/**
* Given a wordset create UYiV yields crossing them and return the best one
* @param ws set of words
* @return best UYiV
*/
shared_ptr<UYiV> Dencoder::getBestUYiV(unordered_set< shared_ptr<Word> > * ws) const
{
shared_ptr<Sequence> seq = getSeq();
shared_ptr<Word> u;
shared_ptr<Word> v;
list <std::pair<int,int> > * wordPosPairs;
WordList* y;
shared_ptr<UYiV> newUYV;
shared_ptr<UYiV> best = NULL;
for(unordered_set< shared_ptr<Word> >::iterator it1 = ws->begin(); it1!= ws->end(); ++it1)
{
u = *it1;
for(unordered_set< shared_ptr<Word> >::iterator it2 = ws->begin(); it2!= ws->end(); ++it2)
{
v = *it2;
//wordPosPairs could be NULL
wordPosPairs = seq->wordPosPairList(u,v);
if(wordPosPairs == NULL || wordPosPairs->size()>=2)
{
y = seq->obtainWordsInside(wordPosPairs, u->getSize());
newUYV = make_shared<UYiV>(u, v, y, wordPosPairs);
if(!(wordPosPairs == NULL || y->getSize() == 0))
{
if(score(best)<=score(newUYV))
{
// cout << "Score best: " << score(best) << endl;
best = newUYV;
// cout << "Score New: " << score(newUYV) << endl;
}
}
}
else
{
delete wordPosPairs;
}
// it = ls.erase(it);
}
}
return best;
}
/**
* Given the current best Yield and a word, try to find a Yield with the suffix
* and prefix of this word that's better than the current one.
* @param best current best yield
* @param mr word
*/
void Dencoder::getUVFromMaximalRepeat(shared_ptr<UYiV> &best, shared_ptr<Word> mr)
{
shared_ptr<Word> u;
shared_ptr<Word> v;
shared_ptr<Sequence> seq = getSeq();
int size = mr->getSize();
shared_ptr<UYiV> newUYV;
WordList* y;
list <std::pair<int,int> > * wordPosPairs;
for(int i = 1; i<size; ++i)
{
u = seq->createSubWord(mr, 0, i);
for(int j = i; j<size; ++j)
{
v = seq->createSubWord(mr, j, size);
wordPosPairs = seq->wordPosPairList(u,v);
clock_t t;
t = clock();
y = seq->obtainWordsInside(wordPosPairs, u->getSize());
t = clock() - t;
printf("obtainWordsInside: It took me %ld clicks (%f seconds).\n",t,((float)t)/CLOCKS_PER_SEC);
newUYV = make_shared<UYiV>(u, v, y, wordPosPairs);
if(!(wordPosPairs == NULL || y->getSize() == 0))
{
if(score(best)<score(newUYV))
{
// cout << "Score best: " << score(best) << endl;
best = newUYV;
// cout << "Score New: " << score(newUYV) << endl;
}
}
}
// cout << "Score primer palabra: " << score(best) << endl;
}
}
/**
* replace the given yield and generate the productions needed
* @param uyv
*/
void Dencoder::replaceYield(shared_ptr<UYiV> uyv)
{
shared_ptr<Word> u = uyv->getU();
int * uInt = u->getSubSequence();
shared_ptr<Word> v = uyv->getV();
int * vInt = v->getSubSequence();
int uSize = u->getSize();
int vSize = v->getSize();
int ySize = 0;
int *yInt;
int currentPos = 0;
WordList* y = uyv->getYList();
assert(y!= NULL);
// set<shared_ptr<Word> > * yWords = y->getWordSet();
vector< pair<shared_ptr<Word>, int> > vecWord = y->getWordVec();
int lenghtInside = y->sumWordsLengthWithoutRepeats();
int I = symbol_->getNonTerminal();
int N = symbol_->getNonTerminal();
std::queue<int> initialQ;
prodElectionOrder_[I] = initialQ ;
replaceNonTerminal(uyv, N);
// U size + I NonTerminal + v Size
int * nRHS = new int[uSize+1+vSize];
// Rule for the insides + y->size() are the pipes
int * iRHS = new int[lenghtInside+vecWord.size()];
//Add u symbols
for(int i = 0; i < uSize; ++i)
{
nRHS[i] = uInt[i];
}
//Add I nonTerminal
nRHS[uSize] = I;
//Add v symbols
for(int i = 0; i < vSize; ++i)
{
nRHS[uSize+1+i] = vInt[i];
}
//Add y simbols
for(vector< pair<shared_ptr<Word>, int> >::iterator it = vecWord.begin(); it != vecWord.end(); ++it)
{
shared_ptr<Word> y = (*it).first;
yInt = y->getSubSequence();
ySize = y->getSize();
//cannot be 0
// assert(ySize!=0);
for (int i = 0; i < ySize; ++i)
{
iRHS[currentPos+i] = yInt[i];
}
iRHS[currentPos + ySize] = 2;
currentPos += ySize+1;
// cout << endl << *y << endl << (*it).second << endl;
}
// int sProdSize = getSeq()->getSize() - y->sumWordsLengthWithRepeats() - n*(uSize+vSize) + n;
//Create Productions:
shared_ptr<Production> p1 = make_shared<Production>(I, iRHS, lenghtInside+y->getWordSet()->size()-1);
shared_ptr<Production> p2 = make_shared<Production>(N, nRHS, uSize+1+vSize);
grammar_->addProd(p1);
grammar_->addProd(p2);
}