mxml-file.c

asterisk 是一个很有知名度开源软件

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

          ch = getc(fp);
	  if (ch != 0xff)
	    return (EOF);

	  *encoding = ENCODE_UTF16BE;

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

          ch = getc(fp);
	  if (ch != 0xfe)
	    return (EOF);

	  *encoding = ENCODE_UTF16LE;

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

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

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

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

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

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

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

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

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

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

	  if ((temp = getc(fp)) == EOF || (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...
	*/

	ch = (ch << 8) | getc(fp);

	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 = (getc(fp) << 8) | getc(fp);

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

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

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

	ch |= (getc(fp) << 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 = getc(fp) | (getc(fp) << 8);

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

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

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

  return (ch);
}


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

static int				/* O - 0 on success, -1 on failure */
mxml_file_putc(int  ch,			/* I - Character to write */
               void *p)			/* I - Pointer to file */
{
  char	buffer[4],			/* Buffer for character */
	*bufptr;			/* Pointer into buffer */
  int	buflen;				/* Number of bytes to write */


  if (ch < 0x80)
    return (putc(ch, (FILE *)p) == EOF ? -1 : 0);

  bufptr = buffer;

  if (ch < 0x800)
  {
   /*
    * Two-byte UTF-8 character...
    */

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

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

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

  buflen = bufptr - buffer;

  return (fwrite(buffer, 1, buflen, (FILE *)p) < buflen ? -1 : 0);
}


/*
 * 'mxml_get_entity()' - Get the character corresponding to an entity...
 */

static int				/* O  - Character value or EOF on error */
mxml_get_entity(mxml_node_t *parent,	/* I  - Parent node */
		void        *p,		/* I  - Pointer to source */
		int         *encoding,	/* IO - Character encoding */
                int         (*getc_cb)(void *, int *))
					/* I  - Get character function */
{
  int	ch;				/* Current character */
  char	entity[64],			/* Entity string */
	*entptr;			/* Pointer into entity */


  entptr = entity;

  while ((ch = (*getc_cb)(p, encoding)) != EOF)
    if (ch > 126 || (!isalnum(ch) && ch != '#'))
      break;
    else if (entptr < (entity + sizeof(entity) - 1))
      *entptr++ = ch;
    else
    {
      mxml_error("Entity name too long under parent <%s>!",
	         parent ? parent->value.element.name : "null");
      break;
    }

  *entptr = '\0';

  if (ch != ';')
  {
    mxml_error("Character entity \"%s\" not terminated under parent <%s>!",
	       entity, parent ? parent->value.element.name : "null");
    return (EOF);
  }

  if (entity[0] == '#')
  {
    if (entity[1] == 'x')
      ch = strtol(entity + 2, NULL, 16);
    else
      ch = strtol(entity + 1, NULL, 10);
  }
  else if ((ch = mxmlEntityGetValue(entity)) < 0)
    mxml_error("Entity name \"%s;\" not supported under parent <%s>!",
	       entity, parent ? parent->value.element.name : "null");

  if (mxml_bad_char(ch))
  {
    mxml_error("Bad control character 0x%02x under parent <%s> not allowed by XML standard!",
               ch, parent ? parent->value.element.name : "null");
    return (EOF);
  }

  return (ch);
}


/*
 * 'mxml_load_data()' - Load data into an XML node tree.
 */

static mxml_node_t *			/* O - First node or NULL if the file could not be read. */
mxml_load_data(mxml_node_t *top,	/* I - Top node */
               void        *p,		/* I - Pointer to data */
               mxml_type_t (*cb)(mxml_node_t *),
					/* I - Callback function or MXML_NO_CALLBACK */
               int         (*getc_cb)(void *, int *))
					/* I - Read function */
{
  mxml_node_t	*node,			/* Current node */
		*first,			/* First node added */
		*parent;		/* Current parent node */
  int		ch,			/* Character from file */
		whitespace;		/* Non-zero if whitespace seen */
  char		*buffer,		/* String buffer */
		*bufptr;		/* Pointer into buffer */
  int		bufsize;		/* Size of buffer */
  mxml_type_t	type;			/* Current node type */
  int		encoding;		/* Character encoding */
  static const char * const types[] =	/* Type strings... */
		{
		  "MXML_ELEMENT",	/* XML element with attributes */
		  "MXML_INTEGER",	/* Integer value */
		  "MXML_OPAQUE",	/* Opaque string */
		  "MXML_REAL",		/* Real value */
		  "MXML_TEXT",		/* Text fragment */
		  "MXML_CUSTOM"		/* Custom data */
		};


 /*
  * Read elements and other nodes from the file...
  */

  if ((buffer = malloc(64)) == NULL)
  {
    mxml_error("Unable to allocate string buffer!");
    return (NULL);
  }

  bufsize    = 64;
  bufptr     = buffer;
  parent     = top;
  first      = NULL;
  whitespace = 0;
  encoding   = ENCODE_UTF8;

  if (cb && parent)
    type = (*cb)(parent);
  else
    type = MXML_TEXT;

  while ((ch = (*getc_cb)(p, &encoding)) != EOF)
  {
    if ((ch == '<' ||
         (isspace(ch) && type != MXML_OPAQUE && type != MXML_CUSTOM)) &&
        bufptr > buffer)
    {
     /*
      * Add a new value node...
      */

      *bufptr = '\0';

      switch (type)
      {
	case MXML_INTEGER :
            node = mxmlNewInteger(parent, strtol(buffer, &bufptr, 0));
	    break;

	case MXML_OPAQUE :
            node = mxmlNewOpaque(parent, buffer);
	    break;

	case MXML_REAL :
            node = mxmlNewReal(parent, strtod(buffer, &bufptr));
	    break;

	case MXML_TEXT :
            node = mxmlNewText(parent, whitespace, buffer);
	    break;

	case MXML_CUSTOM :
	    if (mxml_custom_load_cb)
	    {
	     /*
	      * Use the callback to fill in the custom data...
	      */

              node = mxmlNewCustom(parent, NULL, NULL);

	      if ((*mxml_custom_load_cb)(node, buffer))
	      {
	        mxml_error("Bad custom value '%s' in parent <%s>!",
		           buffer, parent ? parent->value.element.name : "null");
		mxmlDelete(node);
		node = NULL;
	      }
	      break;
	    }

        default : /* Should never happen... */
	    node = NULL;
	    break;
      }	  

      if (*bufptr)
      {
       /*
        * Bad integer/real number value...
	*/