class TlBufferParser : public TlParser {
         c = ' ';
       }
     }
-    if (check_utf8(result)) {
+    if (is_valid_utf8(result)) {
       return result;
     }
-    CHECK(!result.empty());
-    LOG(WARNING) << "Wrong UTF-8 string [[" << result << "]] in " << format::as_hex_dump<4>(parent_->as_slice());
-
-    // trying to remove last character
-    size_t new_size = result.size() - 1;
-    while (new_size != 0 && !is_utf8_character_first_code_unit(static_cast<unsigned char>(result[new_size]))) {
-      new_size--;
-    }
-    result.resize(new_size);
-    if (check_utf8(result)) {
+    result.resize(last_utf8_character_position(result));
+    if (is_valid_utf8(result)) {
       return result;
     }
 
@@ -235,6 +225,10 @@ class TlBufferParser : public TlParser {
   const BufferSlice *parent_;
 
   BufferSlice as_buffer_slice(Slice slice);
+
+  bool is_valid_utf8(CSlice str) const;
+
+  static size_t last_utf8_character_position(Slice str);
 };
 
 template <>