mxml-file.c

适用于嵌入式系统的XML解析库, 规模比libxml2小得多.

	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_load_cb_t  cb,			/* I - Callback function or MXML_NO_CALLBACK */
    _mxml_getc_cb_t getc_cb,		/* I - Read function */
    mxml_sax_cb_t   sax_cb,		/* I - SAX callback or MXML_NO_CALLBACK */
    void            *sax_data)		/* I - SAX user data */
{
  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 */
  _mxml_global_t *global = _mxml_global();
					/* Global data */
  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 == '<' ||
         (mxml_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 (global->custom_load_cb)
	    {
	     /*
	      * Use the callback to fill in the custom data...
	      */

              node = mxmlNewCustom(parent, NULL, NULL);

	      if ((*global->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 : /* Ignore... */
	    node = NULL;
	    break;
      }	  

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

        mxml_error("Bad %s value '%s' in parent <%s>!",
	           type == MXML_INTEGER ? "integer" : "real", buffer,
		   parent ? parent->value.element.name : "null");
	break;
      }

      bufptr     = buffer;
      whitespace = mxml_isspace(ch) && type == MXML_TEXT;

      if (!node && type != MXML_IGNORE)
      {
       /*
	* Print error and return...
	*/

	mxml_error("Unable to add value node of type %s to parent <%s>!",
	           types[type], parent ? parent->value.element.name : "null");
	goto error;
      }

      if (sax_cb)
      {
        (*sax_cb)(node, MXML_SAX_DATA, sax_data);

        if (!mxmlRelease(node))
          node = NULL;
      }

      if (!first && node)
        first = node;
    }
    else if (mxml_isspace(ch) && type == MXML_TEXT)
      whitespace = 1;

   /*
    * Add lone whitespace node if we have an element and existing
    * whitespace...
    */

    if (ch == '<' && whitespace && type == MXML_TEXT)
    {
      node = mxmlNewText(parent, whitespace, "");

      if (sax_cb)
      {
        (*sax_cb)(node, MXML_SAX_DATA, sax_data);

        if (!mxmlRelease(node))
          node = NULL;
      }

      if (!first && node)
        first = node;

      whitespace = 0;
    }

    if (ch == '<')
    {
     /*
      * Start of open/close tag...
      */

      bufptr = buffer;

      while ((ch = (*getc_cb)(p, &encoding)) != EOF)
        if (mxml_isspace(ch) || ch == '>' || (ch == '/' && bufptr > buffer))
	  break;
	else if (ch == '<')
	{
	  mxml_error("Bare < in element!");
	  goto error;
	}
	else if (ch == '&')
	{
	  if ((ch = mxml_get_entity(parent, p, &encoding, getc_cb)) == EOF)
	    goto error;

	  if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
	    goto error;
	}
	else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
	  goto error;
	else if (((bufptr - buffer) == 1 && buffer[0] == '?') ||
	         ((bufptr - buffer) == 3 && !strncmp(buffer, "!--", 3)) ||
	         ((bufptr - buffer) == 8 && !strncmp(buffer, "![CDATA[", 8)))
	  break;

      *bufptr = '\0';

      if (!strcmp(buffer, "!--"))
      {
       /*
        * Gather rest of comment...
	*/

	while ((ch = (*getc_cb)(p, &encoding)) != EOF)
	{
	  if (ch == '>' && bufptr > (buffer + 4) &&
	      bufptr[-3] != '-' && bufptr[-2] == '-' && bufptr[-1] == '-')
	    break;
	  else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
	    goto error;
	}

       /*
        * Error out if we didn't get the whole comment...
	*/

        if (ch != '>')
	{
	 /*
	  * Print error and return...
	  */

	  mxml_error("Early EOF in comment node!");
	  goto error;
	}


       /*
        * Otherwise add this as an element under the current parent...
	*/

	*bufptr = '\0';

	if ((node = mxmlNewElement(parent, buffer)) == NULL)
	{
	 /*
	  * Just print error for now...
	  */

	  mxml_error("Unable to add comment node to parent <%s>!",
	             parent ? parent->value.element.name : "null");
	  break;
	}

        if (sax_cb)
        {
          (*sax_cb)(node, MXML_SAX_COMMENT, sax_data);

          if (!mxmlRelease(node))
            node = NULL;
        }

	if (node && !first)
	  first = node;
      }
      else if (!strcmp(buffer, "![CDATA["))
      {
       /*
        * Gather CDATA section...
	*/

	while ((ch = (*getc_cb)(p, &encoding)) != EOF)
	{
	  if (ch == '>' && !strncmp(bufptr - 2, "]]", 2))
	    break;
	  else if (mxml_add_char(ch, &bufptr, &buffer, &bufsize))
	    goto error;
	}

       /*
        * Error out if we didn't get the whole comment...
	*/

        if (ch != '>')
	{
	 /*
	  * Print error and return...
	  */

	  mxml_error("Early EOF in CDATA node!");
	  goto error;
	}


       /*
        * Otherwise add this as an element under the current parent...
	*/

	*bufptr = '\0';

	if ((node = mxmlNewElement(parent, buffer)) == NULL)
	{
	 /*
	  * Print error and return...
	  */

	  mxml_error("Unable to add CDATA node to parent <%s>!",
	             parent ? parent->value.element.name : "null");
	  goto error;
	}

        if (sax_cb)
        {
          (*sax_cb)(node, MXML_SAX_CDATA, sax_data);

          if (!mxmlRelease(node))
            node = NULL;
        }

	if (node && !first)
	  first = node;
      }
      else if (buffer[0] == '?')
      {
       /*
        * Gather rest of processing instruction...