sax2.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

/**
 * xmlSAX2StartElement:
 * @ctx: the user data (XML parser context)
 * @fullname:  The element name, including namespace prefix
 * @atts:  An array of name/value attributes pairs, NULL terminated
 *
 * called when an opening tag has been processed.
 */
void
xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    xmlNodePtr ret;
    xmlNodePtr parent = ctxt->node;
    xmlNsPtr ns;
    xmlChar *name;
    xmlChar *prefix;
    const xmlChar *att;
    const xmlChar *value;
    int i;

#ifdef DEBUG_SAX
    xmlGenericError(xmlGenericErrorContext,
	    "SAX.xmlSAX2StartElement(%s)\n", fullname);
#endif

    /*
     * First check on validity:
     */
    if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && 
        ((ctxt->myDoc->intSubset == NULL) ||
	 ((ctxt->myDoc->intSubset->notations == NULL) && 
	  (ctxt->myDoc->intSubset->elements == NULL) &&
	  (ctxt->myDoc->intSubset->attributes == NULL) && 
	  (ctxt->myDoc->intSubset->entities == NULL)))) {
	xmlErrValid(ctxt, XML_ERR_NO_DTD,
	  "Validation failed: no DTD found !", NULL, NULL);
	ctxt->validate = 0;
    }
       

    /*
     * Split the full name into a namespace prefix and the tag name
     */
    name = xmlSplitQName(ctxt, fullname, &prefix);


    /*
     * Note : the namespace resolution is deferred until the end of the
     *        attributes parsing, since local namespace can be defined as
     *        an attribute at this level.
     */
    ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, name, NULL);
    if (ret == NULL) {
        if (prefix != NULL)
	    xmlFree(prefix);
	ctxt->errNo = XML_ERR_NO_MEMORY;
	ctxt->instate = XML_PARSER_EOF;
	ctxt->disableSAX = 1;
        return;
    }
    if (ctxt->myDoc->children == NULL) {
#ifdef DEBUG_SAX_TREE
	xmlGenericError(xmlGenericErrorContext, "Setting %s as root\n", name);
#endif
        xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
    } else if (parent == NULL) {
        parent = ctxt->myDoc->children;
    }
    ctxt->nodemem = -1;
    if (ctxt->linenumbers) {
	if (ctxt->input != NULL) {
	    if (ctxt->input->line < 65535)
		ret->line = (short) ctxt->input->line;
	    else
	        ret->line = 65535;
	}
    }

    /*
     * We are parsing a new node.
     */
#ifdef DEBUG_SAX_TREE
    xmlGenericError(xmlGenericErrorContext, "pushing(%s)\n", name);
#endif
    nodePush(ctxt, ret);

    /*
     * Link the child element
     */
    if (parent != NULL) {
        if (parent->type == XML_ELEMENT_NODE) {
#ifdef DEBUG_SAX_TREE
	    xmlGenericError(xmlGenericErrorContext,
		    "adding child %s to %s\n", name, parent->name);
#endif
	    xmlAddChild(parent, ret);
	} else {
#ifdef DEBUG_SAX_TREE
	    xmlGenericError(xmlGenericErrorContext,
		    "adding sibling %s to ", name);
	    xmlDebugDumpOneNode(stderr, parent, 0);
#endif
	    xmlAddSibling(parent, ret);
	}
    }

    /*
     * Insert all the defaulted attributes from the DTD especially namespaces
     */
    if ((!ctxt->html) &&
	((ctxt->myDoc->intSubset != NULL) ||
	 (ctxt->myDoc->extSubset != NULL))) {
	xmlCheckDefaultedAttributes(ctxt, name, prefix, atts);
    }

    /*
     * process all the attributes whose name start with "xmlns"
     */
    if (atts != NULL) {
        i = 0;
	att = atts[i++];
	value = atts[i++];
	if (!ctxt->html) {
	    while ((att != NULL) && (value != NULL)) {
		if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') &&
		    (att[3] == 'n') && (att[4] == 's'))
		    xmlSAX2AttributeInternal(ctxt, att, value, prefix);

		att = atts[i++];
		value = atts[i++];
	    }
	}
    }

    /*
     * Search the namespace, note that since the attributes have been
     * processed, the local namespaces are available.
     */
    ns = xmlSearchNs(ctxt->myDoc, ret, prefix);
    if ((ns == NULL) && (parent != NULL))
	ns = xmlSearchNs(ctxt->myDoc, parent, prefix);
    if ((prefix != NULL) && (ns == NULL)) {
	ns = xmlNewNs(ret, NULL, prefix);
	if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
	    ctxt->sax->warning(ctxt->userData, 
		 "Namespace prefix %s is not defined\n", prefix);
    }

    /*
     * set the namespace node, making sure that if the default namspace
     * is unbound on a parent we simply kee it NULL
     */
    if ((ns != NULL) && (ns->href != NULL) &&
	((ns->href[0] != 0) || (ns->prefix != NULL)))
	xmlSetNs(ret, ns);

    /*
     * process all the other attributes
     */
    if (atts != NULL) {
        i = 0;
	att = atts[i++];
	value = atts[i++];
	if (ctxt->html) {
	    while (att != NULL) {
		xmlSAX2AttributeInternal(ctxt, att, value, NULL);
		att = atts[i++];
		value = atts[i++];
	    }
	} else {
	    while ((att != NULL) && (value != NULL)) {
		if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') ||
		    (att[3] != 'n') || (att[4] != 's'))
		    xmlSAX2AttributeInternal(ctxt, att, value, NULL);

		/*
		 * Next ones
		 */
		att = atts[i++];
		value = atts[i++];
	    }
	}
    }

#ifdef LIBXML_VALID_ENABLED
    /*
     * If it's the Document root, finish the DTD validation and
     * check the document root element for validity
     */
    if ((ctxt->validate) && (ctxt->vctxt.finishDtd == 0)) {
	int chk;

	chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc);
	if (chk <= 0)
	    ctxt->valid = 0;
	if (chk < 0)
	    ctxt->wellFormed = 0;
	ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
	ctxt->vctxt.finishDtd = 1;
    }
#endif /* LIBXML_VALID_ENABLED */

    if (prefix != NULL)
	xmlFree(prefix);

}

/**
 * xmlSAX2EndElement:
 * @ctx: the user data (XML parser context)
 * @name:  The element name
 *
 * called when the end of an element has been detected.
 */
void
xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    xmlParserNodeInfo node_info;
    xmlNodePtr cur = ctxt->node;

#ifdef DEBUG_SAX
    if (name == NULL)
        xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(NULL)\n");
    else
	xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(%s)\n", name);
#endif
    
    /* Capture end position and add node */
    if (cur != NULL && ctxt->record_info) {
      node_info.end_pos = ctxt->input->cur - ctxt->input->base;
      node_info.end_line = ctxt->input->line;
      node_info.node = cur;
      xmlParserAddNodeInfo(ctxt, &node_info);
    }
    ctxt->nodemem = -1;

#ifdef LIBXML_VALID_ENABLED
    if (ctxt->validate && ctxt->wellFormed &&
        ctxt->myDoc && ctxt->myDoc->intSubset)
        ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc,
					     cur);
#endif /* LIBXML_VALID_ENABLED */

    
    /*
     * end of parsing of this node.
     */
#ifdef DEBUG_SAX_TREE
    xmlGenericError(xmlGenericErrorContext, "popping(%s)\n", cur->name);
#endif
    nodePop(ctxt);
}
#endif /* LIBXML_SAX1_ENABLED || LIBXML_HTML_ENABLE */

/*
 * xmlSAX2TextNode:
 * @ctxt:  the parser context
 * @str:  the input string
 * @len: the string length
 * 
 * Remove the entities from an attribute value
 *
 * Returns the newly allocated string or NULL if not needed or error
 */
static xmlNodePtr
xmlSAX2TextNode(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) {
    xmlNodePtr ret;
    const xmlChar *intern = NULL;

    /*
     * Allocate
     */
    if (ctxt->freeElems != NULL) {
	ret = ctxt->freeElems;
	ctxt->freeElems = ret->next;
	ctxt->freeElemsNr--;
    } else {
	ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode));
    }
    if (ret == NULL) {
	if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL))
	    ctxt->sax->error(ctxt->userData, 
		 "SAX.xmlSAX2Characters(): out of memory\n");
	ctxt->errNo = XML_ERR_NO_MEMORY;
	ctxt->instate = XML_PARSER_EOF;
	ctxt->disableSAX = 1;
	return(NULL);
    }
    /*
     * intern the formatting blanks found between tags, or the
     * very short strings
     */
    if (ctxt->dictNames) {
        xmlChar cur = str[len];

	if ((len <= 3) && ((cur == '"') || (cur == '\'') ||
	    ((cur == '<') && (str[len + 1] != '!')))) {
	    intern = xmlDictLookup(ctxt->dict, str, len);
	} else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') &&
	           (str[len + 1] != '!')) {
	    int i;

	    for (i = 1;i < len;i++) {
		if (!IS_BLANK_CH(str[i])) goto skip;
	    }
	    intern = xmlDictLookup(ctxt->dict, str, len);
	}
    }
skip:
    memset(ret, 0, sizeof(xmlNode));
    ret->type = XML_TEXT_NODE;

    ret->name = xmlStringText;
    if (intern == NULL)
	ret->content = xmlStrndup(str, len);
    else
	ret->content = (xmlChar *) intern;

    if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
	xmlRegisterNodeDefaultValue(ret);
    return(ret);
}

#ifdef LIBXML_VALID_ENABLED
/*
 * xmlSAX2DecodeAttrEntities:
 * @ctxt:  the parser context
 * @str:  the input string
 * @len: the string length
 * 
 * Remove the entities from an attribute value
 *
 * Returns the newly allocated string or NULL if not needed or error
 */
static xmlChar *
xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str,
                          const xmlChar *end) {
    const xmlChar *in;
    xmlChar *ret;

    in = str;
    while (in < end)
        if (*in++ == '&')
	    goto decode;
    return(NULL);
decode:
    ctxt->depth++;
    ret = xmlStringLenDecodeEntities(ctxt, str, end - str,
				     XML_SUBSTITUTE_REF, 0,0,0);
    ctxt->depth--;
    return(ret);
}
#endif /* LIBXML_VALID_ENABLED */

/**
 * xmlSAX2AttributeNs:
 * @ctx: the user data (XML parser context)
 * @localname:  the local name of the attribute
 * @prefix:  the attribute namespace prefix if available
 * @URI:  the attribute namespace name if available
 * @value:  Start of the attribute value
 * @valueend: end of the attribute value
 *
 * Handle an attribute that has been read by the parser.
 * The default handling is to convert the attribute into an
 * DOM subtree and past it in a new xmlAttr element added to
 * the element.
 */
static void
xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,
                   const xmlChar * localname,
                   const xmlChar * prefix,
		   const xmlChar * value,
		   const xmlChar * valueend)
{
    xmlAttrPtr ret;
    xmlNsPtr namespace = NULL;
    xmlChar *dup = NULL;

    /*
     * Note: if prefix == NULL, the attribute is not in the default namespace
     */
    if (prefix != NULL)
	namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, prefix);

    /*
     * allocate the node
     */
    if (ctxt->freeAttrs != NULL) {
        ret = ctxt->freeAttrs;
	ctxt->freeAttrs = ret->next;
	ctxt->freeAttrsNr--;
	memset(ret, 0, sizeof(xmlAttr));
	ret->type = XML_ATTRIBUTE_NODE;

	ret->parent = ctxt->node; 
	ret->doc = ctxt->myDoc;
	ret->ns = namespace;

	if (ctxt->dictNames)
	    ret->name = localname;
	else
	    ret->name = xmlStrdup(localname);

        /* link at the end to preserv order, TODO speed up with a last */
	if (ctxt->node->properties == NULL) {
	    ctxt->node->properties = ret;
	} else {
	    xmlAttrPtr prev = ctxt->node->properties;

	    while (prev->next != NULL) prev = prev->next;
	    prev->next = ret;
	    ret->prev = prev;
	}

	if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
	    xmlRegisterNodeDefaultValue((xmlNodePtr)ret);
    } else {
	if (ctxt->dictNames)
	    ret = xmlNewNsPropEatName(ctxt->node, namespace, 
	                              (xmlChar *) localname, NULL);
	else
	    ret = xmlNewNsProp(ctxt->node, namespace, localname, NULL);
	if (ret == NULL) {
	    ctxt->errNo = XML_ERR_NO_MEMORY;
	    ctxt->instate = XML_PARSER_EOF;
	    ctxt->disableSAX = 1;
	    return;
	}
    }

    if ((ctxt->replaceEntities == 0) && (!ctxt->html)) {
	xmlNodePtr tmp;

	/*
	 * We know that if there is an entity reference, then
	 * the string has been dup'ed and terminates with 0
	 * otherwise with ' or "
	 */
	if (*valueend != 0) {
	    tmp = xmlSAX2TextNode(ctxt, value, valueend - value);
	    ret->children = tmp;
	    ret->last = tmp;
	    if (tmp != NULL) {
		tmp->doc = ret->doc;
		tmp->parent = (xmlNodePtr) ret;
	    }
	} else {
	    ret->children = xmlStringLenGetNodeList(ctxt->myDoc, value,
						    valueend - value);
	    tmp = ret->children;
	    while (tmp != NULL) {
		tmp->parent = (xmlNodePtr) ret;
		if (tmp->next == NULL)
		    ret->last = tmp;
		tmp = tmp->next;
	    }
	}
    } else if (value != NULL) {
	xmlNodePtr tmp;

	tmp = xmlSAX2TextNode(ctxt, value, valueend - value);