StdVectorFst Phonetisaurus::makeEpsMapper( ) {
/*
Generate a mapper FST to transform unwanted output symbols
to the epsilon symbol.
This can be used to remove unwanted symbols from the final
result, but in tests was 7x slower than manual removal
via the FstPathFinder object.
*/
StdVectorFst mfst;
mfst.AddState();
mfst.SetStart(0);
set<string>::iterator sit;
for( size_t i=0; i< osyms->NumSymbols(); i++ ) {
string sym = osyms->Find( i );
sit = skipSeqs.find( sym );
if( sit!=skipSeqs.end() )
mfst.AddArc( 0, StdArc( i, 0, 0, 0 ) );
else
mfst.AddArc( 0, StdArc( i, i, 0, 0 ) );
}
mfst.SetFinal(0, 0);
ILabelCompare<StdArc> icomp;
ArcSort( &mfst, icomp );
mfst.SetInputSymbols( osyms );
mfst.SetOutputSymbols( osyms );
return mfst;
}
/*
* CRF_MLFManager::getFst
*
* Input: fname - the filename of the sequence being requested
*
* Returns: OpenFst formated VectorFst of the transcript from the MLF
*
*/
StdVectorFst* CRF_MLFManager::getFst(string fname) {
string key = this->getKey(fname);
if (key == "") {
string errstr="Unable to acquire key from filename "+fname;
throw runtime_error(errstr);
}
int idx=this->fnameTable[key];
StdVectorFst* fst = new StdVectorFst();
StateId startState = fst->AddState();
fst->SetStart(startState);
StateId prevState=startState;
StateId curState;
//cout << key << "\t";
vector<int>::iterator itNode;
for(itNode = this->transcripts.at(idx)->begin(); itNode != this->transcripts.at(idx)->end(); itNode++)
{
curState=fst->AddState();
fst->AddArc(prevState,StdArc(*itNode,*itNode,0,curState));
//cout << *itNode << " ";
prevState=curState;
}
fst->SetFinal(curState,0);
//cout << endl;
return fst;
}
void fstprintpaths::printAllStringsHelper(StdVectorFst& fst, SymbolTable& st, int state, string& str, TropicalWeight cost, string& seqToSkip, ProjectionEnum printInputOrOutput)
{
if(fst.Final(state) != TropicalWeight::Zero())
{
cout << str << " cost " << (Times(cost,fst.Final(state))) << endl;
str = "";
return;
}
for(ArcIterator< StdVectorFst > iter(fst,state); !iter.Done(); iter.Next())
{
StdArc arc = iter.Value();
string symbol = (printInputOrOutput) == fstprintpaths::PROJECT_INPUT ? st.Find( arc.ilabel ) : st.Find( arc.olabel );
if( symbol.compare( seqToSkip ) != 0 ) {
str += symbol;
}
printAllStringsHelper(fst, st, arc.nextstate, str, Times(cost, arc.weight.Value()), seqToSkip, printInputOrOutput);
}
}
StdVectorFst Phonetisaurus::entryToFSA( vector<string> entry ) {
/*
Transform an input spelling/pronunciation into an equivalent
FSA, adding extra arcs as needed to accomodate clusters.
*/
StdVectorFst efst;
efst.AddState();
efst.SetStart(0);
efst.AddState();
efst.AddArc( 0, StdArc( isyms->Find( sb ), isyms->Find( sb ), 0, 1 ));
size_t i=0;
//Build the basic FSA
for( i=0; i<entry.size(); i++) {
efst.AddState();
string ch = entry[i];
efst.AddArc( i+1, StdArc( isyms->Find(ch), isyms->Find(ch), 0, i+2 ));
if( i==0 )
continue;
}
//Add any cluster arcs
map<vector<string>,int>::iterator it_i;
for( it_i=clusters.begin(); it_i!=clusters.end(); it_i++ ) {
vector<string>::iterator it_j;
vector<string>::iterator start = entry.begin();
vector<string> cluster = (*it_i).first;
while( it_j != entry.end() ) {
it_j = search( start, entry.end(), cluster.begin(), cluster.end() );
if( it_j != entry.end() ) {
efst.AddArc( it_j-entry.begin()+1, StdArc(
(*it_i).second, //input symbol
(*it_i).second, //output symbol
0, //weight
it_j-entry.begin()+cluster.size()+1 //destination state
) );
start = it_j+cluster.size();
}
}
}
efst.AddState();
efst.AddArc( i+1, StdArc( isyms->Find( se ), isyms->Find( se ), 0, i+2));
efst.SetFinal( i+2, 0 );
efst.SetInputSymbols( isyms );
efst.SetOutputSymbols( isyms );
return efst;
}
void fstprintpaths::printAllStrings(StdVectorFst& fst, SymbolTable& st, string& seqToSkip, ProjectionEnum printInputOrOutput)
{
string str;
TropicalWeight tw(TropicalWeight::One());
printAllStringsHelper(fst, st, fst.Start(), str, tw, seqToSkip, printInputOrOutput);
}
int main(int argc, char* argv[]) {
StdVectorFst fst;
SymbolTable* isyms;
SymbolTable* osyms;
{
isyms = new SymbolTable("isyms.txt");
osyms = new SymbolTable("osyms.txt");
isyms->AddSymbol("a");
isyms->AddSymbol("b");
isyms->AddSymbol("c");
isyms->Write("isyms.txt");
osyms->AddSymbol("x");
osyms->AddSymbol("y");
osyms->AddSymbol("z");
osyms->Write("osyms.txt");
}
{
fst.SetInputSymbols(isyms);
fst.SetOutputSymbols(osyms);
// Adds state 0 to the initially empty FST and make it the start state.
fst.AddState(); // 1st state will be state 0 (returned by AddState)
fst.SetStart(0); // arg is state ID
// Adds two arcs exiting state 0.
// Arc constructor args: ilabel, olabel, weight, dest state ID.
fst.AddArc(0, StdArc(isyms->Find("a"), osyms->Find("x"), 0.5, 1)); // 1st arg is src state ID
fst.AddArc(0, StdArc(isyms->Find("b"), osyms->Find("y"), 1.5, 1));
// Adds state 1 and its arc.
fst.AddState();
fst.AddArc(1, StdArc(isyms->Find("c"), osyms->Find("z"), 2.5, 2));
// Adds state 2 and set its final weight.
fst.AddState();
fst.SetFinal(2, 3.5); // 1st arg is state ID, 2nd arg weight
fst.Write("example.fst");
}
StdVectorFst search_fst;
{
search_fst.SetInputSymbols(isyms);
search_fst.SetOutputSymbols(osyms);
search_fst.AddState(); // 1st state will be state 0 (returned by AddState)
search_fst.SetStart(0); // arg is state ID
// Adds two arcs exiting state 0.
// Arc constructor args: ilabel, olabel, weight, dest state ID.
search_fst.AddArc(0, StdArc(isyms->Find("a"), osyms->Find("x"), 0.5, 1)); // 1st arg is src state ID
// Adds state 1 and its arc.
search_fst.AddState();
search_fst.AddArc(1, StdArc(isyms->Find("c"), osyms->Find("z"), 2.5, 2));
// Adds state 2 and set its final weight.
search_fst.AddState();
search_fst.SetFinal(2, 3.5); // 1st arg is state ID, 2nd arg weight
}
{
for (StateIterator<StdVectorFst> siter(fst); not siter.Done(); siter.Next()) {
StdIntersectFst::StateId s = siter.Value();
std::cout << "state=" << s << ":";
for (ArcIterator<StdVectorFst> aiter(fst, s); not aiter.Done(); aiter.Next()) {
const StdArc& arc = aiter.Value();
std::cout << arc.ilabel << "/" << arc.olabel << "->" << arc.nextstate << ",";
}
std::cout << std::endl;
}
}
{
Matcher<StdVectorFst> matcher(fst, MATCH_INPUT);
matcher.SetState(0);
StdArc::Label find_label = 1;
if (matcher.Find(find_label)) {
for (; not matcher.Done(); matcher.Next()) {
const StdArc& arc = matcher.Value();
std::cout << "found="
<< arc.ilabel << "/" << arc.olabel << "->" << arc.nextstate << std::endl;
}
}
}
// intersect contains strings in both A and B
{
ArcSort(&fst, StdOLabelCompare());
ArcSort(&search_fst, StdILabelCompare());
/*
ComposeFilter compose_filter;
if (!GetComposeFilter("auto", &compose_filter)) {
LOG(ERROR) << "failed";
exit(1);
}
const fst::IntersectFstOptions<StdArc> opts;
*/
//StdIntersectFst ofst(fst, search_fst, opts);
StdIntersectFst ofst(fst, search_fst);
}
}