Method wchar

src/screen.cpp:429–498

return UCS-4 code of the specified UTF-8 sequence, or 0 for invalid UTF-8, set endptr after the sequence

// return UCS-4 code of the specified UTF-8 sequence, or 0 for invalid UTF-8, set endptr after the sequence
uint32_t Screen::wchar(const char *ptr, const char **endptr)
{
  uint32_t c1, c2, c3, c4;

  c1 = static_cast<unsigned char>(*ptr++);

  if (c1 <= 0x7f)
  {
    if (endptr != NULL)
      *endptr = ptr;
    return c1;
  }

  if (c1 < 0xc2 || c1 > 0xf4)
  {
    if (endptr != NULL)
      *endptr = ptr;
    return 0; // incomplete or invalid UTF-8
  }

  c2 = static_cast<unsigned char>(*ptr++);

  if ((c2 & 0xc0) != 0x80 || (c1 == 0xed && c2 > 0x9f))
  {
    if (endptr != NULL)
      *endptr = ptr;
    return 0; // incomplete UTF-8 or surrogates
  }

  c2 &= 0x3f;

  if (c1 < 0xe0)
  {
    if (endptr != NULL)
      *endptr = ptr;
    return (((c1 & 0x1f) << 6) | c2);
  }

  c3 = static_cast<unsigned char>(*ptr++);

  if ((c3 & 0xc0) != 0x80)
  {
    if (endptr != NULL)
      *endptr = ptr;
    return 0; // incomplete UTF-8
  }

  c3 &= 0x3f;

  if (c1 < 0xf0)
  {
    if (endptr != NULL)
      *endptr = ptr;
    return (((c1 & 0x0f) << 12) | (c2 << 6) | c3);
  }

  c4 = static_cast<unsigned char>(*ptr++);