small cleanup switched DTD validation to use only regexp when configured
* HTMLparser.c: small cleanup
* valid.c xmlregexp.c: switched DTD validation to use only regexp
when configured with them. A bit of debugging around the determinism
checks is still needed
Daniel
diff --git a/valid.c b/valid.c
index adab5af..03b97c4 100644
--- a/valid.c
+++ b/valid.c
@@ -27,6 +27,10 @@
/* #define DEBUG_VALID_ALGO */
+#define TODO \
+ xmlGenericError(xmlGenericErrorContext, \
+ "Unimplemented block at %s:%d\n", \
+ __FILE__, __LINE__);
/*
* Generic function for accessing stacks in the Validity Context
*/
@@ -385,27 +389,49 @@
break;
case XML_ELEMENT_CONTENT_ELEMENT: {
xmlAutomataStatePtr oldstate = ctxt->state;
+ xmlChar *QName = NULL;
+ const xmlChar *fname = content->name;
+
+ if (content->prefix != NULL) {
+ int len;
+
+ len = xmlStrlen(content->name) +
+ xmlStrlen(content->prefix) + 2;
+ QName = xmlMalloc(len);
+ if (QName == NULL) {
+ VERROR(ctxt->userData,
+ "ContentModel %s : alloc failed\n", name);
+ return(0);
+ }
+ snprintf((char *) QName, len, "%s:%s",
+ (char *)content->prefix,
+ (char *)content->name);
+ fname = QName;
+ }
+
switch (content->ocur) {
case XML_ELEMENT_CONTENT_ONCE:
ctxt->state = xmlAutomataNewTransition(ctxt->am,
- ctxt->state, NULL, content->name, NULL);
+ ctxt->state, NULL, fname, NULL);
break;
case XML_ELEMENT_CONTENT_OPT:
ctxt->state = xmlAutomataNewTransition(ctxt->am,
- ctxt->state, NULL, content->name, NULL);
+ ctxt->state, NULL, fname, NULL);
xmlAutomataNewEpsilon(ctxt->am, oldstate, ctxt->state);
break;
case XML_ELEMENT_CONTENT_PLUS:
ctxt->state = xmlAutomataNewTransition(ctxt->am,
- ctxt->state, NULL, content->name, NULL);
+ ctxt->state, NULL, fname, NULL);
xmlAutomataNewTransition(ctxt->am, ctxt->state,
- ctxt->state, content->name, NULL);
+ ctxt->state, fname, NULL);
break;
case XML_ELEMENT_CONTENT_MULT:
xmlAutomataNewTransition(ctxt->am, ctxt->state,
- ctxt->state, content->name, NULL);
+ ctxt->state, fname, NULL);
break;
}
+ if (QName != NULL)
+ xmlFree(QName);
break;
}
case XML_ELEMENT_CONTENT_SEQ: {
@@ -417,11 +443,12 @@
*/
oldstate = ctxt->state;
ocur = content->ocur;
- while (content->type == XML_ELEMENT_CONTENT_SEQ) {
+ do {
xmlValidBuildAContentModel(content->c1, ctxt, name);
content = content->c2;
- }
- xmlValidBuildAContentModel(content->c2, ctxt, name);
+ } while ((content->type == XML_ELEMENT_CONTENT_SEQ) &&
+ (content->ocur == XML_ELEMENT_CONTENT_ONCE));
+ xmlValidBuildAContentModel(content, ctxt, name);
switch (ocur) {
case XML_ELEMENT_CONTENT_ONCE:
break;
@@ -450,14 +477,15 @@
* iterate over the subtypes and remerge the end with an
* epsilon transition
*/
- while (content->type == XML_ELEMENT_CONTENT_OR) {
+ do {
ctxt->state = start;
xmlValidBuildAContentModel(content->c1, ctxt, name);
xmlAutomataNewEpsilon(ctxt->am, ctxt->state, end);
content = content->c2;
- }
+ } while ((content->type == XML_ELEMENT_CONTENT_OR) &&
+ (content->ocur == XML_ELEMENT_CONTENT_ONCE));
ctxt->state = start;
- xmlValidBuildAContentModel(content->c1, ctxt, name);
+ xmlValidBuildAContentModel(content, ctxt, name);
xmlAutomataNewEpsilon(ctxt->am, ctxt->state, end);
ctxt->state = end;
switch (ocur) {
@@ -4280,6 +4308,82 @@
cont = elemDecl->content;
name = elemDecl->name;
+#ifdef LIBXML_REGEXP_ENABLED
+ /* Build the regexp associated to the content model */
+ if (elemDecl->contModel == NULL)
+ ret = xmlValidBuildContentModel(ctxt, elemDecl);
+ if (elemDecl->contModel == NULL) {
+ ret = -1;
+ } else {
+ xmlRegExecCtxtPtr exec;
+
+ exec = xmlRegNewExecCtxt(elemDecl->contModel, NULL, NULL);
+ if (exec != NULL) {
+ cur = child;
+ while (cur != NULL) {
+ switch (cur->type) {
+ case XML_ENTITY_REF_NODE:
+ /*
+ * Push the current node to be able to roll back
+ * and process within the entity
+ */
+ if ((cur->children != NULL) &&
+ (cur->children->children != NULL)) {
+ nodeVPush(ctxt, cur);
+ cur = cur->children->children;
+ continue;
+ }
+ break;
+ case XML_TEXT_NODE:
+ if (xmlIsBlankNode(cur))
+ break;
+ ret = 0;
+ goto fail;
+ case XML_CDATA_SECTION_NODE:
+ TODO
+ ret = 0;
+ goto fail;
+ case XML_ELEMENT_NODE:
+ if ((cur->ns != NULL) && (cur->ns->prefix != NULL)) {
+ xmlChar *QName;
+ int len;
+
+ len = xmlStrlen(cur->name) +
+ xmlStrlen(cur->ns->prefix) + 2;
+ QName = xmlMalloc(len);
+ if (QName == NULL) {
+ ret = -1;
+ goto fail;
+ }
+ snprintf((char *) QName, len, "%s:%s",
+ (char *)cur->ns->prefix,
+ (char *)cur->name);
+ ret = xmlRegExecPushString(exec, QName, NULL);
+ xmlFree(QName);
+ } else {
+ ret = xmlRegExecPushString(exec, cur->name, NULL);
+ }
+ break;
+ default:
+ break;
+ }
+ /*
+ * Switch to next element
+ */
+ cur = cur->next;
+ while (cur == NULL) {
+ cur = nodeVPop(ctxt);
+ if (cur == NULL)
+ break;
+ cur = cur->next;
+ }
+ }
+ ret = xmlRegExecPushString(exec, NULL, NULL);
+fail:
+ xmlRegFreeExecCtxt(exec);
+ }
+ }
+#else /* LIBXML_REGEXP_ENABLED */
/*
* Allocate the stack
*/
@@ -4395,6 +4499,7 @@
STATE = 0;
ret = xmlValidateElementType(ctxt);
}
+#endif /* LIBXML_REGEXP_ENABLED */
if ((warn) && ((ret != 1) && (ret != -3))) {
if ((ctxt != NULL) && (ctxt->warning != NULL)) {
char expr[5000];
@@ -4436,7 +4541,6 @@
if (ret == -3)
ret = 1;
-
done:
/*
* Deallocate the copy if done, and free up the validation stack