htmlparser.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.

    {"head",  200},
    {"body",  200},
    {"html",  220},
    {NULL,    100} /* Default priority */
};

static const char** htmlStartCloseIndex[100];
static int htmlStartCloseIndexinitialized = 0;

/************************************************************************
 *									*
 * 		functions to handle HTML specific data			*
 *									*
 ************************************************************************/

/**
 * htmlInitAutoClose:
 *
 * Initialize the htmlStartCloseIndex for fast lookup of closing tags names.
 * This is not reentrant. Call xmlInitParser() once before processing in
 * case of use in multithreaded programs.
 */
void
htmlInitAutoClose(void) {
    int indx, i = 0;

    if (htmlStartCloseIndexinitialized) return;

    for (indx = 0;indx < 100;indx ++) htmlStartCloseIndex[indx] = NULL;
    indx = 0;
    while ((htmlStartClose[i] != NULL) && (indx < 100 - 1)) {
        htmlStartCloseIndex[indx++] = (const char**) &htmlStartClose[i];
	while (htmlStartClose[i] != NULL) i++;
	i++;
    }
    htmlStartCloseIndexinitialized = 1;
}

/**
 * htmlTagLookup:
 * @tag:  The tag name in lowercase
 *
 * Lookup the HTML tag in the ElementTable
 *
 * Returns the related htmlElemDescPtr or NULL if not found.
 */
const htmlElemDesc *
htmlTagLookup(const xmlChar *tag) {
    unsigned int i;

    for (i = 0; i < (sizeof(html40ElementTable) /
                     sizeof(html40ElementTable[0]));i++) {
        if (!xmlStrcasecmp(tag, BAD_CAST html40ElementTable[i].name))
	    return((htmlElemDescPtr) &html40ElementTable[i]);
    }
    return(NULL);
}

/**
 * htmlGetEndPriority:
 * @name: The name of the element to look up the priority for.
 * 
 * Return value: The "endtag" priority.
 **/
static int
htmlGetEndPriority (const xmlChar *name) {
    int i = 0;

    while ((htmlEndPriority[i].name != NULL) &&
	   (!xmlStrEqual((const xmlChar *)htmlEndPriority[i].name, name)))
	i++;

    return(htmlEndPriority[i].priority);
}


/**
 * htmlCheckAutoClose:
 * @newtag:  The new tag name
 * @oldtag:  The old tag name
 *
 * Checks whether the new tag is one of the registered valid tags for
 * closing old.
 * Initialize the htmlStartCloseIndex for fast lookup of closing tags names.
 *
 * Returns 0 if no, 1 if yes.
 */
static int
htmlCheckAutoClose(const xmlChar * newtag, const xmlChar * oldtag)
{
    int i, indx;
    const char **closed = NULL;

    if (htmlStartCloseIndexinitialized == 0)
        htmlInitAutoClose();

    /* inefficient, but not a big deal */
    for (indx = 0; indx < 100; indx++) {
        closed = htmlStartCloseIndex[indx];
        if (closed == NULL)
            return (0);
        if (xmlStrEqual(BAD_CAST * closed, newtag))
            break;
    }

    i = closed - htmlStartClose;
    i++;
    while (htmlStartClose[i] != NULL) {
        if (xmlStrEqual(BAD_CAST htmlStartClose[i], oldtag)) {
            return (1);
        }
        i++;
    }
    return (0);
}

/**
 * htmlAutoCloseOnClose:
 * @ctxt:  an HTML parser context
 * @newtag:  The new tag name
 * @force:  force the tag closure
 *
 * The HTML DTD allows an ending tag to implicitly close other tags.
 */
static void
htmlAutoCloseOnClose(htmlParserCtxtPtr ctxt, const xmlChar * newtag)
{
    const htmlElemDesc *info;
    int i, priority;

    priority = htmlGetEndPriority(newtag);

    for (i = (ctxt->nameNr - 1); i >= 0; i--) {

        if (xmlStrEqual(newtag, ctxt->nameTab[i]))
            break;
        /*
         * A missplaced endtag can only close elements with lower
         * or equal priority, so if we find an element with higher
         * priority before we find an element with
         * matching name, we just ignore this endtag 
         */
        if (htmlGetEndPriority(ctxt->nameTab[i]) > priority)
            return;
    }
    if (i < 0)
        return;

    while (!xmlStrEqual(newtag, ctxt->name)) {
        info = htmlTagLookup(ctxt->name);
        if ((info != NULL) && (info->endTag == 3)) {
            htmlParseErr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
	                 "Opening and ending tag mismatch: %s and %s\n",
			 newtag, ctxt->name);
        }
        if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
            ctxt->sax->endElement(ctxt->userData, ctxt->name);
	htmlnamePop(ctxt);
    }
}

/**
 * htmlAutoCloseOnEnd:
 * @ctxt:  an HTML parser context
 *
 * Close all remaining tags at the end of the stream
 */
static void
htmlAutoCloseOnEnd(htmlParserCtxtPtr ctxt)
{
    int i;

    if (ctxt->nameNr == 0)
        return;
    for (i = (ctxt->nameNr - 1); i >= 0; i--) {
        if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
            ctxt->sax->endElement(ctxt->userData, ctxt->name);
	htmlnamePop(ctxt);
    }
}

/**
 * htmlAutoClose:
 * @ctxt:  an HTML parser context
 * @newtag:  The new tag name or NULL
 *
 * The HTML DTD allows a tag to implicitly close other tags.
 * The list is kept in htmlStartClose array. This function is
 * called when a new tag has been detected and generates the
 * appropriates closes if possible/needed.
 * If newtag is NULL this mean we are at the end of the resource
 * and we should check 
 */
static void
htmlAutoClose(htmlParserCtxtPtr ctxt, const xmlChar * newtag)
{
    while ((newtag != NULL) && (ctxt->name != NULL) &&
           (htmlCheckAutoClose(newtag, ctxt->name))) {
        if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
            ctxt->sax->endElement(ctxt->userData, ctxt->name);
	htmlnamePop(ctxt);
    }
    if (newtag == NULL) {
        htmlAutoCloseOnEnd(ctxt);
        return;
    }
    while ((newtag == NULL) && (ctxt->name != NULL) &&
           ((xmlStrEqual(ctxt->name, BAD_CAST "head")) ||
            (xmlStrEqual(ctxt->name, BAD_CAST "body")) ||
            (xmlStrEqual(ctxt->name, BAD_CAST "html")))) {
        if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
            ctxt->sax->endElement(ctxt->userData, ctxt->name);
	htmlnamePop(ctxt);
    }
}

/**
 * htmlAutoCloseTag:
 * @doc:  the HTML document
 * @name:  The tag name
 * @elem:  the HTML element
 *
 * The HTML DTD allows a tag to implicitly close other tags.
 * The list is kept in htmlStartClose array. This function checks
 * if the element or one of it's children would autoclose the
 * given tag.
 *
 * Returns 1 if autoclose, 0 otherwise
 */
int
htmlAutoCloseTag(htmlDocPtr doc, const xmlChar *name, htmlNodePtr elem) {
    htmlNodePtr child;

    if (elem == NULL) return(1);
    if (xmlStrEqual(name, elem->name)) return(0);
    if (htmlCheckAutoClose(elem->name, name)) return(1);
    child = elem->children;
    while (child != NULL) {
        if (htmlAutoCloseTag(doc, name, child)) return(1);
	child = child->next;
    }
    return(0);
}

/**
 * htmlIsAutoClosed:
 * @doc:  the HTML document
 * @elem:  the HTML element
 *
 * The HTML DTD allows a tag to implicitly close other tags.
 * The list is kept in htmlStartClose array. This function checks
 * if a tag is autoclosed by one of it's child
 *
 * Returns 1 if autoclosed, 0 otherwise
 */
int
htmlIsAutoClosed(htmlDocPtr doc, htmlNodePtr elem) {
    htmlNodePtr child;

    if (elem == NULL) return(1);
    child = elem->children;
    while (child != NULL) {
	if (htmlAutoCloseTag(doc, elem->name, child)) return(1);
	child = child->next;
    }
    return(0);
}

/**
 * htmlCheckImplied:
 * @ctxt:  an HTML parser context
 * @newtag:  The new tag name
 *
 * The HTML DTD allows a tag to exists only implicitly
 * called when a new tag has been detected and generates the
 * appropriates implicit tags if missing
 */
static void
htmlCheckImplied(htmlParserCtxtPtr ctxt, const xmlChar *newtag) {
    if (!htmlOmittedDefaultValue)
	return;
    if (xmlStrEqual(newtag, BAD_CAST"html"))
	return;
    if (ctxt->nameNr <= 0) {
	htmlnamePush(ctxt, BAD_CAST"html");
	if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL))
	    ctxt->sax->startElement(ctxt->userData, BAD_CAST"html", NULL);
    }
    if ((xmlStrEqual(newtag, BAD_CAST"body")) || (xmlStrEqual(newtag, BAD_CAST"head")))
        return;
    if ((ctxt->nameNr <= 1) && 
        ((xmlStrEqual(newtag, BAD_CAST"script")) ||
	 (xmlStrEqual(newtag, BAD_CAST"style")) ||
	 (xmlStrEqual(newtag, BAD_CAST"meta")) ||
	 (xmlStrEqual(newtag, BAD_CAST"link")) ||
	 (xmlStrEqual(newtag, BAD_CAST"title")) ||
	 (xmlStrEqual(newtag, BAD_CAST"base")))) {
	    /* 
	     * dropped OBJECT ... i you put it first BODY will be
	     * assumed !
	     */
	    htmlnamePush(ctxt, BAD_CAST"head");
	    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL))
		ctxt->sax->startElement(ctxt->userData, BAD_CAST"head", NULL);
    } else if ((!xmlStrEqual(newtag, BAD_CAST"noframes")) &&
	       (!xmlStrEqual(newtag, BAD_CAST"frame")) &&
	       (!xmlStrEqual(newtag, BAD_CAST"frameset"))) {
	int i;
	for (i = 0;i < ctxt->nameNr;i++) {
	    if (xmlStrEqual(ctxt->nameTab[i], BAD_CAST"body")) {
		return;
	    }
	    if (xmlStrEqual(ctxt->nameTab[i], BAD_CAST"head")) {
		return;
	    }
	}
	    
	htmlnamePush(ctxt, BAD_CAST"body");
	if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL))
	    ctxt->sax->startElement(ctxt->userData, BAD_CAST"body", NULL);
    }
}

/**
 * htmlCheckParagraph
 * @ctxt:  an HTML parser context
 *
 * Check whether a p element need to be implied before inserting
 * characters in the current element.
 *
 * Returns 1 if a paragraph has been inserted, 0 if not and -1
 *         in case of error.
 */

static int
htmlCheckParagraph(htmlParserCtxtPtr ctxt) {
    const xmlChar *tag;
    int i;

    if (ctxt == NULL)
	return(-1);
    tag = ctxt->name;