xmlsave.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

	return;
    }
    if (cur->type == XML_PI_NODE) {
	if (cur->content != NULL) {
	    xmlOutputBufferWrite(buf, 2, "<?");
	    xmlOutputBufferWriteString(buf, (const char *)cur->name);
	    if (cur->content != NULL) {
		xmlOutputBufferWrite(buf, 1, " ");
		xmlOutputBufferWriteString(buf, (const char *)cur->content);
	    }
	    xmlOutputBufferWrite(buf, 2, "?>");
	} else {
	    xmlOutputBufferWrite(buf, 2, "<?");
	    xmlOutputBufferWriteString(buf, (const char *)cur->name);
	    xmlOutputBufferWrite(buf, 2, "?>");
	}
	return;
    }
    if (cur->type == XML_COMMENT_NODE) {
	if (cur->content != NULL) {
	    xmlOutputBufferWrite(buf, 4, "<!--");
	    xmlOutputBufferWriteString(buf, (const char *)cur->content);
	    xmlOutputBufferWrite(buf, 3, "-->");
	}
	return;
    }
    if (cur->type == XML_ENTITY_REF_NODE) {
        xmlOutputBufferWrite(buf, 1, "&");
	xmlOutputBufferWriteString(buf, (const char *)cur->name);
        xmlOutputBufferWrite(buf, 1, ";");
	return;
    }
    if (cur->type == XML_CDATA_SECTION_NODE) {
	start = end = cur->content;
	while (*end != '\0') {
	    if (*end == ']' && *(end + 1) == ']' && *(end + 2) == '>') {
		end = end + 2;
		xmlOutputBufferWrite(buf, 9, "<![CDATA[");
		xmlOutputBufferWrite(buf, end - start, (const char *)start);
		xmlOutputBufferWrite(buf, 3, "]]>");
		start = end;
	    }
	    end++;
	}
	if (start != end) {
	    xmlOutputBufferWrite(buf, 9, "<![CDATA[");
	    xmlOutputBufferWriteString(buf, (const char *)start);
	    xmlOutputBufferWrite(buf, 3, "]]>");
	}
	return;
    }

    format = ctxt->format;
    if (format == 1) {
	tmp = cur->children;
	while (tmp != NULL) {
	    if ((tmp->type == XML_TEXT_NODE) || 
		(tmp->type == XML_ENTITY_REF_NODE)) {
		format = 0;
		break;
	    }
	    tmp = tmp->next;
	}
    }
    xmlOutputBufferWrite(buf, 1, "<");
    if ((cur->ns != NULL) && (cur->ns->prefix != NULL)) {
        xmlOutputBufferWriteString(buf, (const char *)cur->ns->prefix);
	xmlOutputBufferWrite(buf, 1, ":");
    }

    xmlOutputBufferWriteString(buf, (const char *)cur->name);
    if (cur->nsDef)
        xmlNsListDumpOutput(buf, cur->nsDef);
    if ((xmlStrEqual(cur->name, BAD_CAST "html") &&
	(cur->ns == NULL) && (cur->nsDef == NULL))) {
	/*
	 * 3.1.1. Strictly Conforming Documents A.3.1.1 3/
	 */
	xmlOutputBufferWriteString(buf,
		" xmlns=\"http://www.w3.org/1999/xhtml\"");
    }
    if (cur->properties != NULL)
        xhtmlAttrListDumpOutput(ctxt, cur->properties);

    if ((cur->type == XML_ELEMENT_NODE) && (cur->children == NULL)) {
	if (((cur->ns == NULL) || (cur->ns->prefix == NULL)) &&
	    (xhtmlIsEmpty(cur) == 1)) {
	    /*
	     * C.2. Empty Elements
	     */
	    xmlOutputBufferWrite(buf, 3, " />");
	} else {
	    /*
	     * C.3. Element Minimization and Empty Element Content
	     */
	    xmlOutputBufferWrite(buf, 3, "></");
	    if ((cur->ns != NULL) && (cur->ns->prefix != NULL)) {
		xmlOutputBufferWriteString(buf, (const char *)cur->ns->prefix);
		xmlOutputBufferWrite(buf, 1, ":");
	    }
	    xmlOutputBufferWriteString(buf, (const char *)cur->name);
	    xmlOutputBufferWrite(buf, 1, ">");
	}
	return;
    }
    xmlOutputBufferWrite(buf, 1, ">");
    if ((cur->type != XML_ELEMENT_NODE) && (cur->content != NULL)) {
	xmlOutputBufferWriteEscape(buf, cur->content, ctxt->escape);
    }

    /*
     * 4.8. Script and Style elements
     */
    if ((cur->type == XML_ELEMENT_NODE) &&
	((xmlStrEqual(cur->name, BAD_CAST "script")) ||
	 (xmlStrEqual(cur->name, BAD_CAST "style"))) &&
	((cur->ns == NULL) ||
	 (xmlStrEqual(cur->ns->href, XHTML_NS_NAME)))) {
	xmlNodePtr child = cur->children;

	while (child != NULL) {
	    if ((child->type == XML_TEXT_NODE) ||
		(child->type == XML_CDATA_SECTION_NODE)) {
		/*
		 * Apparently CDATA escaping for style just break on IE,
		 * mozilla and galeon, so ...
		 */
		if (xmlStrEqual(cur->name, BAD_CAST "style") &&
		    (xmlStrchr(child->content, '<') == NULL) &&
		    (xmlStrchr(child->content, '>') == NULL) &&
		    (xmlStrchr(child->content, '&') == NULL)) {
		    int level = ctxt->level;
		    int indent = ctxt->format;

		    ctxt->level = 0;
		    ctxt->format = 0;
		    xhtmlNodeDumpOutput(ctxt, child);
		    ctxt->level = level;
		    ctxt->format = indent;
		} else {
		    start = end = child->content;
		    while (*end != '\0') {
			if (*end == ']' &&
			    *(end + 1) == ']' &&
			    *(end + 2) == '>') {
			    end = end + 2;
			    xmlOutputBufferWrite(buf, 9, "<![CDATA[");
			    xmlOutputBufferWrite(buf, end - start,
						 (const char *)start);
			    xmlOutputBufferWrite(buf, 3, "]]>");
			    start = end;
			}
			end++;
		    }
		    if (start != end) {
			xmlOutputBufferWrite(buf, 9, "<![CDATA[");
			xmlOutputBufferWrite(buf, end - start,
			                     (const char *)start);
			xmlOutputBufferWrite(buf, 3, "]]>");
		    }
		}
	    } else {
		int level = ctxt->level;
		int indent = ctxt->format;

		ctxt->level = 0;
		ctxt->format = 0;
		xhtmlNodeDumpOutput(ctxt, child);
		ctxt->level = level;
		ctxt->format = indent;
	    }
	    child = child->next;
	}
    } else if (cur->children != NULL) {
	int indent = ctxt->format;
	
	if (format) xmlOutputBufferWrite(buf, 1, "\n");
	if (ctxt->level >= 0) ctxt->level++;
	ctxt->format = format;
	xhtmlNodeListDumpOutput(ctxt, cur->children);
	if (ctxt->level > 0) ctxt->level--;
	ctxt->format = indent;
	if ((xmlIndentTreeOutput) && (format))
	    xmlOutputBufferWrite(buf, ctxt->indent_size *
	                         (ctxt->level > ctxt->indent_nr ? 
				  ctxt->indent_nr : ctxt->level),
				 ctxt->indent);
    }
    xmlOutputBufferWrite(buf, 2, "</");
    if ((cur->ns != NULL) && (cur->ns->prefix != NULL)) {
        xmlOutputBufferWriteString(buf, (const char *)cur->ns->prefix);
	xmlOutputBufferWrite(buf, 1, ":");
    }

    xmlOutputBufferWriteString(buf, (const char *)cur->name);
    xmlOutputBufferWrite(buf, 1, ">");
}
#endif

/************************************************************************
 *									*
 *			Public entry points				*
 *									*
 ************************************************************************/

/**
 * xmlSaveToFd:
 * @fd:  a file descriptor number
 * @encoding:  the encoding name to use or NULL
 * @options:  a set of xmlSaveOptions
 *
 * Create a document saving context serializing to a file descriptor
 * with the encoding and the options given.
 *
 * Returns a new serialization context or NULL in case of error.
 */
xmlSaveCtxtPtr
xmlSaveToFd(int fd, const char *encoding, int options)
{
    xmlSaveCtxtPtr ret;

    ret = xmlNewSaveCtxt(encoding, options);
    if (ret == NULL) return(NULL);
    ret->buf = xmlOutputBufferCreateFd(fd, ret->handler);
    if (ret->buf == NULL) {
	xmlFreeSaveCtxt(ret);
	return(NULL);
    }
    return(ret);
}

/**
 * xmlSaveToFilename:
 * @filename:  a file name or an URL
 * @encoding:  the encoding name to use or NULL
 * @options:  a set of xmlSaveOptions
 *
 * Create a document saving context serializing to a filename or possibly
 * to an URL (but this is less reliable) with the encoding and the options
 * given.
 *
 * Returns a new serialization context or NULL in case of error.
 */
xmlSaveCtxtPtr
xmlSaveToFilename(const char *filename, const char *encoding, int options)
{
    xmlSaveCtxtPtr ret;
    int compression = 0; /* TODO handle compression option */

    ret = xmlNewSaveCtxt(encoding, options);
    if (ret == NULL) return(NULL);
    ret->buf = xmlOutputBufferCreateFilename(filename, ret->handler,
                                             compression);
    if (ret->buf == NULL) {
	xmlFreeSaveCtxt(ret);
	return(NULL);
    }
    return(ret);
}

#if 0
/**
 * xmlSaveToBuffer:
 * @buffer:  a buffer
 * @encoding:  the encoding name to use or NULL
 * @options:  a set of xmlSaveOptions
 *
 * Create a document saving context serializing to a buffer
 * with the encoding and the options given
 *
 * Returns a new serialization context or NULL in case of error.
 */
xmlSaveCtxtPtr
xmlSaveToBuffer(xmlBufferPtr buffer, const char *encoding, int options)
{
    TODO
    return(NULL);
}
#endif

/**
 * xmlSaveToIO:
 * @iowrite:  an I/O write function
 * @ioclose:  an I/O close function
 * @ioctx:  an I/O handler
 * @encoding:  the encoding name to use or NULL
 * @options:  a set of xmlSaveOptions
 *
 * Create a document saving context serializing to a file descriptor
 * with the encoding and the options given
 *
 * Returns a new serialization context or NULL in case of error.
 */
xmlSaveCtxtPtr
xmlSaveToIO(xmlOutputWriteCallback iowrite,
            xmlOutputCloseCallback ioclose,
            void *ioctx, const char *encoding, int options)
{
    xmlSaveCtxtPtr ret;

    ret = xmlNewSaveCtxt(encoding, options);
    if (ret == NULL) return(NULL);
    ret->buf = xmlOutputBufferCreateIO(iowrite, ioclose, ioctx, ret->handler);
    if (ret->buf == NULL) {
	xmlFreeSaveCtxt(ret);
	return(NULL);
    }
    return(ret);
}

/**
 * xmlSaveDoc:
 * @ctxt:  a document saving context
 * @doc:  a document
 *
 * Save a full document to a saving context
 * TODO: The function is not fully implemented yet as it does not return the
 * byte count but 0 instead
 *
 * Returns the number of byte written or -1 in case of error
 */
long
xmlSaveDoc(xmlSaveCtxtPtr ctxt, xmlDocPtr doc)
{
    long ret = 0;

    xmlDocContentDumpOutput(ctxt, doc);
    return(ret);
}

/**
 * xmlSaveTree:
 * @ctxt:  a document saving context
 * @node:  a document
 *
 * Save a subtree starting at the node parameter to a saving context
 * TODO: The function is not fully implemented yet as it does not return the
 * byte count but 0 instead
 *
 * Returns the number of byte written or -1 in case of error
 */
long
xmlSaveTree(xmlSaveCtxtPtr ctxt, xmlNodePtr node)
{
    long ret = 0;

    xmlNodeDumpOutputInternal(ctxt, node);
    return(ret);
}

/**
 * xmlSaveFlush:
 * @ctxt:  a document saving context
 *
 * Flush a document saving context, i.e. make sure that all bytes have
 * been output.
 *
 * Returns the number of byte written or -1 in case of error.
 */
int
xmlSaveFlush(xmlSaveCtxtPtr ctxt)
{
    if (ctxt == NULL) return(-1);
    if (ctxt->buf == NULL) return(-1);
    return(xmlOutputBufferFlush(ctxt->buf));
}

/**
 * xmlSaveClose:
 * @ctxt:  a document saving context
 *
 * Close a document saving context, i.e. make sure that all bytes have
 * been output and free the associated data.
 *
 * Returns the number of byte written or -1 in case of error.
 */
int
xmlSaveClose(xmlSaveCtxtPtr ctxt)
{
    int ret;

    if (ctxt == NULL) return(-1);
    ret = xmlSaveFlush(ctxt);
    xmlFreeSaveCtxt(ctxt);
    return(ret);
}

/**
 * xmlSaveSetEscape:
 * @ctxt:  a document saving context
 * @escape:  the escaping function
 *
 * Set a custom escaping function to be used for text in element content
 *
 * Returns 0 if successful or -1 in case of error.
 */
int
xmlSaveSetEscape(xmlSaveCtxtPtr ctxt, xmlCharEncodingOutputFunc escape)
{
    if (ctxt == NULL) return(-1);
    ctxt->escape = escape;
    return(0);
}

/**
 * xmlSaveSetAttrEscape:
 * @ctxt:  a document saving context
 * @escape:  the escaping function
 *
 * Set a custom escaping function to be used for text in attribute content
 *
 * Returns 0 if successful or -1 in case of error.
 */
int
xmlSaveSetAttrEscape(xmlSaveCtxtPtr ctxt, xmlCharEncodingOutputFunc escape)
{
    if (ctxt == NULL) return(-1);
    ctxt->escapeAttr = escape;
    return(0);
}

/************************************************************************
 *									*
 *		Public entry points based on buffers			*
 *									*
 ************************************************************************/
/**
 * xmlAttrSerializeTxtContent:
 * @buf:  the XML buffer output
 * @doc:  the document
 * @attr: the attribute node
 * @string: the text content
 *
 * Serialize text attribute values to an xml simple buffer
 */
void
xmlAttrSerializeTxtContent(xmlBufferPtr buf, xmlDocPtr doc,
                           xmlAttrPtr attr, const xmlChar * string)
{
    xmlChar *base, *cur;

    if (string == NULL)
        return;
    base = cur = (xmlChar *) string;
    while (*cur != 0) {
        if (*cur == '\n') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&#10;", 5);
            cur++;
            base = cur;
        } else if (*cur == '\r') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&#13;", 5);
            cur++;
            base = cur;
        } else if (*cur == '\t') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&#9;", 4);
            cur++;
            base = cur;
        } else if (*cur == '"') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&quot;", 6);
            cur++;
            base = cur;
        } else if (*cur == '<') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&lt;", 4);
            cur++;
            base = cur;
        } else if (*cur == '>') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&gt;", 4);
            cur++;
            base = cur;
        } else if (*cur == '&') {
            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            xmlBufferAdd(buf, BAD_CAST "&amp;", 5);
            cur++;
            base = cur;
        } else if ((*cur >= 0x80) && ((doc == NULL) ||
                                      (doc->encoding == NULL))) {
            /*
             * We assume we have UTF-8 content.
             */
            unsigned char tmp[10];
            int val = 0, l = 1;

            if (base != cur)
                xmlBufferAdd(buf, base, cur - base);
            if (*cur < 0xC0) {
                xmlSaveErr(XML_SAVE_NOT_UTF8, (xmlNodePtr) attr, NULL);
                if (doc != NULL)
                    doc->encoding = xmlStrdup(BAD_CAST "ISO-8859-1");
		xmlSerializeHexCharRef(tmp, *cur);
                xmlBufferAdd(buf, (xmlChar *) tmp, -1);
                cur++;
                base = cur;
                continue;
            } else if (*cur < 0xE0) {
                val = (cur[0]) & 0x1F;
                val <<= 6;
                val |= (cur[1]) & 0x3F;
                l = 2;
            } else if (*cur < 0xF0) {
                val = (cur[0]) & 0x0F;
                val <<= 6;
                val |= (cur[1]) & 0x3F;
                val <<= 6;
                val |= (cur[2]) & 0x3F;
                l = 3;
            } else if (*cur < 0xF8) {
                val = (cur[0]) & 0x07;
                val <<= 6;
                val |= (cur[1]) & 0x3F;
                val <<= 6;
                val |= (cur[2]) & 0x3F;
                val <<= 6;
                val |= (cur[3]) & 0x3F;
                l = 4;
            }
            if ((l == 1) || (!IS_CHAR(val))) {
                xmlSaveErr(XML_SAVE_CHAR_INVALID, (xmlNodePtr) attr, NULL);
                if (doc != NULL)
                    doc->encoding = xmlStrdup(BAD_CAST "ISO-8859-1");
		
		xmlSerializeHexCharRef(tmp, *cur);
                xmlBufferAdd(buf, (xmlChar *) tmp, -1);
                cur++;
                base = cur;
                continue;
            }
            /*
             * We could do multiple things here. Just save
             * as a char ref
             */
	    xmlSerializeHexCharRef(tmp, val);
            xmlBufferAdd(buf, (xmlChar *) tmp, -1);
            cur += l;
            base = cur;
        } else {