mxml-file.c

适用于嵌入式系统的XML解析库, 规模比libxml2小得多.

  _mxml_global_t *global = _mxml_global();
					/* Global data */


  global->error_cb = cb;
}


/*
 * 'mxmlSetWrapMargin()' - Set the the wrap margin when saving XML data.
 *
 * Wrapping is disabled when "column" is <= 0.
 *
 * @since Mini-XML 2.3@
 */

void
mxmlSetWrapMargin(int column)		/* I - Column for wrapping */
{
  _mxml_global_t *global = _mxml_global();
					/* Global data */


  if (column <= 0)
    global->wrap = 2147483647;
  else
    global->wrap = column;
}


/*
 * 'mxml_add_char()' - Add a character to a buffer, expanding as needed.
 */

static int				/* O  - 0 on success, -1 on error */
mxml_add_char(int  ch,			/* I  - Character to add */
              char **bufptr,		/* IO - Current position in buffer */
	      char **buffer,		/* IO - Current buffer */
	      int  *bufsize)		/* IO - Current buffer size */
{
  char	*newbuffer;			/* New buffer value */


  if (*bufptr >= (*buffer + *bufsize - 4))
  {
   /*
    * Increase the size of the buffer...
    */

    if (*bufsize < 1024)
      (*bufsize) *= 2;
    else
      (*bufsize) += 1024;

    if ((newbuffer = realloc(*buffer, *bufsize)) == NULL)
    {
      free(*buffer);

      mxml_error("Unable to expand string buffer to %d bytes!", *bufsize);

      return (-1);
    }

    *bufptr = newbuffer + (*bufptr - *buffer);
    *buffer = newbuffer;
  }

  if (ch < 0x80)
  {
   /*
    * Single byte ASCII...
    */

    *(*bufptr)++ = ch;
  }
  else if (ch < 0x800)
  {
   /*
    * Two-byte UTF-8...
    */

    *(*bufptr)++ = 0xc0 | (ch >> 6);
    *(*bufptr)++ = 0x80 | (ch & 0x3f);
  }
  else if (ch < 0x10000)
  {
   /*
    * Three-byte UTF-8...
    */

    *(*bufptr)++ = 0xe0 | (ch >> 12);
    *(*bufptr)++ = 0x80 | ((ch >> 6) & 0x3f);
    *(*bufptr)++ = 0x80 | (ch & 0x3f);
  }
  else
  {
   /*
    * Four-byte UTF-8...
    */

    *(*bufptr)++ = 0xf0 | (ch >> 18);
    *(*bufptr)++ = 0x80 | ((ch >> 12) & 0x3f);
    *(*bufptr)++ = 0x80 | ((ch >> 6) & 0x3f);
    *(*bufptr)++ = 0x80 | (ch & 0x3f);
  }

  return (0);
}


/*
 * 'mxml_fd_getc()' - Read a character from a file descriptor.
 */

static int				/* O  - Character or EOF */
mxml_fd_getc(void *p,			/* I  - File descriptor buffer */
             int  *encoding)		/* IO - Encoding */
{
  _mxml_fdbuf_t	*buf;			/* File descriptor buffer */
  int		ch,			/* Current character */
		temp;			/* Temporary character */


 /*
  * Grab the next character in the buffer...
  */

  buf = (_mxml_fdbuf_t *)p;

  if (buf->current >= buf->end)
    if (mxml_fd_read(buf) < 0)
      return (EOF);

  ch = *(buf->current)++;

  switch (*encoding)
  {
    case ENCODE_UTF8 :
       /*
	* Got a UTF-8 character; convert UTF-8 to Unicode and return...
	*/

	if (!(ch & 0x80))
	{
#if DEBUG > 1
          printf("mxml_fd_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch);
#endif /* DEBUG > 1 */

	  if (mxml_bad_char(ch))
	  {
	    mxml_error("Bad control character 0x%02x not allowed by XML standard!",
        	       ch);
	    return (EOF);
	  }

	  return (ch);
        }
	else if (ch == 0xfe)
	{
	 /*
	  * UTF-16 big-endian BOM?
	  */

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  ch = *(buf->current)++;
          
	  if (ch != 0xff)
	    return (EOF);

	  *encoding = ENCODE_UTF16BE;

	  return (mxml_fd_getc(p, encoding));
	}
	else if (ch == 0xff)
	{
	 /*
	  * UTF-16 little-endian BOM?
	  */

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  ch = *(buf->current)++;
          
	  if (ch != 0xfe)
	    return (EOF);

	  *encoding = ENCODE_UTF16LE;

	  return (mxml_fd_getc(p, encoding));
	}
	else if ((ch & 0xe0) == 0xc0)
	{
	 /*
	  * Two-byte value...
	  */

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = ((ch & 0x1f) << 6) | (temp & 0x3f);

	  if (ch < 0x80)
	    return (EOF);
	}
	else if ((ch & 0xf0) == 0xe0)
	{
	 /*
	  * Three-byte value...
	  */

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = ((ch & 0x0f) << 6) | (temp & 0x3f);

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = (ch << 6) | (temp & 0x3f);

	  if (ch < 0x800)
	    return (EOF);
	}
	else if ((ch & 0xf8) == 0xf0)
	{
	 /*
	  * Four-byte value...
	  */

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = ((ch & 0x07) << 6) | (temp & 0x3f);

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = (ch << 6) | (temp & 0x3f);

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  if ((temp & 0xc0) != 0x80)
	    return (EOF);

	  ch = (ch << 6) | (temp & 0x3f);

	  if (ch < 0x10000)
	    return (EOF);
	}
	else
	  return (EOF);
	break;

    case ENCODE_UTF16BE :
       /*
        * Read UTF-16 big-endian char...
	*/

	if (buf->current >= buf->end)
	  if (mxml_fd_read(buf) < 0)
	    return (EOF);

	temp = *(buf->current)++;

	ch = (ch << 8) | temp;

	if (mxml_bad_char(ch))
	{
	  mxml_error("Bad control character 0x%02x not allowed by XML standard!",
        	     ch);
	  return (EOF);
	}
        else if (ch >= 0xd800 && ch <= 0xdbff)
	{
	 /*
	  * Multi-word UTF-16 char...
	  */

          int lch;

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  lch = *(buf->current)++;

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  lch = (lch << 8) | temp;

          if (lch < 0xdc00 || lch >= 0xdfff)
	    return (EOF);

          ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000;
	}
	break;

    case ENCODE_UTF16LE :
       /*
        * Read UTF-16 little-endian char...
	*/

	if (buf->current >= buf->end)
	  if (mxml_fd_read(buf) < 0)
	    return (EOF);

	temp = *(buf->current)++;

	ch |= (temp << 8);

        if (mxml_bad_char(ch))
	{
	  mxml_error("Bad control character 0x%02x not allowed by XML standard!",
        	     ch);
	  return (EOF);
	}
        else if (ch >= 0xd800 && ch <= 0xdbff)
	{
	 /*
	  * Multi-word UTF-16 char...
	  */

          int lch;

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  lch = *(buf->current)++;

	  if (buf->current >= buf->end)
	    if (mxml_fd_read(buf) < 0)
	      return (EOF);

	  temp = *(buf->current)++;

	  lch |= (temp << 8);

          if (lch < 0xdc00 || lch >= 0xdfff)
	    return (EOF);

          ch = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000;
	}
	break;
  }

#if DEBUG > 1
  printf("mxml_fd_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch);
#endif /* DEBUG > 1 */

  return (ch);
}


/*
 * 'mxml_fd_putc()' - Write a character to a file descriptor.
 */

static int				/* O - 0 on success, -1 on error */
mxml_fd_putc(int  ch,			/* I - Character */
             void *p)			/* I - File descriptor buffer */
{
  _mxml_fdbuf_t	*buf;			/* File descriptor buffer */


 /*
  * Flush the write buffer as needed - note above that "end" still leaves
  * 4 characters at the end so that we can avoid a lot of extra tests...
  */

  buf = (_mxml_fdbuf_t *)p;

  if (buf->current >= buf->end)
    if (mxml_fd_write(buf) < 0)
      return (-1);

  if (ch < 0x80)
  {
   /*
    * Write ASCII character directly...
    */

    *(buf->current)++ = ch;
  }
  else if (ch < 0x800)
  {
   /*
    * Two-byte UTF-8 character...
    */

    *(buf->current)++ = 0xc0 | (ch >> 6);
    *(buf->current)++ = 0x80 | (ch & 0x3f);
  }
  else if (ch < 0x10000)
  {
   /*
    * Three-byte UTF-8 character...
    */

    *(buf->current)++ = 0xe0 | (ch >> 12);
    *(buf->current)++ = 0x80 | ((ch >> 6) & 0x3f);
    *(buf->current)++ = 0x80 | (ch & 0x3f);
  }
  else
  {
   /*
    * Four-byte UTF-8 character...
    */

    *(buf->current)++ = 0xf0 | (ch >> 18);
    *(buf->current)++ = 0x80 | ((ch >> 12) & 0x3f);
    *(buf->current)++ = 0x80 | ((ch >> 6) & 0x3f);
    *(buf->current)++ = 0x80 | (ch & 0x3f);
  }

 /*
  * Return successfully...
  */

  return (0);
}


/*
 * 'mxml_fd_read()' - Read a buffer of data from a file descriptor.
 */

static int				/* O - 0 on success, -1 on error */
mxml_fd_read(_mxml_fdbuf_t *buf)		/* I - File descriptor buffer */
{
  int	bytes;				/* Bytes read... */


 /*
  * Range check input...
  */

  if (!buf)
    return (-1);

 /*
  * Read from the file descriptor...
  */

  while ((bytes = read(buf->fd, buf->buffer, sizeof(buf->buffer))) < 0)
#ifdef EINTR
    if (errno != EAGAIN && errno != EINTR)
#else
    if (errno != EAGAIN)
#endif /* EINTR */
      return (-1);

  if (bytes == 0)
    return (-1);

 /*
  * Update the pointers and return success...
  */

  buf->current = buf->buffer;
  buf->end     = buf->buffer + bytes;

  return (0);
}


/*
 * 'mxml_fd_write()' - Write a buffer of data to a file descriptor.
 */

static int				/* O - 0 on success, -1 on error */
mxml_fd_write(_mxml_fdbuf_t *buf)	/* I - File descriptor buffer */
{
  int		bytes;			/* Bytes written */
  unsigned char	*ptr;			/* Pointer into buffer */


 /*
  * Range check...
  */

  if (!buf)
    return (-1);

 /*
  * Return 0 if there is nothing to write...
  */

  if (buf->current == buf->buffer)
    return (0);

 /*
  * Loop until we have written everything...
  */

  for (ptr = buf->buffer; ptr < buf->current; ptr += bytes)
    if ((bytes = write(buf->fd, ptr, buf->current - ptr)) < 0)
      return (-1);

 /*
  * All done, reset pointers and return success...
  */

  buf->current = buf->buffer;

  return (0);
}


/*
 * 'mxml_file_getc()' - Get a character from a file.
 */

static int				/* O  - Character or EOF */
mxml_file_getc(void *p,			/* I  - Pointer to file */
               int  *encoding)		/* IO - Encoding */
{
  int	ch,				/* Character from file */
	temp;				/* Temporary character */
  FILE	*fp;				/* Pointer to file */


 /*
  * Read a character from the file and see if it is EOF or ASCII...
  */

  fp = (FILE *)p;
  ch = getc(fp);

  if (ch == EOF)
    return (EOF);

  switch (*encoding)
  {
    case ENCODE_UTF8 :
       /*
	* Got a UTF-8 character; convert UTF-8 to Unicode and return...
	*/

	if (!(ch & 0x80))
	{
	  if (mxml_bad_char(ch))
	  {
	    mxml_error("Bad control character 0x%02x not allowed by XML standard!",
        	       ch);
	    return (EOF);
	  }

#if DEBUG > 1
          printf("mxml_file_getc: %c (0x%04x)\n", ch < ' ' ? '.' : ch, ch);
#endif /* DEBUG > 1 */

	  return (ch);
        }