private void Compile(Pattern <TData, TOffset> pattern)
{
_fsa = new Fst <TData, TOffset>(EqualityComparer <TOffset> .Default)
{
Direction = _settings.Direction,
Filter = _settings.Filter,
UseUnification = _settings.MatchingMethod == MatchingMethod.Unification
};
_fsa.StartState = _fsa.CreateState();
int nextPriority = 0;
bool hasVariables = GeneratePatternNfa(_fsa.StartState, pattern, null,
new Func <Match <TData, TOffset>, bool> [0], ref nextPriority);
#if FST_GRAPHS
using (var writer = new System.IO.StreamWriter(string.Format("{0}-nfa.dot",
_settings.Direction == Direction.LeftToRight ? "ltor" : "rtol")))
_fsa.ToGraphViz(writer);
#endif
if (!_settings.Nondeterministic && !hasVariables && !_settings.AllSubmatches)
{
_fsa = _fsa.Determinize();
#if FST_GRAPHS
using (var writer = new System.IO.StreamWriter(string.Format("{0}-dfa.dot",
_settings.Direction == Direction.LeftToRight ? "ltor" : "rtol")))
_fsa.ToGraphViz(writer);
#endif
_fsa.Minimize();
#if FST_GRAPHS
using (var writer = new System.IO.StreamWriter(string.Format("{0}-mindfa.dot",
_settings.Direction == Direction.LeftToRight ? "ltor" : "rtol")))
_fsa.ToGraphViz(writer);
#endif
}
else
{
_fsa = _fsa.EpsilonRemoval();
#if FST_GRAPHS
using (var writer = new System.IO.StreamWriter(string.Format("{0}-ernfa.dot",
_settings.Direction == Direction.LeftToRight ? "ltor" : "rtol")))
_fsa.ToGraphViz(writer);
#endif
}
_fsa.IgnoreVariables = !hasVariables;
_fsa.Freeze();
}
public void Transduce()
{
var fst = new Fst<AnnotatedStringData, int>(_operations) {UseUnification = false};
fst.StartState = fst.CreateAcceptingState();
fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas-", "nas?").Value, fst.StartState);
fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Symbol("cor+", "cor-").Value, fst.StartState);
State<AnnotatedStringData, int> s1 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, fst.CreateState());
s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, fst.StartState);
State<AnnotatedStringData, int> s2 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor+").Value, fst.CreateAcceptingState());
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor+").Value, s2);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas-", "nas?").Symbol("cor+", "cor?").Value, fst.StartState);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Symbol("cor+").Value, fst.StartState);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, s1);
Fst<AnnotatedStringData, int> dfst = fst.Determinize();
AnnotatedStringData data = CreateStringData("caNp");
FstResult<AnnotatedStringData, int> result;
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("camp"));
data = CreateStringData("caN");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("can"));
data = CreateStringData("carp");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("carp"));
fst = new Fst<AnnotatedStringData, int>(_operations) {UseUnification = false};
fst.StartState = fst.CreateAcceptingState();
s1 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons+").Value, fst.CreateState())
.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons-").Value, fst.CreateState());
s2 = s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Value, null, fst.CreateState());
State<AnnotatedStringData, int> s3 = s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("voice-").Value, fst.CreateState());
s3.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo(".").Value, s2);
s3.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo("+").Value, fst.CreateState())
.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo(".").Value, s2);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons+").Value, fst.CreateAcceptingState());
dfst = fst.Determinize();
data = CreateStringData("camp");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("cap"));
data = CreateStringData("casp");
IEnumerable<FstResult<AnnotatedStringData, int>> results;
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out results), Is.True);
FstResult<AnnotatedStringData, int>[] resultsArray = results.ToArray();
Assert.That(resultsArray.Length, Is.EqualTo(2));
Assert.That(resultsArray.Select(r => r.Output.String), Is.EquivalentTo(new [] {"cas+.p", "cas.p"}));
}
public void Transduce()
{
var fst = new Fst <AnnotatedStringData, int>(_operations)
{
UseUnification = false
};
fst.StartState = fst.CreateAcceptingState();
fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas-", "nas?").Value, fst.StartState);
fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Symbol("cor+", "cor-").Value, fst.StartState);
State <AnnotatedStringData, int> s1 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, fst.CreateState());
s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, fst.StartState);
State <AnnotatedStringData, int> s2 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor+").Value, fst.CreateAcceptingState());
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor+").Value, s2);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas-", "nas?").Symbol("cor+", "cor?").Value, fst.StartState);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Symbol("cor+").Value, fst.StartState);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cor?").Symbol("nas+").Value, FeatureStruct.New(PhoneticFeatSys).Symbol("cor-").Value, s1);
Fst <AnnotatedStringData, int> dfst = fst.Determinize();
AnnotatedStringData data = CreateStringData("caNp");
FstResult <AnnotatedStringData, int> result;
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("camp"));
data = CreateStringData("caN");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("can"));
data = CreateStringData("carp");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("carp"));
fst = new Fst <AnnotatedStringData, int>(_operations)
{
UseUnification = false
};
fst.StartState = fst.CreateAcceptingState();
s1 = fst.StartState.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons+").Value, fst.CreateState())
.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons-").Value, fst.CreateState());
s2 = s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("nas+").Value, null, fst.CreateState());
State <AnnotatedStringData, int> s3 = s1.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("voice-").Value, fst.CreateState());
s3.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo(".").Value, s2);
s3.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo("+").Value, fst.CreateState())
.Arcs.Add(null, FeatureStruct.New(PhoneticFeatSys).Symbol(Bdry).Feature("strRep").EqualTo(".").Value, s2);
s2.Arcs.Add(FeatureStruct.New(PhoneticFeatSys).Symbol("cons+").Value, fst.CreateAcceptingState());
dfst = fst.Determinize();
data = CreateStringData("camp");
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out result), Is.True);
Assert.That(result.Output.String, Is.EqualTo("cap"));
data = CreateStringData("casp");
IEnumerable <FstResult <AnnotatedStringData, int> > results;
Assert.That(dfst.Transduce(data, data.Annotations.First, null, true, true, true, out results), Is.True);
FstResult <AnnotatedStringData, int>[] resultsArray = results.ToArray();
Assert.That(resultsArray.Length, Is.EqualTo(2));
Assert.That(resultsArray.Select(r => r.Output.String), Is.EquivalentTo(new [] { "cas+.p", "cas.p" }));
}