Compile time checks for size of UTF-16 code units

CMakeLists.txt

@@ -3,7 +3,7 @@
 
 cmake_minimum_required (VERSION 3.14...3.27)
 project (utf8cpp 
-         VERSION 4.0.4
+         VERSION 4.0.6
          LANGUAGES CXX
          DESCRIPTION "C++ portable library for working with utf-8 encoding")
 

source/utf8/core.h

@@ -43,9 +43,14 @@ DEALINGS IN THE SOFTWARE.
 #if UTF_CPP_CPLUSPLUS >= 201103L // C++ 11 or later
     #define UTF_CPP_OVERRIDE override
     #define UTF_CPP_NOEXCEPT noexcept
+    #define UTF_CPP_STATIC_ASSERT(condition) static_assert(condition, "UTFCPP static assert");
 #else // C++ 98/03
     #define UTF_CPP_OVERRIDE
     #define UTF_CPP_NOEXCEPT throw()
+    // Simulate static_assert:
+    template <bool Condition> struct StaticAssert {static void assert() {int static_assert_impl[(Condition ? 1 : -1)];} };
+    template <> struct StaticAssert<true> {static void assert() {}};
+    #define UTF_CPP_STATIC_ASSERT(condition) StaticAssert<condition>::assert();
 #endif // C++ 11 or later
 
 
@@ -215,7 +220,7 @@ namespace internal
 
         UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR(it, end)
 
-        code_point += (*it) & 0x3f;
+        code_point = static_cast<utfchar32_t>(code_point + ((*it) & 0x3f));
 
         return UTF8_OK;
     }
@@ -234,11 +239,11 @@ namespace internal
 
         UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR(it, end)
 
-        code_point += (utf8::internal::mask8(*it) << 6) & 0xfff;
+        code_point = static_cast<utfchar32_t>(code_point + ((utf8::internal::mask8(*it) << 6) & 0xfff));
 
         UTF8_CPP_INCREASE_AND_RETURN_ON_ERROR(it, end)
 
-        code_point += (*it) & 0x3f;
+        code_point = static_cast<utfchar32_t>(code_point + ((*it) & 0x3f));
 
         return UTF8_OK;
     }
@@ -308,6 +313,10 @@ namespace internal
     template <typename word_iterator>
     utf_error validate_next16(word_iterator& it, word_iterator end, utfchar32_t& code_point)
     {
+        // Make sure the iterator dereferences a large enough type
+        typedef typename std::iterator_traits<word_iterator>::value_type word_type;
+        UTF_CPP_STATIC_ASSERT(sizeof(word_type) >= sizeof(utfchar16_t));
+        // Check the edge case:
         if (it == end)
             return NOT_ENOUGH_ROOM;
         // Save the original value of it so we can go back in case of failure
@@ -327,7 +336,7 @@ namespace internal
             else if (is_lead_surrogate(first_word)) {
                 const utfchar16_t second_word = *it++;
                 if (is_trail_surrogate(second_word)) {
-                    code_point = (first_word << 10) + second_word + SURROGATE_OFFSET;
+                    code_point = static_cast<utfchar32_t>(first_word << 10) + second_word + SURROGATE_OFFSET;
                     return UTF8_OK;
                 } else 
                     err = INCOMPLETE_SEQUENCE; 
@@ -395,6 +404,7 @@ namespace internal
     // the word_type.
     template <typename word_iterator, typename word_type>
     word_iterator append16(utfchar32_t cp, word_iterator result) {
+        UTF_CPP_STATIC_ASSERT(sizeof(word_type) >= sizeof(utfchar16_t));
         if (is_in_bmp(cp))
             *(result++) = static_cast<word_type>(cp);
         else {

source/utf8/unchecked.h

@@ -115,15 +115,15 @@ namespace utf8
                     ++it; 
                     cp = ((cp << 12) & 0xffff) + ((utf8::internal::mask8(*it) << 6) & 0xfff);
                     ++it;
-                    cp += (*it) & 0x3f;
+                    cp = static_cast<utfchar32_t>(cp + ((*it) & 0x3f));
                     break;
                 case 4:
                     ++it;
                     cp = ((cp << 18) & 0x1fffff) + ((utf8::internal::mask8(*it) << 12) & 0x3ffff);                
                     ++it;
-                    cp += (utf8::internal::mask8(*it) << 6) & 0xfff;
+                    cp = static_cast<utfchar32_t>(cp + ((utf8::internal::mask8(*it) << 6) & 0xfff));
                     ++it;
-                    cp += (*it) & 0x3f; 
+                    cp = static_cast<utfchar32_t>(cp + ((*it) & 0x3f)); 
                     break;
             }
             ++it;

tests/CMakeLists.txt

@@ -6,6 +6,14 @@ add_library(${PROJECT_NAME} INTERFACE)
 
 include_directories("${PROJECT_SOURCE_DIR}/../source")
 
+if (MSVC)
+    # warning level 4
+    add_compile_options(/W4)
+else()
+    # additional warnings
+    add_compile_options(-Wall -Wextra -Wpedantic -Wconversion -Wsign-conversion)
+endif()
+
 add_executable(negative negative.cpp)
 add_executable(cpp11 test_cpp11.cpp)
 add_executable(cpp17 test_cpp17.cpp)
@@ -14,12 +22,6 @@ add_executable(apitests apitests.cpp)
 
 add_executable(noexceptionstests noexceptionstests.cpp)
 
-
-
-target_compile_options(${PROJECT_NAME} INTERFACE
-  $<$<CXX_COMPILER_ID:MSVC>:/W4>
-  $<$<NOT:$<CXX_COMPILER_ID:MSVC>>:-Wall -Wextra -Wpedantic -Wconversion>)
-
 target_compile_options(noexceptionstests PUBLIC -fno-exceptions)
 
 set_target_properties(negative apitests noexceptionstests