comparison src/charset.c @ 21445:4c0b4a1025cd

(string_to_non_ascii_char): Include garbage bytes (if any) following a multibyte character in *ACTUAL_LEN. (Fcharset_after): New function. (syms_of_charset): Defsubr it. (multibyte_form_length): Modified to be consistent with string_to_non_ascii_char.

21444:0cba6f211d7c

     const unsigned char *str;
     int len, *actual_len;
{
  int charset;
  unsigned char c1, c2;
  register int c;

  if (SPLIT_STRING (str, len, charset, c1, c2) == CHARSET_ASCII)
    {
      if (actual_len)
	*actual_len = 1;
      return (int) *str;
    }

  c = MAKE_NON_ASCII_CHAR (charset, c1, c2);

  if (actual_len)
    *actual_len = (charset == CHARSET_COMPOSITION
		   ? cmpchar_table[COMPOSITE_CHAR_ID (c)]->len
		   : BYTES_BY_CHAR_HEAD (*str));
  return c;
}

/* Return the length of the multi-byte form at string STR of length LEN.  */
int
multibyte_form_length (str, len)
     const unsigned char *str;
     int len;
{
  int charset;
  unsigned char c1, c2;
  register int c;

  if (SPLIT_STRING (str, len, charset, c1, c2) == CHARSET_ASCII)
    return 1;

  return (charset == CHARSET_COMPOSITION
	  ? cmpchar_table[(c1 << 7) | c2]->len
	  : BYTES_BY_CHAR_HEAD (*str));
}

/* Check if string STR of length LEN contains valid multi-byte form of
   a character.  If valid, charset and position codes of the character
   is set at *CHARSET, *C1, and *C2, and return 0.  If not valid,

     Lisp_Object ch;
{
  CHECK_NUMBER (ch, 0);

  return CHARSET_SYMBOL (CHAR_CHARSET (XINT (ch)));
}

DEFUN ("iso-charset", Fiso_charset, Siso_charset, 3, 3, 0,
  "Return charset of ISO's specification DIMENSION, CHARS, and FINAL-CHAR.\n\
\n\

  defsubr (&Sfind_charset_region);
  defsubr (&Sfind_charset_string);
  defsubr (&Smake_char_internal);
  defsubr (&Ssplit_char);
  defsubr (&Schar_charset);
  defsubr (&Siso_charset);
  defsubr (&Schar_valid_p);
  defsubr (&Sunibyte_char_to_multibyte);
  defsubr (&Schar_bytes);
  defsubr (&Schar_width);

21445:4c0b4a1025cd

     const unsigned char *str;
     int len, *actual_len;
{
  int charset;
  unsigned char c1, c2;
  register int c, bytes;

  c = *str;
  bytes = 1;

  if (BASE_LEADING_CODE_P (c))
    {
      while (bytes < len && ! CHAR_HEAD_P (str[bytes])) bytes++;

      if (c == LEADING_CODE_COMPOSITION)
	{
	  int cmpchar_id = str_cmpchar_id (str, bytes);

	  if (cmpchar_id >= 0)
	    c = MAKE_COMPOSITE_CHAR (cmpchar_id);
	}
      else
	{
	  int charset = c, c1, c2 = 0;

	  str++;
	  if (c >= LEADING_CODE_PRIVATE_11)
	    charset = *str++;
	  if (BYTES_BY_CHAR_HEAD (c) <= bytes && CHARSET_DEFINED_P (charset))
	    {
	      c1 = *str++ & 0x7f;
	      if (CHARSET_DIMENSION (charset) == 2)
		c2 = *str & 0x7F;
	      c = MAKE_NON_ASCII_CHAR (charset, c1, c2);
	    }
	}
    }

  if (actual_len)
    *actual_len = bytes;
  return c;
}

/* Return the length of the multi-byte form at string STR of length LEN.  */
int
multibyte_form_length (str, len)
     const unsigned char *str;
     int len;
{
  int bytes = 1;

  if (BASE_LEADING_CODE_P (*str))
    while (bytes < len && ! CHAR_HEAD_P (str[bytes])) bytes++;

  return bytes;
}

/* Check if string STR of length LEN contains valid multi-byte form of
   a character.  If valid, charset and position codes of the character
   is set at *CHARSET, *C1, and *C2, and return 0.  If not valid,

     Lisp_Object ch;
{
  CHECK_NUMBER (ch, 0);

  return CHARSET_SYMBOL (CHAR_CHARSET (XINT (ch)));
}

DEFUN ("charset-after", Fcharset_after, Scharset_after, 0, 1, 0,
  "Return charset of a character in current buffer at position POS.\n\
If POS is nil, it defauls to the current point.")
  (pos)
     Lisp_Object pos;
{
  register int pos_byte, c, charset;
  register unsigned char *p;

  if (NILP (pos))
    pos_byte = PT_BYTE;
  else if (MARKERP (pos))
    pos_byte = marker_byte_position (pos);
  else
    {
      CHECK_NUMBER (pos, 0);
      pos_byte = CHAR_TO_BYTE (XINT (pos));
    }
  p = BYTE_POS_ADDR (pos_byte);
  c = STRING_CHAR (p, Z_BYTE - pos_byte);
  charset = CHAR_CHARSET (c);
  return CHARSET_SYMBOL (charset);
}

DEFUN ("iso-charset", Fiso_charset, Siso_charset, 3, 3, 0,
  "Return charset of ISO's specification DIMENSION, CHARS, and FINAL-CHAR.\n\
\n\

  defsubr (&Sfind_charset_region);
  defsubr (&Sfind_charset_string);
  defsubr (&Smake_char_internal);
  defsubr (&Ssplit_char);
  defsubr (&Schar_charset);
  defsubr (&Scharset_after);
  defsubr (&Siso_charset);
  defsubr (&Schar_valid_p);
  defsubr (&Sunibyte_char_to_multibyte);
  defsubr (&Schar_bytes);
  defsubr (&Schar_width);