Correct UTF-8 encoding of U+FFFF

Overlong (and hence invalid) 4-byte encoding was used for this character
instead of the correct 3-byte 0xEF 0xBF 0xBF sequence.

Fix this by using 3 bytes for the code points up to 0xFFFF included,
instead of excluding it as was done before.

Closes #17920.

docs/changes.txt

@@ -108,6 +108,7 @@ All:
 - Update all bundled 3rd party libraries to their latest versions.
 - Use unique prefix for all zlib symbols to avoid link conflicts.
 - Make wxFile::ReadAll() work for unseekable files too.
+- Correct UTF-8 encoding of U+FFFF (axiom).
 
 All (GUI):
 

src/common/strconv.cpp

@@ -1135,7 +1135,7 @@ wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen,
                 out[0] = 0xC0 | code;
             }
         }
-        else if ( code < 0xFFFF )
+        else if ( code <= 0xFFFF )
         {
             len = 3;
             if ( out )

src/common/ustring.cpp

@@ -431,7 +431,7 @@ wxScopedCharBuffer wxUString::utf8_str() const
         {
             utf8_length += 2;
         }
-        else if ( code < 0xFFFF )
+        else if ( code <= 0xFFFF )
         {
             utf8_length += 3;
         }

tests/mbconv/mbconvtest.cpp

@@ -474,6 +474,20 @@ void MBConvTestCase::UTF8Tests()
     const wchar_t wc = 0xd800;
     CPPUNIT_ASSERT_EQUAL(wxCONV_FAILED, wxConvUTF8.FromWChar(NULL, 0, &wc, 1));
 #endif // SIZEOF_WCHAR_T == 2
+
+    SECTION("UTF-8-FFFF")
+    {
+        const wchar_t wcFFFF = 0xFFFF;
+        REQUIRE(wxConvUTF8.FromWChar(NULL, 0, &wcFFFF, 1) == 3);
+
+        char buf[4];
+        buf[3] = '\0';
+        REQUIRE(wxConvUTF8.FromWChar(buf, 3, &wcFFFF, 1) == 3);
+
+        CHECK(static_cast<unsigned char>(buf[0]) == 0xef);
+        CHECK(static_cast<unsigned char>(buf[1]) == 0xbf);
+        CHECK(static_cast<unsigned char>(buf[2]) == 0xbf);
+    }
 }
 
 void MBConvTestCase::UTF16LETests()