lib/libxml2
1
0
Fork 0
mirror of https://gitlab.gnome.org/GNOME/libxml2.git synced 2025-07-29 11:41:22 +03:00
Code Activity

html: Simplify node info accounting

Browse Source
This commit is contained in:
Nick Wellnhofer
2024-09-07 03:45:09 +02:00
parent ccb61f599e
commit 0d324bde36
1 changed files with 36 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files

160
HTMLparser.c
View File

@ -1417,6 +1417,21 @@ static const elementPriority htmlEndPriority[] = {
* * * *
************************************************************************/ ************************************************************************/
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);
}
}
/** /**
* htmlInitAutoClose: * htmlInitAutoClose:
* *
@ -1546,6 +1561,7 @@ htmlAutoCloseOnClose(htmlParserCtxtPtr ctxt, const xmlChar * newtag)
"Opening and ending tag mismatch: %s and %s\n", "Opening and ending tag mismatch: %s and %s\n",
newtag, ctxt->name); newtag, ctxt->name);
} }
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, ctxt->name); ctxt->sax->endElement(ctxt->userData, ctxt->name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -1566,6 +1582,7 @@ htmlAutoCloseOnEnd(htmlParserCtxtPtr ctxt)
if (ctxt->nameNr == 0) if (ctxt->nameNr == 0)
return; return;
for (i = (ctxt->nameNr - 1); i >= 0; i--) { for (i = (ctxt->nameNr - 1); i >= 0; i--) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, ctxt->name); ctxt->sax->endElement(ctxt->userData, ctxt->name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -1592,6 +1609,7 @@ htmlAutoClose(htmlParserCtxtPtr ctxt, const xmlChar * newtag)
while ((ctxt->name != NULL) && while ((ctxt->name != NULL) &&
(htmlCheckAutoClose(newtag, ctxt->name))) { (htmlCheckAutoClose(newtag, ctxt->name))) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, ctxt->name); ctxt->sax->endElement(ctxt->userData, ctxt->name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -4171,9 +4189,9 @@ htmlParseEndTag(htmlParserCtxtPtr ctxt)
*/ */
oldname = ctxt->name; oldname = ctxt->name;
if ((oldname != NULL) && (xmlStrEqual(oldname, name))) { if ((oldname != NULL) && (xmlStrEqual(oldname, name))) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlNodeInfoPop(ctxt);
htmlnamePop(ctxt); htmlnamePop(ctxt);
ret = 1; ret = 1;
} else { } else {
@ -4251,11 +4269,6 @@ htmlParseReference(htmlParserCtxtPtr ctxt) {
static void static void
htmlParseContent(htmlParserCtxtPtr ctxt) { htmlParseContent(htmlParserCtxtPtr ctxt) {
xmlChar *currentNode;
int depth;
currentNode = xmlStrdup(ctxt->name);
depth = ctxt->nameNr;
while (!PARSER_STOPPED(ctxt)) { while (!PARSER_STOPPED(ctxt)) {
int mode; int mode;
@ -4279,26 +4292,10 @@ htmlParseContent(htmlParserCtxtPtr ctxt) {
* Our tag or one of it's parent or children is ending. * Our tag or one of it's parent or children is ending.
*/ */
if ((CUR == '<') && (NXT(1) == '/')) { if ((CUR == '<') && (NXT(1) == '/')) {
if (htmlParseEndTag(ctxt) && htmlParseEndTag(ctxt);
((currentNode != NULL) || (ctxt->nameNr == 0))) {
if (currentNode != NULL)
xmlFree(currentNode);
return;
}
continue; /* while */ continue; /* while */
} }
/*
* Has this node been popped out during parsing of
* the next element
*/
if ((ctxt->nameNr > 0) && (depth >= ctxt->nameNr) &&
(!xmlStrEqual(currentNode, ctxt->name)))
{
if (currentNode != NULL) xmlFree(currentNode);
return;
}
if ((CUR == '<') && (NXT(1) == '!')) { if ((CUR == '<') && (NXT(1) == '!')) {
/* /*
* Sometimes DOCTYPE arrives in the middle of the document * Sometimes DOCTYPE arrives in the middle of the document
@ -4370,7 +4367,6 @@ done:
SHRINK; SHRINK;
GROW; GROW;
} }
if (currentNode != NULL) xmlFree(currentNode);
} }
/** /**
@ -4426,11 +4422,15 @@ htmlParseElement(htmlParserCtxtPtr ctxt) {
ctxt->endCheckState = info->dataMode; ctxt->endCheckState = info->dataMode;
} }
if (ctxt->record_info)
htmlNodeInfoPush(ctxt, &node_info);
/* /*
* Check for an Empty Element labeled the XML/SGML way * Check for an Empty Element labeled the XML/SGML way
*/ */
if ((CUR == '/') && (NXT(1) == '>')) { if ((CUR == '/') && (NXT(1) == '>')) {
SKIP(2); SKIP(2);
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -4447,20 +4447,11 @@ htmlParseElement(htmlParserCtxtPtr ctxt) {
* end of parsing of this node. * end of parsing of this node.
*/ */
if (xmlStrEqual(name, ctxt->name)) { if (xmlStrEqual(name, ctxt->name)) {
htmlParserFinishElementParsing(ctxt);
nodePop(ctxt); nodePop(ctxt);
htmlnamePop(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);
}
return; return;
} }
@ -4468,6 +4459,7 @@ htmlParseElement(htmlParserCtxtPtr ctxt) {
* Check for an Empty Element from DTD definition * Check for an Empty Element from DTD definition
*/ */
if ((info != NULL) && (info->empty)) { if ((info != NULL) && (info->empty)) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -4504,24 +4496,6 @@ htmlParseElement(htmlParserCtxtPtr ctxt) {
xmlFree(currentNode); xmlFree(currentNode);
} }
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 (CUR == 0) {
htmlAutoCloseOnEnd(ctxt);
}
}
/** /**
* htmlParseElementInternal: * htmlParseElementInternal:
* @ctxt: an HTML parser context * @ctxt: an HTML parser context
@ -4569,11 +4543,15 @@ htmlParseElementInternal(htmlParserCtxtPtr ctxt) {
ctxt->endCheckState = info->dataMode; ctxt->endCheckState = info->dataMode;
} }
if (ctxt->record_info)
htmlNodeInfoPush(ctxt, &node_info);
/* /*
* Check for an Empty Element labeled the XML/SGML way * Check for an Empty Element labeled the XML/SGML way
*/ */
if ((CUR == '/') && (NXT(1) == '>')) { if ((CUR == '/') && (NXT(1) == '>')) {
SKIP(2); SKIP(2);
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -4590,13 +4568,10 @@ htmlParseElementInternal(htmlParserCtxtPtr ctxt) {
* end of parsing of this node. * end of parsing of this node.
*/ */
if (xmlStrEqual(name, ctxt->name)) { if (xmlStrEqual(name, ctxt->name)) {
htmlParserFinishElementParsing(ctxt);
nodePop(ctxt); nodePop(ctxt);
htmlnamePop(ctxt); htmlnamePop(ctxt);
} }
if (ctxt->record_info)
htmlNodeInfoPush(ctxt, &node_info);
htmlParserFinishElementParsing(ctxt);
return; return;
} }
@ -4604,14 +4579,12 @@ htmlParseElementInternal(htmlParserCtxtPtr ctxt) {
* Check for an Empty Element from DTD definition * Check for an Empty Element from DTD definition
*/ */
if ((info != NULL) && (info->empty)) { if ((info != NULL) && (info->empty)) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
return; return;
} }
if (ctxt->record_info)
htmlNodeInfoPush(ctxt, &node_info);
} }
/** /**
@ -4624,19 +4597,6 @@ htmlParseElementInternal(htmlParserCtxtPtr ctxt) {
static void static void
htmlParseContentInternal(htmlParserCtxtPtr ctxt) { htmlParseContentInternal(htmlParserCtxtPtr ctxt) {
xmlChar *currentNode;
int depth;
depth = ctxt->nameNr;
if (depth <= 0) {
currentNode = NULL;
} else {
currentNode = xmlStrdup(ctxt->name);
if (currentNode == NULL) {
htmlErrMemory(ctxt);
return;
}
}
while (PARSER_STOPPED(ctxt) == 0) { while (PARSER_STOPPED(ctxt) == 0) {
int mode; int mode;
@ -4660,48 +4620,10 @@ htmlParseContentInternal(htmlParserCtxtPtr ctxt) {
* Our tag or one of it's parent or children is ending. * Our tag or one of it's parent or children is ending.
*/ */
if ((CUR == '<') && (NXT(1) == '/')) { if ((CUR == '<') && (NXT(1) == '/')) {
if (htmlParseEndTag(ctxt) && htmlParseEndTag(ctxt);
((currentNode != NULL) || (ctxt->nameNr == 0))) {
if (currentNode != NULL)
xmlFree(currentNode);
depth = ctxt->nameNr;
if (depth <= 0) {
currentNode = NULL;
} else {
currentNode = xmlStrdup(ctxt->name);
if (currentNode == NULL) {
htmlErrMemory(ctxt);
break;
}
}
}
continue; /* while */ continue; /* while */
} }
/*
* Has this node been popped out during parsing of
* the next element
*/
if ((ctxt->nameNr > 0) && (depth >= ctxt->nameNr) &&
(!xmlStrEqual(currentNode, ctxt->name)))
{
htmlParserFinishElementParsing(ctxt);
if (currentNode != NULL) xmlFree(currentNode);
if (ctxt->name == NULL) {
currentNode = NULL;
} else {
currentNode = xmlStrdup(ctxt->name);
if (currentNode == NULL) {
htmlErrMemory(ctxt);
break;
}
}
depth = ctxt->nameNr;
continue;
}
if ((CUR == '<') && (NXT(1) == '!')) { if ((CUR == '<') && (NXT(1) == '!')) {
/* /*
* Sometimes DOCTYPE arrives in the middle of the document * Sometimes DOCTYPE arrives in the middle of the document
@ -4738,18 +4660,6 @@ htmlParseContentInternal(htmlParserCtxtPtr ctxt) {
*/ */
else if ((CUR == '<') && IS_ASCII_LETTER(NXT(1))) { else if ((CUR == '<') && IS_ASCII_LETTER(NXT(1))) {
htmlParseElementInternal(ctxt); htmlParseElementInternal(ctxt);
if (currentNode != NULL) xmlFree(currentNode);
if (ctxt->name == NULL) {
currentNode = NULL;
} else {
currentNode = xmlStrdup(ctxt->name);
if (currentNode == NULL) {
htmlErrMemory(ctxt);
break;
}
}
depth = ctxt->nameNr;
} }
else if (CUR == '<') { else if (CUR == '<') {
if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
@ -4785,7 +4695,6 @@ done:
SHRINK; SHRINK;
GROW; GROW;
} }
if (currentNode != NULL) xmlFree(currentNode);
} }
xmlNodePtr xmlNodePtr
@ -5686,6 +5595,7 @@ htmlParseTryOrFinish(htmlParserCtxtPtr ctxt, int terminate) {
*/ */
if ((CUR == '/') && (NXT(1) == '>')) { if ((CUR == '/') && (NXT(1) == '>')) {
SKIP(2); SKIP(2);
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);
@ -5704,6 +5614,7 @@ htmlParseTryOrFinish(htmlParserCtxtPtr ctxt, int terminate) {
* end of parsing of this node. * end of parsing of this node.
*/ */
if (xmlStrEqual(name, ctxt->name)) { if (xmlStrEqual(name, ctxt->name)) {
htmlParserFinishElementParsing(ctxt);
nodePop(ctxt); nodePop(ctxt);
htmlnamePop(ctxt); htmlnamePop(ctxt);
} }
@ -5719,6 +5630,7 @@ htmlParseTryOrFinish(htmlParserCtxtPtr ctxt, int terminate) {
* Check for an Empty Element from DTD definition * Check for an Empty Element from DTD definition
*/ */
if ((info != NULL) && (info->empty)) { if ((info != NULL) && (info->empty)) {
htmlParserFinishElementParsing(ctxt);
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL)) if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, name); ctxt->sax->endElement(ctxt->userData, name);
htmlnamePop(ctxt); htmlnamePop(ctxt);