commit 228eda2f77bb9a7cd86e9fc6a5120865980b0875 Author: Eugene Pimenov Date: Fri Feb 12 07:58:12 2010 +0300 Make html parser non-recursive. * HTMLParser.c: make htmlParseElement non-recursive. * include/libxml/parser.h: add the nodeInfo stack * parserInternals.c: free nodeInfo stack in xmlFreeParserCtxt diff --git a/HTMLparser.c b/HTMLparser.c index 3d4831c..b1b17b9 100644 --- a/HTMLparser.c +++ b/HTMLparser.c @@ -209,6 +209,59 @@ htmlnamePop(htmlParserCtxtPtr ctxt) return (ret); } +/** + * htmlNodeInfoPush: + * @ctxt: an HTML parser context + * @value: the node info + * + * Pushes a new element name on top of the node info stack + * + * Returns 0 in case of error, the index in the stack otherwise + */ +static int +htmlNodeInfoPush(htmlParserCtxtPtr ctxt, htmlParserNodeInfo *value) +{ + if (ctxt->nodeInfoNr >= ctxt->nodeInfoMax) { + if (ctxt->nodeInfoMax == 0) + ctxt->nodeInfoMax = 5; + ctxt->nodeInfoMax *= 2; + ctxt->nodeInfoTab = (htmlParserNodeInfo *) + xmlRealloc((htmlParserNodeInfo *)ctxt->nodeInfoTab, + ctxt->nodeInfoMax * + sizeof(ctxt->nodeInfoTab[0])); + if (ctxt->nodeInfoTab == NULL) { + htmlErrMemory(ctxt, NULL); + return (0); + } + } + ctxt->nodeInfoTab[ctxt->nodeInfoNr] = *value; + ctxt->nodeInfo = &ctxt->nodeInfoTab[ctxt->nodeInfoNr]; + return (ctxt->nodeInfoNr++); +} + +/** + * htmlNodeInfoPop: + * @ctxt: an HTML parser context + * + * Pops the top element name from the node info stack + * + * Returns 0 in case of error, the pointer to NodeInfo otherwise + */ +static htmlParserNodeInfo * +htmlNodeInfoPop(htmlParserCtxtPtr ctxt) +{ + if (ctxt->nodeInfoNr <= 0) + return (NULL); + ctxt->nodeInfoNr--; + if (ctxt->nodeInfoNr < 0) + return (NULL); + if (ctxt->nodeInfoNr > 0) + ctxt->nodeInfo = &ctxt->nodeInfoTab[ctxt->nodeInfoNr - 1]; + else + ctxt->nodeInfo = NULL; + return &ctxt->nodeInfoTab[ctxt->nodeInfoNr]; +} + /* * Macros for accessing the content. Those should be used only by the parser, * and not exported. @@ -3922,6 +3975,24 @@ htmlParseReference(htmlParserCtxtPtr ctxt) { } } + +static void htmlParserFinishElementParsing(htmlParserCtxtPtr ctxt) { + /* + * Capture end position and add node + */ + if ( ctxt->node != NULL && ctxt->record_info ) { + ctxt->nodeInfo->end_pos = ctxt->input->consumed + + (CUR_PTR - ctxt->input->base); + ctxt->nodeInfo->end_line = ctxt->input->line; + ctxt->nodeInfo->node = ctxt->node; + xmlParserAddNodeInfo(ctxt, ctxt->nodeInfo); + htmlNodeInfoPop(ctxt); + } + if (!IS_CHAR_CH(CUR)) { + htmlAutoCloseOnEnd(ctxt); + } +} + /** * htmlParseContent: * @ctxt: an HTML parser context @@ -3953,7 +4024,9 @@ htmlParseContent(htmlParserCtxtPtr ctxt) { ((currentNode != NULL) || (ctxt->nameNr == 0))) { if (currentNode != NULL) xmlFree(currentNode); - return; + + currentNode = xmlStrdup(ctxt->name); + depth = ctxt->nameNr; } continue; /* while */ } @@ -3967,12 +4040,16 @@ htmlParseContent(htmlParserCtxtPtr ctxt) { "htmlParseStartTag: invalid element name\n", NULL, NULL); /* Dump the bogus tag like browsers do */ - while ((IS_CHAR_CH(CUR)) && (CUR != '>')) + while ((IS_CHAR_CH(CUR)) && (CUR != '>')) NEXT; + htmlParserFinishElementParsing(ctxt); if (currentNode != NULL) xmlFree(currentNode); - return; + + currentNode = xmlStrdup(ctxt->name); + depth = ctxt->nameNr; + continue; } if (ctxt->name != NULL) { @@ -3990,8 +4067,12 @@ htmlParseContent(htmlParserCtxtPtr ctxt) { if ((ctxt->nameNr > 0) && (depth >= ctxt->nameNr) && (!xmlStrEqual(currentNode, ctxt->name))) { + htmlParserFinishElementParsing(ctxt); if (currentNode != NULL) xmlFree(currentNode); - return; + + currentNode = xmlStrdup(ctxt->name); + depth = ctxt->nameNr; + continue; } if ((CUR != 0) && ((xmlStrEqual(currentNode, BAD_CAST"script")) || @@ -4035,6 +4116,10 @@ htmlParseContent(htmlParserCtxtPtr ctxt) { */ else if (CUR == '<') { htmlParseElement(ctxt); + if (currentNode != NULL) xmlFree(currentNode); + + currentNode = xmlStrdup(ctxt->name); + depth = ctxt->nameNr; } /* @@ -4087,11 +4172,12 @@ __htmlParseContent(void *ctxt) { htmlParseContent((htmlParserCtxtPtr) ctxt); } + /** * htmlParseElement: * @ctxt: an HTML parser context * - * parse an HTML element, this is highly recursive + * parse an HTML element * * [39] element ::= EmptyElemTag | STag content ETag * @@ -4101,7 +4187,6 @@ __htmlParseContent(void *ctxt) { void htmlParseElement(htmlParserCtxtPtr ctxt) { const xmlChar *name; - xmlChar *currentNode = NULL; const htmlElemDesc * info; htmlParserNodeInfo node_info; int failed; @@ -4166,16 +4251,9 @@ htmlParseElement(htmlParserCtxtPtr ctxt) { htmlnamePop(ctxt); } - /* - * Capture end position and add node - */ - if (ctxt->record_info) { - node_info.end_pos = ctxt->input->consumed + - (CUR_PTR - ctxt->input->base); - node_info.end_line = ctxt->input->line; - node_info.node = ctxt->node; - xmlParserAddNodeInfo(ctxt, &node_info); - } + if (ctxt->record_info) + htmlNodeInfoPush(ctxt, &node_info); + htmlParserFinishElementParsing(ctxt); return; } @@ -4189,34 +4267,8 @@ htmlParseElement(htmlParserCtxtPtr ctxt) { return; } - /* - * Parse the content of the element: - */ - currentNode = xmlStrdup(ctxt->name); - depth = ctxt->nameNr; - while (IS_CHAR_CH(CUR)) { - oldptr = ctxt->input->cur; - htmlParseContent(ctxt); - if (oldptr==ctxt->input->cur) break; - if (ctxt->nameNr < depth) break; - } - - /* - * Capture end position and add node - */ - if ( currentNode != NULL && ctxt->record_info ) { - node_info.end_pos = ctxt->input->consumed + - (CUR_PTR - ctxt->input->base); - node_info.end_line = ctxt->input->line; - node_info.node = ctxt->node; - xmlParserAddNodeInfo(ctxt, &node_info); - } - if (!IS_CHAR_CH(CUR)) { - htmlAutoCloseOnEnd(ctxt); - } - - if (currentNode != NULL) - xmlFree(currentNode); + if (ctxt->record_info) + htmlNodeInfoPush(ctxt, &node_info); } /** @@ -4440,6 +4492,10 @@ htmlInitParserCtxt(htmlParserCtxtPtr ctxt) ctxt->nameMax = 10; ctxt->name = NULL; + ctxt->nodeInfoTab = NULL; + ctxt->nodeInfoNr = 0; + ctxt->nodeInfoMax = 0; + if (sax == NULL) ctxt->sax = (xmlSAXHandlerPtr) &htmlDefaultSAXHandler; else { ctxt->sax = sax; diff --git a/include/libxml/parser.h b/include/libxml/parser.h index 148ee03..b9ccaca 100644 --- a/include/libxml/parser.h +++ b/include/libxml/parser.h @@ -209,6 +209,10 @@ struct _xmlParserCtxt { int record_info; /* Whether node info should be kept */ xmlParserNodeInfoSeq node_seq; /* info about each node parsed */ + xmlParserNodeInfo *nodeInfo; /* Current NodeInfo */ + int nodeInfoNr; /* Depth of the parsing stack */ + int nodeInfoMax; /* Max depth of the parsing stack */ + xmlParserNodeInfo *nodeInfoTab; /* array of nodeInfos */ int errNo; /* error code */ diff --git a/parserInternals.c b/parserInternals.c index ff20435..2404ddf 100644 --- a/parserInternals.c +++ b/parserInternals.c @@ -1782,6 +1782,7 @@ xmlFreeParserCtxt(xmlParserCtxtPtr ctxt) if (ctxt->spaceTab != NULL) xmlFree(ctxt->spaceTab); if (ctxt->nameTab != NULL) xmlFree((xmlChar * *)ctxt->nameTab); if (ctxt->nodeTab != NULL) xmlFree(ctxt->nodeTab); + if (ctxt->nodeInfoTab != NULL) xmlFree(ctxt->nodeInfoTab); if (ctxt->inputTab != NULL) xmlFree(ctxt->inputTab); if (ctxt->version != NULL) xmlFree((char *) ctxt->version); if (ctxt->encoding != NULL) xmlFree((char *) ctxt->encoding);