shaba/libxml2
1
0
Fork 0
mirror of https://gitlab.gnome.org/GNOME/libxml2.git synced 2025-08-30 05:49:22 +03:00
Code Activity

parser: Fix quadratic behavior when copying entities

Browse Source 
Process the first and last text node with the SAX handler to make the
text merging optimization kick in.

Fixes #657.
This commit is contained in:
Nick Wellnhofer
2024-01-07 01:29:37 +01:00
parent d2b55a7a02
commit e1153832b0
1 changed files with 67 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

85
parser.c
View File

@ -7430,12 +7430,14 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
if (xmlParserEntityCheck(ctxt, ent->expandedSize)) if (xmlParserEntityCheck(ctxt, ent->expandedSize))
return; return;
if ((ctxt->sax == NULL) || (ctxt->disableSAX))
return;
if (ctxt->replaceEntities == 0) { if (ctxt->replaceEntities == 0) {
/* /*
* Create a reference * Create a reference
*/ */
if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && if (ctxt->sax->reference != NULL)
(!ctxt->disableSAX))
ctxt->sax->reference(ctxt->userData, ent->name); ctxt->sax->reference(ctxt->userData, ent->name);
} else if ((ent->children != NULL) && (ctxt->node != NULL)) { } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
xmlNodePtr copy, cur; xmlNodePtr copy, cur;
@ -7444,7 +7446,53 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
* Seems we are generating the DOM content, copy the tree * Seems we are generating the DOM content, copy the tree
*/ */
cur = ent->children; cur = ent->children;
/*
* Handle first text node with SAX to coalesce text efficiently
*/
if ((cur->type == XML_TEXT_NODE) ||
(cur->type == XML_CDATA_SECTION_NODE)) {
int len = xmlStrlen(cur->content);
if (cur->type == XML_TEXT_NODE) {
if (ctxt->sax->characters != NULL)
ctxt->sax->characters(ctxt, cur->content, len);
} else {
if (ctxt->sax->cdataBlock != NULL)
ctxt->sax->cdataBlock(ctxt, cur->content, len);
}
cur = cur->next;
}
while (cur != NULL) { while (cur != NULL) {
xmlNodePtr last;
/*
* Handle last text node with SAX to coalesce text efficiently
*/
if ((cur->next == NULL) &&
((cur->type == XML_TEXT_NODE) ||
(cur->type == XML_CDATA_SECTION_NODE))) {
int len = xmlStrlen(cur->content);
if (cur->type == XML_TEXT_NODE) {
if (ctxt->sax->characters != NULL)
ctxt->sax->characters(ctxt, cur->content, len);
} else {
if (ctxt->sax->cdataBlock != NULL)
ctxt->sax->cdataBlock(ctxt, cur->content, len);
}
break;
}
/*
* Reset coalesce buffer stats only for non-text nodes.
*/
ctxt->nodemem = 0;
ctxt->nodelen = 0;
copy = xmlDocCopyNode(cur, ctxt->myDoc, 1); copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
if (copy == NULL) { if (copy == NULL) {
@ -7459,22 +7507,18 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
copy->_private = cur->_private; copy->_private = cur->_private;
} }
/* copy->parent = ctxt->node;
* We have to call xmlAddChild to coalesce text nodes last = ctxt->node->last;
*/ if (last == NULL) {
copy = xmlAddChild(ctxt->node, copy); ctxt->node->children = copy;
if (copy == NULL) } else {
xmlErrMemory(ctxt); last->next = copy;
copy->prev = last;
}
ctxt->node->last = copy;
cur = cur->next; cur = cur->next;
} }
/*
* This is to avoid a nasty side effect, see
* characters() in SAX.c
*/
ctxt->nodemem = 0;
ctxt->nodelen = 0;
} }
} }
@ -11925,9 +11969,6 @@ xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
if (xmlPushInput(ctxt, input) < 0) if (xmlPushInput(ctxt, input) < 0)
goto error; goto error;
ctxt->nodelen = 0;
ctxt->nodemem = 0;
nameNsPush(ctxt, rootName, NULL, NULL, 0, 0); nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
spacePush(ctxt, -1); spacePush(ctxt, -1);
@ -12003,6 +12044,7 @@ xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
int isExternal; int isExternal;
int buildTree; int buildTree;
int oldMinNsIndex; int oldMinNsIndex;
int oldNodelen, oldNodemem;
isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY); isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
buildTree = (ctxt->node != NULL); buildTree = (ctxt->node != NULL);
@ -12032,6 +12074,11 @@ xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
if (buildTree) if (buildTree)
ctxt->nsdb->minNsIndex = ctxt->nsNr; ctxt->nsdb->minNsIndex = ctxt->nsNr;
oldNodelen = ctxt->nodelen;
oldNodemem = ctxt->nodemem;
ctxt->nodelen = 0;
ctxt->nodemem = 0;
/* /*
* Parse content * Parse content
* *
@ -12056,6 +12103,8 @@ xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
ent->flags &= ~XML_ENT_EXPANDING; ent->flags &= ~XML_ENT_EXPANDING;
ctxt->nsdb->minNsIndex = oldMinNsIndex; ctxt->nsdb->minNsIndex = oldMinNsIndex;
ctxt->nodelen = oldNodelen;
ctxt->nodemem = oldNodemem;
/* /*
* Entity size accounting * Entity size accounting