//Replace the Entry(_enitty_id)'s EntityBitSet with _bitset
//Entry of _entity_id must exists
bool
VSTree::replaceEntry(int _entity_id, const EntityBitSet& _bitset)
{
VNode* leafNodePtr = this->getLeafNodeByEntityID(_entity_id);
if (leafNodePtr == NULL)
{
cerr << "error, can not find the mapping leaf node. @VSTree::replaceEntry" << endl;
return false;
}
// find the mapping child entry, update it and refresh signature.
int childNum = leafNodePtr->getChildNum();
bool findFlag = false;
for (int i = 0; i < childNum; i++)
{
const SigEntry& entry = leafNodePtr->getChildEntry(i);
if (entry.getEntityId() == _entity_id)
{
SigEntry newEntry(EntitySig(_bitset), _entity_id);
leafNodePtr->setChildEntry(i, newEntry);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
findFlag = true;
break;
}
}
if (!findFlag)
{
cerr << "error, can not find the mapping child entry in the leaf node. @VSTree::replaceEntry" << endl;
return false;
}
return true;
}
//Replace the Entry(_enitty_id)'s EntityBitSet with _bitset
//Entry of _entity_id must exists
bool
VSTree::replaceEntry(int _entity_id, const EntityBitSet& _bitset)
{
//cout<<"begin replaceEntry()"<<endl;
VNode* leafNodePtr = this->getLeafNodeByEntityID(_entity_id);
if (leafNodePtr == NULL)
{
cerr << "error, can not find the mapping leaf node. @VSTree::replaceEntry" << endl;
return false;
}
// find the mapping child entry, update it and refresh signature.
int childNum = leafNodePtr->getChildNum();
//cout<<"get child num, now to loop"<<endl;
bool findFlag = false;
for (int i = 0; i < childNum; i++)
{
const SigEntry& entry = leafNodePtr->getChildEntry(i);
if (entry.getEntityId() == _entity_id)
{
//cout<<"find the entityid in pos "<<i<<endl;
SigEntry newEntry(EntitySig(_bitset), _entity_id);
leafNodePtr->setChildEntry(i, newEntry);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
findFlag = true;
break;
}
}
//cout<<"root file line: "<<this->root_file_line<<" "<<"max nid num: "<<this->max_nid_alloc<<endl;
//cout<<"node num: "<<this->node_num<<" "<<"file line: "<<leafNodePtr->getFileLine()<<" "<<"child num: "<<childNum<<endl;
//for(int j = 1; j < 4; ++j)
//{
//VNode* tmp = this->getNode(j);
//for(int i = 0; i < tmp->getChildNum(); ++i)
//{
//const SigEntry& entry = tmp->getChildEntry(i);
//cout << entry.getEntityId() << " ";
//}
//cout << endl;
//}
if (!findFlag)
{
cerr << "error, can not find the mapping child entry in the leaf node. @VSTree::replaceEntry" << endl;
return false;
}
return true;
}
//Incrementally update bitset of _entity_id
//conduct OR operation on Entry(_entity_id)'s EntityBitSet with _bitset
//Entry of _entity_id must exists
bool VSTree::updateEntry(int _entity_id, const EntityBitSet& _bitset)
{
VNode* leafNodePtr = this->getLeafNodeByEntityID(_entity_id);
if (leafNodePtr == NULL)
{
cerr << "error, can not find the mapping leaf node. @VSTree::updateEntry" << endl;
return false;
}
// find the mapping child entry, update it and refresh signature.
int childNum = leafNodePtr->getChildNum();
bool findFlag = false;
for (int i = 0; i < childNum; i++)
{
const SigEntry& entry = leafNodePtr->getChildEntry(i);
if (entry.getEntityId() == _entity_id)
{
SigEntry newEntry = entry;
newEntry |= SigEntry(EntitySig(_bitset), _entity_id);
//debug
// {
// if (_entity_id == 10)
// {
// stringstream _ss;
// _ss << "lead node line: " << leafNodePtr->getFileLine() << endl;
// _ss << "old entry:\n " << Signature::BitSet2str(entry.getEntitySig().entityBitSet) << endl;
// _ss << "new entry:\n " << Signature::BitSet2str(newEntry.getEntitySig().entityBitSet) << endl;
// Util::logging(_ss.str());
// }
// }
leafNodePtr->setChildEntry(i, newEntry);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
findFlag = true;
break;
}
}
if (!findFlag)
{
cerr<< "error, can not find the mapping child entry in the leaf node. @VSTree::updateEntry" << endl;
return false;
}
return true;
}
//remove an existed Entry(_entity_id) from VSTree
bool
VSTree::removeEntry(int _entity_id)
{
VNode* leafNodePtr = this->getLeafNodeByEntityID(_entity_id);
if (leafNodePtr == NULL)
{
cerr<< "error, can not find the mapping leaf node. @VSTree::removeEntry" << endl;
return false;
}
// seek the entry index of the leaf node.
int entryIndex = -1;
int childNum = leafNodePtr->getChildNum();
for (int i=0;i<childNum;i++)
{
if (leafNodePtr->getChildEntry(i).getEntityId() == _entity_id)
{
entryIndex = i;
break;
}
}
if (entryIndex == -1)
{
cerr << "error, can not find the entry in leaf node. @VSTree::removeEntry" << endl;
return false;
}
// remove the entry in this leaf node and refresh itself and its ancestors' signature.
leafNodePtr->removeChild(entryIndex);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
this->entry_num --;
//we do not consider the situation which the leaf node is to be empty by now...
//in a better way, if the leaf node is empty after removing entry, we should delete it. and recursively judge whether its
//father is empty, and delete its father node if true. to make the VSTree more balanced, we should combine two nodes if
//their child number are less than the MIN_CHILD_NUM. when deleting one node from the tree, we should also remove it from
//tree node file in hard disk by doing some operations on the node_buffer.
return true;
}
void VNode::refreshAncestorSignature(LRUCache& _nodeBuffer)
{
// refresh self node's signature.
this->refreshSignature();
// refresh father node's signature.
VNode* fatherNodePtr = this->getFather(_nodeBuffer);
if (fatherNodePtr == NULL)
{
if (!this->isRoot())
cerr << "error, can not find father node. @VNode::refreshSignature" << endl;
return;
}
int rank = this->getIndexInFatherNode(_nodeBuffer);
if (fatherNodePtr->getChildEntry(rank).getEntitySig() != this->entry.getEntitySig())
{
fatherNodePtr->setChildEntry(rank, this->entry);
fatherNodePtr->refreshAncestorSignature(_nodeBuffer);
}
}
void
VSTree::split(VNode* _p_node_being_split, const SigEntry& _insert_entry, VNode* _p_insert_node)
{
#ifdef DEBUG_VSTREE
stringstream _ss;
_ss << "**********************split happen at "
<< _p_node_being_split->getFileLine() << endl;
_ss << _p_node_being_split->to_str() << endl;
Util::logging(_ss.str());
#endif
// first, add the new child node(if not leaf) or child entry(if leaf) to the full node.
bool just_insert_entry = (_p_insert_node == NULL);
if(just_insert_entry)
{
_p_node_being_split->addChildEntry(_insert_entry, true);
}
else
{
_p_node_being_split->addChildNode(_p_insert_node, true);
}
SigEntry entryA, entryB;
//BETTER: use hanming, xor result or the vector included angle to guess the distince.
//And then also use the farest two as seeds.
//
//two seeds to generate two new nodes.
//seedA kernel: the SigEntry with the minimal count of signature.
//seedB kernel: the SigEntry with the maximal count of signature.
int maxCount = 0; // record the minimal signature count.
int entryA_index = 0; // record the seedA kernel index.
for(int i = 0; i < VNode::MAX_CHILD_NUM; i++)
{
int currentCount = (int) _p_node_being_split->getChildEntry(i).getSigCount();
if(maxCount < currentCount)
{
maxCount = currentCount;
entryA_index = i;
}
}
entryA = _p_node_being_split->getChildEntry(entryA_index);
maxCount = 0;
int entryB_index = 0; // record the seedB kernel index.
for(int i = 0; i < VNode::MAX_CHILD_NUM; i++)
{
//NOTICE:I think xOR should be used here to choose the farest two
int currentCount = entryA.xOR(_p_node_being_split->getChildEntry(i));
//int currentCount = entryA.xEpsilen(_p_node_being_split->getChildEntry(i));
if(i != entryA_index && maxCount <= currentCount)
{
maxCount = currentCount;
entryB_index = i;
}
}
entryB = _p_node_being_split->getChildEntry(entryB_index);
// AEntryIndex: the entry index near seedA.
// BEntryIndex: the entry index near seedB.
std::vector<int> entryIndex_nearA, entryIndex_nearB;
entryIndex_nearA.clear();
entryIndex_nearB.clear();
entryIndex_nearA.push_back(entryA_index);
entryIndex_nearB.push_back(entryB_index);
int nearA_max_size, nearB_max_size;
bool nearA_tooSmall, nearB_tooSmall;
for(int i = 0; i < VNode::MAX_CHILD_NUM; i++)
{
if(i == entryA_index || i == entryB_index) continue;
//should guarantee that each new node has at least MIN_CHILD_NUM children.
nearA_max_size = VNode::MAX_CHILD_NUM - entryIndex_nearB.size();
nearA_tooSmall = (nearA_max_size <= VNode::MIN_CHILD_NUM);
if(nearA_tooSmall)
{
for(; i < VNode::MAX_CHILD_NUM; i++)
{
if (i == entryA_index || i == entryB_index) continue;
entryIndex_nearA.push_back(i);
}
break;
}
nearB_max_size = VNode::MAX_CHILD_NUM - entryIndex_nearA.size();
nearB_tooSmall = (nearB_max_size <= VNode::MIN_CHILD_NUM);
if(nearB_tooSmall)
{
for(; i < VNode::MAX_CHILD_NUM; i++)
{
if(i == entryA_index || i == entryB_index) continue;
entryIndex_nearB.push_back(i);
}
break;
}
//calculate the distance from
//the i-th child entry signature to seedA(or seedB).
//NOTICE:we should expect that the candidate can be almost contained!
//However, the precondition there are not too many 1s
int disToSeedA = entryA.xEpsilen(_p_node_being_split->getChildEntry(i));
int disToSeedB = entryB.xEpsilen(_p_node_being_split->getChildEntry(i));
// choose the near one seed to add into
if(disToSeedA <= disToSeedB)
{
entryIndex_nearA.push_back(i);
}
else
{
entryIndex_nearB.push_back(i);
}
}
// then create a new node to act as BEntryIndex's father.
VNode* newNodePtr = this->createNode();
#ifdef DEBUG_VSTREE
stringstream _ss2;
_ss2 << "new Node is :[" << newNodePtr->getFileLine() << "]" << endl;
Util::logging(_ss2.str());
#endif
// the old one acts as AEntryIndex's father.
VNode* oldNodePtr = _p_node_being_split;
// if the old node is leaf, set the new node as a leaf.
if(oldNodePtr->isLeaf())
{
newNodePtr->setAsLeaf(true);
}
//add all the entries in BEntryIndex into the new node child entry array,
//and calculate the new node's entry.
for(unsigned i = 0; i < entryIndex_nearB.size(); i++)
{
if(oldNodePtr->isLeaf())
{
newNodePtr->addChildEntry(oldNodePtr->getChildEntry(entryIndex_nearB[i]), false);
}
else
{
//debug target 2
VNode* childPtr = oldNodePtr->getChild(entryIndex_nearB[i], *(this->node_buffer));
newNodePtr->addChildNode(childPtr);
}
}
newNodePtr->refreshSignature();
//label the child being removed with -1,
//and update the old node's entry.
sort(entryIndex_nearA.begin(), entryIndex_nearA.end(), less<int>());
#ifdef DEBUG_VSTREE
stringstream _ss1;
{
_ss1 << "nearA: ";
for(unsigned i = 0; i < entryIndex_nearA.size(); i++)
{
_ss1 << entryIndex_nearA[i] << " ";
}
_ss1 << endl;
_ss1 << "nearB: ";
for(unsigned i = 0; i < entryIndex_nearB.size(); i++)
{
_ss1 << entryIndex_nearB[i] << " ";
}
_ss1 << endl;
}
Util::logging(_ss1.str());
#endif
for(unsigned i = 0; i < entryIndex_nearA.size(); i++)
{
oldNodePtr->setChildEntry(i, oldNodePtr->getChildEntry(entryIndex_nearA[i]));
oldNodePtr->setChildFileLine(i, oldNodePtr->getChildFileLine(entryIndex_nearA[i]));
}
oldNodePtr->setChildNum(entryIndex_nearA.size());
oldNodePtr->refreshSignature();
int oldNode_index = oldNodePtr->getIndexInFatherNode(*(this->node_buffer));
// full node's father pointer.
VNode* oldNodeFatherPtr = oldNodePtr->getFather(*(this->node_buffer));
if(oldNodePtr->isRoot())
{
//if the old node is root,
//split the root, create a new root,
//and the tree height will be increased.
VNode* RootNewPtr = this->createNode();
//change the old root node to not-root node,
//and set the RootNew to root node.
oldNodePtr->setAsRoot(false);
RootNewPtr->setAsRoot(true);
//set the split two node(old node and new node) as the new root's child,
//and update signatures.
RootNewPtr->addChildNode(oldNodePtr);
RootNewPtr->addChildNode(newNodePtr);
RootNewPtr->refreshSignature();
//debug
// {
// stringstream _ss;
// _ss << "create new root:" << endl;
// _ss << "before swap file line, two sons are: " << oldNodePtr->getFileLine() << " " << newNodePtr->getFileLine() << endl;
// Util::logging(_ss.str());
// }
//should keep the root node always being
//at the first line(line zero) of the tree node file.
this->swapNodeFileLine(RootNewPtr, oldNodePtr);
this->height++;
//debug
// {
// stringstream _ss;
// _ss << "create new root:" << endl;
// _ss << "two sons are: " << oldNodePtr->getFileLine() << " " << newNodePtr->getFileLine() << endl;
// _ss << Signature::BitSet2str(oldNodePtr->getEntry().getEntitySig().entityBitSet) << endl;
// _ss << RootNewPtr->to_str() << endl;
// Util::logging(_ss.str());
// }
}
else
{
//if the (OldNode) is not Root,
//change the old node's signature to A's signature.
oldNodeFatherPtr->setChildEntry(oldNode_index, oldNodePtr->getEntry());
if(oldNodeFatherPtr->isFull())
{
oldNodeFatherPtr->refreshAncestorSignature(*(this->node_buffer));
this->split(oldNodeFatherPtr, newNodePtr->getEntry(), newNodePtr);
}
else
{
oldNodeFatherPtr->addChildNode(newNodePtr);
oldNodeFatherPtr->refreshAncestorSignature(*(this->node_buffer));
}
}
//debug
// if (!oldNodePtr->checkState())
// {
// stringstream _ss;
// _ss << "node " << oldNodePtr->getFileLine() << " childFileLine error. oldNode when split" << endl;
// Util::logging(_ss.str());
// }
// if (!newNodePtr->checkState())
// {
// stringstream _ss;
// _ss << "node " << newNodePtr->getFileLine() << " childFileLine error. newNode when split" << endl;
// Util::logging(_ss.str());
// }
// update the entityID2FileLineMap by these two nodes.
this->updateEntityID2FileLineMap(oldNodePtr);
this->updateEntityID2FileLineMap(newNodePtr);
}
//retrieve the candidate entity ID which signature can cover the _entity_bit_set, and add them to the _p_id_list.
void
VSTree::retrieveEntity(const EntityBitSet& _entity_bit_set, IDList* _p_id_list)
{
Util::logging("IN retrieveEntity");
EntitySig filterSig(_entity_bit_set);
#ifdef DEBUG_VSTREE
cerr << "the filter signature: " << filterSig.to_str() << endl;
#endif
queue<int> nodeQueue; //searching node file line queue.
//debug
{
stringstream _ss;
_ss << "filterSig=" << Signature::BitSet2str(filterSig.entityBitSet) << endl;
Util::logging(_ss.str());
}
const SigEntry& root_entry = (this->getRoot())->getEntry();
Util::logging("Get Root Entry");
if(root_entry.cover(filterSig))
{
nodeQueue.push(this->getRoot()->getFileLine());
Util::logging("root cover the filter_sig");
}
else
{
Util::logging("warning: root is not cover the filter_sig");
}
//debug
// {
// Util::logging(this->getRoot()->to_str());
// Util::logging("Before BFS");
// }
//using BFS algorithm to traverse the VSTree and retrieve the entry.
while (!nodeQueue.empty())
{
int currentNodeFileLine = nodeQueue.front();
nodeQueue.pop();
VNode* currentNodePtr = this->getNode(currentNodeFileLine);
int childNum = currentNodePtr->getChildNum();
//debug
// {
// std::stringstream _ss;
// _ss << "childNum of ["
// << currentNodePtr->getFileLine()
// << "] is " << childNum << endl;
//
// for (int i=0;i<childNum;i++)
// {
// _ss << currentNodePtr->getChildFileLine(i) << " ";
// }
// _ss << endl;
//
// Util::logging(_ss.str());
// }
int valid = 0;
for (int i = 0; i < childNum; i++)
{
const SigEntry& entry = currentNodePtr->getChildEntry(i);
#ifdef DEBUG_VSTREE
//cerr << "current entry: " << entry.to_str() << endl;
#endif
if (entry.cover(filterSig))
{
valid++;
if (currentNodePtr->isLeaf())
{
// if leaf node, add the satisfying entries' entity id to result list.
_p_id_list->addID(entry.getEntityId());
//debug
// {
// stringstream _ss;
// _ss << "child_" << i << " cover filter sig" << endl;
// _ss << Signature::BitSet2str(entry.getEntitySig().entityBitSet)<< endl;
// Util::logging(_ss.str());
// }
}
else
{
// if non-leaf node, add the child node pointer to the searching queue.
//VNode* childPtr = currentNodePtr->getChild(i, *(this->node_buffer));
// if non-leaf node, add the child node file line to the searching queue.
int childNodeFileLine = currentNodePtr->getChildFileLine(i);
nodeQueue.push(childNodeFileLine);
//debug
// {
// stringstream _ss;
// _ss << "child[" << childPtr->getFileLine() << "] cover filter sig" << endl;
// Util::logging(_ss.str());
// }
}
}
}
#ifdef DEBUG_VSTREE
//cerr << "child num: " << childNum << " valid num: " << valid << endl;
#endif
}
Util::logging("OUT retrieveEntity");
}
//the _entry_index in _child is to be removed.
//node can only be deleted in this function.
void
VSTree::coalesce(VNode* _child, int _entry_index)
{
#ifdef DEBUG
cout << "coalesce happen" <<endl;
#endif
//found the father and index
VNode* _father = _child->getFather(*(this->node_buffer));
int cn = _child->getChildNum();
if(_father == NULL) //this is already root
{
//NOTICE:when root is leaf, at least one key, otherwise the tree is empty
//But when root is internal, at least two key, if one key then shrink
//(1-key internal root is not permitted)
//
//Notice that leaf-root case has been discussed in upper function removeEntry()
//so here the root must be internal node
_child->removeChild(_entry_index);
if(cn == 2)
{
//only one key after remove, shrink root
VNode* newRoot = _child->getChild(0, *(this->node_buffer));
newRoot->setAsRoot(true);
cout<<"shrink root in coalesce() -- to swap node file"<<endl;
this->swapNodeFileLine(newRoot, _child);
this->root_file_line = newRoot->getFileLine();
this->height--;
this->removeNode(_child);
}
return;
}
if(cn > VNode::MIN_CHILD_NUM)
{
cout<<"no need to move or union in coalesce()"<<endl;
_child->removeChild(_entry_index);
_child->refreshAncestorSignature(*(this->node_buffer));
return;
}
int fn = _father->getChildNum();
int i, _child_index = -1;
for (i = 0; i < fn; ++i)
{
if (_father->getChildFileLine(i) == _child->getFileLine())
{
break;
}
}
if(i == fn)
{
cerr << "not found the leaf node in VSTree::coalesce()" << endl;
return;
}
else
{
_child_index = i;
}
//_child->removeChild(_entry_index);
//_child->setChildNum(cn);
//NOTICE:we do not consider the efficiency here, so just ensure the operation is right
//BETTER:find good way to ensure signatures are separated(maybe similar ones together)
int ccase = 0;
VNode* p = NULL;
int n = 0;
if(_child_index < fn - 1)
{
p = _father->getChild(_child_index+1, *(this->node_buffer));
n = p->getChildNum();
if(n > VNode::MIN_CHILD_NUM)
{
ccase = 2;
}
else
{
ccase = 1;
}
}
if(_child_index > 0)
{
VNode* tp = _father->getChild(_child_index-1, *(this->node_buffer));
int tn = tp->getChildNum();
if(ccase < 2)
{
if(ccase == 0)
ccase = 3;
if(tn > VNode::MIN_CHILD_NUM)
ccase = 4;
}
if(ccase > 2)
{
p = tp;
n = tn;
}
}
VNode* tmp = NULL;
int child_no = _child->getFileLine();
int father_no = _father->getFileLine();
#ifdef DEBUG
if(ccase == 1 || ccase == 3)
{
cout << "union happened" << endl;
}
else if(ccase == 2 || ccase == 4)
{
cout << "move happened" << endl;
}
cout<< "father num: "<<fn<<" child num: "<<cn<<" neighbor num: "<<n<<endl;
cout<<"child file line "<<child_no<<endl;
cout<<"neighbor file line "<<p->getFileLine()<<endl;
#endif
switch(ccase)
{
case 1: //union right to this
if(_child->isLeaf())
{
_child->setChildFileLine(_entry_index, p->getChildFileLine(0));
_child->setChildEntry(_entry_index, p->getChildEntry(0));
for(int i = 1; i < n; ++i)
{
_child->setChildFileLine(cn+i-1, p->getChildFileLine(i));
_child->addChildEntry(p->getChildEntry(i));
}
}
else
{
_child->setChildFileLine(_entry_index, p->getChildFileLine(0));
_child->setChildEntry(_entry_index, p->getChildEntry(0));
tmp = p->getChild(0, *(this->node_buffer));
tmp->setFatherFileLine(child_no);
for(int i = 1; i < n; ++i)
{
tmp = p->getChild(i, *(this->node_buffer));
//cout<<i<<" "<<_child->getChildNum()<<endl;
_child->addChildNode(tmp);
//_child->setChildNum(cn+i);
//tmp->setFatherFileLine(child_no);
}
}
this->removeNode(p);
_child->refreshSignature();
//recursive:to remove child index+1 in father
this->coalesce(_father, _child_index+1);
break;
case 2: //move one from right
_child->setChildFileLine(_entry_index, p->getChildFileLine(n-1));
_child->setChildEntry(_entry_index, p->getChildEntry(n-1));
_child->refreshSignature();
if(!_child->isLeaf())
{
tmp = p->getChild(n-1, *(this->node_buffer));
tmp->setFatherFileLine(child_no);
}
p->removeChild(n-1);
p->refreshSignature();
_father->setChildEntry(_child_index, _child->getEntry());
_father->setChildEntry(_child_index+1, p->getEntry());
_father->refreshAncestorSignature(*(this->node_buffer));
break;
case 3: //union left to this
if(_child->isLeaf())
{
_child->setChildFileLine(_entry_index, p->getChildFileLine(0));
_child->setChildEntry(_entry_index, p->getChildEntry(0));
for(int i = 1; i < n; ++i)
{
_child->setChildFileLine(cn+i-1, p->getChildFileLine(i));
_child->addChildEntry(p->getChildEntry(i));
}
}
else
{
_child->setChildFileLine(_entry_index, p->getChildFileLine(0));
_child->setChildEntry(_entry_index, p->getChildEntry(0));
tmp = p->getChild(0, *(this->node_buffer));
tmp->setFatherFileLine(child_no);
for(int i = 1; i < n; ++i)
{
tmp = p->getChild(i, *(this->node_buffer));
//cout<<i<<" "<<_child->getChildNum()<<endl;
_child->addChildNode(tmp);
//_child->setChildNum(cn+i);
//tmp->setFatherFileLine(child_no);
}
}
this->removeNode(p);
_child->refreshSignature();
//recursive:to remove child index-1 in father
this->coalesce(_father, _child_index-1);
break;
case 4: //move one from left
_child->setChildFileLine(_entry_index, p->getChildFileLine(n-1));
_child->setChildEntry(_entry_index, p->getChildEntry(n-1));
_child->refreshSignature();
if(!_child->isLeaf())
{
VNode* tmp = p->getChild(n-1, *(this->node_buffer));
tmp->setFatherFileLine(child_no);
}
p->removeChild(n-1);
p->refreshSignature();
_father->setChildEntry(_child_index, _child->getEntry());
_father->setChildEntry(_child_index-1, p->getEntry());
_father->refreshAncestorSignature(*(this->node_buffer));
break;
default:
cout << "error in coalesce: Invalid case!";
break;
}
//BETTER:this maybe very costly because many entity no need to update
if(_child->isLeaf())
{
this->updateEntityID2FileLineMap(_child);
if(ccase == 2 || ccase == 4)
{
this->updateEntityID2FileLineMap(p);
}
}
}
//remove an existed Entry(_entity_id) from VSTree
bool
VSTree::removeEntry(int _entity_id)
{
VNode* leafNodePtr = this->getLeafNodeByEntityID(_entity_id);
if (leafNodePtr == NULL)
{
cerr<< "error, can not find the mapping leaf node. @VSTree::removeEntry" << endl;
return false;
}
// seek the entry index of the leaf node.
int entryIndex = -1;
int childNum = leafNodePtr->getChildNum();
//cout<<"root file line: "<<this->root_file_line<<" "<<"max nid num: "<<this->max_nid_alloc<<endl;
//cout<<"node num: "<<this->node_num<<" "<<"file line: "<<leafNodePtr->getFileLine()<<" "<<"child num: "<<childNum<<endl;
for(int i = 0; i < childNum; i++)
{
if(leafNodePtr->getChildEntry(i).getEntityId() == _entity_id)
{
entryIndex = i;
break;
}
}
if(entryIndex == -1)
{
cerr << "error, can not find the entry in leaf node. @VSTree::removeEntry" << endl;
return false;
}
//BETTER?:consider up->bopttom to deal, not find leaf and recursively
if(leafNodePtr->isRoot())
{
if(childNum == 1)
{ //the tree is empty now
leafNodePtr->removeChild(entryIndex);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
this->removeNode(leafNodePtr);
this->root_file_line = -1;
this->height = 0;
this->entry_num = 0;
this->node_num = 0;
}
else
{
leafNodePtr->removeChild(entryIndex);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
}
}
else
{
if(childNum <= VNode::MIN_CHILD_NUM)
{
//cerr << "In VSTree::remove() -- the node is invalid" << endl;
//TODO+BETTER:this may search again, too costly
//VNode* fatherNodePtr = leafNodePtr->getFather(*(this->node_buffer));
////int index = leafNodePtr->getIndexInFatherNode(*(this->node_buffer));
//int n = fatherNodePtr->getChildNum();
//for (int i = 0; i < n; ++i)
//{
//if (fatherNodePtr->getChildFileLine(i) == leafNodePtr->getFileLine())
//{
//this->coalesce(fatherNodePtr, i, leafNodePtr, entryIndex);
//break;
//}
//}
//return false;
this->coalesce(leafNodePtr, entryIndex);
}
else
{
leafNodePtr->removeChild(entryIndex);
leafNodePtr->refreshAncestorSignature(*(this->node_buffer));
}
}
this->entry_num--;
this->entityID2FileLineMap.erase(_entity_id);
//NOTICE:insert is costly but can keep balance.
//However, remove is not too costly but can not keep balance at all.
//And remove maybe error if without coalesce!
//If remove and insert are both wonderful, update/replace can both be balanced using delete and insert again.
//(not care the balance now, if insert/delete many times, rebuilding is suggested)
//we do not consider the situation which the leaf node is to be empty by now...
//in a better way, if the leaf node is empty after removing entry, we should delete it. and recursively judge whether its
//father is empty, and delete its father node if true. to make the VSTree more balanced, we should combine two nodes if
//their child number are less than the MIN_CHILD_NUM. when deleting one node from the tree, we should also remove it from
//tree node file in hard disk by doing some operations on the node_buffer.
return true;
}
void VSTree::split(VNode* _p_node_being_split, const SigEntry& _insert_entry, VNode* _p_insert_node)
{
//debug
// {
// stringstream _ss;
// _ss << "**********************split happen at "
// << _p_node_being_split->getFileLine() << endl;
// _ss << _p_node_being_split->to_str() << endl;
// Util::logging(_ss.str());
// }
// first, add the new child node(if not leaf) or child entry(if leaf) to the full node.
bool just_insert_entry = (_p_insert_node == NULL);
if (just_insert_entry)
{
_p_node_being_split->addChildEntry(_insert_entry, true);
}
else
{
_p_node_being_split->addChildNode(_p_insert_node, true);
}
SigEntry entryA, entryB;
/* two seeds to generate two new nodes.
* seedA kernel: the SigEntry with the minimal count of signature.
* seedB kernel: the SigEntry with the second minimal count of signature.
* */
int minCount = 0; // record the minimal signature count.
int entryA_index = 0; // record the seedA kernel index.
for (int i=0;i<VNode::MAX_CHILD_NUM;i++)
{
int currentCount = (int) _p_node_being_split->getChildEntry(i).getSigCount();
if (minCount < currentCount)
{
minCount = currentCount;
entryA_index = i;
}
}
entryA = _p_node_being_split->getChildEntry(entryA_index);
minCount = 0;
int entryB_index = 0; // record the seedB kernel index.
for (int i=0;i<VNode::MAX_CHILD_NUM;i++)
{
int currentCount = entryA.xEpsilen(_p_node_being_split->getChildEntry(i));
if (i != entryA_index && minCount <= currentCount)
{
minCount = currentCount;
entryB_index = i;
}
}
entryB = _p_node_being_split->getChildEntry(entryB_index);
// AEntryIndex: the entry index near seedA.
// BEntryIndex: the entry index near seedB.
std::vector<int> entryIndex_nearA, entryIndex_nearB;
entryIndex_nearA.clear();
entryIndex_nearB.clear();
entryIndex_nearA.push_back(entryA_index);
entryIndex_nearB.push_back(entryB_index);
/* just tmp variables, for more readibility */
int nearA_max_size, nearB_max_size;
bool nearA_tooSmall, nearB_tooSmall;
for (int i=0;i<VNode::MAX_CHILD_NUM;i++)
{
if (i == entryA_index || i == entryB_index) continue;
/* should guarantee that each new node has at least MIN_CHILD_NUM children. */
nearA_max_size = VNode::MAX_CHILD_NUM - entryIndex_nearB.size();
nearA_tooSmall = (nearA_max_size <= VNode::MIN_CHILD_NUM);
if (nearA_tooSmall)
{
for (;i<VNode::MAX_CHILD_NUM;i++)
{
if (i == entryA_index || i == entryB_index) continue;
entryIndex_nearA.push_back(i);
}
break;
}
nearB_max_size = VNode::MAX_CHILD_NUM - entryIndex_nearA.size();
nearB_tooSmall = (nearB_max_size <= VNode::MIN_CHILD_NUM);
if (nearB_tooSmall)
{
for (;i<VNode::MAX_CHILD_NUM;i++)
{
if (i == entryA_index || i == entryB_index) continue;
entryIndex_nearB.push_back(i);
}
break;
}
/* calculate the distance from
* the i-th child entry signature to seedA(or seedB).*/
/*debug target 1*/
int disToSeedA = entryA.xEpsilen(_p_node_being_split->getChildEntry(i));
int disToSeedB = entryB.xEpsilen(_p_node_being_split->getChildEntry(i));
// choose the near one seed to add into
if (disToSeedA <= disToSeedB)
{
entryIndex_nearA.push_back(i);
}
else
{
entryIndex_nearB.push_back(i);
}
}
// then create a new node to act as BEntryIndex's father.
VNode* newNodePtr = this->createNode();
//debug
// {
// stringstream _ss;
// _ss << "new Node is :[" << newNodePtr->getFileLine() << "]" << endl;
// Util::logging(_ss.str());
// }
// the old one acts as AEntryIndex's father.
VNode* oldNodePtr = _p_node_being_split;
// if the old node is leaf, set the new node as a leaf.
if (oldNodePtr->isLeaf())
{
newNodePtr->setAsLeaf(true);
}
/* add all the entries in BEntryIndex into the new node child entry array,
and calculate the new node's entry.*/
for (unsigned i=0;i<entryIndex_nearB.size();i++)
{
if (oldNodePtr->isLeaf())
{
newNodePtr->addChildEntry(oldNodePtr->getChildEntry(entryIndex_nearB[i]), false);
}
else
{
/*debug target 2*/
VNode* childPtr = oldNodePtr->getChild(entryIndex_nearB[i], *(this->node_buffer));
newNodePtr->addChildNode(childPtr);
}
}
newNodePtr->refreshSignature();
/* label the child being removed with -1,
* and update the old node's entry.*/
std::sort(entryIndex_nearA.begin(), entryIndex_nearA.end(), less<int>());
//debug
// {
// stringstream _ss;
// {
// _ss << "nearA: ";
// for(int i = 0; i < entryIndex_nearA.size(); i ++)
// {
// _ss << entryIndex_nearA[i] << " ";
// }
// _ss << endl;
//
// _ss << "nearB: ";
// for(int i = 0; i < entryIndex_nearB.size(); i ++)
// {
// _ss << entryIndex_nearB[i] << " ";
// }
// _ss << endl;
// }
// Util::logging(_ss.str());
// }
for (unsigned i=0;i<entryIndex_nearA.size();i++)
{
oldNodePtr->setChildEntry(i, oldNodePtr->getChildEntry(entryIndex_nearA[i]));
oldNodePtr->setChildFileLine(i, oldNodePtr->getChildFileLine(entryIndex_nearA[i]));
}
oldNodePtr->setChildNum(entryIndex_nearA.size());
oldNodePtr->refreshSignature();
int oldNode_index = oldNodePtr->getIndexInFatherNode(*(this->node_buffer));
// full node's father pointer.
VNode* oldNodeFatherPtr = oldNodePtr->getFather(*(this->node_buffer));
if (oldNodePtr->isRoot())
{
/* if the old node is root,
* split the root, create a new root,
* and the tree height will be increased.*/
VNode* RootNewPtr = this->createNode();
/* change the old root node to not-root node,
* and set the RootNew to root node.*/
oldNodePtr->setAsRoot(false);
RootNewPtr->setAsRoot(true);
/* set the split two node(old node and new node) as the new root's child,
* and update signatures.*/
RootNewPtr->addChildNode(oldNodePtr);
RootNewPtr->addChildNode(newNodePtr);
RootNewPtr->refreshSignature();
//debug
// {
// stringstream _ss;
// _ss << "create new root:" << endl;
// _ss << "before swap file line, two sons are: " << oldNodePtr->getFileLine() << " " << newNodePtr->getFileLine() << endl;
// Util::logging(_ss.str());
// }
/* should keep the root node always being
* at the first line(line zero) of the tree node file.*/
this->swapNodeFileLine(RootNewPtr, oldNodePtr);
this->height ++;
//debug
// {
// stringstream _ss;
// _ss << "create new root:" << endl;
// _ss << "two sons are: " << oldNodePtr->getFileLine() << " " << newNodePtr->getFileLine() << endl;
// _ss << Signature::BitSet2str(oldNodePtr->getEntry().getEntitySig().entityBitSet) << endl;
// _ss << RootNewPtr->to_str() << endl;
// Util::logging(_ss.str());
// }
}
else
{
/* if the (OldNode) is not Root,
* change the old node's signature to A's signature.*/
oldNodeFatherPtr->setChildEntry(oldNode_index, oldNodePtr->getEntry());
if (oldNodeFatherPtr->isFull())
{
oldNodeFatherPtr->refreshAncestorSignature(*(this->node_buffer));
this->split(oldNodeFatherPtr, newNodePtr->getEntry(), newNodePtr);
}
else
{
oldNodeFatherPtr->addChildNode(newNodePtr);
oldNodeFatherPtr->refreshAncestorSignature(*(this->node_buffer));
}
}
//debug
// if (!oldNodePtr->checkState())
// {
// stringstream _ss;
// _ss << "node " << oldNodePtr->getFileLine() << " childFileLine error. oldNode when split" << endl;
// Util::logging(_ss.str());
// }
// if (!newNodePtr->checkState())
// {
// stringstream _ss;
// _ss << "node " << newNodePtr->getFileLine() << " childFileLine error. newNode when split" << endl;
// Util::logging(_ss.str());
// }
// update the entityID2FileLineMap by these two nodes.
this->updateEntityID2FileLineMap(oldNodePtr);
this->updateEntityID2FileLineMap(newNodePtr);
}