working on DTD validation on top of xml reader interfaces. Allows to
* testReader.c xmlreader.c valid.c include/libxml/tree.h
include/libxml/valid.h include/libxml/xmlreader.h: working on
DTD validation on top of xml reader interfaces. Allows to
validate arbitrary large instances. This required some extensions
to the valid module interface and augmenting the size of xmlID
and xmlRef structs a bit.
* uri.c xmlregexp.c: simple cleanup.
Daniel
diff --git a/valid.c b/valid.c
index 421a29a..7fa634c 100644
--- a/valid.c
+++ b/valid.c
@@ -33,8 +33,90 @@
"Unimplemented block at %s:%d\n", \
__FILE__, __LINE__);
+#define VERROR \
+ if ((ctxt != NULL) && (ctxt->error != NULL)) ctxt->error
-#ifndef LIBXML_REGEXP_ENABLED
+#define VWARNING \
+ if ((ctxt != NULL) && (ctxt->warning != NULL)) ctxt->warning
+
+
+#ifdef LIBXML_REGEXP_ENABLED
+/*
+ * If regexp are enabled we can do continuous validation without the
+ * need of a tree to validate the content model. this is done in each
+ * callbacks.
+ * Each xmlValidState represent the validation state associated to the
+ * set of nodes currently open from the document root to the current element.
+ */
+
+
+typedef struct _xmlValidState {
+ xmlElementPtr elemDecl; /* pointer to the content model */
+ xmlNodePtr node; /* pointer to the current node */
+ xmlRegExecCtxtPtr exec; /* regexp runtime */
+} _xmlValidState;
+
+
+static int
+vstateVPush(xmlValidCtxtPtr ctxt, xmlElementPtr elemDecl, xmlNodePtr node) {
+ if (ctxt->vstateMax == 0) {
+ ctxt->vstateMax = 10;
+ ctxt->vstateTab = (xmlValidState *) xmlMalloc(ctxt->vstateMax *
+ sizeof(ctxt->vstateTab[0]));
+ if (ctxt->vstateTab == NULL) {
+ VERROR(ctxt->userData, "realloc failed !n");
+ return(-1);
+ }
+ }
+
+ if (ctxt->vstateNr >= ctxt->vstateMax) {
+ ctxt->vstateMax *= 2;
+ ctxt->vstateTab = (xmlValidState *) xmlRealloc(ctxt->vstateTab,
+ ctxt->vstateMax * sizeof(ctxt->vstateTab[0]));
+ if (ctxt->vstateTab == NULL) {
+ VERROR(ctxt->userData, "realloc failed !n");
+ return(-1);
+ }
+ }
+ ctxt->vstate = &ctxt->vstateTab[ctxt->vstateNr];
+ ctxt->vstateTab[ctxt->vstateNr].elemDecl = elemDecl;
+ ctxt->vstateTab[ctxt->vstateNr].node = node;
+ if ((elemDecl != NULL) && (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)) {
+ if (elemDecl->contModel == NULL)
+ xmlValidBuildContentModel(ctxt, elemDecl);
+ if (elemDecl->contModel != NULL) {
+ ctxt->vstateTab[ctxt->vstateNr].exec =
+ xmlRegNewExecCtxt(elemDecl->contModel, NULL, NULL);
+ } else {
+ ctxt->vstateTab[ctxt->vstateNr].exec = NULL;
+ VERROR(ctxt->userData,
+ "Failed to build content model regexp for %s", node->name);
+ }
+ }
+ return(ctxt->vstateNr++);
+}
+
+static int
+vstateVPop(xmlValidCtxtPtr ctxt) {
+ xmlElementPtr elemDecl;
+
+ if (ctxt->vstateNr <= 1) return(-1);
+ ctxt->vstateNr--;
+ elemDecl = ctxt->vstateTab[ctxt->vstateNr].elemDecl;
+ ctxt->vstateTab[ctxt->vstateNr].elemDecl = NULL;
+ ctxt->vstateTab[ctxt->vstateNr].node = NULL;
+ if ((elemDecl != NULL) && (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)) {
+ xmlRegFreeExecCtxt(ctxt->vstateTab[ctxt->vstateNr].exec);
+ }
+ ctxt->vstateTab[ctxt->vstateNr].exec = NULL;
+ if (ctxt->vstateNr >= 1)
+ ctxt->vstate = &ctxt->vstateTab[ctxt->vstateNr - 1];
+ else
+ ctxt->vstate = NULL;
+ return(ctxt->vstateNr);
+}
+
+#else /* not LIBXML_REGEXP_ENABLED */
/*
* If regexp are not enabled, it uses a home made algorithm less
* complex and easier to
@@ -346,12 +428,6 @@
(int) (long) node->content); \
}
-#define VERROR \
- if ((ctxt != NULL) && (ctxt->error != NULL)) ctxt->error
-
-#define VWARNING \
- if ((ctxt != NULL) && (ctxt->warning != NULL)) ctxt->warning
-
#define CHECK_DTD \
if (doc == NULL) return(0); \
else if ((doc->intSubset == NULL) && \
@@ -2079,6 +2155,8 @@
if (id == NULL) return;
if (id->value != NULL)
xmlFree((xmlChar *) id->value);
+ if (id->name != NULL)
+ xmlFree((xmlChar *) id->name);
xmlFree(id);
}
@@ -2138,7 +2216,17 @@
* fill the structure.
*/
ret->value = xmlStrdup(value);
- ret->attr = attr;
+ if ((ctxt != NULL) && (ctxt->vstateNr != 0)) {
+ /*
+ * Operating in streaming mode, attr is gonna disapear
+ */
+ ret->name = xmlStrdup(attr->name);
+ ret->attr = NULL;
+ } else {
+ ret->attr = attr;
+ ret->name = NULL;
+ }
+ ret->lineno = xmlGetLineNo(attr->parent);
if (xmlHashAddEntry(table, value, ret) < 0) {
/*
@@ -2292,6 +2380,13 @@
id = xmlHashLookup(table, ID);
if (id == NULL)
return(NULL);
+ if (id->attr == NULL) {
+ /*
+ * We are operating on a stream, return a well known reference
+ * since the attribute node doesn't exist anymore
+ */
+ return((xmlAttrPtr) doc);
+ }
return(id->attr);
}
@@ -2341,6 +2436,8 @@
if (ref == NULL) return;
if (ref->value != NULL)
xmlFree((xmlChar *)ref->value);
+ if (ref->name != NULL)
+ xmlFree((xmlChar *)ref->name);
xmlFree(ref);
}
@@ -2434,7 +2531,17 @@
* fill the structure.
*/
ret->value = xmlStrdup(value);
- ret->attr = attr;
+ if ((ctxt != NULL) && (ctxt->vstateNr != 0)) {
+ /*
+ * Operating in streaming mode, attr is gonna disapear
+ */
+ ret->name = xmlStrdup(attr->name);
+ ret->attr = NULL;
+ } else {
+ ret->name = NULL;
+ ret->attr = attr;
+ }
+ ret->lineno = xmlGetLineNo(attr->parent);
/* To add a reference :-
* References are maintained as a list of references,
@@ -4601,7 +4708,7 @@
ret = 0;
goto fail;
case XML_CDATA_SECTION_NODE:
- TODO
+ /* TODO */
ret = 0;
goto fail;
case XML_ELEMENT_NODE:
@@ -4897,6 +5004,311 @@
}
/**
+ * xmlValidateCheckMixed:
+ * @ctxt: the validation context
+ * @cont: the mixed content model
+ * @qname: the qualified name as appearing in the serialization
+ *
+ * Check if the given node is part of the content model.
+ *
+ * Returns 1 if yes, 0 if no, -1 in case of error
+ */
+static int
+xmlValidateCheckMixed(xmlValidCtxtPtr ctxt ATTRIBUTE_UNUSED,
+ xmlElementContentPtr cont, const xmlChar *qname) {
+ while (cont != NULL) {
+ if (cont->type == XML_ELEMENT_CONTENT_ELEMENT) {
+ if (xmlStrEqual(cont->name, qname))
+ return(1);
+ } else if ((cont->type == XML_ELEMENT_CONTENT_OR) &&
+ (cont->c1 != NULL) &&
+ (cont->c1->type == XML_ELEMENT_CONTENT_ELEMENT)){
+ if (xmlStrEqual(cont->c1->name, qname))
+ return(1);
+ } else if ((cont->type != XML_ELEMENT_CONTENT_OR) ||
+ (cont->c1 == NULL) ||
+ (cont->c1->type != XML_ELEMENT_CONTENT_PCDATA)){
+ /* Internal error !!! */
+ xmlGenericError(xmlGenericErrorContext,
+ "Internal: MIXED struct bad\n");
+ break;
+ }
+ cont = cont->c2;
+ }
+ return(0);
+}
+
+/**
+ * xmlValidGetElemDecl:
+ * @ctxt: the validation context
+ * @doc: a document instance
+ * @elem: an element instance
+ * @extsubset: pointer, (out) indicate if the declaration was found
+ * in the external subset.
+ *
+ * Finds a declaration associated to an element in the document.
+ *
+ * returns the pointer to the declaration or NULL if not found.
+ */
+static xmlElementPtr
+xmlValidGetElemDecl(xmlValidCtxtPtr ctxt, xmlDocPtr doc,
+ xmlNodePtr elem, int *extsubset) {
+ xmlElementPtr elemDecl = NULL;
+ const xmlChar *prefix = NULL;
+
+ if ((elem == NULL) || (elem->name == NULL)) return(NULL);
+ if (extsubset != NULL)
+ *extsubset = 0;
+
+ /*
+ * Fetch the declaration for the qualified name
+ */
+ if ((elem->ns != NULL) && (elem->ns->prefix != NULL))
+ prefix = elem->ns->prefix;
+
+ if (prefix != NULL) {
+ elemDecl = xmlGetDtdQElementDesc(doc->intSubset,
+ elem->name, prefix);
+ if ((elemDecl == NULL) && (doc->extSubset != NULL)) {
+ elemDecl = xmlGetDtdQElementDesc(doc->extSubset,
+ elem->name, prefix);
+ if ((elemDecl != NULL) && (extsubset != NULL))
+ *extsubset = 1;
+ }
+ }
+
+ /*
+ * Fetch the declaration for the non qualified name
+ * This is "non-strict" validation should be done on the
+ * full QName but in that case being flexible makes sense.
+ */
+ if (elemDecl == NULL) {
+ elemDecl = xmlGetDtdElementDesc(doc->intSubset, elem->name);
+ if ((elemDecl == NULL) && (doc->extSubset != NULL)) {
+ elemDecl = xmlGetDtdElementDesc(doc->extSubset, elem->name);
+ if ((elemDecl != NULL) && (extsubset != NULL))
+ *extsubset = 1;
+ }
+ }
+ if (elemDecl == NULL) {
+ VECTXT(ctxt, elem);
+ VERROR(ctxt->userData, "No declaration for element %s\n",
+ elem->name);
+ }
+ return(elemDecl);
+}
+
+/**
+ * xmlValidatePushElement:
+ * @ctxt: the validation context
+ * @doc: a document instance
+ * @elem: an element instance
+ * @qname: the qualified name as appearing in the serialization
+ *
+ * Push a new element start on the validation stack.
+ *
+ * returns 1 if no validation problem was found or 0 otherwise
+ */
+int
+xmlValidatePushElement(xmlValidCtxtPtr ctxt, xmlDocPtr doc,
+ xmlNodePtr elem, const xmlChar *qname) {
+ int ret = 1;
+ xmlElementPtr eDecl;
+ int extsubset = 0;
+
+ if ((ctxt->vstateNr > 0) && (ctxt->vstate != NULL)) {
+ xmlValidStatePtr state = ctxt->vstate;
+ xmlElementPtr elemDecl;
+
+ /*
+ * Check the new element agaisnt the content model of the new elem.
+ */
+ if (state->elemDecl != NULL) {
+ elemDecl = state->elemDecl;
+
+ switch(elemDecl->etype) {
+ case XML_ELEMENT_TYPE_UNDEFINED:
+ ret = 0;
+ break;
+ case XML_ELEMENT_TYPE_EMPTY:
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s was declared EMPTY this one has content\n",
+ state->node->name);
+ ret = 0;
+ break;
+ case XML_ELEMENT_TYPE_ANY:
+ /* I don't think anything is required then */
+ break;
+ case XML_ELEMENT_TYPE_MIXED:
+ /* simple case of declared as #PCDATA */
+ if ((elemDecl->content != NULL) &&
+ (elemDecl->content->type ==
+ XML_ELEMENT_CONTENT_PCDATA)) {
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s was declared #PCDATA but contains non text nodes\n",
+ state->node->name);
+ ret = 0;
+ } else {
+ ret = xmlValidateCheckMixed(ctxt, elemDecl->content,
+ qname);
+ if (ret != 1) {
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s is not declared in %s list of possible children\n",
+ qname, state->node->name);
+ }
+ }
+ break;
+ case XML_ELEMENT_TYPE_ELEMENT:
+ /*
+ * TODO:
+ * VC: Standalone Document Declaration
+ * - element types with element content, if white space
+ * occurs directly within any instance of those types.
+ */
+ if (state->exec != NULL) {
+ ret = xmlRegExecPushString(state->exec, qname, NULL);
+ if (ret < 0) {
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s content does not follow the DTD\nMisplaced %s\n",
+ state->node->name, qname);
+ ret = 0;
+ } else {
+ ret = 1;
+ }
+ }
+ break;
+ }
+ }
+ }
+ eDecl = xmlValidGetElemDecl(ctxt, doc, elem, &extsubset);
+ vstateVPush(ctxt, eDecl, elem);
+ return(ret);
+}
+
+/**
+ * xmlValidatePushCData:
+ * @ctxt: the validation context
+ * @data: some character data read
+ * @len: the lenght of the data
+ *
+ * check the CData parsed for validation in the current stack
+ *
+ * returns 1 if no validation problem was found or 0 otherwise
+ */
+int
+xmlValidatePushCData(xmlValidCtxtPtr ctxt, const xmlChar *data, int len) {
+ int ret = 1;
+
+ if (len <= 0)
+ return(ret);
+ if ((ctxt->vstateNr > 0) && (ctxt->vstate != NULL)) {
+ xmlValidStatePtr state = ctxt->vstate;
+ xmlElementPtr elemDecl;
+
+ /*
+ * Check the new element agaisnt the content model of the new elem.
+ */
+ if (state->elemDecl != NULL) {
+ elemDecl = state->elemDecl;
+
+ switch(elemDecl->etype) {
+ case XML_ELEMENT_TYPE_UNDEFINED:
+ ret = 0;
+ break;
+ case XML_ELEMENT_TYPE_EMPTY:
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s was declared EMPTY this one has content\n",
+ state->node->name);
+ ret = 0;
+ break;
+ case XML_ELEMENT_TYPE_ANY:
+ break;
+ case XML_ELEMENT_TYPE_MIXED:
+ break;
+ case XML_ELEMENT_TYPE_ELEMENT:
+ if (len > 0) {
+ int i;
+
+ for (i = 0;i < len;i++) {
+ if (!IS_BLANK(data[i])) {
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s content does not follow the DTD\nText not allowed\n",
+ state->node->name);
+ ret = 0;
+ goto done;
+ }
+ }
+ /*
+ * TODO:
+ * VC: Standalone Document Declaration
+ * element types with element content, if white space
+ * occurs directly within any instance of those types.
+ */
+ }
+ break;
+ }
+ }
+ }
+done:
+ return(ret);
+}
+
+/**
+ * xmlValidatePopElement:
+ * @ctxt: the validation context
+ * @doc: a document instance
+ * @elem: an element instance
+ * @qname: the qualified name as appearing in the serialization
+ *
+ * Pop the element end from the validation stack.
+ *
+ * returns 1 if no validation problem was found or 0 otherwise
+ */
+int
+xmlValidatePopElement(xmlValidCtxtPtr ctxt, xmlDocPtr doc ATTRIBUTE_UNUSED,
+ xmlNodePtr elem, const xmlChar *qname ATTRIBUTE_UNUSED) {
+ int ret = 1;
+
+ if ((ctxt->vstateNr > 0) && (ctxt->vstate != NULL)) {
+ xmlValidStatePtr state = ctxt->vstate;
+ xmlElementPtr elemDecl;
+
+ /*
+ * Check the new element agaisnt the content model of the new elem.
+ */
+ if (state->elemDecl != NULL) {
+ elemDecl = state->elemDecl;
+
+ if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT) {
+ if (state->exec != NULL) {
+ ret = xmlRegExecPushString(state->exec, NULL, NULL);
+ if (ret == 0) {
+ VECTXT(ctxt, state->node);
+ VERROR(ctxt->userData,
+ "Element %s content does not follow the DTD\nExpecting more child\n",
+ state->node->name);
+ } else {
+ /*
+ * previous validation errors should not generate
+ * a new one here
+ */
+ ret = 1;
+ }
+ }
+ }
+ }
+ vstateVPop(ctxt);
+ }
+ return(ret);
+}
+
+/**
* xmlValidateOneElement:
* @ctxt: the validation context
* @doc: a document instance
@@ -4923,7 +5335,6 @@
xmlNodePtr child;
int ret = 1, tmp;
const xmlChar *name;
- const xmlChar *prefix = NULL;
int extsubset = 0;
CHECK_DTD;
@@ -5002,45 +5413,19 @@
"unknown element type %d\n", elem->type);
return(0);
}
- if (elem->name == NULL) return(0);
/*
- * Fetch the declaration for the qualified name
+ * Fetch the declaration
*/
- if ((elem->ns != NULL) && (elem->ns->prefix != NULL))
- prefix = elem->ns->prefix;
-
- if (prefix != NULL) {
- elemDecl = xmlGetDtdQElementDesc(doc->intSubset,
- elem->name, prefix);
- if ((elemDecl == NULL) && (doc->extSubset != NULL)) {
- elemDecl = xmlGetDtdQElementDesc(doc->extSubset,
- elem->name, prefix);
- if (elemDecl != NULL)
- extsubset = 1;
- }
- }
-
- /*
- * Fetch the declaration for the non qualified name
- * This is "non-strict" validation should be done on the
- * full QName but in that case being flexible makes sense.
- */
- if (elemDecl == NULL) {
- elemDecl = xmlGetDtdElementDesc(doc->intSubset, elem->name);
- if ((elemDecl == NULL) && (doc->extSubset != NULL)) {
- elemDecl = xmlGetDtdElementDesc(doc->extSubset, elem->name);
- if (elemDecl != NULL)
- extsubset = 1;
- }
- }
- if (elemDecl == NULL) {
- VECTXT(ctxt, elem);
- VERROR(ctxt->userData, "No declaration for element %s\n",
- elem->name);
+ elemDecl = xmlValidGetElemDecl(ctxt, doc, elem, &extsubset);
+ if (elemDecl == NULL)
return(0);
- }
+ /*
+ * If vstateNr is not zero that means continuous validation is
+ * activated, do not try to check the content model at that level.
+ */
+ if (ctxt->vstateNr == 0) {
/* Check that the element content matches the definition */
switch (elemDecl->etype) {
case XML_ELEMENT_TYPE_UNDEFINED:
@@ -5168,6 +5553,7 @@
ret = tmp;
break;
}
+ } /* not continuous */
/* [ VC: Required Attribute ] */
attr = elemDecl->attributes;
@@ -5433,15 +5819,42 @@
if (ref == NULL)
return;
- attr = ref->attr;
- if (attr == NULL)
+ if ((ref->attr == NULL) && (ref->name == NULL))
return;
- if (attr->atype == XML_ATTRIBUTE_IDREF) {
+ attr = ref->attr;
+ if (attr == NULL) {
+ xmlChar *dup, *str = NULL, *cur, save;
+
+ dup = xmlStrdup(name);
+ if (dup == NULL) {
+ ctxt->valid = 0;
+ return;
+ }
+ cur = dup;
+ while (*cur != 0) {
+ str = cur;
+ while ((*cur != 0) && (!IS_BLANK(*cur))) cur++;
+ save = *cur;
+ *cur = 0;
+ id = xmlGetID(ctxt->doc, str);
+ if (id == NULL) {
+ VERROR(ctxt->userData,
+ "attribute %s line %d references an unknown ID \"%s\"\n",
+ ref->name, ref->lineno, str);
+ ctxt->valid = 0;
+ }
+ if (save == 0)
+ break;
+ *cur = save;
+ while (IS_BLANK(*cur)) cur++;
+ }
+ xmlFree(dup);
+ } else if (attr->atype == XML_ATTRIBUTE_IDREF) {
id = xmlGetID(ctxt->doc, name);
if (id == NULL) {
VECTXT(ctxt, attr->parent);
VERROR(ctxt->userData,
- "IDREF attribute %s references an unknown ID \"%s\"\n",
+ "IDREF attribute %s references an unknown ID \"%s\"\n",
attr->name, name);
ctxt->valid = 0;
}
@@ -5463,7 +5876,7 @@
if (id == NULL) {
VECTXT(ctxt, attr->parent);
VERROR(ctxt->userData,
- "IDREFS attribute %s references an unknown ID \"%s\"\n",
+ "IDREFS attribute %s references an unknown ID \"%s\"\n",
attr->name, str);
ctxt->valid = 0;
}