Issue in the new implementation of Utf8ToWideChar

[syagrius]

Hi @synopse,

I think I found an issue in the new implemenation of Utf8ToWideChar.
I use Delphi 12.1, target Win32

I'm using the JSONToDataSet() function and this function fails with the latest revisions of Mormot2.

After tracing the code I've found an Issue:

Utf8ToWideChar() truncates the destination string.
The resulting string is shortened by 1 character.

How to reproduce

procedure  TTestCoreBase._Utf8ToWideChar;
var
  dest: array[0..9] of WideChar;
begin
  Utf8ToWideChar(@dest, '1234567890', 10, sizeof(dest));
  check( WideCharToString(Dest)='1234567890')   //<<   Returns '123456789'
end;

Workaround

Note: It seems that this fixes the problem, but I'm not certain about my correction

In unit mormot.core.unicode.pas

I replace lines 2870 and 2899:
if PtrUInt(@dest[1]) >= PtrUInt(endDest) then
with
if PtrUInt(@dest[1]) > PtrUInt(endDest) then

[synopse]

The documentation states that
a WideChar(#0) is added at the end (if something is written)
So your destination should be defined as:

var
  dest: array[0..10] of WideChar;

For me, it works as expected and defined.
Especially you NEED to follow this requirement:
enough place must be available in dest buffer (guess is sourceBytes*3+2)
which was not the case in your code.