bool XSAXMLScanner::scanStartTag(bool& gotData)
{
// Assume we will still have data until proven otherwise. It will only
// ever be false if this is the root and its empty.
gotData = true;
// Reset element content
fContent.reset();
// The current position is after the open bracket, so we need to read in
// in the element name.
int prefixColonPos;
if (!fReaderMgr.getQName(fQNameBuf, &prefixColonPos))
{
if (fQNameBuf.isEmpty())
emitError(XMLErrs::ExpectedElementName);
else
emitError(XMLErrs::InvalidElementName, fQNameBuf.getRawBuffer());
fReaderMgr.skipToChar(chOpenAngle);
return false;
}
// See if its the root element
const bool isRoot = fElemStack.isEmpty();
// Skip any whitespace after the name
fReaderMgr.skipPastSpaces();
// First we have to do the rawest attribute scan. We don't do any
// normalization of them at all, since we don't know yet what type they
// might be (since we need the element decl in order to do that.)
const XMLCh* qnameRawBuf = fQNameBuf.getRawBuffer();
bool isEmpty;
unsigned int attCount = rawAttrScan(qnameRawBuf, *fRawAttrList, isEmpty);
// save the contentleafname and currentscope before addlevel, for later use
ContentLeafNameTypeVector* cv = 0;
XMLContentModel* cm = 0;
int currentScope = Grammar::TOP_LEVEL_SCOPE;
bool laxThisOne = false;
if (!isRoot)
{
// schema validator will have correct type if validating
SchemaElementDecl* tempElement = (SchemaElementDecl*)
fElemStack.topElement()->fThisElement;
SchemaElementDecl::ModelTypes modelType = tempElement->getModelType();
ComplexTypeInfo *currType = 0;
if (fValidate)
{
currType = ((SchemaValidator*)fValidator)->getCurrentTypeInfo();
if (currType)
modelType = (SchemaElementDecl::ModelTypes)currType->getContentType();
else // something must have gone wrong
modelType = SchemaElementDecl::Any;
}
else {
currType = tempElement->getComplexTypeInfo();
}
if ((modelType == SchemaElementDecl::Mixed_Simple)
|| (modelType == SchemaElementDecl::Mixed_Complex)
|| (modelType == SchemaElementDecl::Children))
{
cm = currType->getContentModel();
cv = cm->getContentLeafNameTypeVector();
currentScope = fElemStack.getCurrentScope();
}
else if (modelType == SchemaElementDecl::Any) {
laxThisOne = true;
}
}
// Now, since we might have to update the namespace map for this element,
// but we don't have the element decl yet, we just tell the element stack
// to expand up to get ready.
unsigned int elemDepth = fElemStack.addLevel();
fElemStack.setValidationFlag(fValidate);
fElemStack.setPrefixColonPos(prefixColonPos);
// Make an initial pass through the list and find any xmlns attributes or
// schema attributes.
if (attCount)
scanRawAttrListforNameSpaces(attCount);
// Resolve the qualified name to a URI and name so that we can look up
// the element decl for this element. We have now update the prefix to
// namespace map so we should get the correct element now.
unsigned int uriId = resolveQNameWithColon
(
qnameRawBuf, fPrefixBuf, ElemStack::Mode_Element, prefixColonPos
);
//if schema, check if we should lax or skip the validation of this element
bool parentValidation = fValidate;
if (cv) {
QName element(fPrefixBuf.getRawBuffer(), &qnameRawBuf[prefixColonPos + 1], uriId, fMemoryManager);
// elementDepth will be > 0, as cv is only constructed if element is not
// root.
laxThisOne = laxElementValidation(&element, cv, cm, elemDepth - 1);
}
// Look up the element now in the grammar. This will get us back a
// generic element decl object. We tell him to fault one in if he does
// not find it.
bool wasAdded = false;
const XMLCh* nameRawBuf = &qnameRawBuf[prefixColonPos + 1];
XMLElementDecl* elemDecl = fGrammar->getElemDecl
(
uriId, nameRawBuf, qnameRawBuf, currentScope
);
if (!elemDecl)
{
// URI is different, so we try to switch grammar
if (uriId != fURIStringPool->getId(fGrammar->getTargetNamespace())) {
switchGrammar(getURIText(uriId), laxThisOne);
}
// look for a global element declaration
elemDecl = fGrammar->getElemDecl(
uriId, nameRawBuf, qnameRawBuf, Grammar::TOP_LEVEL_SCOPE
);
if (!elemDecl)
{
// if still not found, look in list of undeclared elements
elemDecl = fElemNonDeclPool->getByKey(
nameRawBuf, uriId, Grammar::TOP_LEVEL_SCOPE);
if (!elemDecl)
{
elemDecl = new (fMemoryManager) SchemaElementDecl
(
fPrefixBuf.getRawBuffer(), nameRawBuf, uriId
, SchemaElementDecl::Any, Grammar::TOP_LEVEL_SCOPE
, fMemoryManager
);
elemDecl->setId
(
fElemNonDeclPool->put
(
(void*)elemDecl->getBaseName(), uriId
, Grammar::TOP_LEVEL_SCOPE, (SchemaElementDecl*)elemDecl
)
);
wasAdded = true;
}
}
}
// We do something different here according to whether we found the
// element or not.
if (wasAdded || !elemDecl->isDeclared())
{
if (laxThisOne) {
fValidate = false;
fElemStack.setValidationFlag(fValidate);
}
// If validating then emit an error
if (fValidate)
{
// This is to tell the reuse Validator that this element was
// faulted-in, was not an element in the grammar pool originally
elemDecl->setCreateReason(XMLElementDecl::JustFaultIn);
fValidator->emitError
(
XMLValid::ElementNotDefined, elemDecl->getFullName()
);
}
}
// Now we can update the element stack to set the current element
// decl. We expanded the stack above, but couldn't store the element
// decl because we didn't know it yet.
fElemStack.setElement(elemDecl, fReaderMgr.getCurrentReaderNum());
fElemStack.setCurrentURI(uriId);
if (isRoot) {
fRootElemName = XMLString::replicate(qnameRawBuf, fMemoryManager);
}
// Validate the element
if (fValidate) {
fValidator->validateElement(elemDecl);
}
// squirrel away the element's QName, so that we can do an efficient
// end-tag match
fElemStack.setCurrentSchemaElemName(fQNameBuf.getRawBuffer());
ComplexTypeInfo* typeinfo = (fValidate)
? ((SchemaValidator*)fValidator)->getCurrentTypeInfo()
: ((SchemaElementDecl*) elemDecl)->getComplexTypeInfo();
if (typeinfo)
{
currentScope = typeinfo->getScopeDefined();
// switch grammar if the typeinfo has a different grammar
XMLCh* typeName = typeinfo->getTypeName();
int comma = XMLString::indexOf(typeName, chComma);
if (comma > 0)
{
XMLBufBid bbPrefix(&fBufMgr);
XMLBuffer& prefixBuf = bbPrefix.getBuffer();
prefixBuf.append(typeName, comma);
switchGrammar(prefixBuf.getRawBuffer(), laxThisOne);
}
}
fElemStack.setCurrentScope(currentScope);
// Set element next state
if (elemDepth >= fElemStateSize) {
resizeElemState();
}
fElemState[elemDepth] = 0;
fElemStack.setCurrentGrammar(fGrammar);
// If this is the first element and we are validating, check the root
// element.
if (!isRoot && parentValidation) {
fElemStack.addChild(elemDecl->getElementName(), true);
}
// Now lets get the fAttrList filled in. This involves faulting in any
// defaulted and fixed attributes and normalizing the values of any that
// we got explicitly.
//
// We update the attCount value with the total number of attributes, but
// it goes in with the number of values we got during the raw scan of
// explictly provided attrs above.
attCount = buildAttList(*fRawAttrList, attCount, elemDecl, *fAttrList);
if(attCount)
{
// clean up after ourselves:
// clear the map used to detect duplicate attributes
fUndeclaredAttrRegistryNS->removeAll();
}
// Since the element may have default values, call start tag now regardless if it is empty or not
// If we have a document handler, then tell it about this start tag
if (fDocHandler)
{
fDocHandler->startElement
(
*elemDecl, uriId, fPrefixBuf.getRawBuffer(), *fAttrList
, attCount, false, isRoot
);
} // may be where we output something...
// If empty, validate content right now if we are validating and then
// pop the element stack top. Else, we have to update the current stack
// top's namespace mapping elements.
if (isEmpty)
{
// Pop the element stack back off since it'll never be used now
fElemStack.popTop();
// If validating, then insure that its legal to have no content
if (fValidate)
{
const int res = fValidator->checkContent(elemDecl, 0, 0);
if (res >= 0)
{
// REVISIT: in the case of xsi:type, this may
// return the wrong string...
fValidator->emitError
(
XMLValid::ElementNotValidForContent
, elemDecl->getFullName()
, elemDecl->getFormattedContentModel()
);
}
}
// If we have a doc handler, tell it about the end tag
if (fDocHandler)
{
fDocHandler->endElement
(
*elemDecl, uriId, isRoot, fPrefixBuf.getRawBuffer()
);
}
// If the elem stack is empty, then it was an empty root
if (isRoot) {
gotData = false;
}
else
{
// Restore the grammar
fGrammar = fElemStack.getCurrentGrammar();
fGrammarType = fGrammar->getGrammarType();
fValidator->setGrammar(fGrammar);
// Restore the validation flag
fValidate = fElemStack.getValidationFlag();
}
}
return true;
}
// ---------------------------------------------------------------------------
// XSDDOMParser: Implementation of the XMLDocumentHandler interface
// ---------------------------------------------------------------------------
void XSDDOMParser::startElement( const XMLElementDecl& elemDecl
, const unsigned int urlId
, const XMLCh* const elemPrefix
, const RefVectorOf<XMLAttr>& attrList
, const unsigned int attrCount
, const bool isEmpty
, const bool isRoot)
{
fDepth++;
// while it is true that non-whitespace character data
// may only occur in appInfo or documentation
// elements, it's certainly legal for comments and PI's to
// occur as children of annotation; we need
// to account for these here.
if (fAnnotationDepth == -1)
{
if (XMLString::equals(elemDecl.getBaseName(), SchemaSymbols::fgELT_ANNOTATION) &&
XMLString::equals(getURIText(urlId), SchemaSymbols::fgURI_SCHEMAFORSCHEMA))
{
fAnnotationDepth = fDepth;
startAnnotation(elemDecl, attrList, attrCount);
}
}
else if (fDepth == fAnnotationDepth+1)
{
fInnerAnnotationDepth = fDepth;
startAnnotationElement(elemDecl, attrList, attrCount);
}
else
{
startAnnotationElement(elemDecl, attrList, attrCount);
// avoid falling through; don't call startElement in this case
return;
}
DOMElement *elem;
if (urlId != fScanner->getEmptyNamespaceId()) //TagName has a prefix
{
if (elemPrefix && *elemPrefix)
{
XMLBufBid elemQName(&fBufMgr);
elemQName.set(elemPrefix);
elemQName.append(chColon);
elemQName.append(elemDecl.getBaseName());
elem = createElementNSNode(
fScanner->getURIText(urlId), elemQName.getRawBuffer());
}
else {
elem = createElementNSNode(
fScanner->getURIText(urlId), elemDecl.getBaseName());
}
}
else {
elem = createElementNSNode(0, elemDecl.getBaseName());
}
DOMElementImpl *elemImpl = (DOMElementImpl *) elem;
for (unsigned int index = 0; index < attrCount; ++index)
{
const XMLAttr* oneAttrib = attrList.elementAt(index);
unsigned int attrURIId = oneAttrib->getURIId();
const XMLCh* namespaceURI = 0;
//for xmlns=...
if (XMLString::equals(oneAttrib->getName(), XMLUni::fgXMLNSString))
attrURIId = fScanner->getXMLNSNamespaceId();
//TagName has a prefix
if (attrURIId != fScanner->getEmptyNamespaceId())
namespaceURI = fScanner->getURIText(attrURIId); //get namespaceURI
// revisit. Optimize to init the named node map to the
// right size up front.
DOMAttrImpl *attr = (DOMAttrImpl *)
fDocument->createAttributeNS(namespaceURI, oneAttrib->getQName());
attr->setValue(oneAttrib -> getValue());
DOMNode* remAttr = elemImpl->setAttributeNodeNS(attr);
if (remAttr)
remAttr->release();
// Attributes of type ID. If this is one, add it to the hashtable of IDs
// that is constructed for use by GetElementByID().
if (oneAttrib->getType()==XMLAttDef::ID)
{
if (fDocument->fNodeIDMap == 0)
fDocument->fNodeIDMap = new (fDocument) DOMNodeIDMap(500, fDocument);
fDocument->fNodeIDMap->add(attr);
attr->fNode.isIdAttr(true);
}
attr->setSpecified(oneAttrib->getSpecified());
}
// set up the default attributes
if (elemDecl.hasAttDefs())
{
XMLAttDefList* defAttrs = &elemDecl.getAttDefList();
XMLAttDef* attr = 0;
DOMAttrImpl * insertAttr = 0;
while (defAttrs->hasMoreElements())
{
attr = &defAttrs->nextElement();
const XMLAttDef::DefAttTypes defType = attr->getDefaultType();
if ((defType == XMLAttDef::Default)
|| (defType == XMLAttDef::Fixed))
{
// DOM Level 2 wants all namespace declaration attributes
// to be bound to "http://www.w3.org/2000/xmlns/"
// So as long as the XML parser doesn't do it, it needs to
// done here.
const XMLCh* qualifiedName = attr->getFullName();
XMLBufBid bbPrefixQName(&fBufMgr);
XMLBuffer& prefixBuf = bbPrefixQName.getBuffer();
int colonPos = -1;
unsigned int uriId = fScanner->resolveQName(qualifiedName, prefixBuf, ElemStack::Mode_Attribute, colonPos);
const XMLCh* namespaceURI = 0;
if (XMLString::equals(qualifiedName, XMLUni::fgXMLNSString))
uriId = fScanner->getXMLNSNamespaceId();
//TagName has a prefix
if (uriId != fScanner->getEmptyNamespaceId())
namespaceURI = fScanner->getURIText(uriId);
insertAttr = (DOMAttrImpl *) fDocument->createAttributeNS(
namespaceURI, qualifiedName);
DOMAttr* remAttr = elemImpl->setDefaultAttributeNodeNS(insertAttr);
if (remAttr)
remAttr->release();
if (attr->getValue() != 0)
{
insertAttr->setValue(attr->getValue());
insertAttr->setSpecified(false);
}
}
insertAttr = 0;
attr->reset();
}
}
fCurrentParent->appendChild(elem);
fNodeStack->push(fCurrentParent);
fCurrentParent = elem;
fCurrentNode = elem;
fWithinElement = true;
// If an empty element, do end right now (no endElement() will be called)
if (isEmpty)
endElement(elemDecl, urlId, isRoot, elemPrefix);
}
