1
0
mirror of https://gitlab.gnome.org/GNOME/libxml2.git synced 2025-03-21 22:50:08 +03:00

parser: Fix general entity parsing

Clear namespace database.

Ignore non-fatal errors.
This commit is contained in:
Nick Wellnhofer 2023-12-26 22:37:26 +01:00
parent ecfbcc8a52
commit f3fa34dcad
12 changed files with 69 additions and 248 deletions

116
parser.c
View File

@ -12357,21 +12357,43 @@ xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) {
* *
************************************************************************/
static xmlNodePtr
static xmlParserErrors
xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
int hasTextDecl) {
int hasTextDecl, xmlNodePtr *list) {
xmlParserInputPtr oldInput;
xmlNodePtr oldNode, root = NULL;
int oldNameNr, oldSpaceNr, oldNodeNr;
int oldWellFormed;
int oldProgressive;
int oldNodeLen, oldNodeMem;
xmlParserNsData *nsdb, *oldNsdb;
int ret;
if (list != NULL)
*list = NULL;
if (((ctxt->depth > 40) && ((ctxt->options & XML_PARSE_HUGE) == 0)) ||
(ctxt->depth > 100)) {
xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_LOOP,
"Maximum entity nesting depth exceeded");
return(NULL);
return(XML_ERR_ENTITY_LOOP);
}
root = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL);
if (root == NULL) {
xmlErrMemory(ctxt);
return(XML_ERR_NO_MEMORY);
}
/*
* We need to reset the namespace database, so that entities don't
* pick up namespaces from the parent of a reference.
*/
nsdb = xmlParserNsCreate();
if (nsdb == NULL) {
xmlErrMemory(ctxt);
xmlFreeNode(root);
return(XML_ERR_NO_MEMORY);
}
oldNode = ctxt->node;
@ -12380,17 +12402,12 @@ xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
oldProgressive = ctxt->progressive;
oldNodeLen = ctxt->nodelen;
oldNodeMem = ctxt->nodemem;
oldNsdb = ctxt->nsdb;
oldNameNr = ctxt->nameNr;
oldSpaceNr = ctxt->spaceNr;
oldNodeNr = ctxt->nodeNr;
root = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL);
if (root == NULL) {
xmlErrMemory(ctxt);
return(NULL);
}
/*
* TODO: It would be nice if we could simply push the input onto
* the input stack. But there's still some code that assumes that
@ -12400,6 +12417,7 @@ xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
ctxt->progressive = 0;
ctxt->nodelen = 0;
ctxt->nodemem = 0;
ctxt->nsdb = nsdb;
nameNsPush(ctxt, root->name, NULL, NULL, 0, 0);
spacePush(ctxt, -1);
@ -12436,10 +12454,28 @@ xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL);
}
if ((ctxt->wellFormed == 0) &&
((ctxt->recovery == 0) || (ctxt->errNo == XML_ERR_NO_MEMORY))) {
xmlFreeNode(root);
root = NULL;
if ((ctxt->wellFormed) ||
((ctxt->recovery) && (ctxt->errNo != XML_ERR_NO_MEMORY))) {
if (list != NULL) {
xmlNodePtr cur;
/*
* Return the newly created nodeset after unlinking it from
* its pseudo parent.
*/
cur = root->children;
*list = cur;
while (cur != NULL) {
cur->parent = NULL;
cur = cur->next;
}
root->children = NULL;
root->last = NULL;
}
ret = XML_ERR_OK;
} else {
ret = ctxt->errNo;
}
/*
@ -12469,8 +12505,12 @@ xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
ctxt->progressive = oldProgressive;
ctxt->nodelen = oldNodeLen;
ctxt->nodemem = oldNodeMem;
ctxt->nsdb = oldNsdb;
return(root);
xmlParserNsFree(nsdb);
xmlFreeNode(root);
return(ret);
}
/**
@ -12494,8 +12534,8 @@ int
xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL,
const xmlChar *ID, xmlNodePtr *list) {
xmlParserInputPtr input;
xmlNodePtr root;
unsigned long consumed;
int ret;
if (list != NULL)
*list = NULL;
@ -12507,24 +12547,7 @@ xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL,
if (input == NULL)
return(ctxt->errNo);
root = xmlCtxtParseContent(ctxt, input, 1 /* hasTextDecl */);
if ((list != NULL) && (root != NULL)) {
xmlNodePtr cur;
/*
* Return the newly created nodeset after unlinking it from
* their pseudo parent.
*/
cur = root->children;
*list = cur;
while (cur != NULL) {
cur->parent = NULL;
cur = cur->next;
}
root->children = NULL;
root->last = NULL;
}
ret = xmlCtxtParseContent(ctxt, input, /* hasTextDecl */ 1, list);
/*
* Also record the size of the entity parsed
@ -12535,10 +12558,9 @@ xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL,
xmlSaturatedAdd(&ctxt->sizeentities, consumed);
xmlSaturatedAdd(&ctxt->sizeentcopy, consumed);
xmlFreeNode(root);
xmlFreeInputStream(input);
return(ctxt->errNo);
return(ret);
}
#ifdef LIBXML_SAX1_ENABLED
@ -12638,8 +12660,8 @@ static xmlParserErrors
xmlCtxtParseInternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *string,
xmlNodePtr *list) {
xmlParserInputPtr input;
xmlNodePtr root;
unsigned long consumed;
int ret;
if (list != NULL)
*list = NULL;
@ -12651,24 +12673,7 @@ xmlCtxtParseInternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *string,
if (input == NULL)
return(ctxt->errNo);
root = xmlCtxtParseContent(ctxt, input, 0 /* hasTextDecl */);
if ((list != NULL) && (root != NULL)) {
xmlNodePtr cur;
/*
* Return the newly created nodeset after unlinking it from
* their pseudo parent.
*/
cur = root->children;
*list = cur;
while (cur != NULL) {
cur->parent = NULL;
cur = cur->next;
}
root->children = NULL;
root->last = NULL;
}
ret = xmlCtxtParseContent(ctxt, input, /* hasTextDecl */ 0, list);
/*
* Also record the size of the entity parsed
@ -12678,10 +12683,9 @@ xmlCtxtParseInternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *string,
xmlSaturatedAdd(&ctxt->sizeentcopy, consumed);
xmlFreeNode(root);
xmlFreeInputStream(input);
return(ctxt->errNo);
return(ret);
}
/**

View File

@ -0,0 +1,3 @@
namespace error : Namespace prefix ns on elem is not defined
<ns:elem/>
^

View File

@ -0,0 +1,3 @@
namespace error : Namespace prefix ns on elem is not defined
<ns:elem/>
^

View File

@ -0,0 +1,3 @@
namespace error : Namespace prefix ns on elem is not defined
<ns:elem/>
^

View File

@ -1,11 +0,0 @@
<?xml version="1.0"?>
<!DOCTYPE doc [
<!ENTITY ent1 "<elem/>">
<!ENTITY ent2 "<ns:elem/>">
]>
<doc>
<a xmlns="urn:a"><elem/></a>
<b xmlns="urn:b"><default:elem xmlns:default="urn:a"/></b>
<a xmlns:ns="urn:a"><ns:elem/></a>
<b xmlns:ns="urn:b"><ns:elem xmlns:ns="urn:a"/></b>
</doc>

View File

@ -1,41 +0,0 @@
SAX.setDocumentLocator()
SAX.startDocument()
SAX.internalSubset(doc, , )
SAX.entityDecl(ent1, 1, (null), (null), <elem/>)
SAX.getEntity(ent1)
SAX.entityDecl(ent2, 1, (null), (null), <ns:elem/>)
SAX.getEntity(ent2)
SAX.externalSubset(doc, , )
SAX.startElementNs(doc, NULL, NULL, 0, 0, 0)
SAX.characters(
, 5)
SAX.startElementNs(a, NULL, 'urn:a', 1, xmlns='urn:a', 0, 0)
SAX.getEntity(ent1)
SAX.startElementNs(elem, NULL, 'urn:a', 0, 0, 0)
SAX.endElementNs(elem, NULL, 'urn:a')
SAX.endElementNs(a, NULL, 'urn:a')
SAX.characters(
, 5)
SAX.startElementNs(b, NULL, 'urn:b', 1, xmlns='urn:b', 0, 0)
SAX.getEntity(ent1)
SAX.startElementNs(elem, NULL, 'urn:b', 0, 0, 0)
SAX.endElementNs(elem, NULL, 'urn:b')
SAX.endElementNs(b, NULL, 'urn:b')
SAX.characters(
, 5)
SAX.startElementNs(a, NULL, NULL, 1, xmlns:ns='urn:a', 0, 0)
SAX.getEntity(ent2)
SAX.startElementNs(elem, ns, 'urn:a', 0, 0, 0)
SAX.endElementNs(elem, ns, 'urn:a')
SAX.endElementNs(a, NULL, NULL)
SAX.characters(
, 5)
SAX.startElementNs(b, NULL, NULL, 1, xmlns:ns='urn:b', 0, 0)
SAX.getEntity(ent2)
SAX.startElementNs(elem, ns, 'urn:b', 0, 0, 0)
SAX.endElementNs(elem, ns, 'urn:b')
SAX.endElementNs(b, NULL, NULL)
SAX.characters(
, 1)
SAX.endElementNs(doc, NULL, NULL)
SAX.endDocument()

View File

@ -1,25 +0,0 @@
0 10 doc 0 0
0 1 doc 0 0
1 14 #text 0 1
1 1 a 0 0
2 1 default:elem 1 0
1 15 a 0 0
1 14 #text 0 1
1 1 b 0 0
2 1 default:elem 1 0
1 15 b 0 0
1 14 #text 0 1
1 1 a 0 0
2 1 ns:elem 1 0
1 15 a 0 0
1 14 #text 0 1
1 1 b 0 0
2 1 ns:elem 1 0
1 15 b 0 0
1 14 #text 0 1
0 15 doc 0 0

View File

@ -1,25 +0,0 @@
0 10 doc 0 0
0 1 doc 0 0
1 14 #text 0 1
1 1 a 0 0
2 5 ent1 0 0
1 15 a 0 0
1 14 #text 0 1
1 1 b 0 0
2 5 ent1 0 0
1 15 b 0 0
1 14 #text 0 1
1 1 a 0 0
2 5 ent2 0 0
1 15 a 0 0
1 14 #text 0 1
1 1 b 0 0
2 5 ent2 0 0
1 15 b 0 0
1 14 #text 0 1
0 15 doc 0 0

View File

@ -1,45 +0,0 @@
SAX.setDocumentLocator()
SAX.startDocument()
SAX.internalSubset(doc, , )
SAX.entityDecl(ent1, 1, (null), (null), <elem/>)
SAX.getEntity(ent1)
SAX.entityDecl(ent2, 1, (null), (null), <ns:elem/>)
SAX.getEntity(ent2)
SAX.externalSubset(doc, , )
SAX.startElement(doc)
SAX.characters(
, 5)
SAX.startElement(a, xmlns='urn:a')
SAX.getEntity(ent1)
SAX.startElement(elem)
SAX.endElement(elem)
SAX.reference(ent1)
SAX.endElement(a)
SAX.characters(
, 5)
SAX.startElement(b, xmlns='urn:b')
SAX.getEntity(ent1)
SAX.startElement(elem)
SAX.endElement(elem)
SAX.reference(ent1)
SAX.endElement(b)
SAX.characters(
, 5)
SAX.startElement(a, xmlns:ns='urn:a')
SAX.getEntity(ent2)
SAX.startElement(ns:elem)
SAX.endElement(ns:elem)
SAX.reference(ent2)
SAX.endElement(a)
SAX.characters(
, 5)
SAX.startElement(b, xmlns:ns='urn:b')
SAX.getEntity(ent2)
SAX.startElement(ns:elem)
SAX.endElement(ns:elem)
SAX.reference(ent2)
SAX.endElement(b)
SAX.characters(
, 1)
SAX.endElement(doc)
SAX.endDocument()

View File

@ -1,45 +0,0 @@
SAX.setDocumentLocator()
SAX.startDocument()
SAX.internalSubset(doc, , )
SAX.entityDecl(ent1, 1, (null), (null), <elem/>)
SAX.getEntity(ent1)
SAX.entityDecl(ent2, 1, (null), (null), <ns:elem/>)
SAX.getEntity(ent2)
SAX.externalSubset(doc, , )
SAX.startElementNs(doc, NULL, NULL, 0, 0, 0)
SAX.characters(
, 5)
SAX.startElementNs(a, NULL, 'urn:a', 1, xmlns='urn:a', 0, 0)
SAX.getEntity(ent1)
SAX.startElementNs(elem, NULL, 'urn:a', 0, 0, 0)
SAX.endElementNs(elem, NULL, 'urn:a')
SAX.reference(ent1)
SAX.endElementNs(a, NULL, 'urn:a')
SAX.characters(
, 5)
SAX.startElementNs(b, NULL, 'urn:b', 1, xmlns='urn:b', 0, 0)
SAX.getEntity(ent1)
SAX.startElementNs(elem, NULL, 'urn:b', 0, 0, 0)
SAX.endElementNs(elem, NULL, 'urn:b')
SAX.reference(ent1)
SAX.endElementNs(b, NULL, 'urn:b')
SAX.characters(
, 5)
SAX.startElementNs(a, NULL, NULL, 1, xmlns:ns='urn:a', 0, 0)
SAX.getEntity(ent2)
SAX.startElementNs(elem, ns, 'urn:a', 0, 0, 0)
SAX.endElementNs(elem, ns, 'urn:a')
SAX.reference(ent2)
SAX.endElementNs(a, NULL, NULL)
SAX.characters(
, 5)
SAX.startElementNs(b, NULL, NULL, 1, xmlns:ns='urn:b', 0, 0)
SAX.getEntity(ent2)
SAX.startElementNs(elem, ns, 'urn:b', 0, 0, 0)
SAX.endElementNs(elem, ns, 'urn:b')
SAX.reference(ent2)
SAX.endElementNs(b, NULL, NULL)
SAX.characters(
, 1)
SAX.endElementNs(doc, NULL, NULL)
SAX.endDocument()