comparison src/charset.c @ 20933:5ceea9d50194

(invalid_character): New function. (chars_in_text): Make the behaviour consistent with INC_POS. (multibyte_chars_in_text): Likewise. (str_cmpchar_id): Likewise. (non_ascii_char_to_string): Call invalid_character. (Fchar_direction): Likewise.

20932:3c2c8431c51d

Lisp_Object Vnonascii_translate_table;

#define min(X, Y) ((X) < (Y) ? (X) : (Y))
#define max(X, Y) ((X) > (Y) ? (X) : (Y))

/* Set STR a pointer to the multi-byte form of the character C.  If C
   is not a composite character, the multi-byte form is set in WORKBUF
   and STR points WORKBUF.  The caller should allocate at least 4-byte
   area at WORKBUF in advance.  Returns the length of the multi-byte
   form.  If C is an invalid character to have a multi-byte form,

	  *str = cmpchar_table[cmpchar_id]->data;
	  return cmpchar_table[cmpchar_id]->len;
	}
      else
	{
	  error ("Invalid character: %d", c);
	}
    }

  SPLIT_NON_ASCII_CHAR (c, charset, c1, c2);
  if (!charset
      || ! CHARSET_DEFINED_P (charset)
      || c1 >= 0 && c1 < 32
      || c2 >= 0 && c2 < 32)
    error ("Invalid character: %d", c);

  *str = workbuf;
  *workbuf++ = CHARSET_LEADING_CODE_BASE (charset);
  if (*workbuf = CHARSET_LEADING_CODE_EXT (charset))
    workbuf++;

  int charset;

  CHECK_NUMBER (ch, 0);
  charset = CHAR_CHARSET (XFASTINT (ch));
  if (!CHARSET_DEFINED_P (charset))
    error ("Invalid character: %d", XINT (ch));
  return CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX);
}

DEFUN ("chars-in-region", Fchars_in_region, Schars_in_region, 2, 2, 0,
  "Return number of characters between BEG and END.")

int
chars_in_text (ptr, nbytes)
     unsigned char *ptr;
     int nbytes;
{
  unsigned char *endp;
  int chars;

  /* current_buffer is null at early stages of Emacs initialization.  */
  if (current_buffer == 0
      || NILP (current_buffer->enable_multibyte_characters))

  endp = ptr + nbytes;
  chars = 0;

  while (ptr < endp)
    {
      if (*ptr == LEADING_CODE_COMPOSITION)
	{
	  ptr++;
	  while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++;
	}
      else
	ptr += BYTES_BY_CHAR_HEAD (*ptr);
      chars++;
    }

  return chars;
}

int
multibyte_chars_in_text (ptr, nbytes)
     unsigned char *ptr;
     int nbytes;
{
  unsigned char *endp;
  int chars;

  endp = ptr + nbytes;
  chars = 0;

  while (ptr < endp)
    {
      if (*ptr == LEADING_CODE_COMPOSITION)
	{
	  ptr++;
	  while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++;
	}
      else
	ptr += BYTES_BY_CHAR_HEAD (*ptr);
      chars++;
    }

  return chars;
}

  int embedded_rule;		/* 1 if composition rule is embedded.  */
  int chars;			/* number of components.  */
  int i;
  struct cmpchar_info *cmpcharp;

  if (len < 5)
    /* Any composite char have at least 3-byte length.  */
    return -1;

  /* The second byte 0xFF means compostion rule is embedded.  */
  embedded_rule = (str[1] == 0xFF);

  /* At first, get the actual length of the composite character.  */
  {
    unsigned char *p, *endp = str + 1, *lastp = str + len;
    int bytes;

    while (endp < lastp && ! CHAR_HEAD_P (*endp)) endp++;
    chars = 0;
    p = str + 1 + embedded_rule;
    while (p < endp)
      {
	/* No need of checking if *P is 0xA0 because
	 BYTES_BY_CHAR_HEAD (0x80) surely returns 2.  */
	p += (bytes = BYTES_BY_CHAR_HEAD (*p - 0x20) + embedded_rule);
	chars++;
      }
    len = (p -= embedded_rule) - str;
    if (p > endp)
      len -= - bytes, chars--;

    if (chars < 2 || chars > MAX_COMPONENT_COUNT)
      /* Invalid number of components.  */
      return -1;
  }
  hash_idx = hash_string (str, len) % CMPCHAR_HASH_TABLE_SIZE;
  hashp = cmpchar_hash_table[hash_idx];

  /* Then, look into the hash table.  */

20933:5ceea9d50194

Lisp_Object Vnonascii_translate_table;

#define min(X, Y) ((X) < (Y) ? (X) : (Y))
#define max(X, Y) ((X) > (Y) ? (X) : (Y))

void
invalid_character (c)
     int c;
{
  error ("Invalid character: %o, %d, 0x%x", c);
}


/* Set STR a pointer to the multi-byte form of the character C.  If C
   is not a composite character, the multi-byte form is set in WORKBUF
   and STR points WORKBUF.  The caller should allocate at least 4-byte
   area at WORKBUF in advance.  Returns the length of the multi-byte
   form.  If C is an invalid character to have a multi-byte form,

	  *str = cmpchar_table[cmpchar_id]->data;
	  return cmpchar_table[cmpchar_id]->len;
	}
      else
	{
	  invalid_character (c);
	}
    }

  SPLIT_NON_ASCII_CHAR (c, charset, c1, c2);
  if (!charset
      || ! CHARSET_DEFINED_P (charset)
      || c1 >= 0 && c1 < 32
      || c2 >= 0 && c2 < 32)
    invalid_character (c);

  *str = workbuf;
  *workbuf++ = CHARSET_LEADING_CODE_BASE (charset);
  if (*workbuf = CHARSET_LEADING_CODE_EXT (charset))
    workbuf++;

  int charset;

  CHECK_NUMBER (ch, 0);
  charset = CHAR_CHARSET (XFASTINT (ch));
  if (!CHARSET_DEFINED_P (charset))
    invalid_character (XINT (ch));
  return CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX);
}

DEFUN ("chars-in-region", Fchars_in_region, Schars_in_region, 2, 2, 0,
  "Return number of characters between BEG and END.")

int
chars_in_text (ptr, nbytes)
     unsigned char *ptr;
     int nbytes;
{
  unsigned char *endp, c;
  int chars;

  /* current_buffer is null at early stages of Emacs initialization.  */
  if (current_buffer == 0
      || NILP (current_buffer->enable_multibyte_characters))

  endp = ptr + nbytes;
  chars = 0;

  while (ptr < endp)
    {
      c = *ptr++;

      if (BASE_LEADING_CODE_P (c))
	while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++;
      chars++;
    }

  return chars;
}

int
multibyte_chars_in_text (ptr, nbytes)
     unsigned char *ptr;
     int nbytes;
{
  unsigned char *endp, c;
  int chars;

  endp = ptr + nbytes;
  chars = 0;

  while (ptr < endp)
    {
      c = *ptr++;

      if (BASE_LEADING_CODE_P (c))
	while (ptr < endp && ! CHAR_HEAD_P (*ptr)) ptr++;
      chars++;
    }

  return chars;
}

  int embedded_rule;		/* 1 if composition rule is embedded.  */
  int chars;			/* number of components.  */
  int i;
  struct cmpchar_info *cmpcharp;

  /* The second byte 0xFF means compostion rule is embedded.  */
  embedded_rule = (str[1] == 0xFF);

  /* At first, get the actual length of the composite character.  */
  {
    unsigned char *p, *endp = str + 1, *lastp = str + len;
    int bytes;

    while (endp < lastp && ! CHAR_HEAD_P (*endp)) endp++;
    if (endp - str < 5)
      /* Any composite char have at least 5-byte length.  */
      return -1;

    chars = 0;
    p = str + 1;
    while (p < endp)
      {
	if (embedded_rule) p++;
	/* No need of checking if *P is 0xA0 because
	   BYTES_BY_CHAR_HEAD (0x80) surely returns 2.  */
	p += BYTES_BY_CHAR_HEAD (*p - 0x20);
	chars++;
      }
    if (p > endp || chars < 2 || chars > MAX_COMPONENT_COUNT)
      /* Invalid components.  */
      return -1;
    len = p - str;
  }
  hash_idx = hash_string (str, len) % CMPCHAR_HASH_TABLE_SIZE;
  hashp = cmpchar_hash_table[hash_idx];

  /* Then, look into the hash table.  */