void ExtractTask::saveAllHieroPhrases( int startT, int endT, int startS, int endS, HoleCollection &holeColl, int countS)
{
LabelIndex labelIndex,labelCount;
// number of target head labels
int numLabels = m_options.targetSyntax ? m_sentence.targetTree.GetNodes(startT,endT).size() : 1;
if (m_options.targetSyntacticPreferences && !numLabels) {
numLabels++;
}
labelCount.push_back(numLabels);
labelIndex.push_back(0);
// number of source head labels
numLabels = m_options.sourceSyntax ? m_sentence.sourceTree.GetNodes(startS,endS).size() : 1;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
// number of target hole labels
for( HoleList::const_iterator hole = holeColl.GetHoles().begin();
hole != holeColl.GetHoles().end(); hole++ ) {
int numLabels = m_options.targetSyntax ? m_sentence.targetTree.GetNodes(hole->GetStart(1),hole->GetEnd(1)).size() : 1 ;
if (m_options.targetSyntacticPreferences && !numLabels) {
numLabels++;
}
labelCount.push_back(numLabels);
labelIndex.push_back(0);
}
// number of source hole labels
holeColl.SortSourceHoles();
for( vector<Hole*>::iterator i = holeColl.GetSortedSourceHoles().begin();
i != holeColl.GetSortedSourceHoles().end(); i++ ) {
const Hole &hole = **i;
int numLabels = m_options.sourceSyntax ? m_sentence.sourceTree.GetNodes(hole.GetStart(0),hole.GetEnd(0)).size() : 1 ;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
}
// loop through the holes
bool done = false;
while(!done) {
saveHieroPhrase( startT, endT, startS, endS, holeColl, labelIndex, countS );
for(unsigned int i=0; i<labelIndex.size(); i++) {
labelIndex[i]++;
if(labelIndex[i] == labelCount[i]) {
labelIndex[i] = 0;
if (i == labelIndex.size()-1)
done = true;
} else {
break;
}
}
}
}
void printAllHieroPhrases( SentenceAlignmentWithSyntax &sentence
, int startT, int endT, int startS, int endS
, HoleCollection &holeColl)
{
LabelIndex labelIndex,labelCount;
// number of target head labels
int numLabels = options.targetSyntax ? sentence.targetTree.GetNodes(startT,endT).size() : 1;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
// number of source head labels
numLabels = options.sourceSyntax ? sentence.sourceTree.GetNodes(startS,endS).size() : 1;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
// number of target hole labels
for( HoleList::const_iterator hole = holeColl.GetHoles().begin();
hole != holeColl.GetHoles().end(); hole++ ) {
int numLabels = options.targetSyntax ? sentence.targetTree.GetNodes(hole->GetStart(1),hole->GetEnd(1)).size() : 1 ;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
}
// number of source hole labels
holeColl.SortSourceHoles();
for( vector<Hole*>::iterator i = holeColl.GetSortedSourceHoles().begin();
i != holeColl.GetSortedSourceHoles().end(); i++ ) {
const Hole &hole = **i;
int numLabels = options.sourceSyntax ? sentence.sourceTree.GetNodes(hole.GetStart(0),hole.GetEnd(0)).size() : 1 ;
labelCount.push_back(numLabels);
labelIndex.push_back(0);
}
// loop through the holes
bool done = false;
while(!done) {
printHieroPhrase( sentence, startT, endT, startS, endS, holeColl, labelIndex );
for(int i=0; i<labelIndex.size(); i++) {
labelIndex[i]++;
if(labelIndex[i] == labelCount[i]) {
labelIndex[i] = 0;
if (i == labelIndex.size()-1)
done = true;
} else {
break;
}
}
}
}
// this function is called recursively
// it pokes a new hole into the phrase pair, and then calls itself for more holes
void addHieroRule( SentenceAlignmentWithSyntax &sentence
, int startT, int endT, int startS, int endS
, RuleExist &ruleExist, const HoleCollection &holeColl
, int numHoles, int initStartT, int wordCountT, int wordCountS)
{
// done, if already the maximum number of non-terminals in phrase pair
if (numHoles >= options.maxNonTerm)
return;
// find a hole...
for (int startHoleT = initStartT; startHoleT <= endT; ++startHoleT) {
for (int endHoleT = startHoleT+(options.minHoleTarget-1); endHoleT <= endT; ++endHoleT) {
// if last non-terminal, enforce word count limit
if (numHoles == options.maxNonTerm-1 && wordCountT - (endHoleT-startT+1) + (numHoles+1) > options.maxSymbolsTarget)
continue;
// determine the number of remaining target words
const int newWordCountT = wordCountT - (endHoleT-startHoleT+1);
// always enforce min word count limit
if (newWordCountT < options.minWords)
continue;
// except the whole span
if (startHoleT == startT && endHoleT == endT)
continue;
// does a phrase cover this target span?
// if it does, then there should be a list of mapped source phrases
// (multiple possible due to unaligned words)
const HoleList &sourceHoles = ruleExist.GetSourceHoles(startHoleT, endHoleT);
// loop over sub phrase pairs
HoleList::const_iterator iterSourceHoles;
for (iterSourceHoles = sourceHoles.begin(); iterSourceHoles != sourceHoles.end(); ++iterSourceHoles) {
const Hole &sourceHole = *iterSourceHoles;
const int sourceHoleSize = sourceHole.GetEnd(0)-sourceHole.GetStart(0)+1;
// enforce minimum hole size
if (sourceHoleSize < options.minHoleSource)
continue;
// determine the number of remaining source words
const int newWordCountS = wordCountS - sourceHoleSize;
// if last non-terminal, enforce word count limit
if (numHoles == options.maxNonTerm-1 && newWordCountS + (numHoles+1) > options.maxSymbolsSource)
continue;
// enforce min word count limit
if (newWordCountS < options.minWords)
continue;
// hole must be subphrase of the source phrase
// (may be violated if subphrase contains additional unaligned source word)
if (startS > sourceHole.GetStart(0) || endS < sourceHole.GetEnd(0))
continue;
// make sure target side does not overlap with another hole
if (holeColl.OverlapSource(sourceHole))
continue;
// if consecutive non-terminals are not allowed, also check for source
if (!options.nonTermConsecSource && holeColl.ConsecSource(sourceHole) )
continue;
// check that rule scope would not exceed limit if sourceHole
// were added
if (holeColl.Scope(sourceHole) > options.maxScope)
continue;
// require that at least one aligned word is left (unless there are no words at all)
if (options.requireAlignedWord && (newWordCountS > 0 || newWordCountT > 0)) {
HoleList::const_iterator iterHoleList = holeColl.GetHoles().begin();
bool foundAlignedWord = false;
// loop through all word positions
for(int pos = startT; pos <= endT && !foundAlignedWord; pos++) {
// new hole? moving on...
if (pos == startHoleT) {
pos = endHoleT;
}
// covered by hole? moving on...
else if (iterHoleList != holeColl.GetHoles().end() && iterHoleList->GetStart(1) == pos) {
pos = iterHoleList->GetEnd(1);
++iterHoleList;
}
// covered by word? check if it is aligned
else {
if (sentence.alignedToT[pos].size() > 0)
foundAlignedWord = true;
}
}
if (!foundAlignedWord)
continue;
}
// update list of holes in this phrase pair
HoleCollection copyHoleColl(holeColl);
copyHoleColl.Add(startHoleT, endHoleT, sourceHole.GetStart(0), sourceHole.GetEnd(0));
// now some checks that disallow this phrase pair, but not further recursion
bool allowablePhrase = true;
// maximum words count violation?
if (newWordCountS + (numHoles+1) > options.maxSymbolsSource)
allowablePhrase = false;
if (newWordCountT + (numHoles+1) > options.maxSymbolsTarget)
allowablePhrase = false;
// passed all checks...
if (allowablePhrase)
printAllHieroPhrases(sentence, startT, endT, startS, endS, copyHoleColl);
// recursively search for next hole
int nextInitStartT = options.nonTermConsecTarget ? endHoleT + 1 : endHoleT + 2;
addHieroRule(sentence, startT, endT, startS, endS
, ruleExist, copyHoleColl, numHoles + 1, nextInitStartT
, newWordCountT, newWordCountS);
}
}
}
}