OnDiskPt::PhrasePtr Tokenize(SourcePhrase &sourcePhrase, TargetPhrase &targetPhrase, const std::string &lineStr, OnDiskWrapper &onDiskWrapper, int numScores, vector<float> &misc)
{
char line[lineStr.size() + 1];
strcpy(line, lineStr.c_str());
stringstream sparseFeatures, property;
size_t scoreInd = 0;
// MAIN LOOP
size_t stage = 0;
/* 0 = source phrase
1 = target phrase
2 = scores
3 = align
4 = count
7 = properties
*/
char *tok = strtok (line," ");
OnDiskPt::PhrasePtr out(new Phrase());
while (tok != NULL) {
if (0 == strcmp(tok, "|||")) {
++stage;
} else {
switch (stage) {
case 0: {
WordPtr w = Tokenize(sourcePhrase, tok, true, true, onDiskWrapper, 1);
if (w != NULL)
out->AddWord(w);
break;
}
case 1: {
Tokenize(targetPhrase, tok, false, true, onDiskWrapper, 0);
break;
}
case 2: {
float score = Moses::Scan<float>(tok);
targetPhrase.SetScore(score, scoreInd);
++scoreInd;
break;
}
case 3: {
//targetPhrase.Create1AlignFromString(tok);
targetPhrase.CreateAlignFromString(tok);
break;
}
case 4: {
// store only the 3rd one (rule count)
float val = Moses::Scan<float>(tok);
misc[0] = val;
break;
}
case 5: {
// sparse features
sparseFeatures << tok << " ";
break;
}
case 6: {
property << tok << " ";
break;
}
default:
cerr << "ERROR in line " << line << endl;
assert(false);
break;
}
}
tok = strtok (NULL, " ");
} // while (tok != NULL)
assert(scoreInd == numScores);
targetPhrase.SetSparseFeatures(Moses::Trim(sparseFeatures.str()));
targetPhrase.SetProperty(Moses::Trim(property.str()));
targetPhrase.SortAlign();
return out;
} // Tokenize()
void ChartParserUnknown::Process(const Word &sourceWord, const WordsRange &range, ChartParserCallback &to)
{
// unknown word, add as trans opt
const StaticData &staticData = StaticData::Instance();
const UnknownWordPenaltyProducer &unknownWordPenaltyProducer = UnknownWordPenaltyProducer::Instance();
size_t isDigit = 0;
if (staticData.GetDropUnknown()) {
const Factor *f = sourceWord[0]; // TODO hack. shouldn't know which factor is surface
const StringPiece s = f->GetString();
isDigit = s.find_first_of("0123456789");
if (isDigit == string::npos)
isDigit = 0;
else
isDigit = 1;
// modify the starting bitmap
}
Phrase* unksrc = new Phrase(1);
unksrc->AddWord() = sourceWord;
Word &newWord = unksrc->GetWord(0);
newWord.SetIsOOV(true);
m_unksrcs.push_back(unksrc);
//TranslationOption *transOpt;
if (! staticData.GetDropUnknown() || isDigit) {
// loop
const UnknownLHSList &lhsList = staticData.GetUnknownLHS();
UnknownLHSList::const_iterator iterLHS;
for (iterLHS = lhsList.begin(); iterLHS != lhsList.end(); ++iterLHS) {
const string &targetLHSStr = iterLHS->first;
float prob = iterLHS->second;
// lhs
//const Word &sourceLHS = staticData.GetInputDefaultNonTerminal();
Word *targetLHS = new Word(true);
targetLHS->CreateFromString(Output, staticData.GetOutputFactorOrder(), targetLHSStr, true);
UTIL_THROW_IF2(targetLHS->GetFactor(0) == NULL, "Null factor for target LHS");
// add to dictionary
TargetPhrase *targetPhrase = new TargetPhrase();
Word &targetWord = targetPhrase->AddWord();
targetWord.CreateUnknownWord(sourceWord);
// scores
float unknownScore = FloorScore(TransformScore(prob));
targetPhrase->GetScoreBreakdown().Assign(&unknownWordPenaltyProducer, unknownScore);
targetPhrase->Evaluate(*unksrc);
targetPhrase->SetTargetLHS(targetLHS);
targetPhrase->SetAlignmentInfo("0-0");
if (staticData.IsDetailedTreeFragmentsTranslationReportingEnabled()) {
targetPhrase->SetProperty("Tree","[ " + (*targetLHS)[0]->GetString().as_string() + " "+sourceWord[0]->GetString().as_string()+" ]");
}
// chart rule
to.AddPhraseOOV(*targetPhrase, m_cacheTargetPhraseCollection, range);
} // for (iterLHS
} else {
// drop source word. create blank trans opt
float unknownScore = FloorScore(-numeric_limits<float>::infinity());
TargetPhrase *targetPhrase = new TargetPhrase();
// loop
const UnknownLHSList &lhsList = staticData.GetUnknownLHS();
UnknownLHSList::const_iterator iterLHS;
for (iterLHS = lhsList.begin(); iterLHS != lhsList.end(); ++iterLHS) {
const string &targetLHSStr = iterLHS->first;
//float prob = iterLHS->second;
Word *targetLHS = new Word(true);
targetLHS->CreateFromString(Output, staticData.GetOutputFactorOrder(), targetLHSStr, true);
UTIL_THROW_IF2(targetLHS->GetFactor(0) == NULL, "Null factor for target LHS");
targetPhrase->GetScoreBreakdown().Assign(&unknownWordPenaltyProducer, unknownScore);
targetPhrase->Evaluate(*unksrc);
targetPhrase->SetTargetLHS(targetLHS);
// chart rule
to.AddPhraseOOV(*targetPhrase, m_cacheTargetPhraseCollection, range);
}
}
}
