void
PhraseDictionaryMemory::
GetTargetPhraseCollectionBatch(const InputPathList &inputPathQueue) const
{
InputPathList::const_iterator iter;
for (iter = inputPathQueue.begin(); iter != inputPathQueue.end(); ++iter) {
InputPath &node = **iter;
const Phrase &phrase = node.GetPhrase();
const InputPath *prevPath = node.GetPrevPath();
const PhraseDictionaryNodeMemory *prevPtNode = NULL;
if (prevPath) {
prevPtNode = static_cast<const PhraseDictionaryNodeMemory*>(prevPath->GetPtNode(*this));
} else {
// Starting subphrase.
assert(phrase.GetSize() == 1);
prevPtNode = &GetRootNode();
}
if (prevPtNode) {
Word lastWord = phrase.GetWord(phrase.GetSize() - 1);
lastWord.OnlyTheseFactors(m_inputFactors);
const PhraseDictionaryNodeMemory *ptNode = prevPtNode->GetChild(lastWord);
if (ptNode) {
const TargetPhraseCollection &targetPhrases = ptNode->GetTargetPhraseCollection();
node.SetTargetPhrases(*this, &targetPhrases, ptNode);
} else {
node.SetTargetPhrases(*this, NULL, NULL);
}
}
}
}
void PhraseDictionaryOnDisk::GetTargetPhraseCollectionBatch(const InputPathList &inputPathQueue) const
{
InputPathList::const_iterator iter;
for (iter = inputPathQueue.begin(); iter != inputPathQueue.end(); ++iter) {
InputPath &inputPath = **iter;
GetTargetPhraseCollectionBatch(inputPath);
}
}
void PhraseDictionary::GetTargetPhraseCollectionBatch(const InputPathList &phraseDictionaryQueue) const
{
InputPathList::const_iterator iter;
for (iter = phraseDictionaryQueue.begin(); iter != phraseDictionaryQueue.end(); ++iter) {
InputPath &node = **iter;
const Phrase &phrase = node.GetPhrase();
const TargetPhraseCollection *targetPhrases = this->GetTargetPhraseCollectionLEGACY(phrase);
node.SetTargetPhrases(*this, targetPhrases, NULL);
}
}
void PhraseDictionaryOnDisk::GetTargetPhraseCollectionBatch(const InputPathList &phraseDictionaryQueue) const
{
OnDiskPt::OnDiskWrapper &wrapper = const_cast<OnDiskPt::OnDiskWrapper&>(GetImplementation());
InputPathList::const_iterator iter;
for (iter = phraseDictionaryQueue.begin(); iter != phraseDictionaryQueue.end(); ++iter) {
InputPath &node = **iter;
const Phrase &phrase = node.GetPhrase();
const InputPath *prevNode = node.GetPrevNode();
const OnDiskPt::PhraseNode *prevPtNode = NULL;
if (prevNode) {
prevPtNode = static_cast<const OnDiskPt::PhraseNode*>(prevNode->GetPtNode(*this));
} else {
// Starting subphrase.
assert(phrase.GetSize() == 1);
prevPtNode = &wrapper.GetRootSourceNode();
}
if (prevPtNode) {
Word lastWord = phrase.GetWord(phrase.GetSize() - 1);
lastWord.OnlyTheseFactors(m_inputFactors);
OnDiskPt::Word *lastWordOnDisk = wrapper.ConvertFromMoses(m_input, lastWord);
if (lastWordOnDisk == NULL) {
// OOV according to this phrase table. Not possible to extend
node.SetTargetPhrases(*this, NULL, NULL);
} else {
const OnDiskPt::PhraseNode *ptNode = prevPtNode->GetChild(*lastWordOnDisk, wrapper);
if (ptNode) {
vector<float> weightT = StaticData::Instance().GetWeights(this);
OnDiskPt::Vocab &vocab = wrapper.GetVocab();
const OnDiskPt::TargetPhraseCollection *targetPhrasesOnDisk = ptNode->GetTargetPhraseCollection(m_tableLimit, wrapper);
TargetPhraseCollection *targetPhrases
= targetPhrasesOnDisk->ConvertToMoses(m_input, m_output, *this, weightT, vocab, false);
node.SetTargetPhrases(*this, targetPhrases, ptNode);
delete targetPhrasesOnDisk;
} else {
node.SetTargetPhrases(*this, NULL, NULL);
}
delete lastWordOnDisk;
}
}
}
}
void SkeletonPT::GetTargetPhraseCollectionBatch(const InputPathList &phraseDictionaryQueue) const
{
InputPathList::const_iterator iter;
for (iter = phraseDictionaryQueue.begin(); iter != phraseDictionaryQueue.end(); ++iter) {
InputPath &inputPath = **iter;
TargetPhrase *tp = CreateTargetPhrase(inputPath.GetPhrase());
TargetPhraseCollection *tpColl = new TargetPhraseCollection();
tpColl->Add(tp);
m_allTPColl.push_back(tpColl);
inputPath.SetTargetPhrases(*this, tpColl, NULL);
}
}
void PhraseDictionaryOnDisk::GetTargetPhraseCollectionBatch(const InputPathList &inputPathQueue) const
{
InputPathList::const_iterator iter;
for (iter = inputPathQueue.begin(); iter != inputPathQueue.end(); ++iter) {
InputPath &inputPath = **iter;
GetTargetPhraseCollectionBatch(inputPath);
}
// delete nodes that's been saved
for (iter = inputPathQueue.begin(); iter != inputPathQueue.end(); ++iter) {
InputPath &inputPath = **iter;
const OnDiskPt::PhraseNode *ptNode = static_cast<const OnDiskPt::PhraseNode*>(inputPath.GetPtNode(*this));
delete ptNode;
}
}
void PhraseDictionaryTransliteration::GetTargetPhraseCollectionBatch(const InputPathList &inputPathQueue) const
{
InputPathList::const_iterator iter;
for (iter = inputPathQueue.begin(); iter != inputPathQueue.end(); ++iter) {
InputPath &inputPath = **iter;
if (!SatisfyBackoff(inputPath)) {
continue;
}
const Phrase &sourcePhrase = inputPath.GetPhrase();
if (sourcePhrase.GetSize() != 1) {
// only translit single words. A limitation of the translit script
continue;
}
GetTargetPhraseCollection(inputPath);
}
}
//Find top n translations of source, and send them to output
static void outputTopN(const StringPiece& sourcePhraseString, PhraseDictionary* phraseTable, const std::vector<FactorType> &input, ostream& out) {
//get list of target phrases
Phrase sourcePhrase;
sourcePhrase.CreateFromString(Input,input,sourcePhraseString,NULL);
InputPath inputPath(sourcePhrase, NonTerminalSet(), WordsRange(0,sourcePhrase.GetSize()-1),NULL,NULL);
InputPathList inputPaths;
inputPaths.push_back(&inputPath);
phraseTable->GetTargetPhraseCollectionBatch(inputPaths);
const TargetPhraseCollection* targetPhrases = inputPath.GetTargetPhrases(*phraseTable);
//print phrases
const std::vector<FactorType>& output = StaticData::Instance().GetOutputFactorOrder();
if (targetPhrases) {
//if (targetPhrases->GetSize() > 10) cerr << "src " << sourcePhrase << " tgt count " << targetPhrases->GetSize() << endl;
for (TargetPhraseCollection::const_iterator i = targetPhrases->begin(); i != targetPhrases->end(); ++i) {
const TargetPhrase* targetPhrase = *i;
out << sourcePhrase.GetStringRep(input);
out << " ||| ";
out << targetPhrase->GetStringRep(output);
out << " ||| ";
const ScoreComponentCollection scores = targetPhrase->GetScoreBreakdown();
vector<float> phraseScores = scores.GetScoresForProducer(phraseTable);
for (size_t j = 0; j < phraseScores.size(); ++j) {
out << exp(phraseScores[j]) << " ";
}
out << "||| ";
const AlignmentInfo& align = targetPhrase->GetAlignTerm();
for (AlignmentInfo::const_iterator j = align.begin(); j != align.end(); ++j) {
out << j->first << "-" << j->second << " ";
}
out << endl;
}
}
}