xmlio.c

xml开源解析代码.版本为libxml2-2.6.29,可支持GB3212.网络消息发送XML时很有用.

 * @len:  number of bytes to write
 *
 * Write @len bytes from @buffer to the xml buffer
 *
 * Returns the number of bytes written
 */
static int
xmlBufferWrite (void * context, const char * buffer, int len) {
    int ret;

    ret = xmlBufferAdd((xmlBufferPtr) context, (const xmlChar *) buffer, len);
    if (ret != 0)
        return(-1);
    return(len);
}
#endif

#ifdef HAVE_ZLIB_H
/************************************************************************
 *									*
 *		I/O for compressed file accesses			*
 *									*
 ************************************************************************/
/**
 * xmlGzfileMatch:
 * @filename:  the URI for matching
 *
 * input from compressed file test
 *
 * Returns 1 if matches, 0 otherwise
 */
static int
xmlGzfileMatch (const char *filename ATTRIBUTE_UNUSED) {
    return(1);
}

/**
 * xmlGzfileOpen_real:
 * @filename:  the URI for matching
 *
 * input from compressed file open
 * if @filename is " " then the standard input is used
 *
 * Returns an I/O context or NULL in case of error
 */
static void *
xmlGzfileOpen_real (const char *filename) {
    const char *path = NULL;
    gzFile fd;

    if (!strcmp(filename, "-")) {
        fd = gzdopen(dup(0), "rb");
	return((void *) fd);
    }

    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST "file://localhost/", 17))
#if defined (_WIN32) || defined (__DJGPP__) && !defined(__CYGWIN__)
	path = &filename[17];
#else
	path = &filename[16];
#endif
    else if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST "file:///", 8)) {
#if defined (_WIN32) || defined (__DJGPP__) && !defined(__CYGWIN__)
	path = &filename[8];
#else
	path = &filename[7];
#endif
    } else 
	path = filename;

    if (path == NULL)
	return(NULL);
    if (!xmlCheckFilename(path))
        return(NULL);

    fd = gzopen(path, "rb");
    return((void *) fd);
}

/**
 * xmlGzfileOpen:
 * @filename:  the URI for matching
 *
 * Wrapper around xmlGzfileOpen if the open fais, it will
 * try to unescape @filename
 */
static void *
xmlGzfileOpen (const char *filename) {
    char *unescaped;
    void *retval;

    retval = xmlGzfileOpen_real(filename);
    if (retval == NULL) {
	unescaped = xmlURIUnescapeString(filename, 0, NULL);
	if (unescaped != NULL) {
	    retval = xmlGzfileOpen_real(unescaped);
	}
	xmlFree(unescaped);
    }
    return retval;
}

#ifdef LIBXML_OUTPUT_ENABLED
/**
 * xmlGzfileOpenW:
 * @filename:  the URI for matching
 * @compression:  the compression factor (0 - 9 included)
 *
 * input from compressed file open
 * if @filename is " " then the standard input is used
 *
 * Returns an I/O context or NULL in case of error
 */
static void *
xmlGzfileOpenW (const char *filename, int compression) {
    const char *path = NULL;
    char mode[15];
    gzFile fd;

    snprintf(mode, sizeof(mode), "wb%d", compression);
    if (!strcmp(filename, "-")) {
        fd = gzdopen(dup(1), mode);
	return((void *) fd);
    }

    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST "file://localhost/", 17))
#if defined (_WIN32) || defined (__DJGPP__) && !defined(__CYGWIN__)
	path = &filename[17];
#else
	path = &filename[16];
#endif
    else if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST "file:///", 8)) {
#if defined (_WIN32) || defined (__DJGPP__) && !defined(__CYGWIN__)
	path = &filename[8];
#else
	path = &filename[7];
#endif
    } else 
	path = filename;

    if (path == NULL)
	return(NULL);

    fd = gzopen(path, mode);
    return((void *) fd);
}
#endif /* LIBXML_OUTPUT_ENABLED */

/**
 * xmlGzfileRead:
 * @context:  the I/O context
 * @buffer:  where to drop data
 * @len:  number of bytes to write
 *
 * Read @len bytes to @buffer from the compressed I/O channel.
 *
 * Returns the number of bytes written
 */
static int
xmlGzfileRead (void * context, char * buffer, int len) {
    int ret;

    ret = gzread((gzFile) context, &buffer[0], len);
    if (ret < 0) xmlIOErr(0, "gzread()");
    return(ret);
}

#ifdef LIBXML_OUTPUT_ENABLED
/**
 * xmlGzfileWrite:
 * @context:  the I/O context
 * @buffer:  where to drop data
 * @len:  number of bytes to write
 *
 * Write @len bytes from @buffer to the compressed I/O channel.
 *
 * Returns the number of bytes written
 */
static int
xmlGzfileWrite (void * context, const char * buffer, int len) {
    int ret;

    ret = gzwrite((gzFile) context, (char *) &buffer[0], len);
    if (ret < 0) xmlIOErr(0, "gzwrite()");
    return(ret);
}
#endif /* LIBXML_OUTPUT_ENABLED */

/**
 * xmlGzfileClose:
 * @context:  the I/O context
 *
 * Close a compressed I/O channel
 */
static int
xmlGzfileClose (void * context) {
    int ret;

    ret =  (gzclose((gzFile) context) == Z_OK ) ? 0 : -1;
    if (ret < 0) xmlIOErr(0, "gzclose()");
    return(ret);
}
#endif /* HAVE_ZLIB_H */

#ifdef LIBXML_HTTP_ENABLED
/************************************************************************
 *									*
 *			I/O for HTTP file accesses			*
 *									*
 ************************************************************************/

#ifdef LIBXML_OUTPUT_ENABLED
typedef struct xmlIOHTTPWriteCtxt_
{
    int			compression;

    char *		uri;

    void *		doc_buff;

} xmlIOHTTPWriteCtxt, *xmlIOHTTPWriteCtxtPtr;

#ifdef HAVE_ZLIB_H

#define DFLT_WBITS		( -15 )
#define DFLT_MEM_LVL		( 8 )
#define GZ_MAGIC1		( 0x1f )
#define GZ_MAGIC2		( 0x8b )
#define LXML_ZLIB_OS_CODE	( 0x03 )
#define INIT_HTTP_BUFF_SIZE	( 32768 )
#define DFLT_ZLIB_RATIO		( 5 )

/*
**  Data structure and functions to work with sending compressed data
**  via HTTP.
*/

typedef struct xmlZMemBuff_
{
   unsigned long	size;
   unsigned long	crc;

   unsigned char *	zbuff;
   z_stream		zctrl;

} xmlZMemBuff, *xmlZMemBuffPtr;

/**
 * append_reverse_ulong
 * @buff:  Compressed memory buffer
 * @data:  Unsigned long to append
 *
 * Append a unsigned long in reverse byte order to the end of the
 * memory buffer.
 */
static void
append_reverse_ulong( xmlZMemBuff * buff, unsigned long data ) {

    int		idx;

    if ( buff == NULL )
	return;

    /*
    **  This is plagiarized from putLong in gzio.c (zlib source) where
    **  the number "4" is hardcoded.  If zlib is ever patched to 
    **  support 64 bit file sizes, this code would need to be patched
    **  as well.
    */

    for ( idx = 0; idx < 4; idx++ ) {
	*buff->zctrl.next_out = ( data & 0xff );
	data >>= 8;
	buff->zctrl.next_out++;
    }

    return;
}

/**
 *
 * xmlFreeZMemBuff
 * @buff:  The memory buffer context to clear
 *
 * Release all the resources associated with the compressed memory buffer.
 */
static void
xmlFreeZMemBuff( xmlZMemBuffPtr buff ) {

#ifdef DEBUG_HTTP
    int z_err;
#endif

    if ( buff == NULL )
	return;

    xmlFree( buff->zbuff );
#ifdef DEBUG_HTTP
    z_err = deflateEnd( &buff->zctrl );
    if ( z_err != Z_OK )
	xmlGenericError( xmlGenericErrorContext,
			"xmlFreeZMemBuff:  Error releasing zlib context:  %d\n",
			z_err );
#else
    deflateEnd( &buff->zctrl );
#endif

    xmlFree( buff );
    return;
}

/**
 * xmlCreateZMemBuff
 *@compression:	Compression value to use
 *
 * Create a memory buffer to hold the compressed XML document.  The
 * compressed document in memory will end up being identical to what
 * would be created if gzopen/gzwrite/gzclose were being used to 
 * write the document to disk.  The code for the header/trailer data to
 * the compression is plagiarized from the zlib source files.
 */
static void *
xmlCreateZMemBuff( int compression ) {

    int			z_err;
    int			hdr_lgth;
    xmlZMemBuffPtr	buff = NULL;

    if ( ( compression < 1 ) || ( compression > 9 ) )
	return ( NULL );

    /*  Create the control and data areas  */

    buff = xmlMalloc( sizeof( xmlZMemBuff ) );
    if ( buff == NULL ) {
	xmlIOErrMemory("creating buffer context");
	return ( NULL );
    }

    (void)memset( buff, 0, sizeof( xmlZMemBuff ) );
    buff->size = INIT_HTTP_BUFF_SIZE;
    buff->zbuff = xmlMalloc( buff->size );
    if ( buff->zbuff == NULL ) {
	xmlFreeZMemBuff( buff );
	xmlIOErrMemory("creating buffer");
	return ( NULL );
    }

    z_err = deflateInit2( &buff->zctrl, compression, Z_DEFLATED,
			    DFLT_WBITS, DFLT_MEM_LVL, Z_DEFAULT_STRATEGY );
    if ( z_err != Z_OK ) {
	xmlChar msg[500];
	xmlFreeZMemBuff( buff );
	buff = NULL;
	xmlStrPrintf(msg, 500,
		    (const xmlChar *) "xmlCreateZMemBuff:  %s %d\n",
		    "Error initializing compression context.  ZLIB error:",
		    z_err );
	xmlIOErr(XML_IO_WRITE, (const char *) msg);
	return ( NULL );
    }

    /*  Set the header data.  The CRC will be needed for the trailer  */
    buff->crc = crc32( 0L, NULL, 0 );
    hdr_lgth = snprintf( (char *)buff->zbuff, buff->size,
			"%c%c%c%c%c%c%c%c%c%c",
			GZ_MAGIC1, GZ_MAGIC2, Z_DEFLATED, 
			0, 0, 0, 0, 0, 0, LXML_ZLIB_OS_CODE );
    buff->zctrl.next_out  = buff->zbuff + hdr_lgth;
    buff->zctrl.avail_out = buff->size - hdr_lgth;

    return ( buff );
}

/**
 * xmlZMemBuffExtend
 * @buff:  Buffer used to compress and consolidate data.
 * @ext_amt:   Number of bytes to extend the buffer.
 *
 * Extend the internal buffer used to store the compressed data by the
 * specified amount.
 *
 * Returns 0 on success or -1 on failure to extend the buffer.  On failure
 * the original buffer still exists at the original size.
 */
static int
xmlZMemBuffExtend( xmlZMemBuffPtr buff, size_t ext_amt ) {

    int			rc = -1;
    size_t		new_size;
    size_t		cur_used;

    unsigned char *	tmp_ptr = NULL;

    if ( buff == NULL )
	return ( -1 );

    else if ( ext_amt == 0 )
	return ( 0 );

    cur_used = buff->zctrl.next_out - buff->zbuff;
    new_size = buff->size + ext_amt;

#ifdef DEBUG_HTTP
    if ( cur_used > new_size ) 
	xmlGenericError( xmlGenericErrorContext,
			"xmlZMemBuffExtend:  %s\n%s %d bytes.\n",
			"Buffer overwrite detected during compressed memory",
			"buffer extension.  Overflowed by", 
			(cur_used - new_size ) );
#endif

    tmp_ptr = xmlRealloc( buff->zbuff, new_size );
    if ( tmp_ptr != NULL ) {
	rc = 0;
	buff->size  = new_size;
	buff->zbuff = tmp_ptr;
	buff->zctrl.next_out  = tmp_ptr + cur_used;
	buff->zctrl.avail_out = new_size - cur_used;
    }
    else {
	xmlChar msg[500];
	xmlStrPrintf(msg, 500,
		    (const xmlChar *) "xmlZMemBuffExtend:  %s %lu bytes.\n",
		    "Allocation failure extending output buffer to",
		    new_size );
	xmlIOErr(XML_IO_WRITE, (const char *) msg);
    }

    return ( rc );
}

/**
 * xmlZMemBuffAppend
 * @buff:  Buffer used to compress and consolidate data
 * @src:   Uncompressed source content to append to buffer
 * @len:   Length of source data to append to buffer
 *
 * Compress and append data to the internal buffer.  The data buffer
 * will be expanded if needed to store the additional data.
 *
 * Returns the number of bytes appended to the buffer or -1 on error.
 */
static int
xmlZMemBuffAppend( xmlZMemBuffPtr buff, const char * src, int len ) {

    int		z_err;
    size_t	min_accept;

    if ( ( buff == NULL ) || ( src == NULL ) )
	return ( -1 );

    buff->zctrl.avail_in = len;
    buff->zctrl.next_in  = (unsigned char *)src;
    while ( buff->zctrl.avail_in > 0 ) {
	/*
	**  Extend the buffer prior to deflate call if a reasonable amount
	**  of output buffer space is not available.
	*/
	min_accept = buff->zctrl.avail_in / DFLT_ZLIB_RATIO;
	if ( buff->zctrl.avail_out <= min_accept ) {
	    if ( xmlZMemBuffExtend( buff, buff->size ) == -1 )
		return ( -1 );
	}

	z_err = deflate( &buff->zctrl, Z_NO_FLUSH );
	if ( z_err != Z_OK ) {
	    xmlChar msg[500];
	    xmlStrPrintf(msg, 500,
			(const xmlChar *) "xmlZMemBuffAppend:  %s %d %s - %d",
			"Compression error while appending",
			len, "bytes to buffer.  ZLIB error", z_err );
	    xmlIOErr(XML_IO_WRITE, (const char *) msg);
	    return ( -1 );
	}
    }

    buff->crc = crc32( buff->crc, (unsigned char *)src, len );

    return ( len );
}

/**
 * xmlZMemBuffGetContent
 * @buff:  Compressed memory content buffer
 * @data_ref:  Pointer reference to point to compressed content
 *
 * Flushes the compression buffers, appends gzip file trailers and
 * returns the compressed content and length of the compressed data.
 * NOTE:  The gzip trailer code here is plagiarized from zlib source.
 *
 * Returns the length of the compressed data or -1 on error.
 */
static int
xmlZMemBuffGetContent( xmlZMemBuffPtr buff, char ** data_ref ) {

    int		zlgth = -1;
    int		z_err;

    if ( ( buff == NULL ) || ( data_ref == NULL ) )
	return ( -1 );

    /*  Need to loop until compression output buffers are flushed  */

    do
    {
	z_err = deflate( &buff->zctrl, Z_FINISH );
	if ( z_err == Z_OK ) {
	    /*  In this case Z_OK means more buffer space needed  */

	    if ( xmlZMemBuffExtend( buff, buff->size ) == -1 )
		return ( -1 );
	}
    }
    while ( z_err == Z_OK );

    /*  If the compression state is not Z_STREAM_END, some error occurred  */

    if ( z_err == Z_STREAM_END ) {

	/*  Need to append the gzip data trailer  */

	if ( buff->zctrl.avail_out < ( 2 * sizeof( unsigned long ) ) ) {
	    if ( xmlZMemBuffExtend(buff, (2 * sizeof(unsigned long))) == -1 )
		return ( -1 );
	}

	/*