Function mbs_logical_substr

src/expr.c:165–205

Extract the substring of S, from logical character position POS and LEN characters. first character position is 1. POS and LEN refer to logical characters, not octets. Upon exit, sets v->s to the new string. The new string might be empty if POS/LEN are invalid. */

   Upon exit, sets v->s to the new string.
   The new string might be empty if POS/LEN are invalid. */
static char *
mbs_logical_substr (char const *s, size_t pos, size_t len)
{
  size_t mb_cur_max = MB_CUR_MAX;
  idx_t llen = mb_cur_max <= 1 ? strlen (s) : mbslen (s); /* logical length */

  /* characters to copy */
  size_t vlen = MIN (len, pos <= llen ? llen - pos + 1 : 0);

  char const *substart = s;
  idx_t sublen = 0;
  if (pos == 0 || len == SIZE_MAX)
    {
      /* The request is invalid.  Silently yield an empty string.  */
    }
  else if (mb_cur_max <= 1)
    {
      substart += pos - 1;
      sublen = vlen;
    }
  else
    for (idx_t idx = 1; *s && vlen; idx++)
      {
        idx_t char_bytes = mcel_scanz (s).len;

        /* Skip until we reach the starting position.  */
        if (pos <= idx)
          {
            if (pos == idx)
              substart = s;

            /* Add one character's length in bytes.  */
            vlen--;
            sublen += char_bytes;
          }

        s += char_bytes;
      }

  return ximemdup0 (substart, sublen);
}

/* Return the number of logical characters (possibly multibyte)
   that are in string S in the first OFS octets.