wordID_t Vocab::GetWordID(const std::string& word_str) {
FactorList factors;
factors.push_back(0);
Word word;
word.CreateFromString(Input, factors, word_str, false);
return GetWordID(word);
}
std::string Word::GetString(const FactorList &factorTypes) const
{
assert(factorTypes.size());
std::stringstream ret;
ret << m_factors[factorTypes[0]]->GetString();
for (size_t i = 1; i < factorTypes.size(); ++i) {
FactorType factorType = factorTypes[i];
ret << "|" << m_factors[factorType];
}
return ret.str();
}
const Word Vocab::InitSpecialWord( const std::string& word_str)
{
FactorList factors;
factors.push_back(0); // store the special word string as the first factor
Word word;
// define special word as Input word with one factor and isNonTerminal=false
word.CreateFromString( Input, factors, word_str, false ); // Input is enum defined in ../typedef.h
// TODO not sure if this will work properly:
// - word comparison can fail because the last parameter (isNonTerminal)
// in function CreateFromString may not match properly created words
// - special word is Input word but what about Output words?
// - currently Input/Output variable is not stored in class Word, but in the future???
return word;
}
void PhraseBoundaryFeature::AddFeatures(
const Word* leftWord, const Word* rightWord, const FactorList& factors, const string& side,
ScoreComponentCollection* scores) const
{
for (size_t i = 0; i < factors.size(); ++i) {
ostringstream name;
name << side << ":";
name << factors[i];
name << ":";
if (leftWord) {
name << leftWord->GetFactor(factors[i])->GetString();
} else {
name << BOS_;
}
name << ":";
if (rightWord) {
name << rightWord->GetFactor(factors[i])->GetString();
} else {
name << EOS_;
}
scores->PlusEquals(this,name.str(),1);
}
}
bool Vocab::Load(FileHandler* vcbin)
{
FactorList factors;
factors.push_back(0);
return Load(vcbin, Input, factors);
}
