TargetPhrase*
BilingualDynSuffixArray::
GetMosesFactorIDs(const SAPhrase& phrase, const Phrase& sourcePhrase) const
{
TargetPhrase* targetPhrase = new TargetPhrase();
for(size_t i=0; i < phrase.words.size(); ++i) { // look up trg words
Word& word = m_trgVocab->GetWord( phrase.words[i]);
CHECK(word != m_trgVocab->GetkOOVWord());
targetPhrase->AddWord(word);
}
targetPhrase->SetSourcePhrase(sourcePhrase);
// scoring
return targetPhrase;
}
bool RuleTableLoaderStandard::Load(FormatType format
, const std::vector<FactorType> &input
, const std::vector<FactorType> &output
, const std::string &inFile
, const std::vector<float> &weight
, size_t /* tableLimit */
, const LMList &languageModels
, const WordPenaltyProducer* wpProducer
, RuleTableTrie &ruleTable)
{
PrintUserTime(string("Start loading text SCFG phrase table. ") + (format==MosesFormat?"Moses ":"Hiero ") + " format");
const StaticData &staticData = StaticData::Instance();
const std::string& factorDelimiter = staticData.GetFactorDelimiter();
string lineOrig;
size_t count = 0;
std::ostream *progress = NULL;
IFVERBOSE(1) progress = &std::cerr;
util::FilePiece in(inFile.c_str(), progress);
// reused variables
vector<float> scoreVector;
StringPiece line;
std::string hiero_before, hiero_after;
while(true) {
try {
line = in.ReadLine();
} catch (const util::EndOfFileException &e) { break; }
if (format == HieroFormat) { // inefficiently reformat line
hiero_before.assign(line.data(), line.size());
ReformatHieroRule(hiero_before, hiero_after);
line = hiero_after;
}
util::TokenIter<util::MultiCharacter> pipes(line, "|||");
StringPiece sourcePhraseString(*pipes);
StringPiece targetPhraseString(*++pipes);
StringPiece scoreString(*++pipes);
StringPiece alignString(*++pipes);
// TODO(bhaddow) efficiently handle default instead of parsing this string every time.
StringPiece ruleCountString = ++pipes ? *pipes : StringPiece("1 1");
if (++pipes) {
stringstream strme;
strme << "Syntax error at " << ruleTable.GetFilePath() << ":" << count;
UserMessage::Add(strme.str());
abort();
}
bool isLHSEmpty = (sourcePhraseString.find_first_not_of(" \t", 0) == string::npos);
if (isLHSEmpty && !staticData.IsWordDeletionEnabled()) {
TRACE_ERR( ruleTable.GetFilePath() << ":" << count << ": pt entry contains empty target, skipping\n");
continue;
}
scoreVector.clear();
for (util::TokenIter<util::AnyCharacter, true> s(scoreString, " \t"); s; ++s) {
char *err_ind;
scoreVector.push_back(strtod(s->data(), &err_ind));
UTIL_THROW_IF(err_ind == s->data(), util::Exception, "Bad score " << *s << " on line " << count);
}
const size_t numScoreComponents = ruleTable.GetFeature()->GetNumScoreComponents();
if (scoreVector.size() != numScoreComponents) {
stringstream strme;
strme << "Size of scoreVector != number (" << scoreVector.size() << "!="
<< numScoreComponents << ") of score components on line " << count;
UserMessage::Add(strme.str());
abort();
}
// parse source & find pt node
// constituent labels
Word sourceLHS, targetLHS;
// source
Phrase sourcePhrase( 0);
sourcePhrase.CreateFromStringNewFormat(Input, input, sourcePhraseString, factorDelimiter, sourceLHS);
// create target phrase obj
TargetPhrase *targetPhrase = new TargetPhrase(Output);
targetPhrase->CreateFromStringNewFormat(Output, output, targetPhraseString, factorDelimiter, targetLHS);
targetPhrase->SetSourcePhrase(sourcePhrase);
// rest of target phrase
targetPhrase->SetAlignmentInfo(alignString, sourcePhrase);
targetPhrase->SetTargetLHS(targetLHS);
targetPhrase->SetRuleCount(ruleCountString, scoreVector[0]);
//targetPhrase->SetDebugOutput(string("New Format pt ") + line);
// component score, for n-best output
std::transform(scoreVector.begin(),scoreVector.end(),scoreVector.begin(),TransformScore);
std::transform(scoreVector.begin(),scoreVector.end(),scoreVector.begin(),FloorScore);
targetPhrase->SetScoreChart(ruleTable.GetFeature(), scoreVector, weight, languageModels,wpProducer);
TargetPhraseCollection &phraseColl = GetOrCreateTargetPhraseCollection(ruleTable, sourcePhrase, *targetPhrase, sourceLHS);
phraseColl.Add(targetPhrase);
count++;
}
// sort and prune each target phrase collection
SortAndPrune(ruleTable);
return true;
}
TargetPhraseVectorPtr PhraseDecoder::DecodeCollection(
TargetPhraseVectorPtr tpv, BitWrapper<> &encodedBitStream,
const Phrase &sourcePhrase, bool topLevel)
{
bool extending = tpv->size();
size_t bitsLeft = encodedBitStream.TellFromEnd();
typedef std::pair<size_t, size_t> AlignPointSizeT;
std::vector<int> sourceWords;
if(m_coding == REnc)
{
for(size_t i = 0; i < sourcePhrase.GetSize(); i++)
{
std::string sourceWord
= sourcePhrase.GetWord(i).GetString(*m_input, false);
unsigned idx = GetSourceSymbolId(sourceWord);
sourceWords.push_back(idx);
}
}
unsigned phraseStopSymbol = 0;
AlignPoint alignStopSymbol(-1, -1);
std::vector<float> scores;
std::set<AlignPointSizeT> alignment;
enum DecodeState { New, Symbol, Score, Alignment, Add } state = New;
size_t srcSize = sourcePhrase.GetSize();
TargetPhrase* targetPhrase = NULL;
while(encodedBitStream.TellFromEnd())
{
if(state == New)
{
// Creating new TargetPhrase on the heap
tpv->push_back(TargetPhrase(Output));
targetPhrase = &tpv->back();
targetPhrase->SetSourcePhrase(sourcePhrase);
alignment.clear();
scores.clear();
state = Symbol;
}
if(state == Symbol)
{
unsigned symbol = m_symbolTree->Read(encodedBitStream);
if(symbol == phraseStopSymbol)
{
state = Score;
}
else
{
if(m_coding == REnc)
{
std::string wordString;
size_t type = GetREncType(symbol);
if(type == 1)
{
unsigned decodedSymbol = DecodeREncSymbol1(symbol);
wordString = GetTargetSymbol(decodedSymbol);
}
else if (type == 2)
{
size_t rank = DecodeREncSymbol2Rank(symbol);
size_t srcPos = DecodeREncSymbol2Position(symbol);
if(srcPos >= sourceWords.size())
return TargetPhraseVectorPtr();
wordString = GetTargetSymbol(GetTranslation(sourceWords[srcPos], rank));
if(m_phraseDictionary.m_useAlignmentInfo)
{
size_t trgPos = targetPhrase->GetSize();
alignment.insert(AlignPoint(srcPos, trgPos));
}
}
else if(type == 3)
{
size_t rank = DecodeREncSymbol3(symbol);
size_t srcPos = targetPhrase->GetSize();
if(srcPos >= sourceWords.size())
return TargetPhraseVectorPtr();
wordString = GetTargetSymbol(GetTranslation(sourceWords[srcPos], rank));
if(m_phraseDictionary.m_useAlignmentInfo)
{
size_t trgPos = srcPos;
alignment.insert(AlignPoint(srcPos, trgPos));
}
}
Word word;
word.CreateFromString(Output, *m_output, wordString, false);
targetPhrase->AddWord(word);
}
else if(m_coding == PREnc)
{
// if the symbol is just a word
if(GetPREncType(symbol) == 1)
{
unsigned decodedSymbol = DecodePREncSymbol1(symbol);
Word word;
word.CreateFromString(Output, *m_output,
GetTargetSymbol(decodedSymbol), false);
targetPhrase->AddWord(word);
}
// if the symbol is a subphrase pointer
else
{
int left = DecodePREncSymbol2Left(symbol);
int right = DecodePREncSymbol2Right(symbol);
unsigned rank = DecodePREncSymbol2Rank(symbol);
int srcStart = left + targetPhrase->GetSize();
int srcEnd = srcSize - right - 1;
// false positive consistency check
if(0 > srcStart || srcStart > srcEnd || unsigned(srcEnd) >= srcSize)
return TargetPhraseVectorPtr();
// false positive consistency check
if(m_maxRank && rank > m_maxRank)
return TargetPhraseVectorPtr();
// set subphrase by default to itself
TargetPhraseVectorPtr subTpv = tpv;
// if range smaller than source phrase retrieve subphrase
if(unsigned(srcEnd - srcStart + 1) != srcSize)
{
Phrase subPhrase = sourcePhrase.GetSubString(WordsRange(srcStart, srcEnd));
subTpv = CreateTargetPhraseCollection(subPhrase, false);
}
// false positive consistency check
if(subTpv != NULL && rank < subTpv->size())
{
// insert the subphrase into the main target phrase
TargetPhrase& subTp = subTpv->at(rank);
if(m_phraseDictionary.m_useAlignmentInfo)
{
// reconstruct the alignment data based on the alignment of the subphrase
for(AlignmentInfo::const_iterator it = subTp.GetAlignmentInfo().begin();
it != subTp.GetAlignmentInfo().end(); it++)
{
alignment.insert(AlignPointSizeT(srcStart + it->first,
targetPhrase->GetSize() + it->second));
}
}
targetPhrase->Append(subTp);
}
else
return TargetPhraseVectorPtr();
}
}
else
{
Word word;
word.CreateFromString(Output, *m_output,
GetTargetSymbol(symbol), false);
targetPhrase->AddWord(word);
}
}
}
else if(state == Score)
{
size_t idx = m_multipleScoreTrees ? scores.size() : 0;
float score = m_scoreTrees[idx]->Read(encodedBitStream);
scores.push_back(score);
if(scores.size() == m_numScoreComponent)
{
targetPhrase->SetScore(m_feature, scores, ScoreComponentCollection() /*sparse*/,*m_weight, m_weightWP, *m_languageModels);
if(m_containsAlignmentInfo)
state = Alignment;
else
state = Add;
}
}
else if(state == Alignment)
{
AlignPoint alignPoint = m_alignTree->Read(encodedBitStream);
if(alignPoint == alignStopSymbol)
{
state = Add;
}
else
{
if(m_phraseDictionary.m_useAlignmentInfo)
alignment.insert(AlignPointSizeT(alignPoint));
}
}
if(state == Add)
{
if(m_phraseDictionary.m_useAlignmentInfo)
targetPhrase->SetAlignmentInfo(alignment);
if(m_coding == PREnc)
{
if(!m_maxRank || tpv->size() <= m_maxRank)
bitsLeft = encodedBitStream.TellFromEnd();
if(!topLevel && m_maxRank && tpv->size() >= m_maxRank)
break;
}
if(encodedBitStream.TellFromEnd() <= 8)
break;
state = New;
}
}
if(m_coding == PREnc && !extending)
{
bitsLeft = bitsLeft > 8 ? bitsLeft : 0;
m_decodingCache.Cache(sourcePhrase, tpv, bitsLeft, m_maxRank);
}
return tpv;
}
bool RuleTableLoaderCompact::LoadRuleSection(
LineReader &reader,
const std::vector<Word> &vocab,
const std::vector<Phrase> &sourcePhrases,
const std::vector<Phrase> &targetPhrases,
const std::vector<size_t> &targetLhsIds,
const std::vector<const AlignmentInfo *> &alignmentSets,
RuleTableTrie &ruleTable)
{
// Read rule count.
reader.ReadLine();
const size_t ruleCount = std::atoi(reader.m_line.c_str());
// Read rules and add to table.
const size_t numScoreComponents = ruleTable.GetNumScoreComponents();
std::vector<float> scoreVector(numScoreComponents);
std::vector<size_t> tokenPositions;
for (size_t i = 0; i < ruleCount; ++i) {
reader.ReadLine();
tokenPositions.clear();
FindTokens(tokenPositions, reader.m_line);
const char *charLine = reader.m_line.c_str();
// The first three tokens are IDs for the source phrase, target phrase,
// and alignment set.
const int sourcePhraseId = std::atoi(charLine+tokenPositions[0]);
const int targetPhraseId = std::atoi(charLine+tokenPositions[1]);
const int alignmentSetId = std::atoi(charLine+tokenPositions[2]);
const Phrase &sourcePhrase = sourcePhrases[sourcePhraseId];
const Phrase &targetPhrasePhrase = targetPhrases[targetPhraseId];
const Word *targetLhs = new Word(vocab[targetLhsIds[targetPhraseId]]);
Word sourceLHS("X"); // TODO not implemented for compact
const AlignmentInfo *alignNonTerm = alignmentSets[alignmentSetId];
// Then there should be one score for each score component.
for (size_t j = 0; j < numScoreComponents; ++j) {
float score = std::atof(charLine+tokenPositions[3+j]);
scoreVector[j] = FloorScore(TransformScore(score));
}
if (reader.m_line[tokenPositions[3+numScoreComponents]] != ':') {
std::stringstream msg;
msg << "Size of scoreVector != number ("
<< scoreVector.size() << "!=" << numScoreComponents
<< ") of score components on line " << reader.m_lineNum;
UserMessage::Add(msg.str());
return false;
}
// The remaining columns are currently ignored.
// Create and score target phrase.
TargetPhrase *targetPhrase = new TargetPhrase(targetPhrasePhrase);
targetPhrase->SetAlignNonTerm(alignNonTerm);
targetPhrase->SetTargetLHS(targetLhs);
targetPhrase->SetSourcePhrase(sourcePhrase);
targetPhrase->Evaluate(sourcePhrase, ruleTable.GetFeaturesToApply());
// Insert rule into table.
TargetPhraseCollection &coll = GetOrCreateTargetPhraseCollection(
ruleTable, sourcePhrase, *targetPhrase, &sourceLHS);
coll.Add(targetPhrase);
}
return true;
}
const TargetPhraseCollection*
PhraseDictionaryInterpolated::GetTargetPhraseCollection(const Phrase& src) const {
delete m_targetPhrases;
m_targetPhrases = new TargetPhraseCollection();
PhraseSet allPhrases;
vector<PhraseSet> phrasesByTable(m_dictionaries.size());
for (size_t i = 0; i < m_dictionaries.size(); ++i) {
const TargetPhraseCollection* phrases = m_dictionaries[i]->GetTargetPhraseCollection(src);
if (phrases) {
for (TargetPhraseCollection::const_iterator j = phrases->begin();
j != phrases->end(); ++j) {
allPhrases.insert(*j);
phrasesByTable[i].insert(*j);
}
}
}
ScoreComponentCollection sparseVector;
for (PhraseSet::const_iterator i = allPhrases.begin(); i != allPhrases.end(); ++i) {
TargetPhrase* combinedPhrase = new TargetPhrase((Phrase)**i);
//combinedPhrase->ResetScore();
//cerr << *combinedPhrase << " " << combinedPhrase->GetScoreBreakdown() << endl;
combinedPhrase->SetSourcePhrase((*i)->GetSourcePhrase());
combinedPhrase->SetAlignTerm(&((*i)->GetAlignTerm()));
combinedPhrase->SetAlignNonTerm(&((*i)->GetAlignTerm()));
Scores combinedScores(GetFeature()->GetNumScoreComponents());
for (size_t j = 0; j < phrasesByTable.size(); ++j) {
PhraseSet::const_iterator tablePhrase = phrasesByTable[j].find(combinedPhrase);
if (tablePhrase != phrasesByTable[j].end()) {
Scores tableScores = (*tablePhrase)->GetScoreBreakdown()
.GetScoresForProducer(GetFeature());
//cerr << "Scores from " << j << " table: ";
for (size_t k = 0; k < tableScores.size()-1; ++k) {
//cerr << tableScores[k] << "(" << exp(tableScores[k]) << ") ";
combinedScores[k] += m_weights[k][j] * exp(tableScores[k]);
//cerr << m_weights[k][j] * exp(tableScores[k]) << " ";
}
//cerr << endl;
}
}
//map back to log space
//cerr << "Combined ";
for (size_t k = 0; k < combinedScores.size()-1; ++k) {
//cerr << combinedScores[k] << " ";
combinedScores[k] = log(combinedScores[k]);
//cerr << combinedScores[k] << " ";
}
//cerr << endl;
combinedScores.back() = 1; //assume last is penalty
combinedPhrase->SetScore(
GetFeature(),
combinedScores,
sparseVector,
m_weightT,
m_weightWP,
*m_languageModels);
//cerr << *combinedPhrase << " " << combinedPhrase->GetScoreBreakdown() << endl;
m_targetPhrases->Add(combinedPhrase);
}
m_targetPhrases->Prune(true,m_tableLimit);
return m_targetPhrases;
}
