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());
}
}
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;
}
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;
}