// static
MatchExpression* CanonicalQuery::normalizeTree(MatchExpression* root) {
// root->isLogical() is true now. We care about AND, OR, and NOT. NOR currently scares us.
if (MatchExpression::AND == root->matchType() || MatchExpression::OR == root->matchType()) {
// We could have AND of AND of AND. Make sure we clean up our children before merging
// them.
// UNITTEST 11738048
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
// If any of our children are of the same logical operator that we are, we remove the
// child's children and append them to ourselves after we examine all children.
std::vector<MatchExpression*> absorbedChildren;
for (size_t i = 0; i < root->numChildren();) {
MatchExpression* child = root->getChild(i);
if (child->matchType() == root->matchType()) {
// AND of an AND or OR of an OR. Absorb child's children into ourself.
for (size_t j = 0; j < child->numChildren(); ++j) {
absorbedChildren.push_back(child->getChild(j));
}
// TODO(opt): this is possibly n^2-ish
root->getChildVector()->erase(root->getChildVector()->begin() + i);
child->getChildVector()->clear();
// Note that this only works because we cleared the child's children
delete child;
// Don't increment 'i' as the current child 'i' used to be child 'i+1'
} else {
++i;
}
}
root->getChildVector()->insert(
root->getChildVector()->end(), absorbedChildren.begin(), absorbedChildren.end());
// AND of 1 thing is the thing, OR of 1 thing is the thing.
if (1 == root->numChildren()) {
MatchExpression* ret = root->getChild(0);
root->getChildVector()->clear();
delete root;
return ret;
}
} else if (MatchExpression::NOT == root->matchType()) {
// Normalize the rest of the tree hanging off this NOT node.
NotMatchExpression* nme = static_cast<NotMatchExpression*>(root);
MatchExpression* child = nme->releaseChild();
// normalizeTree(...) takes ownership of 'child', and then
// transfers ownership of its return value to 'nme'.
nme->resetChild(normalizeTree(child));
} else if (MatchExpression::ELEM_MATCH_VALUE == root->matchType()) {
// Just normalize our children.
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
}
return root;
}
Status CanonicalQuery::init(LiteParsedQuery* lpq,
const MatchExpressionParser::WhereCallback& whereCallback,
MatchExpression* root) {
_pq.reset(lpq);
// Normalize, sort and validate tree.
root = normalizeTree(root);
sortTree(root);
_root.reset(root);
Status validStatus = isValid(root, *_pq);
if (!validStatus.isOK()) {
return validStatus;
}
// Validate the projection if there is one.
if (!_pq->getProj().isEmpty()) {
ParsedProjection* pp;
Status projStatus =
ParsedProjection::make(_pq->getProj(), _root.get(), &pp, whereCallback);
if (!projStatus.isOK()) {
return projStatus;
}
_proj.reset(pp);
}
return Status::OK();
}
Status CanonicalQuery::init(LiteParsedQuery* lpq) {
_pq.reset(lpq);
// Build a parse tree from the BSONObj in the parsed query.
StatusWithMatchExpression swme = MatchExpressionParser::parse(_pq->getFilter());
if (!swme.isOK()) {
return swme.getStatus();
}
MatchExpression* root = swme.getValue();
root = normalizeTree(root);
Status validStatus = isValid(root);
if (!validStatus.isOK()) {
return validStatus;
}
_root.reset(root);
// Validate the projection if there is one.
if (!_pq->getProj().isEmpty()) {
ParsedProjection* pp;
Status projStatus = ParsedProjection::make(_pq->getProj(), root, &pp);
if (!projStatus.isOK()) {
return projStatus;
}
_proj.reset(pp);
}
return Status::OK();
}
Status CanonicalQuery::init(LiteParsedQuery* lpq,
const ExtensionsCallback& extensionsCallback,
MatchExpression* root) {
_pq.reset(lpq);
_hasNoopExtensions = extensionsCallback.hasNoopExtensions();
_isIsolated = LiteParsedQuery::isQueryIsolated(lpq->getFilter());
// Normalize, sort and validate tree.
root = normalizeTree(root);
sortTree(root);
_root.reset(root);
Status validStatus = isValid(root, *_pq);
if (!validStatus.isOK()) {
return validStatus;
}
// Validate the projection if there is one.
if (!_pq->getProj().isEmpty()) {
ParsedProjection* pp;
Status projStatus =
ParsedProjection::make(_pq->getProj(), _root.get(), &pp, extensionsCallback);
if (!projStatus.isOK()) {
return projStatus;
}
_proj.reset(pp);
}
if (_proj && _proj->wantSortKey() && _pq->getSort().isEmpty()) {
return Status(ErrorCodes::BadValue, "cannot use sortKey $meta projection without a sort");
}
return Status::OK();
}
Status CanonicalQuery::init(LiteParsedQuery* lpq) {
_pq.reset(lpq);
// Build a parse tree from the BSONObj in the parsed query.
StatusWithMatchExpression swme = MatchExpressionParser::parse(_pq->getFilter());
if (!swme.isOK()) { return swme.getStatus(); }
MatchExpression* root = swme.getValue();
root = normalizeTree(root);
Status validStatus = isValid(root);
if (!validStatus.isOK()) {
return validStatus;
}
_root.reset(root);
if (!_pq->getProj().isEmpty()) {
LiteProjection* liteProj = NULL;
Status liteProjStatus = LiteProjection::make(_pq->getFilter(), _pq->getProj(), &liteProj);
if (!liteProjStatus.isOK()) {
return liteProjStatus;
}
_liteProj.reset(liteProj);
}
return Status::OK();
}
// static
MatchExpression* CanonicalQuery::normalizeTree(MatchExpression* root) {
// root->isLogical() is true now. We care about AND and OR. Negations currently scare us.
if (MatchExpression::AND == root->matchType() || MatchExpression::OR == root->matchType()) {
// We could have AND of AND of AND. Make sure we clean up our children before merging
// them.
// UNITTEST 11738048
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
// If any of our children are of the same logical operator that we are, we remove the
// child's children and append them to ourselves after we examine all children.
vector<MatchExpression*> absorbedChildren;
for (size_t i = 0; i < root->numChildren();) {
MatchExpression* child = root->getChild(i);
if (child->matchType() == root->matchType()) {
// AND of an AND or OR of an OR. Absorb child's children into ourself.
for (size_t j = 0; j < child->numChildren(); ++j) {
absorbedChildren.push_back(child->getChild(j));
}
// TODO(opt): this is possibly n^2-ish
root->getChildVector()->erase(root->getChildVector()->begin() + i);
child->getChildVector()->clear();
// Note that this only works because we cleared the child's children
delete child;
// Don't increment 'i' as the current child 'i' used to be child 'i+1'
}
else {
++i;
}
}
root->getChildVector()->insert(root->getChildVector()->end(),
absorbedChildren.begin(),
absorbedChildren.end());
// AND of 1 thing is the thing, OR of 1 thing is the thing.
if (1 == root->numChildren()) {
MatchExpression* ret = root->getChild(0);
root->getChildVector()->clear();
delete root;
return ret;
}
}
return root;
}
// static
// XXX TODO: This does not belong here at all.
MatchExpression* CanonicalQuery::logicalRewrite(MatchExpression* tree) {
// Only thing we do is pull an OR up at the root.
if (MatchExpression::AND != tree->matchType()) {
return tree;
}
// We want to bail out ASAP if we have nothing to do here.
size_t numOrs = 0;
for (size_t i = 0; i < tree->numChildren(); ++i) {
if (MatchExpression::OR == tree->getChild(i)->matchType()) {
++numOrs;
}
}
// Only do this for one OR right now.
if (1 != numOrs) {
return tree;
}
// Detach the OR from the root.
invariant(NULL != tree->getChildVector());
std::vector<MatchExpression*>& rootChildren = *tree->getChildVector();
MatchExpression* orChild = NULL;
for (size_t i = 0; i < rootChildren.size(); ++i) {
if (MatchExpression::OR == rootChildren[i]->matchType()) {
orChild = rootChildren[i];
rootChildren.erase(rootChildren.begin() + i);
break;
}
}
// AND the existing root with each or child.
invariant(NULL != orChild);
invariant(NULL != orChild->getChildVector());
std::vector<MatchExpression*>& orChildren = *orChild->getChildVector();
for (size_t i = 0; i < orChildren.size(); ++i) {
AndMatchExpression* ama = new AndMatchExpression();
ama->add(orChildren[i]);
ama->add(tree->shallowClone());
orChildren[i] = ama;
}
delete tree;
// Clean up any consequences from this tomfoolery.
return normalizeTree(orChild);
}
//void AssociationTree::solve(std::vector<std::pair<unsigned int, unsigned int> > & _pairs, std::vector<unsigned int> & _unassoc)
void AssociationTree::solve(std::vector<std::pair<unsigned int, unsigned int> > & _pairs, std::vector<bool> & _associated_mask)
{
std::list<AssociationNode*>::iterator best_node;
bool rootReached = false;
AssociationNode *anPtr;
//grows tree exploring all likely hypothesis
growTree();
//computes tree probs
computeTree();
//normalizes tree probs
normalizeTree();
//if terminus_node_list_ is empty exit withou pairing
if ( terminus_node_list_.empty() ) return;
//choose best node based on best tree probability
chooseBestTerminus(best_node);
//resize _associated_mask and resets it to false
_associated_mask.resize(nd_,false);
//set pairs
anPtr = *best_node; //init pointer
int ii=0;
while( ! anPtr->isRoot() ) //set pairs
{
// if ( anPtr->getTargetIndex() == nt_) //detection with void target -> unassociated detection
// {
// _unassoc.push_back(anPtr->getDetectionIndex());
// }
// else
// {
// _pairs.push_back( std::pair<unsigned int, unsigned int>(anPtr->getDetectionIndex(), anPtr->getTargetIndex()) );
// }
if ( anPtr->getTargetIndex() < nt_ ) //association pair
{
_associated_mask.at(anPtr->getDetectionIndex()) = true;
_pairs.push_back( std::pair<unsigned int, unsigned int>(anPtr->getDetectionIndex(), anPtr->getTargetIndex()) );
}
anPtr = anPtr->upNodePtr();
}
}
Status CanonicalQuery::init(LiteParsedQuery* lpq) {
_pq.reset(lpq);
// Build a parse tree from the BSONObj in the parsed query.
StatusWithMatchExpression swme = MatchExpressionParser::parse(_pq->getFilter());
if (!swme.isOK()) { return swme.getStatus(); }
MatchExpression* root = swme.getValue();
normalizeTree(root);
_root.reset(root);
if (!_pq->getProj().isEmpty()) {
ParsedProjection* proj;
Status projStatus = ProjectionParser::parseFindSyntax(_pq->getProj(), &proj);
if (!projStatus.isOK()) {
return projStatus;
}
_proj.reset(proj);
}
return Status::OK();
}
Status CanonicalQuery::init(std::unique_ptr<QueryRequest> qr,
const ExtensionsCallback& extensionsCallback,
MatchExpression* root,
std::unique_ptr<CollatorInterface> collator) {
_qr = std::move(qr);
_collator = std::move(collator);
_hasNoopExtensions = extensionsCallback.hasNoopExtensions();
_isIsolated = QueryRequest::isQueryIsolated(_qr->getFilter());
// Normalize, sort and validate tree.
root = normalizeTree(root);
sortTree(root);
_root.reset(root);
Status validStatus = isValid(root, *_qr);
if (!validStatus.isOK()) {
return validStatus;
}
// Validate the projection if there is one.
if (!_qr->getProj().isEmpty()) {
ParsedProjection* pp;
Status projStatus =
ParsedProjection::make(_qr->getProj(), _root.get(), &pp, extensionsCallback);
if (!projStatus.isOK()) {
return projStatus;
}
_proj.reset(pp);
}
if (_proj && _proj->wantSortKey() && _qr->getSort().isEmpty()) {
return Status(ErrorCodes::BadValue, "cannot use sortKey $meta projection without a sort");
}
return Status::OK();
}
Status CanonicalQuery::normalize(MatchExpression* root) {
_root.reset(normalizeTree(root));
sortTree(_root.get());
return isValid(_root.get());
}
// static
MatchExpression* CanonicalQuery::normalizeTree(MatchExpression* root) {
if (MatchExpression::AND == root->matchType() || MatchExpression::OR == root->matchType()) {
// We could have AND of AND of AND. Make sure we clean up our children before merging them.
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
// If any of our children are of the same logical operator that we are, we remove the
// child's children and append them to ourselves after we examine all children.
std::vector<MatchExpression*> absorbedChildren;
for (size_t i = 0; i < root->numChildren();) {
MatchExpression* child = root->getChild(i);
if (child->matchType() == root->matchType()) {
// AND of an AND or OR of an OR. Absorb child's children into ourself.
for (size_t j = 0; j < child->numChildren(); ++j) {
absorbedChildren.push_back(child->getChild(j));
}
// TODO(opt): this is possibly n^2-ish
root->getChildVector()->erase(root->getChildVector()->begin() + i);
child->getChildVector()->clear();
// Note that this only works because we cleared the child's children
delete child;
// Don't increment 'i' as the current child 'i' used to be child 'i+1'
} else {
++i;
}
}
root->getChildVector()->insert(
root->getChildVector()->end(), absorbedChildren.begin(), absorbedChildren.end());
// AND of 1 thing is the thing, OR of 1 thing is the thing.
if (1 == root->numChildren()) {
MatchExpression* ret = root->getChild(0);
root->getChildVector()->clear();
delete root;
return ret;
}
} else if (MatchExpression::NOR == root->matchType()) {
// First clean up children.
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
// NOR of one thing is NOT of the thing.
if (1 == root->numChildren()) {
// Detach the child and assume ownership.
std::unique_ptr<MatchExpression> child(root->getChild(0));
root->getChildVector()->clear();
// Delete the root when this goes out of scope.
std::unique_ptr<NorMatchExpression> ownedRoot(static_cast<NorMatchExpression*>(root));
// Make a NOT to be the new root and transfer ownership of the child to it.
auto newRoot = stdx::make_unique<NotMatchExpression>();
newRoot->init(child.release()).transitional_ignore();
return newRoot.release();
}
} else if (MatchExpression::NOT == root->matchType()) {
// Normalize the rest of the tree hanging off this NOT node.
NotMatchExpression* nme = static_cast<NotMatchExpression*>(root);
MatchExpression* child = nme->releaseChild();
// normalizeTree(...) takes ownership of 'child', and then
// transfers ownership of its return value to 'nme'.
nme->resetChild(normalizeTree(child));
} else if (MatchExpression::ELEM_MATCH_OBJECT == root->matchType()) {
// Normalize the rest of the tree hanging off this ELEM_MATCH_OBJECT node.
ElemMatchObjectMatchExpression* emome = static_cast<ElemMatchObjectMatchExpression*>(root);
auto child = emome->releaseChild();
// normalizeTree(...) takes ownership of 'child', and then
// transfers ownership of its return value to 'emome'.
emome->resetChild(std::unique_ptr<MatchExpression>(normalizeTree(child.release())));
} else if (MatchExpression::ELEM_MATCH_VALUE == root->matchType()) {
// Just normalize our children.
for (size_t i = 0; i < root->getChildVector()->size(); ++i) {
(*root->getChildVector())[i] = normalizeTree(root->getChild(i));
}
} else if (MatchExpression::MATCH_IN == root->matchType()) {
std::unique_ptr<InMatchExpression> in(static_cast<InMatchExpression*>(root));
// IN of 1 regex is the regex.
if (in->getRegexes().size() == 1 && in->getEqualities().empty()) {
RegexMatchExpression* childRe = in->getRegexes().begin()->get();
invariant(!childRe->getTag());
// Create a new RegexMatchExpression, because 'childRe' does not have a path.
auto re = stdx::make_unique<RegexMatchExpression>();
re->init(in->path(), childRe->getString(), childRe->getFlags()).transitional_ignore();
if (in->getTag()) {
re->setTag(in->getTag()->clone());
}
return normalizeTree(re.release());
}
// IN of 1 equality is the equality.
if (in->getEqualities().size() == 1 && in->getRegexes().empty()) {
auto eq = stdx::make_unique<EqualityMatchExpression>();
eq->init(in->path(), *(in->getEqualities().begin())).transitional_ignore();
eq->setCollator(in->getCollator());
if (in->getTag()) {
eq->setTag(in->getTag()->clone());
}
return eq.release();
}
return in.release();
}
return root;
}