Function build_wcs_upper_buffer

compat/regex/regex_internal.c:277–488

/* Build wide character buffer PSTR->WCS like build_wcs_buffer,
   but for REG_ICASE.  */

static reg_errcode_t
internal_function
build_wcs_upper_buffer (re_string_t *pstr)
{
  mbstate_t prev_st;
  int src_idx, byte_idx, end_idx, remain_len;
  size_t mbclen;
#ifdef _LIBC
  char buf[MB_LEN_MAX];
  assert (MB_LEN_MAX >= pstr->mb_cur_max);
#else
  char buf[64];
#endif

  byte_idx = pstr->valid_len;
  end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len;

  /* The following optimization assumes that ASCII characters can be
     mapped to wide characters with a simple cast.  */
  if (! pstr->map_notascii && pstr->trans == NULL && !pstr->offsets_needed)
    {
      while (byte_idx < end_idx)
	{
	  wchar_t wc;

	  if (isascii (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx])
	      && mbsinit (&pstr->cur_state))
	    {
	      /* In case of a singlebyte character.  */
	      pstr->mbs[byte_idx]
		= toupper (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]);
	      /* The next step uses the assumption that wchar_t is encoded
		 ASCII-safe: all ASCII values can be converted like this.  */
	      pstr->wcs[byte_idx] = (wchar_t) pstr->mbs[byte_idx];
	      ++byte_idx;
	      continue;
	    }

	  remain_len = end_idx - byte_idx;
	  prev_st = pstr->cur_state;
	  mbclen = __mbrtowc (&wc,
			      ((const char *) pstr->raw_mbs + pstr->raw_mbs_idx
			       + byte_idx), remain_len, &pstr->cur_state);
	  if (BE (mbclen + 2 > 2, 1))
	    {
	      wchar_t wcu = wc;
	      if (iswlower (wc))
		{
		  size_t mbcdlen;

		  wcu = towupper (wc);
		  mbcdlen = wcrtomb (buf, wcu, &prev_st);
		  if (BE (mbclen == mbcdlen, 1))
		    memcpy (pstr->mbs + byte_idx, buf, mbclen);
		  else
		    {
		      src_idx = byte_idx;