void XsdSchemaDebugger::dumpParticle(const XsdParticle::Ptr &particle, int level)
{
QString prefix; prefix.fill(QLatin1Char(' '), level);
qDebug("%s min=%s max=%s", qPrintable(prefix), qPrintable(QString::number(particle->minimumOccurs())),
qPrintable(particle->maximumOccursUnbounded() ? QLatin1String("unbounded") : QString::number(particle->maximumOccurs())));
if (particle->term()->isElement()) {
qDebug("%selement (%s)", qPrintable(prefix), qPrintable(XsdElement::Ptr(particle->term())->displayName(m_namePool)));
} else if (particle->term()->isModelGroup()) {
const XsdModelGroup::Ptr group(particle->term());
if (group->compositor() == XsdModelGroup::SequenceCompositor) {
qDebug("%ssequence", qPrintable(prefix));
} else if (group->compositor() == XsdModelGroup::AllCompositor) {
qDebug("%sall", qPrintable(prefix));
} else if (group->compositor() == XsdModelGroup::ChoiceCompositor) {
qDebug("%schoice", qPrintable(prefix));
}
for (int i = 0; i < group->particles().count(); ++i)
dumpParticle(group->particles().at(i), level + 5);
} else if (particle->term()->isWildcard()) {
XsdWildcard::Ptr wildcard(particle->term());
qDebug("%swildcard (process=%d)", qPrintable(prefix), wildcard->processContents());
}
}
/**
* Internal helper method that checks if the given @p particle contains an element with the
* same name and type twice.
*/
static bool hasDuplicatedElementsInternal(const XsdParticle::Ptr &particle, const NamePool::Ptr &namePool, ElementHash &hash, XsdElement::Ptr &conflictingElement)
{
const XsdTerm::Ptr term = particle->term();
if (term->isElement()) {
const XsdElement::Ptr mainElement(term);
XsdElement::WeakList substGroups = mainElement->substitutionGroups();
if (substGroups.isEmpty())
substGroups << mainElement.data();
for (int i = 0; i < substGroups.count(); ++i) {
const XsdElement::Ptr element(substGroups.at(i));
if (hash.contains(element->name(namePool))) {
if (element->type()->name(namePool) != hash.value(element->name(namePool))->type()->name(namePool)) {
conflictingElement = element;
return true;
}
} else {
hash.insert(element->name(namePool), element);
}
}
} else if (term->isModelGroup()) {
const XsdModelGroup::Ptr group(term);
const XsdParticle::List particles = group->particles();
for (int i = 0; i < particles.count(); ++i) {
if (hasDuplicatedElementsInternal(particles.at(i), namePool, hash, conflictingElement))
return true;
}
}
return false;
}
bool XsdSchemaChecker::particleEqualsRecursively(const XsdParticle::Ptr &particle, const XsdParticle::Ptr &otherParticle) const
{
// @see http://www.w3.org/TR/xmlschema11-1/#cos-particle-extend
//TODO: find out what 'properties' of a particle should be checked here...
if (particle->minimumOccurs() != otherParticle->minimumOccurs())
return false;
if (particle->maximumOccursUnbounded() != otherParticle->maximumOccursUnbounded())
return false;
if (particle->maximumOccurs() != otherParticle->maximumOccurs())
return false;
const XsdTerm::Ptr term = particle->term();
const XsdTerm::Ptr otherTerm = otherParticle->term();
if (term->isElement() && !(otherTerm->isElement()))
return false;
if (term->isModelGroup() && !(otherTerm->isModelGroup()))
return false;
if (term->isWildcard() && !(otherTerm->isWildcard()))
return false;
if (term->isElement()) {
const XsdElement::Ptr element = term;
const XsdElement::Ptr otherElement = otherTerm;
if (element->name(m_namePool) != otherElement->name(m_namePool))
return false;
if (element->type()->name(m_namePool) != otherElement->type()->name(m_namePool))
return false;
}
if (term->isModelGroup()) {
const XsdModelGroup::Ptr group = term;
const XsdModelGroup::Ptr otherGroup = otherTerm;
if (group->particles().count() != otherGroup->particles().count())
return false;
for (int i = 0; i < group->particles().count(); ++i) {
if (!particleEqualsRecursively(group->particles().at(i), otherGroup->particles().at(i)))
return false;
}
}
if (term->isWildcard()) {
}
return true;
}
bool XsdParticleChecker::isUPAConformXsdAll(const XsdParticle::Ptr &particle, const NamePool::Ptr &namePool)
{
/**
* see http://www.w3.org/TR/xmlschema-1/#non-ambig
*/
const XsdModelGroup::Ptr group(particle->term());
const XsdParticle::List particles = group->particles();
const int count = particles.count();
for (int left = 0; left < count; ++left) {
for (int right = left+1; right < count; ++right) {
if (termMatches(particles.at(left)->term(), particles.at(right)->term(), namePool))
return false;
}
}
return true;
}
QSet<XsdElement::Ptr> collectAllElements(const XsdParticle::Ptr &particle)
{
QSet<XsdElement::Ptr> elements;
const XsdTerm::Ptr term(particle->term());
if (term->isElement()) {
elements.insert(XsdElement::Ptr(term));
} else if (term->isModelGroup()) {
const XsdModelGroup::Ptr group(term);
for (int i = 0; i < group->particles().count(); ++i)
elements.unite(collectAllElements(group->particles().at(i)));
}
return elements;
}
/*
* Create the FSA according to Algorithm Tp(S) from http://www.ltg.ed.ac.uk/~ht/XML_Europe_2003.html
*/
XsdStateMachine<XsdTerm::Ptr>::StateId XsdStateMachineBuilder::buildParticle(const XsdParticle::Ptr &particle, XsdStateMachine<XsdTerm::Ptr>::StateId endState)
{
XsdStateMachine<XsdTerm::Ptr>::StateId currentStartState = endState;
XsdStateMachine<XsdTerm::Ptr>::StateId currentEndState = endState;
// 2
if (particle->maximumOccursUnbounded()) {
const XsdStateMachine<XsdTerm::Ptr>::StateId t = m_stateMachine->addState(XsdStateMachine<XsdTerm::Ptr>::InternalState);
const XsdStateMachine<XsdTerm::Ptr>::StateId n = buildTerm(particle->term(), t);
m_stateMachine->addEpsilonTransition(t, n);
m_stateMachine->addEpsilonTransition(n, endState);
currentEndState = t;
currentStartState = t;
} else { // 3
int count = (particle->maximumOccurs() - particle->minimumOccurs());
if (count > 100)
count = 100;
for (int i = 0; i < count; ++i) {
currentStartState = buildTerm(particle->term(), currentEndState);
m_stateMachine->addEpsilonTransition(currentStartState, endState);
currentEndState = currentStartState;
}
}
int minOccurs = particle->minimumOccurs();
if (minOccurs > 100)
minOccurs = 100;
for (int i = 0; i < minOccurs; ++i) {
currentStartState = buildTerm(particle->term(), currentEndState);
currentEndState = currentStartState;
}
return currentStartState;
}
/*
* Create the FSA according to Algorithm Tt(S) from http://www.ltg.ed.ac.uk/~ht/XML_Europe_2003.html
*/
XsdStateMachine<XsdTerm::Ptr>::StateId XsdStateMachineBuilder::buildTerm(const XsdTerm::Ptr &term, XsdStateMachine<XsdTerm::Ptr>::StateId endState)
{
if (term->isWildcard()) { // 1
const XsdStateMachine<XsdTerm::Ptr>::StateId b = m_stateMachine->addState(XsdStateMachine<XsdTerm::Ptr>::InternalState);
m_stateMachine->addTransition(b, term, endState);
return b;
} else if (term->isElement()) { // 2
const XsdStateMachine<XsdTerm::Ptr>::StateId b = m_stateMachine->addState(XsdStateMachine<XsdTerm::Ptr>::InternalState);
m_stateMachine->addTransition(b, term, endState);
const XsdElement::Ptr element(term);
if (m_mode == CheckingMode) {
const XsdElement::WeakList substGroups = element->substitutionGroups();
for (int i = 0; i < substGroups.count(); ++i)
m_stateMachine->addTransition(b, XsdElement::Ptr(substGroups.at(i)), endState);
} else if (m_mode == ValidatingMode) {
const XsdElement::WeakList substGroups = element->substitutionGroups();
for (int i = 0; i < substGroups.count(); ++i) {
if (XsdSchemaHelper::substitutionGroupOkTransitive(element, XsdElement::Ptr(substGroups.at(i)), m_namePool))
m_stateMachine->addTransition(b, XsdElement::Ptr(substGroups.at(i)), endState);
}
}
return b;
} else if (term->isModelGroup()) {
const XsdModelGroup::Ptr group(term);
if (group->compositor() == XsdModelGroup::ChoiceCompositor) { // 3
const XsdStateMachine<XsdTerm::Ptr>::StateId b = m_stateMachine->addState(XsdStateMachine<XsdTerm::Ptr>::InternalState);
for (int i = 0; i < group->particles().count(); ++i) {
const XsdParticle::Ptr particle(group->particles().at(i));
if (particle->maximumOccurs() != 0) {
const XsdStateMachine<XsdTerm::Ptr>::StateId state = buildParticle(particle, endState);
m_stateMachine->addEpsilonTransition(b, state);
}
}
return b;
} else if (group->compositor() == XsdModelGroup::SequenceCompositor) { // 4
XsdStateMachine<XsdTerm::Ptr>::StateId currentStartState = endState;
XsdStateMachine<XsdTerm::Ptr>::StateId currentEndState = endState;
for (int i = (group->particles().count() - 1); i >= 0; --i) { // iterate reverse
const XsdParticle::Ptr particle(group->particles().at(i));
if (particle->maximumOccurs() != 0) {
currentStartState = buildParticle(particle, currentEndState);
currentEndState = currentStartState;
}
}
return currentStartState;
} else if (group->compositor() == XsdModelGroup::AllCompositor) {
const XsdStateMachine<XsdTerm::Ptr>::StateId newStartState = m_stateMachine->addState(XsdStateMachine<XsdTerm::Ptr>::InternalState);
const QList<XsdParticle::List> list = allCombinations(group->particles());
for (int i = 0; i < list.count(); ++i) {
XsdStateMachine<XsdTerm::Ptr>::StateId currentStartState = endState;
XsdStateMachine<XsdTerm::Ptr>::StateId currentEndState = endState;
const XsdParticle::List particles = list.at(i);
for (int j = (particles.count() - 1); j >= 0; --j) { // iterate reverse
const XsdParticle::Ptr particle(particles.at(j));
if (particle->maximumOccurs() != 0) {
currentStartState = buildParticle(particle, currentEndState);
currentEndState = currentStartState;
}
}
m_stateMachine->addEpsilonTransition(newStartState, currentStartState);
}
if (list.isEmpty())
return endState;
else
return newStartState;
}
}
Q_ASSERT(false);
return 0;
}
bool XsdParticleChecker::isUPAConform(const XsdParticle::Ptr &particle, const NamePool::Ptr &namePool)
{
/**
* In case we encounter an <xsd:all> element, don't construct a state machine, but use the approach
* described at http://www.w3.org/TR/xmlschema-1/#non-ambig
* Reason: For n elements inside the <xsd:all>, represented in the NDA, the state machine
* constructs n! states in the DFA, which does not scale.
*/
if (particle->term()->isModelGroup()) {
const XsdModelGroup::Ptr group(particle->term());
if (group->compositor() == XsdModelGroup::AllCompositor)
return isUPAConformXsdAll(particle, namePool);
}
/**
* The algorithm is implemented like described in http://www.ltg.ed.ac.uk/~ht/XML_Europe_2003.html#S2.2
*/
// create a state machine for the given particle
XsdStateMachine<XsdTerm::Ptr> stateMachine(namePool);
XsdStateMachineBuilder builder(&stateMachine, namePool);
const XsdStateMachine<XsdTerm::Ptr>::StateId endState = builder.reset();
const XsdStateMachine<XsdTerm::Ptr>::StateId startState = builder.buildParticle(particle, endState);
builder.addStartState(startState);
/*
static int counter = 0;
{
QFile file(QString("/tmp/file_upa%1.dot").arg(counter));
file.open(QIODevice::WriteOnly);
stateMachine.outputGraph(&file, "Base");
file.close();
}
::system(QString("dot -Tpng /tmp/file_upa%1.dot -o/tmp/file_upa%1.png").arg(counter).toLatin1().data());
*/
const XsdStateMachine<XsdTerm::Ptr> dfa = stateMachine.toDFA();
/*
{
QFile file(QString("/tmp/file_upa%1dfa.dot").arg(counter));
file.open(QIODevice::WriteOnly);
dfa.outputGraph(&file, "Base");
file.close();
}
::system(QString("dot -Tpng /tmp/file_upa%1dfa.dot -o/tmp/file_upa%1dfa.png").arg(counter).toLatin1().data());
*/
const QHash<XsdStateMachine<XsdTerm::Ptr>::StateId, XsdStateMachine<XsdTerm::Ptr>::StateType> states = dfa.states();
const QHash<XsdStateMachine<XsdTerm::Ptr>::StateId, QHash<XsdTerm::Ptr, QVector<XsdStateMachine<XsdTerm::Ptr>::StateId> > > transitions = dfa.transitions();
// the basic idea of that algorithm is to iterate over all states of that machine and check that no two edges
// that match on the same term leave a state, so for a given term it should always be obvious which edge to take
QHashIterator<XsdStateMachine<XsdTerm::Ptr>::StateId, XsdStateMachine<XsdTerm::Ptr>::StateType> stateIt(states);
while (stateIt.hasNext()) { // iterate over all states
stateIt.next();
// fetch all transitions the current state allows
const QHash<XsdTerm::Ptr, QVector<XsdStateMachine<XsdTerm::Ptr>::StateId> > currentTransitions = transitions.value(stateIt.key());
QHashIterator<XsdTerm::Ptr, QVector<XsdStateMachine<XsdTerm::Ptr>::StateId> > transitionIt(currentTransitions);
while (transitionIt.hasNext()) { // iterate over all transitions
transitionIt.next();
if (transitionIt.value().size() > 1) {
// we have one state with two edges leaving it, that means
// the XsdTerm::Ptr exists twice, that is an error
return false;
}
QHashIterator<XsdTerm::Ptr, QVector<XsdStateMachine<XsdTerm::Ptr>::StateId> > innerTransitionIt(currentTransitions);
while (innerTransitionIt.hasNext()) { // iterate over all transitions again, as we have to compare all transitions with all
innerTransitionIt.next();
if (transitionIt.key() == innerTransitionIt.key()) // do no compare with ourself
continue;
// use the helper method termMatches to check if both term matches
if (termMatches(transitionIt.key(), innerTransitionIt.key(), namePool))
return false;
}
}
}
return true;
}
bool XsdSchemaChecker::elementSequenceAccepted(const XsdModelGroup::Ptr &sequence, const XsdParticle::Ptr &particle) const
{
// @see http://www.w3.org/TR/xmlschema11-1/#cvc-accept
if (particle->term()->isWildcard()) { // 1
const XsdWildcard::Ptr wildcard(particle->term());
// 1.1
if ((unsigned int)sequence->particles().count() < particle->minimumOccurs())
return false;
// 1.2
if (!particle->maximumOccursUnbounded()) {
if ((unsigned int)sequence->particles().count() > particle->maximumOccurs())
return false;
}
// 1.3
const XsdParticle::List particles(sequence->particles());
for (int i = 0; i < particles.count(); ++i) {
if (particles.at(i)->term()->isElement()) {
if (!XsdSchemaHelper::wildcardAllowsExpandedName(XsdElement::Ptr(particles.at(i)->term())->name(m_namePool), wildcard, m_namePool))
return false;
}
}
} else if (particle->term()->isElement()) { // 2
const XsdElement::Ptr element(particle->term());
// 2.1
if ((unsigned int)sequence->particles().count() < particle->minimumOccurs())
return false;
// 2.2
if (!particle->maximumOccursUnbounded()) {
if ((unsigned int)sequence->particles().count() > particle->maximumOccurs())
return false;
}
// 2.3
const XsdParticle::List particles(sequence->particles());
for (int i = 0; i < particles.count(); ++i) {
bool isValid = false;
if (particles.at(i)->term()->isElement()) {
const XsdElement::Ptr seqElement(particles.at(i)->term());
// 2.3.1
if (element->name(m_namePool) == seqElement->name(m_namePool))
isValid = true;
// 2.3.2
if (element->scope() && element->scope()->variety() == XsdElement::Scope::Global) {
if (!(element->disallowedSubstitutions() & NamedSchemaComponent::SubstitutionConstraint)) {
//TODO: continue
}
}
}
}
}
return true;
}
bool XsdSchemaChecker::isValidParticleExtension(const XsdParticle::Ptr &extension, const XsdParticle::Ptr &base) const
{
// @see http://www.w3.org/TR/xmlschema11-1/#cos-particle-extend
// 1
if (extension == base)
return true;
// 2
if (extension->minimumOccurs() == 1 && extension->maximumOccurs() == 1 && extension->maximumOccursUnbounded() == false) {
if (extension->term()->isModelGroup()) {
const XsdModelGroup::Ptr modelGroup = extension->term();
if (modelGroup->compositor() == XsdModelGroup::SequenceCompositor) {
if (particleEqualsRecursively(modelGroup->particles().first(), base))
return true;
}
}
}
// 3
if (extension->minimumOccurs() == base->minimumOccurs()) { // 3.1
if (extension->term()->isModelGroup() && base->term()->isModelGroup()) {
const XsdModelGroup::Ptr extensionGroup(extension->term());
const XsdModelGroup::Ptr baseGroup(base->term());
if (extensionGroup->compositor() == XsdModelGroup::AllCompositor && baseGroup->compositor() == XsdModelGroup::AllCompositor) {
const XsdParticle::List extensionParticles = extensionGroup->particles();
const XsdParticle::List baseParticles = baseGroup->particles();
for (int i = 0; i < baseParticles.count() && i < extensionParticles.count(); ++i) {
if (baseParticles.at(i) != extensionParticles.at(i))
return false;
}
}
}
}
return false;
}