comparison src/editfns.c @ 23554:e06e84c477fa

(Fsubst_char_in_region): Correctly handle the case that byte combining before happens. (Ftranslate_region): Likewise.

23553:b6d888e0dfcc

{
  register int pos, pos_byte, stop, i, len, end_byte;
  int changed = 0;
  unsigned char fromwork[4], *fromstr, towork[4], *tostr, *p;
  int count = specpdl_ptr - specpdl;

  validate_region (&start, &end);
  CHECK_NUMBER (fromchar, 2);
  CHECK_NUMBER (tochar, 3);

  if (! NILP (current_buffer->enable_multibyte_characters))
    {
      len = CHAR_STRING (XFASTINT (fromchar), fromwork, fromstr);
      if (CHAR_STRING (XFASTINT (tochar), towork, tostr) != len)
	error ("Characters in subst-char-in-region have different byte-lengths");
    }
  else
    {
      len = 1;
      fromwork[0] = XFASTINT (fromchar), fromstr = fromwork;

  if (pos_byte < GPT_BYTE)
    stop = min (stop, GPT_BYTE);
  while (1)
    {
      if (pos_byte >= stop)
	{
	  if (pos_byte >= end_byte) break;
	  stop = end_byte;
	}
      p = BYTE_POS_ADDR (pos_byte);
      if (p[0] == fromstr[0]
	  && (len == 1
	      || (p[1] == fromstr[1]
		  && (len == 2 || (p[2] == fromstr[2]
				 && (len == 3 || p[3] == fromstr[3]))))))
	{

	      changed = 1;
	    }

	  /* Take care of the case where the new character
	     combines with neighboring bytes.  */ 
	  if (len == 1
	      && ((! CHAR_HEAD_P (tostr[0])
		   && pos_byte > BEGV_BYTE
		   && ! ASCII_BYTE_P (FETCH_BYTE (pos_byte - 1)))
		  ||
		  (! ASCII_BYTE_P (tostr[0])
		   && pos_byte + 1 < ZV_BYTE
		   && ! CHAR_HEAD_P (FETCH_BYTE (pos_byte + 1)))))
	    {
	      Lisp_Object tem, string;

	      struct gcpro gcpro1;

	      tem = current_buffer->undo_list;
	      GCPRO1 (tem);

	      /* Make a multibyte string containing this
		 single-byte character.  */
	      string = Fmake_string (make_number (1),
				     make_number (tochar));
	      SET_STRING_BYTES (XSTRING (string), 1);
	      /* replace_range is less efficient, because it moves the gap,
		 but it handles combining correctly.  */
	      replace_range (pos, pos + 1, string,
			     0, 0, 1);
	      if (! NILP (noundo))
		current_buffer->undo_list = tem;

	      UNGCPRO;
	    }
	  else
	    {
	      if (NILP (noundo))
		record_change (pos, 1);
	      for (i = 0; i < len; i++) *p++ = tostr[i];
	    }
	}
      INC_BOTH (pos, pos_byte);
    }

  if (changed)
    signal_after_change (XINT (start),
			 XINT (end) - XINT (start), XINT (end) - XINT (start));

  for (; pos_byte < stop; )
    {
      register unsigned char *p = BYTE_POS_ADDR (pos_byte);
      int len;
      int oc;

      oc = STRING_CHAR_AND_LENGTH (p, stop - pos_byte, len);
      if (oc < size && len == 1)
	{
	  nc = tt[oc];
	  if (nc != oc)
	    {
	      /* Take care of the case where the new character
		 combines with neighboring bytes.  */ 
	      if ((! CHAR_HEAD_P (nc)
		   && pos_byte > BEGV_BYTE
		   && ! ASCII_BYTE_P (FETCH_BYTE (pos_byte - 1)))
		  ||
		  (! ASCII_BYTE_P (nc)
		   && pos_byte + 1 < ZV_BYTE
		   && ! CHAR_HEAD_P (FETCH_BYTE (pos_byte + 1))))
		{
		  Lisp_Object string;

		  string = Fmake_string (make_number (1),
					 make_number (nc));
		  SET_STRING_BYTES (XSTRING (string), 1);

		  /* This is less efficient, because it moves the gap,
		     but it handles combining correctly.  */
		  replace_range (pos, pos + 1, string,
				 1, 0, 0);
		}
	      else
		{
		  record_change (pos, 1);
		  *p = nc;
		  signal_after_change (pos, 1, 1);
		}
	      ++cnt;
	    }
	}
      pos_byte += len;
      pos++;
    }

  return make_number (cnt);
}

23554:e06e84c477fa

{
  register int pos, pos_byte, stop, i, len, end_byte;
  int changed = 0;
  unsigned char fromwork[4], *fromstr, towork[4], *tostr, *p;
  int count = specpdl_ptr - specpdl;
  int maybe_byte_combining = 0;

  validate_region (&start, &end);
  CHECK_NUMBER (fromchar, 2);
  CHECK_NUMBER (tochar, 3);

  if (! NILP (current_buffer->enable_multibyte_characters))
    {
      len = CHAR_STRING (XFASTINT (fromchar), fromwork, fromstr);
      if (CHAR_STRING (XFASTINT (tochar), towork, tostr) != len)
	error ("Characters in subst-char-in-region have different byte-lengths");
      if (len == 1)
	/* If *TOSTR is in the range 0x80..0x9F, it may be combined
           with the after bytes.  If it is in the range 0xA0..0xFF, it
           may be combined with the before bytes.  */
	maybe_byte_combining = !ASCII_BYTE_P (*tostr);
    }
  else
    {
      len = 1;
      fromwork[0] = XFASTINT (fromchar), fromstr = fromwork;

  if (pos_byte < GPT_BYTE)
    stop = min (stop, GPT_BYTE);
  while (1)
    {
      int pos_byte_next = pos_byte;

      if (pos_byte >= stop)
	{
	  if (pos_byte >= end_byte) break;
	  stop = end_byte;
	}
      p = BYTE_POS_ADDR (pos_byte);
      INC_POS (pos_byte_next);
      if (pos_byte_next - pos_byte == len
	  && p[0] == fromstr[0]
	  && (len == 1
	      || (p[1] == fromstr[1]
		  && (len == 2 || (p[2] == fromstr[2]
				 && (len == 3 || p[3] == fromstr[3]))))))
	{

	      changed = 1;
	    }

	  /* Take care of the case where the new character
	     combines with neighboring bytes.  */ 
	  if (maybe_byte_combining
	      && (CHAR_HEAD_P (*tostr)
		  ? ! CHAR_HEAD_P (FETCH_BYTE (pos_byte + 1))
		  : (pos_byte > BEGV_BYTE
		     && ! ASCII_BYTE_P (FETCH_BYTE (pos_byte - 1)))))
	    {
	      Lisp_Object tem, string;

	      struct gcpro gcpro1;

	      tem = current_buffer->undo_list;
	      GCPRO1 (tem);

	      /* Make a multibyte string containing this single-byte
		  character.  */
	      string = make_multibyte_string (tostr, 1, 1);
	      /* replace_range is less efficient, because it moves the gap,
		 but it handles combining correctly.  */
	      replace_range (pos, pos + 1, string,
			     0, 0, 1);
	      pos_byte_next = CHAR_TO_BYTE (pos);
	      if (pos_byte_next > pos_byte)
		/* Before combining happened.  We should not increment
		   POS because now it points the next character.  */
		pos_byte = pos_byte_next;
	      else
		{
		  pos++;
		  INC_POS (pos_byte_next);
		}
		
	      if (! NILP (noundo))
		current_buffer->undo_list = tem;

	      UNGCPRO;
	    }
	  else
	    {
	      if (NILP (noundo))
		record_change (pos, 1);
	      for (i = 0; i < len; i++) *p++ = tostr[i];
	      pos_byte = pos_byte_next;
	      pos++;
	    }
	}
      else
	{
	  pos_byte = pos_byte_next;
	  pos++;
	}
    }

  if (changed)
    signal_after_change (XINT (start),
			 XINT (end) - XINT (start), XINT (end) - XINT (start));

  for (; pos_byte < stop; )
    {
      register unsigned char *p = BYTE_POS_ADDR (pos_byte);
      int len;
      int oc;
      int pos_byte_next;

      oc = STRING_CHAR_AND_LENGTH (p, stop - pos_byte, len);
      pos_byte_next = pos_byte + len;
      if (oc < size && len == 1)
	{
	  nc = tt[oc];
	  if (nc != oc)
	    {
	      /* Take care of the case where the new character
		 combines with neighboring bytes.  */ 
	      if (!ASCII_BYTE_P (nc)
		  && (CHAR_HEAD_P (nc)
		      ? ! CHAR_HEAD_P (FETCH_BYTE (pos_byte + 1))
		      : (pos_byte > BEGV_BYTE
			 && ! ASCII_BYTE_P (FETCH_BYTE (pos_byte - 1)))))
		{
		  Lisp_Object string;

		  string = make_multibyte_string (tt + oc, 1, 1);
		  /* This is less efficient, because it moves the gap,
		     but it handles combining correctly.  */
		  replace_range (pos, pos + 1, string,
				 1, 0, 1);
		  pos_byte_next = CHAR_TO_BYTE (pos);
		  if (pos_byte_next > pos_byte)
		    /* Before combining happened.  We should not
		       increment POS because now it points the next
		       character.  */
		    pos_byte = pos_byte_next;
		  else
		    {
		      pos++;
		      INC_POS (pos_byte_next);
		    }
		}
	      else
		{
		  record_change (pos, 1);
		  *p = nc;
		  signal_after_change (pos, 1, 1);
		  pos_byte++;
		  pos++;
		}
	      ++cnt;
	    }
	  else
	    {
	      pos_byte++;
	      pos++;
	    }
	}
      else
	{
	  pos_byte += len;
	  pos++;
	}
    }

  return make_number (cnt);
}