On May 31, 2017, at 2:59 PM, Akira Kakuto <kakuto@fuk.kindai.ac.jp> wrote:

Hi Karl,

I realize you're reporting a separate bug, that the value gets
misinterpreted

\pdfglyphtounicode{Z}{D835DC81}

<5A> <36E537DC81>

I confirmed that Ross's
\pdfglyphtounicode{Z}{D835 DC81}
with a space works ok.

In the case of
\pdfglyphtounicode{Z}{D835DC81}
I encountered an assertion error because
long code = 0XD835DC81 < 0 in my case, where
sizeof(long) = 4.

Ross obtained erroneously vh = 0X36E537, vl = 0XDC81
because long code = 0XD835DC81 > 0, if sizeof(long) = 8.

Is assert(code >= 0 && code <= 0X10FFFF) OK or not OK?

(from tounicode.c)
static char *utf16be_str(long code)
{
  static char buf[SMALL_BUF_SIZE];
  long v;
  unsigned vh, vl;

  assert(code >= 0);

  if (code <= 0xFFFF)
      sprintf(buf, "%04lX", code);
  else {
      v = code - 0x10000;
      vh = v / 0x400 + 0xD800;
      vl = v % 0x400 + 0xDC00;
      sprintf(buf, "%04X%04X", vh, vl);
  }
  return buf;
}

Best,
Akira